id	sid	tid	token	lemma	pos
dad-10670	1	1	dialogue	dialogue	NOUN
dad-10670	1	2	and	and	CCONJ
dad-10670	1	3	discourse	discourse	NOUN
dad-10670	1	4	1(3	1(3	NUM
dad-10670	1	5	)	)	PUNCT
dad-10670	1	6	(	(	PUNCT
dad-10670	1	7	2010	2010	NUM
dad-10670	1	8	)	)	PUNCT
dad-10670	1	9	1	1	NUM
dad-10670	1	10	-	-	SYM
dad-10670	1	11	33	33	NUM
dad-10670	1	12	doi	doi	NOUN
dad-10670	1	13	:	:	PUNCT
dad-10670	1	14	10.5087	10.5087	NUM
dad-10670	1	15	/	/	SYM
dad-10670	1	16	dad.2010.003	dad.2010.003	PROPN
dad-10670	1	17	hilda	hilda	VERB
dad-10670	1	18	:	:	PUNCT
dad-10670	1	19	a	a	DET
dad-10670	1	20	discourse	discourse	NOUN
dad-10670	1	21	parser	parser	NOUN
dad-10670	1	22	using	use	VERB
dad-10670	1	23	support	support	NOUN
dad-10670	1	24	vector	vector	NOUN
dad-10670	1	25	machine	machine	NOUN
dad-10670	1	26	classification∗	classification∗	NOUN
dad-10670	1	27	hugo	hugo	PROPN
dad-10670	1	28	hernault	hernault	PROPN
dad-10670	1	29	hugo@mi.ci.i.u-tokyo.ac.jp	hugo@mi.ci.i.u-tokyo.ac.jp	ADJ
dad-10670	1	30	graduate	graduate	ADJ
dad-10670	1	31	school	school	NOUN
dad-10670	1	32	of	of	ADP
dad-10670	1	33	information	information	NOUN
dad-10670	1	34	science	science	PROPN
dad-10670	1	35	&	&	CCONJ
dad-10670	1	36	technology	technology	PROPN
dad-10670	1	37	the	the	DET
dad-10670	1	38	university	university	NOUN
dad-10670	1	39	of	of	ADP
dad-10670	1	40	tokyo	tokyo	PROPN
dad-10670	1	41	7	7	NUM
dad-10670	1	42	-	-	SYM
dad-10670	1	43	3	3	NUM
dad-10670	1	44	-	-	SYM
dad-10670	1	45	1	1	NUM
dad-10670	1	46	hongo	hongo	PROPN
dad-10670	1	47	,	,	PUNCT
dad-10670	1	48	bunkyo	bunkyo	PROPN
dad-10670	1	49	-	-	PUNCT
dad-10670	1	50	ku	ku	PROPN
dad-10670	1	51	,	,	PUNCT
dad-10670	1	52	tokyo	tokyo	PROPN
dad-10670	1	53	113	113	NUM
dad-10670	1	54	-	-	SYM
dad-10670	1	55	8656	8656	NUM
dad-10670	1	56	,	,	PUNCT
dad-10670	1	57	japan	japan	PROPN
dad-10670	1	58	helmut	helmut	PROPN
dad-10670	1	59	prendinger	prendinger	PROPN
dad-10670	1	60	helmut@nii.ac.jp	helmut@nii.ac.jp	PROPN
dad-10670	1	61	national	national	PROPN
dad-10670	1	62	institute	institute	PROPN
dad-10670	1	63	of	of	ADP
dad-10670	1	64	informatics	informatics	PROPN
dad-10670	1	65	2	2	NUM
dad-10670	1	66	-	-	SYM
dad-10670	1	67	1	1	NUM
dad-10670	1	68	-	-	SYM
dad-10670	1	69	2	2	NUM
dad-10670	1	70	hitotsubashi	hitotsubashi	NOUN
dad-10670	1	71	,	,	PUNCT
dad-10670	1	72	chiyoda	chiyoda	PROPN
dad-10670	1	73	-	-	PUNCT
dad-10670	1	74	ku	ku	PROPN
dad-10670	1	75	,	,	PUNCT
dad-10670	1	76	tokyo	tokyo	PROPN
dad-10670	1	77	101	101	NUM
dad-10670	1	78	-	-	PUNCT
dad-10670	1	79	8430	8430	NUM
dad-10670	1	80	,	,	PUNCT
dad-10670	1	81	japan	japan	PROPN
dad-10670	1	82	david	david	PROPN
dad-10670	1	83	a.	a.	PROPN
dad-10670	1	84	duverle	duverle	PROPN
dad-10670	1	85	dave@kuicr.kyoto-u.ac.jp	dave@kuicr.kyoto-u.ac.jp	PROPN
dad-10670	1	86	bioinformatics	bioinformatic	NOUN
dad-10670	1	87	center	center	NOUN
dad-10670	1	88	,	,	PUNCT
dad-10670	1	89	institute	institute	NOUN
dad-10670	1	90	for	for	ADP
dad-10670	1	91	chemical	chemical	PROPN
dad-10670	1	92	research	research	PROPN
dad-10670	1	93	kyoto	kyoto	PROPN
dad-10670	1	94	university	university	PROPN
dad-10670	1	95	gokasho	gokasho	PROPN
dad-10670	1	96	,	,	PUNCT
dad-10670	1	97	uji	uji	PROPN
dad-10670	1	98	,	,	PUNCT
dad-10670	1	99	kyoto	kyoto	NOUN
dad-10670	1	100	611	611	NUM
dad-10670	1	101	-	-	SYM
dad-10670	1	102	0011	0011	NUM
dad-10670	1	103	,	,	PUNCT
dad-10670	1	104	japan	japan	PROPN
dad-10670	1	105	mitsuru	mitsuru	PROPN
dad-10670	1	106	ishizuka	ishizuka	PROPN
dad-10670	1	107	ishizuka@i.u-tokyo.ac.jp	ishizuka@i.u-tokyo.ac.jp	PROPN
dad-10670	1	108	graduate	graduate	PROPN
dad-10670	1	109	school	school	NOUN
dad-10670	1	110	of	of	ADP
dad-10670	1	111	information	information	NOUN
dad-10670	1	112	science	science	PROPN
dad-10670	1	113	&	&	CCONJ
dad-10670	1	114	technology	technology	PROPN
dad-10670	1	115	the	the	DET
dad-10670	1	116	university	university	NOUN
dad-10670	1	117	of	of	ADP
dad-10670	1	118	tokyo	tokyo	PROPN
dad-10670	1	119	7	7	NUM
dad-10670	1	120	-	-	SYM
dad-10670	1	121	3	3	NUM
dad-10670	1	122	-	-	SYM
dad-10670	1	123	1	1	NUM
dad-10670	1	124	hongo	hongo	PROPN
dad-10670	1	125	,	,	PUNCT
dad-10670	1	126	bunkyo	bunkyo	PROPN
dad-10670	1	127	-	-	PUNCT
dad-10670	1	128	ku	ku	PROPN
dad-10670	1	129	,	,	PUNCT
dad-10670	1	130	tokyo	tokyo	PROPN
dad-10670	1	131	113	113	NUM
dad-10670	1	132	-	-	SYM
dad-10670	1	133	8656	8656	NUM
dad-10670	1	134	,	,	PUNCT
dad-10670	1	135	japan	japan	PROPN
dad-10670	1	136	editor	editor	PROPN
dad-10670	1	137	:	:	PUNCT
dad-10670	1	138	tim	tim	PROPN
dad-10670	1	139	paek	paek	PROPN
dad-10670	1	140	abstract	abstract	PROPN
dad-10670	1	141	discourse	discourse	NOUN
dad-10670	1	142	structures	structure	NOUN
dad-10670	1	143	have	have	VERB
dad-10670	1	144	a	a	DET
dad-10670	1	145	central	central	ADJ
dad-10670	1	146	role	role	NOUN
dad-10670	1	147	in	in	ADP
dad-10670	1	148	several	several	ADJ
dad-10670	1	149	computational	computational	ADJ
dad-10670	1	150	tasks	task	NOUN
dad-10670	1	151	,	,	PUNCT
dad-10670	1	152	such	such	ADJ
dad-10670	1	153	as	as	ADP
dad-10670	1	154	question	question	NOUN
dad-10670	1	155	–	–	PUNCT
dad-10670	1	156	answering	answering	NOUN
dad-10670	1	157	or	or	CCONJ
dad-10670	1	158	dialogue	dialogue	NOUN
dad-10670	1	159	generation	generation	NOUN
dad-10670	1	160	.	.	PUNCT
dad-10670	2	1	in	in	ADP
dad-10670	2	2	particular	particular	ADJ
dad-10670	2	3	,	,	PUNCT
dad-10670	2	4	the	the	DET
dad-10670	2	5	framework	framework	NOUN
dad-10670	2	6	of	of	ADP
dad-10670	2	7	the	the	DET
dad-10670	2	8	rhetorical	rhetorical	ADJ
dad-10670	2	9	structure	structure	NOUN
dad-10670	2	10	theory	theory	NOUN
dad-10670	2	11	(	(	PUNCT
dad-10670	2	12	rst	rst	PROPN
dad-10670	2	13	)	)	PUNCT
dad-10670	2	14	offers	offer	VERB
dad-10670	2	15	a	a	DET
dad-10670	2	16	sound	sound	ADJ
dad-10670	2	17	formalism	formalism	NOUN
dad-10670	2	18	for	for	ADP
dad-10670	2	19	hierarchical	hierarchical	ADJ
dad-10670	2	20	text	text	NOUN
dad-10670	2	21	organization	organization	NOUN
dad-10670	2	22	.	.	PUNCT
dad-10670	3	1	in	in	ADP
dad-10670	3	2	this	this	DET
dad-10670	3	3	article	article	NOUN
dad-10670	3	4	,	,	PUNCT
dad-10670	3	5	we	we	PRON
dad-10670	3	6	present	present	VERB
dad-10670	3	7	hilda	hilda	PROPN
dad-10670	3	8	,	,	PUNCT
dad-10670	3	9	an	an	DET
dad-10670	3	10	implemented	implement	VERB
dad-10670	3	11	discourse	discourse	NOUN
dad-10670	3	12	parser	parser	NOUN
dad-10670	3	13	based	base	VERB
dad-10670	3	14	on	on	ADP
dad-10670	3	15	rst	rst	PROPN
dad-10670	3	16	and	and	CCONJ
dad-10670	3	17	support	support	VERB
dad-10670	3	18	vector	vector	NOUN
dad-10670	3	19	machine	machine	NOUN
dad-10670	3	20	(	(	PUNCT
dad-10670	3	21	svm	svm	ADJ
dad-10670	3	22	)	)	PUNCT
dad-10670	3	23	classification	classification	NOUN
dad-10670	3	24	.	.	PUNCT
dad-10670	4	1	svm	svm	PROPN
dad-10670	4	2	classifiers	classifier	NOUN
dad-10670	4	3	are	be	AUX
dad-10670	4	4	trained	train	VERB
dad-10670	4	5	and	and	CCONJ
dad-10670	4	6	applied	apply	VERB
dad-10670	4	7	to	to	AUX
dad-10670	4	8	discourse	discourse	VERB
dad-10670	4	9	segmentation	segmentation	NOUN
dad-10670	4	10	and	and	CCONJ
dad-10670	4	11	relation	relation	NOUN
dad-10670	4	12	labeling	labeling	NOUN
dad-10670	4	13	.	.	PUNCT
dad-10670	5	1	by	by	ADP
dad-10670	5	2	combining	combine	VERB
dad-10670	5	3	labeling	labeling	NOUN
dad-10670	5	4	with	with	ADP
dad-10670	5	5	a	a	DET
dad-10670	5	6	greedy	greedy	ADJ
dad-10670	5	7	bottom	bottom	ADJ
dad-10670	5	8	-	-	PUNCT
dad-10670	5	9	up	up	ADP
dad-10670	5	10	tree	tree	NOUN
dad-10670	5	11	building	building	NOUN
dad-10670	5	12	approach	approach	NOUN
dad-10670	5	13	,	,	PUNCT
dad-10670	5	14	we	we	PRON
dad-10670	5	15	are	be	AUX
dad-10670	5	16	able	able	ADJ
dad-10670	5	17	to	to	PART
dad-10670	5	18	create	create	VERB
dad-10670	5	19	accurate	accurate	ADJ
dad-10670	5	20	discourse	discourse	NOUN
dad-10670	5	21	trees	tree	NOUN
dad-10670	5	22	in	in	ADP
dad-10670	5	23	linear	linear	ADJ
dad-10670	5	24	time	time	NOUN
dad-10670	5	25	complexity	complexity	NOUN
dad-10670	5	26	with	with	ADP
dad-10670	5	27	respect	respect	NOUN
dad-10670	5	28	to	to	ADP
dad-10670	5	29	the	the	DET
dad-10670	5	30	length	length	NOUN
dad-10670	5	31	of	of	ADP
dad-10670	5	32	the	the	DET
dad-10670	5	33	input	input	NOUN
dad-10670	5	34	text	text	NOUN
dad-10670	5	35	.	.	PUNCT
dad-10670	6	1	importantly	importantly	ADV
dad-10670	6	2	,	,	PUNCT
dad-10670	6	3	our	our	PRON
dad-10670	6	4	parser	parser	NOUN
dad-10670	6	5	can	can	AUX
dad-10670	6	6	parse	parse	VERB
dad-10670	6	7	entire	entire	ADJ
dad-10670	6	8	texts	text	NOUN
dad-10670	6	9	,	,	PUNCT
dad-10670	6	10	whereas	whereas	SCONJ
dad-10670	6	11	the	the	DET
dad-10670	6	12	publicly	publicly	ADV
dad-10670	6	13	available	available	ADJ
dad-10670	6	14	parser	parser	NOUN
dad-10670	6	15	spade	spade	NOUN
dad-10670	6	16	(	(	PUNCT
dad-10670	6	17	soricut	soricut	VERB
dad-10670	6	18	and	and	CCONJ
dad-10670	6	19	marcu	marcu	NOUN
dad-10670	6	20	,	,	PUNCT
dad-10670	6	21	2003	2003	NUM
dad-10670	6	22	)	)	PUNCT
dad-10670	6	23	is	be	AUX
dad-10670	6	24	limited	limit	VERB
dad-10670	6	25	to	to	ADP
dad-10670	6	26	sentence	sentence	NOUN
dad-10670	6	27	level	level	NOUN
dad-10670	6	28	analysis	analysis	NOUN
dad-10670	6	29	.	.	PUNCT
dad-10670	7	1	hilda	hilda	PROPN
dad-10670	7	2	outperforms	outperform	VERB
dad-10670	7	3	other	other	ADJ
dad-10670	7	4	discourse	discourse	NOUN
dad-10670	7	5	parsers	parser	NOUN
dad-10670	7	6	for	for	ADP
dad-10670	7	7	tree	tree	NOUN
dad-10670	7	8	structure	structure	NOUN
dad-10670	7	9	construction	construction	NOUN
dad-10670	7	10	and	and	CCONJ
dad-10670	7	11	discourse	discourse	NOUN
dad-10670	7	12	relation	relation	NOUN
dad-10670	7	13	labeling	labeling	NOUN
dad-10670	7	14	.	.	PUNCT
dad-10670	8	1	for	for	ADP
dad-10670	8	2	the	the	DET
dad-10670	8	3	discourse	discourse	NOUN
dad-10670	8	4	parsing	parsing	NOUN
dad-10670	8	5	task	task	NOUN
dad-10670	8	6	,	,	PUNCT
dad-10670	8	7	our	our	PRON
dad-10670	8	8	system	system	NOUN
dad-10670	8	9	reaches	reach	VERB
dad-10670	8	10	78.3	78.3	NUM
dad-10670	8	11	%	%	NOUN
dad-10670	8	12	of	of	ADP
dad-10670	8	13	the	the	DET
dad-10670	8	14	performance	performance	NOUN
dad-10670	8	15	level	level	NOUN
dad-10670	8	16	of	of	ADP
dad-10670	8	17	human	human	ADJ
dad-10670	8	18	annotators	annotator	NOUN
dad-10670	8	19	.	.	PUNCT
dad-10670	9	1	compared	compare	VERB
dad-10670	9	2	to	to	ADP
dad-10670	9	3	a	a	DET
dad-10670	9	4	state	state	NOUN
dad-10670	9	5	-	-	PUNCT
dad-10670	9	6	of	of	ADP
dad-10670	9	7	-	-	PUNCT
dad-10670	9	8	the	the	DET
dad-10670	9	9	-	-	PUNCT
dad-10670	9	10	art	art	NOUN
dad-10670	9	11	rule	rule	NOUN
dad-10670	9	12	-	-	PUNCT
dad-10670	9	13	based	base	VERB
dad-10670	9	14	discourse	discourse	NOUN
dad-10670	9	15	parser	parser	NOUN
dad-10670	9	16	,	,	PUNCT
dad-10670	9	17	our	our	PRON
dad-10670	9	18	system	system	NOUN
dad-10670	9	19	achieves	achieve	VERB
dad-10670	9	20	an	an	DET
dad-10670	9	21	performance	performance	NOUN
dad-10670	9	22	increase	increase	NOUN
dad-10670	9	23	of	of	ADP
dad-10670	9	24	11.6	11.6	NUM
dad-10670	9	25	%	%	NOUN
dad-10670	9	26	.	.	PUNCT
dad-10670	10	1	keywords	keyword	NOUN
dad-10670	10	2	:	:	PUNCT
dad-10670	10	3	discourse	discourse	NOUN
dad-10670	10	4	parser	parser	NOUN
dad-10670	10	5	,	,	PUNCT
dad-10670	10	6	rhetorical	rhetorical	ADJ
dad-10670	10	7	structure	structure	NOUN
dad-10670	10	8	theory	theory	NOUN
dad-10670	10	9	,	,	PUNCT
dad-10670	10	10	support	support	VERB
dad-10670	10	11	vector	vector	NOUN
dad-10670	10	12	machines	machine	NOUN
dad-10670	10	13	1	1	NUM
dad-10670	10	14	.	.	PUNCT
dad-10670	11	1	introduction	introduction	NOUN
dad-10670	11	2	in	in	ADP
dad-10670	11	3	the	the	DET
dad-10670	11	4	last	last	ADJ
dad-10670	11	5	twenty	twenty	NUM
dad-10670	11	6	years	year	NOUN
dad-10670	11	7	,	,	PUNCT
dad-10670	11	8	the	the	DET
dad-10670	11	9	study	study	NOUN
dad-10670	11	10	of	of	ADP
dad-10670	11	11	discourse	discourse	NOUN
dad-10670	11	12	has	have	AUX
dad-10670	11	13	received	receive	VERB
dad-10670	11	14	continuous	continuous	ADJ
dad-10670	11	15	attention	attention	NOUN
dad-10670	11	16	from	from	ADP
dad-10670	11	17	the	the	DET
dad-10670	11	18	natural	natural	ADJ
dad-10670	11	19	language	language	NOUN
dad-10670	11	20	processing	processing	NOUN
dad-10670	11	21	community	community	NOUN
dad-10670	11	22	(	(	PUNCT
dad-10670	11	23	moore	moore	NOUN
dad-10670	11	24	and	and	CCONJ
dad-10670	11	25	wiemer	wiemer	NOUN
dad-10670	11	26	-	-	PUNCT
dad-10670	11	27	hastings	hastings	PROPN
dad-10670	11	28	,	,	PUNCT
dad-10670	11	29	2003	2003	NUM
dad-10670	11	30	)	)	PUNCT
dad-10670	11	31	.	.	PUNCT
dad-10670	12	1	discourse	discourse	NOUN
dad-10670	12	2	structure	structure	NOUN
dad-10670	12	3	is	be	AUX
dad-10670	12	4	fundamental	fundamental	ADJ
dad-10670	12	5	to	to	ADP
dad-10670	12	6	many	many	ADJ
dad-10670	12	7	text	text	NOUN
dad-10670	12	8	-	-	PUNCT
dad-10670	12	9	based	base	VERB
dad-10670	12	10	applications	application	NOUN
dad-10670	12	11	,	,	PUNCT
dad-10670	12	12	such	such	ADJ
dad-10670	12	13	as	as	ADP
dad-10670	12	14	question	question	NOUN
dad-10670	12	15	–	–	PUNCT
dad-10670	12	16	answering	answering	NOUN
dad-10670	12	17	(	(	PUNCT
dad-10670	12	18	chai	chai	NOUN
dad-10670	12	19	and	and	CCONJ
dad-10670	12	20	jin	jin	NOUN
dad-10670	12	21	,	,	PUNCT
dad-10670	12	22	2004	2004	NUM
dad-10670	12	23	)	)	PUNCT
dad-10670	12	24	or	or	CCONJ
dad-10670	12	25	∗.	∗.	NOUN
dad-10670	12	26	this	this	DET
dad-10670	12	27	article	article	NOUN
dad-10670	12	28	is	be	AUX
dad-10670	12	29	a	a	DET
dad-10670	12	30	significantly	significantly	ADV
dad-10670	12	31	improved	improve	VERB
dad-10670	12	32	and	and	CCONJ
dad-10670	12	33	extended	extend	VERB
dad-10670	12	34	version	version	NOUN
dad-10670	12	35	of	of	ADP
dad-10670	12	36	duverle	duverle	NOUN
dad-10670	12	37	and	and	CCONJ
dad-10670	12	38	prendinger	prendinger	NOUN
dad-10670	12	39	(	(	PUNCT
dad-10670	12	40	2009	2009	NUM
dad-10670	12	41	)	)	PUNCT
dad-10670	12	42	.	.	PUNCT
dad-10670	13	1	c	c	X
dad-10670	13	2	©	©	PROPN
dad-10670	13	3	2010	2010	NUM
dad-10670	13	4	hugo	hugo	PROPN
dad-10670	13	5	hernault	hernault	NOUN
dad-10670	13	6	,	,	PUNCT
dad-10670	13	7	helmut	helmut	PROPN
dad-10670	13	8	prendinger	prendinger	NOUN
dad-10670	13	9	,	,	PUNCT
dad-10670	13	10	david	david	PROPN
dad-10670	13	11	a.	a.	NOUN
dad-10670	13	12	duverle	duverle	PROPN
dad-10670	13	13	,	,	PUNCT
dad-10670	13	14	and	and	CCONJ
dad-10670	13	15	mitsuru	mitsuru	PROPN
dad-10670	13	16	ishizuka	ishizuka	PROPN
dad-10670	13	17	submitted	submit	VERB
dad-10670	13	18	2/10	2/10	NUM
dad-10670	13	19	;	;	PUNCT
dad-10670	13	20	accepted	accept	VERB
dad-10670	13	21	11/10	11/10	NUM
dad-10670	13	22	;	;	PUNCT
dad-10670	13	23	published	publish	VERB
dad-10670	13	24	online	online	ADJ
dad-10670	13	25	12/10	12/10	NUM
dad-10670	13	26	hernault	hernault	NOUN
dad-10670	13	27	,	,	PUNCT
dad-10670	13	28	prendinger	prendinger	NOUN
dad-10670	13	29	,	,	PUNCT
dad-10670	13	30	duverle	duverle	NOUN
dad-10670	13	31	,	,	PUNCT
dad-10670	13	32	and	and	CCONJ
dad-10670	13	33	ishizuka	ishizuka	PROPN
dad-10670	13	34	dialogue	dialogue	NOUN
dad-10670	13	35	generation	generation	NOUN
dad-10670	13	36	(	(	PUNCT
dad-10670	13	37	prendinger	prendinger	NOUN
dad-10670	13	38	et	et	PROPN
dad-10670	13	39	al	al	PROPN
dad-10670	13	40	.	.	PROPN
dad-10670	13	41	,	,	PUNCT
dad-10670	13	42	2007	2007	NUM
dad-10670	13	43	)	)	PUNCT
dad-10670	13	44	.	.	PUNCT
dad-10670	14	1	those	those	DET
dad-10670	14	2	applications	application	NOUN
dad-10670	14	3	require	require	VERB
dad-10670	14	4	the	the	DET
dad-10670	14	5	availability	availability	NOUN
dad-10670	14	6	of	of	ADP
dad-10670	14	7	robust	robust	ADJ
dad-10670	14	8	and	and	CCONJ
dad-10670	14	9	efficient	efficient	ADJ
dad-10670	14	10	discourse	discourse	NOUN
dad-10670	14	11	parsers	parser	NOUN
dad-10670	14	12	.	.	PUNCT
dad-10670	15	1	several	several	ADJ
dad-10670	15	2	attempts	attempt	NOUN
dad-10670	15	3	have	have	AUX
dad-10670	15	4	been	be	AUX
dad-10670	15	5	made	make	VERB
dad-10670	15	6	to	to	PART
dad-10670	15	7	create	create	VERB
dad-10670	15	8	discourse	discourse	NOUN
dad-10670	15	9	parsers	parser	NOUN
dad-10670	15	10	in	in	ADP
dad-10670	15	11	the	the	DET
dad-10670	15	12	framework	framework	NOUN
dad-10670	15	13	of	of	ADP
dad-10670	15	14	the	the	DET
dad-10670	15	15	rhetorical	rhetorical	ADJ
dad-10670	15	16	structure	structure	NOUN
dad-10670	15	17	theory	theory	NOUN
dad-10670	15	18	(	(	PUNCT
dad-10670	15	19	mann	mann	PROPN
dad-10670	15	20	and	and	CCONJ
dad-10670	15	21	thompson	thompson	PROPN
dad-10670	15	22	,	,	PUNCT
dad-10670	15	23	1988	1988	NUM
dad-10670	15	24	)	)	PUNCT
dad-10670	15	25	,	,	PUNCT
dad-10670	15	26	which	which	PRON
dad-10670	15	27	is	be	AUX
dad-10670	15	28	one	one	NUM
dad-10670	15	29	of	of	ADP
dad-10670	15	30	the	the	DET
dad-10670	15	31	most	most	ADV
dad-10670	15	32	widely	widely	ADV
dad-10670	15	33	used	use	VERB
dad-10670	15	34	theories	theory	NOUN
dad-10670	15	35	of	of	ADP
dad-10670	15	36	text	text	NOUN
dad-10670	15	37	organization	organization	NOUN
dad-10670	15	38	.	.	PUNCT
dad-10670	16	1	in	in	ADP
dad-10670	16	2	rst	rst	PROPN
dad-10670	16	3	,	,	PUNCT
dad-10670	16	4	a	a	DET
dad-10670	16	5	text	text	NOUN
dad-10670	16	6	is	be	AUX
dad-10670	16	7	first	first	ADV
dad-10670	16	8	divided	divide	VERB
dad-10670	16	9	into	into	ADP
dad-10670	16	10	non	non	ADJ
dad-10670	16	11	-	-	ADJ
dad-10670	16	12	overlapping	overlapping	ADJ
dad-10670	16	13	text	text	NOUN
dad-10670	16	14	chunks	chunk	NOUN
dad-10670	16	15	,	,	PUNCT
dad-10670	16	16	called	call	VERB
dad-10670	16	17	elementary	elementary	ADJ
dad-10670	16	18	discourse	discourse	NOUN
dad-10670	16	19	units	unit	NOUN
dad-10670	16	20	(	(	PUNCT
dad-10670	16	21	abbreviated	abbreviate	VERB
dad-10670	16	22	:	:	PUNCT
dad-10670	16	23	edus	edus	NOUN
dad-10670	16	24	)	)	PUNCT
dad-10670	16	25	.	.	PUNCT
dad-10670	17	1	for	for	ADP
dad-10670	17	2	instance	instance	NOUN
dad-10670	17	3	,	,	PUNCT
dad-10670	17	4	the	the	DET
dad-10670	17	5	following	follow	VERB
dad-10670	17	6	sentence	sentence	NOUN
dad-10670	17	7	,	,	PUNCT
dad-10670	17	8	taken	take	VERB
dad-10670	17	9	from	from	ADP
dad-10670	17	10	the	the	DET
dad-10670	17	11	rst	rst	PROPN
dad-10670	17	12	discourse	discourse	NOUN
dad-10670	17	13	treebank	treebank	PROPN
dad-10670	17	14	corpus	corpus	PROPN
dad-10670	17	15	(	(	PUNCT
dad-10670	17	16	carlson	carlson	PROPN
dad-10670	17	17	et	et	PROPN
dad-10670	17	18	al	al	PROPN
dad-10670	17	19	.	.	PROPN
dad-10670	17	20	,	,	PUNCT
dad-10670	17	21	2001	2001	NUM
dad-10670	17	22	):	):	PUNCT
dad-10670	17	23	farm	farm	NOUN
dad-10670	17	24	lending	lending	NOUN
dad-10670	17	25	was	be	AUX
dad-10670	17	26	enacted	enact	VERB
dad-10670	17	27	to	to	PART
dad-10670	17	28	correct	correct	VERB
dad-10670	17	29	this	this	DET
dad-10670	17	30	problem	problem	NOUN
dad-10670	17	31	by	by	ADP
dad-10670	17	32	providing	provide	VERB
dad-10670	17	33	a	a	DET
dad-10670	17	34	reliable	reliable	ADJ
dad-10670	17	35	flow	flow	NOUN
dad-10670	17	36	of	of	ADP
dad-10670	17	37	lendable	lendable	ADJ
dad-10670	17	38	funds	fund	NOUN
dad-10670	17	39	.	.	PUNCT
dad-10670	18	1	can	can	AUX
dad-10670	18	2	be	be	AUX
dad-10670	18	3	segmented	segment	VERB
dad-10670	18	4	into	into	ADP
dad-10670	18	5	edus	edus	NOUN
dad-10670	18	6	as	as	SCONJ
dad-10670	18	7	shown	show	VERB
dad-10670	18	8	in	in	ADP
dad-10670	18	9	figure	figure	NOUN
dad-10670	18	10	1	1	NUM
dad-10670	18	11	.	.	PUNCT
dad-10670	19	1	[	[	X
dad-10670	19	2	farm	farm	NOUN
dad-10670	19	3	lending	lending	NOUN
dad-10670	19	4	was	be	AUX
dad-10670	19	5	enacted]1a	enacted]1a	NOUN
dad-10670	19	6	[	[	X
dad-10670	19	7	to	to	PART
dad-10670	19	8	correct	correct	VERB
dad-10670	19	9	this	this	DET
dad-10670	19	10	problem]1b	problem]1b	NOUN
dad-10670	19	11	[	[	X
dad-10670	19	12	by	by	ADP
dad-10670	19	13	providing	provide	VERB
dad-10670	19	14	a	a	DET
dad-10670	19	15	reliable	reliable	ADJ
dad-10670	19	16	flow	flow	NOUN
dad-10670	19	17	of	of	ADP
dad-10670	19	18	lendable	lendable	ADJ
dad-10670	19	19	funds.]1c	funds.]1c	NOUN
dad-10670	19	20	(	(	PUNCT
dad-10670	19	21	wsj1131	wsj1131	NOUN
dad-10670	19	22	)	)	PUNCT
dad-10670	19	23	figure	figure	NOUN
dad-10670	19	24	1	1	NUM
dad-10670	19	25	:	:	PUNCT
dad-10670	19	26	segmentation	segmentation	NOUN
dad-10670	19	27	of	of	ADP
dad-10670	19	28	a	a	DET
dad-10670	19	29	sentence	sentence	NOUN
dad-10670	19	30	into	into	ADP
dad-10670	19	31	edus	edus	NOUN
dad-10670	19	32	next	next	ADV
dad-10670	19	33	,	,	PUNCT
dad-10670	19	34	consecutive	consecutive	ADJ
dad-10670	19	35	edus	edus	NOUN
dad-10670	19	36	are	be	AUX
dad-10670	19	37	put	put	VERB
dad-10670	19	38	in	in	ADP
dad-10670	19	39	relation	relation	NOUN
dad-10670	19	40	with	with	ADP
dad-10670	19	41	each	each	DET
dad-10670	19	42	other	other	ADJ
dad-10670	19	43	,	,	PUNCT
dad-10670	19	44	using	use	VERB
dad-10670	19	45	a	a	DET
dad-10670	19	46	pre	pre	ADJ
dad-10670	19	47	-	-	ADJ
dad-10670	19	48	defined	define	VERB
dad-10670	19	49	set	set	NOUN
dad-10670	19	50	of	of	ADP
dad-10670	19	51	rhetorical	rhetorical	ADJ
dad-10670	19	52	,	,	PUNCT
dad-10670	19	53	or	or	CCONJ
dad-10670	19	54	discourse	discourse	NOUN
dad-10670	19	55	,	,	PUNCT
dad-10670	19	56	relations	relation	NOUN
dad-10670	19	57	.	.	PUNCT
dad-10670	20	1	the	the	DET
dad-10670	20	2	final	final	ADJ
dad-10670	20	3	goal	goal	NOUN
dad-10670	20	4	of	of	ADP
dad-10670	20	5	the	the	DET
dad-10670	20	6	discourse	discourse	NOUN
dad-10670	20	7	parser	parser	NOUN
dad-10670	20	8	is	be	AUX
dad-10670	20	9	to	to	PART
dad-10670	20	10	produce	produce	VERB
dad-10670	20	11	a	a	DET
dad-10670	20	12	tree	tree	NOUN
dad-10670	20	13	structure	structure	NOUN
dad-10670	20	14	as	as	ADP
dad-10670	20	15	a	a	DET
dad-10670	20	16	representation	representation	NOUN
dad-10670	20	17	of	of	ADP
dad-10670	20	18	how	how	SCONJ
dad-10670	20	19	all	all	DET
dad-10670	20	20	units	unit	NOUN
dad-10670	20	21	of	of	ADP
dad-10670	20	22	the	the	DET
dad-10670	20	23	text	text	NOUN
dad-10670	20	24	relate	relate	VERB
dad-10670	20	25	to	to	ADP
dad-10670	20	26	each	each	DET
dad-10670	20	27	other	other	ADJ
dad-10670	20	28	.	.	PUNCT
dad-10670	21	1	figure	figure	NOUN
dad-10670	21	2	2	2	NUM
dad-10670	21	3	shows	show	VERB
dad-10670	21	4	two	two	NUM
dad-10670	21	5	types	type	NOUN
dad-10670	21	6	of	of	ADP
dad-10670	21	7	discourse	discourse	NOUN
dad-10670	21	8	relations	relation	NOUN
dad-10670	21	9	in	in	ADP
dad-10670	21	10	rst	rst	ADJ
dad-10670	21	11	:	:	PUNCT
dad-10670	21	12	hypotactic	hypotactic	ADJ
dad-10670	21	13	(	(	PUNCT
dad-10670	21	14	‘	'	PUNCT
dad-10670	21	15	mononuclear	mononuclear	ADJ
dad-10670	21	16	’	'	PUNCT
dad-10670	21	17	)	)	PUNCT
dad-10670	21	18	and	and	CCONJ
dad-10670	21	19	paratactic	paratactic	NOUN
dad-10670	21	20	(	(	PUNCT
dad-10670	21	21	‘	'	PUNCT
dad-10670	21	22	multi	multi	ADJ
dad-10670	21	23	-	-	ADJ
dad-10670	21	24	nuclear	nuclear	ADJ
dad-10670	21	25	’	'	PUNCT
dad-10670	21	26	)	)	PUNCT
dad-10670	21	27	.	.	PUNCT
dad-10670	22	1	the	the	DET
dad-10670	22	2	left	left	ADJ
dad-10670	22	3	-	-	PUNCT
dad-10670	22	4	hand	hand	NOUN
dad-10670	22	5	tree	tree	NOUN
dad-10670	22	6	is	be	AUX
dad-10670	22	7	an	an	DET
dad-10670	22	8	example	example	NOUN
dad-10670	22	9	of	of	ADP
dad-10670	22	10	a	a	DET
dad-10670	22	11	mononuclear	mononuclear	ADJ
dad-10670	22	12	relation	relation	NOUN
dad-10670	22	13	.	.	PUNCT
dad-10670	23	1	here	here	ADV
dad-10670	23	2	the	the	DET
dad-10670	23	3	two	two	NUM
dad-10670	23	4	discourse	discourse	NOUN
dad-10670	23	5	units	unit	NOUN
dad-10670	23	6	connected	connect	VERB
dad-10670	23	7	by	by	ADP
dad-10670	23	8	the	the	DET
dad-10670	23	9	relation	relation	NOUN
dad-10670	23	10	have	have	VERB
dad-10670	23	11	different	different	ADJ
dad-10670	23	12	status	status	NOUN
dad-10670	23	13	,	,	PUNCT
dad-10670	23	14	which	which	PRON
dad-10670	23	15	is	be	AUX
dad-10670	23	16	indicated	indicate	VERB
dad-10670	23	17	by	by	ADP
dad-10670	23	18	the	the	DET
dad-10670	23	19	direction	direction	NOUN
dad-10670	23	20	of	of	ADP
dad-10670	23	21	the	the	DET
dad-10670	23	22	arrow	arrow	NOUN
dad-10670	23	23	.	.	PUNCT
dad-10670	24	1	the	the	DET
dad-10670	24	2	endpoint	endpoint	NOUN
dad-10670	24	3	of	of	ADP
dad-10670	24	4	the	the	DET
dad-10670	24	5	arrow	arrow	NOUN
dad-10670	24	6	denotes	denote	VERB
dad-10670	24	7	the	the	DET
dad-10670	24	8	‘	'	PUNCT
dad-10670	24	9	nucleus	nucleus	NOUN
dad-10670	24	10	’	'	PUNCT
dad-10670	24	11	of	of	ADP
dad-10670	24	12	the	the	DET
dad-10670	24	13	relation	relation	NOUN
dad-10670	24	14	,	,	PUNCT
dad-10670	24	15	whereas	whereas	SCONJ
dad-10670	24	16	the	the	DET
dad-10670	24	17	unit	unit	NOUN
dad-10670	24	18	at	at	ADP
dad-10670	24	19	the	the	DET
dad-10670	24	20	other	other	ADJ
dad-10670	24	21	end	end	NOUN
dad-10670	24	22	is	be	AUX
dad-10670	24	23	referred	refer	VERB
dad-10670	24	24	to	to	ADP
dad-10670	24	25	as	as	ADP
dad-10670	24	26	the	the	DET
dad-10670	24	27	‘	'	PUNCT
dad-10670	24	28	satellite	satellite	NOUN
dad-10670	24	29	’	'	PUNCT
dad-10670	24	30	.	.	PUNCT
dad-10670	25	1	the	the	DET
dad-10670	25	2	nucleus	nucleus	NOUN
dad-10670	25	3	is	be	AUX
dad-10670	25	4	considered	consider	VERB
dad-10670	25	5	as	as	ADP
dad-10670	25	6	more	more	ADV
dad-10670	25	7	prominent	prominent	ADJ
dad-10670	25	8	than	than	ADP
dad-10670	25	9	the	the	DET
dad-10670	25	10	satellite	satellite	NOUN
dad-10670	25	11	.	.	PUNCT
dad-10670	26	1	for	for	ADP
dad-10670	26	2	example	example	NOUN
dad-10670	26	3	,	,	PUNCT
dad-10670	26	4	condition	condition	NOUN
dad-10670	26	5	is	be	AUX
dad-10670	26	6	defined	define	VERB
dad-10670	26	7	in	in	ADP
dad-10670	26	8	carlson	carlson	PROPN
dad-10670	26	9	et	et	PROPN
dad-10670	26	10	al	al	PROPN
dad-10670	26	11	.	.	PROPN
dad-10670	27	1	(	(	PUNCT
dad-10670	27	2	2001	2001	NUM
dad-10670	27	3	)	)	PUNCT
dad-10670	28	1	as	as	SCONJ
dad-10670	28	2	:	:	PUNCT
dad-10670	28	3	“	"	PUNCT
dad-10670	28	4	in	in	ADP
dad-10670	28	5	a	a	DET
dad-10670	28	6	condition	condition	NOUN
dad-10670	28	7	relation	relation	NOUN
dad-10670	28	8	,	,	PUNCT
dad-10670	28	9	the	the	DET
dad-10670	28	10	truth	truth	NOUN
dad-10670	28	11	of	of	ADP
dad-10670	28	12	the	the	DET
dad-10670	28	13	proposition	proposition	NOUN
dad-10670	28	14	associated	associate	VERB
dad-10670	28	15	with	with	ADP
dad-10670	28	16	the	the	DET
dad-10670	28	17	nucleus	nucleus	NOUN
dad-10670	28	18	is	be	AUX
dad-10670	28	19	a	a	DET
dad-10670	28	20	consequence	consequence	NOUN
dad-10670	28	21	of	of	ADP
dad-10670	28	22	the	the	DET
dad-10670	28	23	fulfillment	fulfillment	NOUN
dad-10670	28	24	of	of	ADP
dad-10670	28	25	the	the	DET
dad-10670	28	26	condition	condition	NOUN
dad-10670	28	27	in	in	ADP
dad-10670	28	28	the	the	DET
dad-10670	28	29	satellite	satellite	NOUN
dad-10670	28	30	.	.	PUNCT
dad-10670	29	1	the	the	DET
dad-10670	29	2	satellite	satellite	NOUN
dad-10670	29	3	presents	present	VERB
dad-10670	29	4	a	a	DET
dad-10670	29	5	situation	situation	NOUN
dad-10670	29	6	that	that	PRON
dad-10670	29	7	is	be	AUX
dad-10670	29	8	not	not	PART
dad-10670	29	9	realized	realize	VERB
dad-10670	29	10	.	.	PUNCT
dad-10670	29	11	”	"	PUNCT
dad-10670	30	1	other	other	ADJ
dad-10670	30	2	mononuclear	mononuclear	ADJ
dad-10670	30	3	discourse	discourse	NOUN
dad-10670	30	4	relations	relation	NOUN
dad-10670	30	5	described	describe	VERB
dad-10670	30	6	in	in	ADP
dad-10670	30	7	(	(	PUNCT
dad-10670	30	8	mann	mann	PROPN
dad-10670	30	9	and	and	CCONJ
dad-10670	30	10	thompson	thompson	PROPN
dad-10670	30	11	,	,	PUNCT
dad-10670	30	12	1988	1988	NUM
dad-10670	30	13	;	;	PUNCT
dad-10670	30	14	carlson	carlson	PROPN
dad-10670	30	15	et	et	PROPN
dad-10670	30	16	al	al	PROPN
dad-10670	30	17	.	.	PROPN
dad-10670	30	18	,	,	PUNCT
dad-10670	30	19	2001	2001	NUM
dad-10670	30	20	)	)	PUNCT
dad-10670	30	21	include	include	VERB
dad-10670	30	22	:	:	PUNCT
dad-10670	30	23	background	background	NOUN
dad-10670	30	24	,	,	PUNCT
dad-10670	30	25	circumstance	circumstance	NOUN
dad-10670	30	26	,	,	PUNCT
dad-10670	30	27	elaboration	elaboration	NOUN
dad-10670	30	28	and	and	CCONJ
dad-10670	30	29	purpose	purpose	NOUN
dad-10670	30	30	.	.	PUNCT
dad-10670	31	1	paratactic	paratactic	NOUN
dad-10670	31	2	(	(	PUNCT
dad-10670	31	3	‘	'	PUNCT
dad-10670	31	4	multi	multi	ADJ
dad-10670	31	5	-	-	ADJ
dad-10670	31	6	nuclear	nuclear	ADJ
dad-10670	31	7	’	'	PUNCT
dad-10670	31	8	)	)	PUNCT
dad-10670	31	9	relations	relation	NOUN
dad-10670	31	10	,	,	PUNCT
dad-10670	31	11	on	on	ADP
dad-10670	31	12	the	the	DET
dad-10670	31	13	other	other	ADJ
dad-10670	31	14	hand	hand	NOUN
dad-10670	31	15	,	,	PUNCT
dad-10670	31	16	have	have	VERB
dad-10670	31	17	no	no	DET
dad-10670	31	18	distinguished	distinguished	ADJ
dad-10670	31	19	nucleus	nucleus	NOUN
dad-10670	31	20	and	and	CCONJ
dad-10670	31	21	the	the	DET
dad-10670	31	22	relation	relation	NOUN
dad-10670	31	23	can	can	AUX
dad-10670	31	24	connect	connect	VERB
dad-10670	31	25	an	an	DET
dad-10670	31	26	arbitrary	arbitrary	ADJ
dad-10670	31	27	number	number	NOUN
dad-10670	31	28	of	of	ADP
dad-10670	31	29	discourse	discourse	NOUN
dad-10670	31	30	units	unit	NOUN
dad-10670	31	31	.	.	PUNCT
dad-10670	32	1	the	the	DET
dad-10670	32	2	relation	relation	NOUN
dad-10670	32	3	to	to	ADP
dad-10670	32	4	the	the	DET
dad-10670	32	5	right	right	NOUN
dad-10670	32	6	in	in	ADP
dad-10670	32	7	figure	figure	NOUN
dad-10670	32	8	2	2	NUM
dad-10670	32	9	is	be	AUX
dad-10670	32	10	multi	multi	ADJ
dad-10670	32	11	-	-	ADJ
dad-10670	32	12	nuclear	nuclear	ADJ
dad-10670	32	13	.	.	PUNCT
dad-10670	33	1	for	for	ADP
dad-10670	33	2	example	example	NOUN
dad-10670	33	3	,	,	PUNCT
dad-10670	33	4	list	list	NOUN
dad-10670	33	5	is	be	AUX
dad-10670	33	6	defined	define	VERB
dad-10670	33	7	as	as	ADP
dad-10670	33	8	“	"	PUNCT
dad-10670	33	9	[	[	X
dad-10670	33	10	...	...	X
dad-10670	33	11	]	]	X
dad-10670	33	12	a	a	DET
dad-10670	33	13	multinuclear	multinuclear	ADJ
dad-10670	33	14	relation	relation	NOUN
dad-10670	33	15	whose	whose	DET
dad-10670	33	16	elements	element	NOUN
dad-10670	33	17	can	can	AUX
dad-10670	33	18	be	be	AUX
dad-10670	33	19	listed	list	VERB
dad-10670	33	20	,	,	PUNCT
dad-10670	33	21	but	but	CCONJ
dad-10670	33	22	which	which	PRON
dad-10670	33	23	are	be	AUX
dad-10670	33	24	not	not	PART
dad-10670	33	25	in	in	ADP
dad-10670	33	26	a	a	DET
dad-10670	33	27	comparison	comparison	NOUN
dad-10670	33	28	[	[	X
dad-10670	33	29	...	...	PUNCT
dad-10670	33	30	]	]	X
dad-10670	33	31	”	"	PUNCT
dad-10670	33	32	(	(	PUNCT
dad-10670	33	33	carlson	carlson	PROPN
dad-10670	33	34	et	et	PROPN
dad-10670	33	35	al	al	PROPN
dad-10670	33	36	.	.	PROPN
dad-10670	33	37	,	,	PUNCT
dad-10670	33	38	2001	2001	NUM
dad-10670	33	39	)	)	PUNCT
dad-10670	33	40	.	.	PUNCT
dad-10670	34	1	the	the	DET
dad-10670	34	2	following	follow	VERB
dad-10670	34	3	discourse	discourse	NOUN
dad-10670	34	4	relations	relation	NOUN
dad-10670	34	5	are	be	AUX
dad-10670	34	6	also	also	ADV
dad-10670	34	7	multi	multi	ADJ
dad-10670	34	8	-	-	ADJ
dad-10670	34	9	nuclear	nuclear	ADJ
dad-10670	34	10	(	(	PUNCT
dad-10670	34	11	see	see	VERB
dad-10670	34	12	mann	mann	PROPN
dad-10670	34	13	and	and	CCONJ
dad-10670	34	14	thompson	thompson	PROPN
dad-10670	34	15	(	(	PUNCT
dad-10670	34	16	1988	1988	NUM
dad-10670	34	17	)	)	PUNCT
dad-10670	34	18	;	;	PUNCT
dad-10670	35	1	carlson	carlson	PROPN
dad-10670	35	2	et	et	PROPN
dad-10670	35	3	al	al	PROPN
dad-10670	35	4	.	.	PROPN
dad-10670	35	5	(	(	PUNCT
dad-10670	35	6	2001	2001	NUM
dad-10670	35	7	)	)	PUNCT
dad-10670	35	8	):	):	PUNCT
dad-10670	35	9	contrast	contrast	NOUN
dad-10670	35	10	,	,	PUNCT
dad-10670	35	11	disjunction	disjunction	NOUN
dad-10670	35	12	,	,	PUNCT
dad-10670	35	13	sequence	sequence	NOUN
dad-10670	35	14	,	,	PUNCT
dad-10670	35	15	and	and	CCONJ
dad-10670	35	16	topic	topic	NOUN
dad-10670	35	17	-	-	PUNCT
dad-10670	35	18	comment	comment	NOUN
dad-10670	35	19	.	.	PUNCT
dad-10670	36	1	r	r	NOUN
dad-10670	36	2	relationname	relationname	NOUN
dad-10670	36	3	satellite	satellite	PROPN
dad-10670	36	4	nucleus	nucleus	PROPN
dad-10670	36	5	relationname	relationname	PROPN
dad-10670	36	6	nucleus	nucleus	PROPN
dad-10670	36	7	nucleus	nucleus	PROPN
dad-10670	36	8	figure	figure	NOUN
dad-10670	36	9	2	2	NUM
dad-10670	36	10	:	:	PUNCT
dad-10670	36	11	the	the	DET
dad-10670	36	12	two	two	NUM
dad-10670	36	13	relation	relation	NOUN
dad-10670	36	14	types	type	NOUN
dad-10670	36	15	in	in	ADP
dad-10670	36	16	rst	rst	PROPN
dad-10670	36	17	.	.	PUNCT
dad-10670	37	1	left	leave	VERB
dad-10670	37	2	:	:	PUNCT
dad-10670	37	3	mononuclear	mononuclear	ADJ
dad-10670	37	4	;	;	PUNCT
dad-10670	37	5	right	right	ADJ
dad-10670	37	6	:	:	PUNCT
dad-10670	37	7	multi	multi	ADJ
dad-10670	37	8	-	-	ADJ
dad-10670	37	9	nuclear	nuclear	ADJ
dad-10670	37	10	figure	figure	NOUN
dad-10670	37	11	3	3	NUM
dad-10670	37	12	shows	show	VERB
dad-10670	37	13	the	the	DET
dad-10670	37	14	tree	tree	NOUN
dad-10670	37	15	representation	representation	NOUN
dad-10670	37	16	of	of	ADP
dad-10670	37	17	the	the	DET
dad-10670	37	18	preceding	precede	VERB
dad-10670	37	19	example	example	NOUN
dad-10670	37	20	(	(	PUNCT
dad-10670	37	21	see	see	VERB
dad-10670	37	22	figure	figure	NOUN
dad-10670	37	23	1	1	NUM
dad-10670	37	24	)	)	PUNCT
dad-10670	37	25	.	.	PUNCT
dad-10670	38	1	2	2	NUM
dad-10670	38	2	hilda	hilda	VERB
dad-10670	38	3	:	:	PUNCT
dad-10670	38	4	a	a	DET
dad-10670	38	5	discourse	discourse	NOUN
dad-10670	38	6	parser	parser	NOUN
dad-10670	38	7	using	use	VERB
dad-10670	38	8	support	support	NOUN
dad-10670	38	9	vector	vector	NOUN
dad-10670	38	10	machine	machine	NOUN
dad-10670	38	11	classification	classification	NOUN
dad-10670	38	12	enablement	enablement	NOUN
dad-10670	38	13	farm	farm	NOUN
dad-10670	38	14	lending	lending	NOUN
dad-10670	38	15	was	be	AUX
dad-10670	38	16	enacted	enact	VERB
dad-10670	38	17	manner	manner	NOUN
dad-10670	38	18	-	-	PUNCT
dad-10670	38	19	means	means	NOUN
dad-10670	38	20	to	to	PART
dad-10670	38	21	correct	correct	VERB
dad-10670	38	22	this	this	DET
dad-10670	38	23	problem	problem	NOUN
dad-10670	38	24	by	by	ADP
dad-10670	38	25	providing	provide	VERB
dad-10670	38	26	a	a	DET
dad-10670	38	27	reliable	reliable	ADJ
dad-10670	38	28	flow	flow	NOUN
dad-10670	38	29	of	of	ADP
dad-10670	38	30	lendable	lendable	ADJ
dad-10670	38	31	funds	fund	NOUN
dad-10670	38	32	.	.	PUNCT
dad-10670	39	1	figure	figure	VERB
dad-10670	39	2	3	3	NUM
dad-10670	39	3	:	:	PUNCT
dad-10670	39	4	a	a	DET
dad-10670	39	5	simple	simple	ADJ
dad-10670	39	6	discourse	discourse	NOUN
dad-10670	39	7	tree	tree	NOUN
dad-10670	39	8	(	(	PUNCT
dad-10670	39	9	wsj1131	wsj1131	PROPN
dad-10670	39	10	)	)	PUNCT
dad-10670	39	11	the	the	DET
dad-10670	39	12	tasks	task	NOUN
dad-10670	39	13	of	of	ADP
dad-10670	39	14	discourse	discourse	NOUN
dad-10670	39	15	segmentation	segmentation	NOUN
dad-10670	39	16	and	and	CCONJ
dad-10670	39	17	relation	relation	NOUN
dad-10670	39	18	labeling	labeling	NOUN
dad-10670	39	19	have	have	AUX
dad-10670	39	20	been	be	AUX
dad-10670	39	21	previously	previously	ADV
dad-10670	39	22	modeled	model	VERB
dad-10670	39	23	using	use	VERB
dad-10670	39	24	rule	rule	NOUN
dad-10670	39	25	-	-	PUNCT
dad-10670	39	26	based	base	VERB
dad-10670	39	27	and	and	CCONJ
dad-10670	39	28	statistical	statistical	ADJ
dad-10670	39	29	methods	method	NOUN
dad-10670	39	30	.	.	PUNCT
dad-10670	40	1	in	in	ADP
dad-10670	40	2	rule	rule	NOUN
dad-10670	40	3	-	-	PUNCT
dad-10670	40	4	based	base	VERB
dad-10670	40	5	systems	system	NOUN
dad-10670	40	6	,	,	PUNCT
dad-10670	40	7	rules	rule	NOUN
dad-10670	40	8	manipulating	manipulate	VERB
dad-10670	40	9	syntactic	syntactic	ADJ
dad-10670	40	10	and	and	CCONJ
dad-10670	40	11	lexical	lexical	ADJ
dad-10670	40	12	information	information	NOUN
dad-10670	40	13	are	be	AUX
dad-10670	40	14	defined	define	VERB
dad-10670	40	15	manually	manually	ADV
dad-10670	40	16	.	.	PUNCT
dad-10670	41	1	when	when	SCONJ
dad-10670	41	2	applied	apply	VERB
dad-10670	41	3	to	to	ADP
dad-10670	41	4	a	a	DET
dad-10670	41	5	text	text	NOUN
dad-10670	41	6	,	,	PUNCT
dad-10670	41	7	these	these	DET
dad-10670	41	8	rules	rule	NOUN
dad-10670	41	9	enable	enable	VERB
dad-10670	41	10	to	to	PART
dad-10670	41	11	detect	detect	VERB
dad-10670	41	12	the	the	DET
dad-10670	41	13	presence	presence	NOUN
dad-10670	41	14	of	of	ADP
dad-10670	41	15	edu	edu	PROPN
dad-10670	41	16	boundaries	boundary	NOUN
dad-10670	41	17	or	or	CCONJ
dad-10670	41	18	select	select	VERB
dad-10670	41	19	the	the	DET
dad-10670	41	20	discourse	discourse	NOUN
dad-10670	41	21	relations	relation	NOUN
dad-10670	41	22	holding	hold	VERB
dad-10670	41	23	between	between	ADP
dad-10670	41	24	edus	edus	NOUN
dad-10670	41	25	.	.	PUNCT
dad-10670	42	1	however	however	ADV
dad-10670	42	2	,	,	PUNCT
dad-10670	42	3	given	give	VERB
dad-10670	42	4	the	the	DET
dad-10670	42	5	heterogeneity	heterogeneity	NOUN
dad-10670	42	6	of	of	ADP
dad-10670	42	7	all	all	DET
dad-10670	42	8	possible	possible	ADJ
dad-10670	42	9	texts	text	NOUN
dad-10670	42	10	,	,	PUNCT
dad-10670	42	11	a	a	DET
dad-10670	42	12	rule	rule	NOUN
dad-10670	42	13	-	-	PUNCT
dad-10670	42	14	based	base	VERB
dad-10670	42	15	model	model	NOUN
dad-10670	42	16	requires	require	VERB
dad-10670	42	17	the	the	DET
dad-10670	42	18	creation	creation	NOUN
dad-10670	42	19	of	of	ADP
dad-10670	42	20	a	a	DET
dad-10670	42	21	large	large	ADJ
dad-10670	42	22	number	number	NOUN
dad-10670	42	23	of	of	ADP
dad-10670	42	24	rules	rule	NOUN
dad-10670	42	25	.	.	PUNCT
dad-10670	43	1	supervised	supervise	VERB
dad-10670	43	2	learning	learn	VERB
dad-10670	43	3	techniques	technique	NOUN
dad-10670	43	4	offer	offer	VERB
dad-10670	43	5	an	an	DET
dad-10670	43	6	interesting	interesting	ADJ
dad-10670	43	7	alternative	alternative	NOUN
dad-10670	43	8	.	.	PUNCT
dad-10670	44	1	with	with	ADP
dad-10670	44	2	those	those	DET
dad-10670	44	3	techniques	technique	NOUN
dad-10670	44	4	,	,	PUNCT
dad-10670	44	5	labeled	label	VERB
dad-10670	44	6	instances	instance	NOUN
dad-10670	44	7	are	be	AUX
dad-10670	44	8	extracted	extract	VERB
dad-10670	44	9	from	from	ADP
dad-10670	44	10	a	a	DET
dad-10670	44	11	large	large	ADJ
dad-10670	44	12	body	body	NOUN
dad-10670	44	13	of	of	ADP
dad-10670	44	14	text	text	NOUN
dad-10670	44	15	.	.	PUNCT
dad-10670	45	1	classifiers	classifier	NOUN
dad-10670	45	2	are	be	AUX
dad-10670	45	3	then	then	ADV
dad-10670	45	4	trained	train	VERB
dad-10670	45	5	to	to	PART
dad-10670	45	6	differentiate	differentiate	VERB
dad-10670	45	7	between	between	ADP
dad-10670	45	8	edu	edu	PROPN
dad-10670	45	9	boundary	boundary	PROPN
dad-10670	45	10	and	and	CCONJ
dad-10670	45	11	non	non	ADJ
dad-10670	45	12	-	-	ADJ
dad-10670	45	13	boundary	boundary	ADJ
dad-10670	45	14	words	word	NOUN
dad-10670	45	15	,	,	PUNCT
dad-10670	45	16	or	or	CCONJ
dad-10670	45	17	assign	assign	VERB
dad-10670	45	18	discourse	discourse	NOUN
dad-10670	45	19	relation	relation	NOUN
dad-10670	45	20	labels	label	NOUN
dad-10670	45	21	to	to	ADP
dad-10670	45	22	edu	edu	PROPN
dad-10670	45	23	pairs	pair	NOUN
dad-10670	45	24	.	.	PUNCT
dad-10670	46	1	the	the	DET
dad-10670	46	2	machine	machine	NOUN
dad-10670	46	3	learning	learning	NOUN
dad-10670	46	4	based	base	VERB
dad-10670	46	5	approach	approach	NOUN
dad-10670	46	6	for	for	ADP
dad-10670	46	7	discourse	discourse	NOUN
dad-10670	46	8	parsing	parsing	NOUN
dad-10670	46	9	was	be	AUX
dad-10670	46	10	greatly	greatly	ADV
dad-10670	46	11	facilitated	facilitate	VERB
dad-10670	46	12	by	by	ADP
dad-10670	46	13	the	the	DET
dad-10670	46	14	release	release	NOUN
dad-10670	46	15	of	of	ADP
dad-10670	46	16	the	the	DET
dad-10670	46	17	rst	rst	PROPN
dad-10670	46	18	discourse	discourse	NOUN
dad-10670	46	19	treebank	treebank	NOUN
dad-10670	46	20	(	(	PUNCT
dad-10670	46	21	rst	rst	NOUN
dad-10670	46	22	-	-	PUNCT
dad-10670	46	23	dt	dt	NOUN
dad-10670	46	24	)	)	PUNCT
dad-10670	46	25	(	(	PUNCT
dad-10670	46	26	carlson	carlson	PROPN
dad-10670	46	27	et	et	PROPN
dad-10670	46	28	al	al	PROPN
dad-10670	46	29	.	.	PROPN
dad-10670	46	30	,	,	PUNCT
dad-10670	46	31	2001	2001	NUM
dad-10670	46	32	)	)	PUNCT
dad-10670	46	33	,	,	PUNCT
dad-10670	46	34	a	a	DET
dad-10670	46	35	corpus	corpus	NOUN
dad-10670	46	36	of	of	ADP
dad-10670	46	37	rst	rst	ADJ
dad-10670	46	38	-	-	ADJ
dad-10670	46	39	annotated	annotate	VERB
dad-10670	46	40	snippets	snippet	NOUN
dad-10670	46	41	from	from	ADP
dad-10670	46	42	the	the	DET
dad-10670	46	43	wall	wall	PROPN
dad-10670	46	44	street	street	PROPN
dad-10670	46	45	journal	journal	PROPN
dad-10670	46	46	.	.	PUNCT
dad-10670	47	1	previous	previous	ADJ
dad-10670	47	2	supervised	supervised	ADJ
dad-10670	47	3	approaches	approach	NOUN
dad-10670	47	4	were	be	AUX
dad-10670	47	5	aimed	aim	VERB
dad-10670	47	6	at	at	ADP
dad-10670	47	7	producing	produce	VERB
dad-10670	47	8	sentence	sentence	NOUN
dad-10670	47	9	level	level	NOUN
dad-10670	47	10	analysis	analysis	NOUN
dad-10670	47	11	(	(	PUNCT
dad-10670	47	12	soricut	soricut	NOUN
dad-10670	47	13	and	and	CCONJ
dad-10670	47	14	marcu	marcu	NOUN
dad-10670	47	15	,	,	PUNCT
dad-10670	47	16	2003	2003	NUM
dad-10670	47	17	)	)	PUNCT
dad-10670	47	18	or	or	CCONJ
dad-10670	47	19	at	at	ADP
dad-10670	47	20	describing	describe	VERB
dad-10670	47	21	partially	partially	ADV
dad-10670	47	22	-	-	PUNCT
dad-10670	47	23	implemented	implement	VERB
dad-10670	47	24	systems	system	NOUN
dad-10670	47	25	(	(	PUNCT
dad-10670	47	26	reitter	reitter	NOUN
dad-10670	47	27	,	,	PUNCT
dad-10670	47	28	2003b	2003b	NUM
dad-10670	47	29	)	)	PUNCT
dad-10670	47	30	.	.	PUNCT
dad-10670	48	1	by	by	ADP
dad-10670	48	2	contrast	contrast	NOUN
dad-10670	48	3	,	,	PUNCT
dad-10670	48	4	our	our	PRON
dad-10670	48	5	work	work	NOUN
dad-10670	48	6	targets	target	NOUN
dad-10670	48	7	discourse	discourse	NOUN
dad-10670	48	8	structure	structure	NOUN
dad-10670	48	9	at	at	ADP
dad-10670	48	10	text	text	NOUN
dad-10670	48	11	level	level	NOUN
dad-10670	48	12	.	.	PUNCT
dad-10670	49	1	specifically	specifically	ADV
dad-10670	49	2	,	,	PUNCT
dad-10670	49	3	we	we	PRON
dad-10670	49	4	created	create	VERB
dad-10670	49	5	a	a	DET
dad-10670	49	6	fully	fully	ADV
dad-10670	49	7	-	-	PUNCT
dad-10670	49	8	implemented	implement	VERB
dad-10670	49	9	,	,	PUNCT
dad-10670	49	10	extensivelyevaluated	extensivelyevaluate	VERB
dad-10670	49	11	system	system	NOUN
dad-10670	49	12	.	.	PUNCT
dad-10670	50	1	in	in	ADP
dad-10670	50	2	this	this	DET
dad-10670	50	3	article	article	NOUN
dad-10670	50	4	,	,	PUNCT
dad-10670	50	5	we	we	PRON
dad-10670	50	6	will	will	AUX
dad-10670	50	7	present	present	VERB
dad-10670	50	8	hilda	hilda	PROPN
dad-10670	50	9	(	(	PUNCT
dad-10670	50	10	high	high	ADJ
dad-10670	50	11	-	-	PUNCT
dad-10670	50	12	level	level	NOUN
dad-10670	50	13	discourse	discourse	NOUN
dad-10670	50	14	analyzer	analyzer	NOUN
dad-10670	50	15	)	)	PUNCT
dad-10670	50	16	,	,	PUNCT
dad-10670	50	17	a	a	DET
dad-10670	50	18	text	text	NOUN
dad-10670	50	19	level	level	NOUN
dad-10670	50	20	discourse	discourse	NOUN
dad-10670	50	21	parser	parser	NOUN
dad-10670	50	22	based	base	VERB
dad-10670	50	23	on	on	ADP
dad-10670	50	24	support	support	NOUN
dad-10670	50	25	vector	vector	NOUN
dad-10670	50	26	machine	machine	NOUN
dad-10670	50	27	classification	classification	NOUN
dad-10670	50	28	.	.	PUNCT
dad-10670	51	1	in	in	ADP
dad-10670	51	2	section	section	NOUN
dad-10670	51	3	2	2	NUM
dad-10670	51	4	,	,	PUNCT
dad-10670	51	5	we	we	PRON
dad-10670	51	6	will	will	AUX
dad-10670	51	7	report	report	VERB
dad-10670	51	8	on	on	ADP
dad-10670	51	9	available	available	ADJ
dad-10670	51	10	approaches	approach	NOUN
dad-10670	51	11	using	use	VERB
dad-10670	51	12	rule	rule	NOUN
dad-10670	51	13	-	-	PUNCT
dad-10670	51	14	based	base	VERB
dad-10670	51	15	and	and	CCONJ
dad-10670	51	16	statistical	statistical	ADJ
dad-10670	51	17	discourse	discourse	NOUN
dad-10670	51	18	segmentation	segmentation	NOUN
dad-10670	51	19	and	and	CCONJ
dad-10670	51	20	parsing	parsing	NOUN
dad-10670	51	21	.	.	PUNCT
dad-10670	52	1	in	in	ADP
dad-10670	52	2	section	section	NOUN
dad-10670	52	3	3	3	NUM
dad-10670	52	4	,	,	PUNCT
dad-10670	52	5	we	we	PRON
dad-10670	52	6	will	will	AUX
dad-10670	52	7	explain	explain	VERB
dad-10670	52	8	the	the	DET
dad-10670	52	9	two	two	NUM
dad-10670	52	10	core	core	NOUN
dad-10670	52	11	tasks	task	NOUN
dad-10670	52	12	for	for	ADP
dad-10670	52	13	an	an	DET
dad-10670	52	14	automated	automate	VERB
dad-10670	52	15	discourse	discourse	NOUN
dad-10670	52	16	parser	parser	NOUN
dad-10670	52	17	(	(	PUNCT
dad-10670	52	18	segmentation	segmentation	NOUN
dad-10670	52	19	and	and	CCONJ
dad-10670	52	20	relation	relation	NOUN
dad-10670	52	21	labeling	labeling	NOUN
dad-10670	52	22	)	)	PUNCT
dad-10670	52	23	and	and	CCONJ
dad-10670	52	24	describe	describe	VERB
dad-10670	52	25	our	our	PRON
dad-10670	52	26	working	working	NOUN
dad-10670	52	27	assumptions	assumption	NOUN
dad-10670	52	28	.	.	PUNCT
dad-10670	53	1	section	section	NOUN
dad-10670	53	2	3.4	3.4	NUM
dad-10670	53	3	is	be	AUX
dad-10670	53	4	dedicated	dedicate	VERB
dad-10670	53	5	to	to	ADP
dad-10670	53	6	the	the	DET
dad-10670	53	7	set	set	NOUN
dad-10670	53	8	of	of	ADP
dad-10670	53	9	lexical	lexical	ADJ
dad-10670	53	10	,	,	PUNCT
dad-10670	53	11	syntactic	syntactic	ADJ
dad-10670	53	12	and	and	CCONJ
dad-10670	53	13	structural	structural	ADJ
dad-10670	53	14	features	feature	NOUN
dad-10670	53	15	used	use	VERB
dad-10670	53	16	in	in	ADP
dad-10670	53	17	our	our	PRON
dad-10670	53	18	model	model	NOUN
dad-10670	53	19	.	.	PUNCT
dad-10670	54	1	in	in	ADP
dad-10670	54	2	section	section	NOUN
dad-10670	54	3	4	4	NUM
dad-10670	54	4	,	,	PUNCT
dad-10670	54	5	we	we	PRON
dad-10670	54	6	will	will	AUX
dad-10670	54	7	provide	provide	VERB
dad-10670	54	8	(	(	PUNCT
dad-10670	54	9	1	1	NUM
dad-10670	54	10	)	)	PUNCT
dad-10670	54	11	a	a	DET
dad-10670	54	12	detailed	detailed	ADJ
dad-10670	54	13	evaluation	evaluation	NOUN
dad-10670	54	14	of	of	ADP
dad-10670	54	15	the	the	DET
dad-10670	54	16	discourse	discourse	NOUN
dad-10670	54	17	segmentation	segmentation	NOUN
dad-10670	54	18	and	and	CCONJ
dad-10670	54	19	relation	relation	NOUN
dad-10670	54	20	labeling	labeling	NOUN
dad-10670	54	21	modules	module	NOUN
dad-10670	54	22	,	,	PUNCT
dad-10670	54	23	(	(	PUNCT
dad-10670	54	24	2	2	X
dad-10670	54	25	)	)	PUNCT
dad-10670	54	26	a	a	DET
dad-10670	54	27	full	full	ADJ
dad-10670	54	28	system	system	NOUN
dad-10670	54	29	evaluation	evaluation	NOUN
dad-10670	54	30	,	,	PUNCT
dad-10670	54	31	and	and	CCONJ
dad-10670	54	32	(	(	PUNCT
dad-10670	54	33	3	3	X
dad-10670	54	34	)	)	PUNCT
dad-10670	54	35	a	a	DET
dad-10670	54	36	comparison	comparison	NOUN
dad-10670	54	37	to	to	ADP
dad-10670	54	38	other	other	ADJ
dad-10670	54	39	discourse	discourse	NOUN
dad-10670	54	40	parsers	parser	NOUN
dad-10670	54	41	.	.	PUNCT
dad-10670	55	1	finally	finally	ADV
dad-10670	55	2	,	,	PUNCT
dad-10670	55	3	in	in	ADP
dad-10670	55	4	section	section	NOUN
dad-10670	55	5	5	5	NUM
dad-10670	55	6	,	,	PUNCT
dad-10670	55	7	we	we	PRON
dad-10670	55	8	will	will	AUX
dad-10670	55	9	conclude	conclude	VERB
dad-10670	55	10	the	the	DET
dad-10670	55	11	article	article	NOUN
dad-10670	55	12	and	and	CCONJ
dad-10670	55	13	discuss	discuss	VERB
dad-10670	55	14	future	future	ADJ
dad-10670	55	15	work	work	NOUN
dad-10670	55	16	.	.	PUNCT
dad-10670	56	1	2	2	X
dad-10670	56	2	.	.	X
dad-10670	56	3	related	relate	VERB
dad-10670	56	4	work	work	NOUN
dad-10670	56	5	since	since	SCONJ
dad-10670	56	6	marcu	marcu	NOUN
dad-10670	56	7	’s	’s	PART
dad-10670	56	8	first	first	ADJ
dad-10670	56	9	attempt	attempt	NOUN
dad-10670	56	10	at	at	ADP
dad-10670	56	11	developing	develop	VERB
dad-10670	56	12	a	a	DET
dad-10670	56	13	rule	rule	NOUN
dad-10670	56	14	-	-	PUNCT
dad-10670	56	15	based	base	VERB
dad-10670	56	16	discourse	discourse	NOUN
dad-10670	56	17	parser	parser	NOUN
dad-10670	56	18	(	(	PUNCT
dad-10670	56	19	marcu	marcu	NOUN
dad-10670	56	20	,	,	PUNCT
dad-10670	56	21	2000	2000	NUM
dad-10670	56	22	)	)	PUNCT
dad-10670	56	23	,	,	PUNCT
dad-10670	56	24	several	several	ADJ
dad-10670	56	25	algorithms	algorithm	NOUN
dad-10670	56	26	for	for	ADP
dad-10670	56	27	discourse	discourse	NOUN
dad-10670	56	28	parsing	parsing	NOUN
dad-10670	56	29	have	have	AUX
dad-10670	56	30	been	be	AUX
dad-10670	56	31	proposed	propose	VERB
dad-10670	56	32	,	,	PUNCT
dad-10670	56	33	both	both	CCONJ
dad-10670	56	34	statistical	statistical	ADJ
dad-10670	56	35	and	and	CCONJ
dad-10670	56	36	rule	rule	NOUN
dad-10670	56	37	-	-	PUNCT
dad-10670	56	38	based	base	VERB
dad-10670	56	39	.	.	PUNCT
dad-10670	57	1	each	each	PRON
dad-10670	57	2	of	of	ADP
dad-10670	57	3	them	they	PRON
dad-10670	57	4	extracts	extract	VERB
dad-10670	57	5	a	a	DET
dad-10670	57	6	different	different	ADJ
dad-10670	57	7	set	set	NOUN
dad-10670	57	8	of	of	ADP
dad-10670	57	9	features	feature	NOUN
dad-10670	57	10	from	from	ADP
dad-10670	57	11	the	the	DET
dad-10670	57	12	input	input	NOUN
dad-10670	57	13	,	,	PUNCT
dad-10670	57	14	and	and	CCONJ
dad-10670	57	15	demonstrates	demonstrate	VERB
dad-10670	57	16	different	different	ADJ
dad-10670	57	17	run	run	NOUN
dad-10670	57	18	-	-	PUNCT
dad-10670	57	19	time	time	NOUN
dad-10670	57	20	complexity	complexity	NOUN
dad-10670	57	21	.	.	PUNCT
dad-10670	58	1	in	in	ADP
dad-10670	58	2	this	this	DET
dad-10670	58	3	section	section	NOUN
dad-10670	58	4	we	we	PRON
dad-10670	58	5	present	present	VERB
dad-10670	58	6	the	the	DET
dad-10670	58	7	most	most	ADV
dad-10670	58	8	notable	notable	ADJ
dad-10670	58	9	ones	one	NOUN
dad-10670	58	10	,	,	PUNCT
dad-10670	58	11	which	which	PRON
dad-10670	58	12	also	also	ADV
dad-10670	58	13	proved	prove	VERB
dad-10670	58	14	instrumental	instrumental	ADJ
dad-10670	58	15	in	in	ADP
dad-10670	58	16	identifying	identify	VERB
dad-10670	58	17	3	3	NUM
dad-10670	58	18	hernault	hernault	ADJ
dad-10670	58	19	,	,	PUNCT
dad-10670	58	20	prendinger	prendinger	NOUN
dad-10670	58	21	,	,	PUNCT
dad-10670	58	22	duverle	duverle	NOUN
dad-10670	58	23	,	,	PUNCT
dad-10670	58	24	and	and	CCONJ
dad-10670	58	25	ishizuka	ishizuka	VERB
dad-10670	58	26	useful	useful	ADJ
dad-10670	58	27	features	feature	NOUN
dad-10670	58	28	for	for	ADP
dad-10670	58	29	our	our	PRON
dad-10670	58	30	own	own	ADJ
dad-10670	58	31	algorithm	algorithm	NOUN
dad-10670	58	32	.	.	PUNCT
dad-10670	59	1	we	we	PRON
dad-10670	59	2	start	start	VERB
dad-10670	59	3	with	with	ADP
dad-10670	59	4	discussing	discuss	VERB
dad-10670	59	5	statistical	statistical	ADJ
dad-10670	59	6	approaches	approach	NOUN
dad-10670	59	7	to	to	PART
dad-10670	59	8	discourse	discourse	VERB
dad-10670	59	9	segmenting	segmenting	NOUN
dad-10670	59	10	and	and	CCONJ
dad-10670	59	11	parsing	parsing	NOUN
dad-10670	59	12	.	.	PUNCT
dad-10670	60	1	spade	spade	NOUN
dad-10670	60	2	(	(	PUNCT
dad-10670	60	3	soricut	soricut	VERB
dad-10670	60	4	and	and	CCONJ
dad-10670	60	5	marcu	marcu	NOUN
dad-10670	60	6	,	,	PUNCT
dad-10670	60	7	2003	2003	NUM
dad-10670	60	8	)	)	PUNCT
dad-10670	60	9	is	be	AUX
dad-10670	60	10	a	a	DET
dad-10670	60	11	sentence	sentence	NOUN
dad-10670	60	12	level	level	NOUN
dad-10670	60	13	discourse	discourse	NOUN
dad-10670	60	14	parser	parser	NOUN
dad-10670	60	15	.	.	PUNCT
dad-10670	61	1	two	two	NUM
dad-10670	61	2	probabilistic	probabilistic	ADJ
dad-10670	61	3	models	model	NOUN
dad-10670	61	4	are	be	AUX
dad-10670	61	5	employed	employ	VERB
dad-10670	61	6	that	that	PRON
dad-10670	61	7	use	use	VERB
dad-10670	61	8	syntactic	syntactic	ADJ
dad-10670	61	9	and	and	CCONJ
dad-10670	61	10	lexical	lexical	ADJ
dad-10670	61	11	information	information	NOUN
dad-10670	61	12	to	to	ADP
dad-10670	61	13	segment	segment	VERB
dad-10670	61	14	and	and	CCONJ
dad-10670	61	15	to	to	PART
dad-10670	61	16	parse	parse	VERB
dad-10670	61	17	text	text	NOUN
dad-10670	61	18	.	.	PUNCT
dad-10670	62	1	issues	issue	NOUN
dad-10670	62	2	of	of	ADP
dad-10670	62	3	algorithmic	algorithmic	ADJ
dad-10670	62	4	complexity	complexity	NOUN
dad-10670	62	5	that	that	PRON
dad-10670	62	6	made	make	VERB
dad-10670	62	7	their	their	PRON
dad-10670	62	8	original	original	ADJ
dad-10670	62	9	algorithm	algorithm	NOUN
dad-10670	62	10	impractical	impractical	ADJ
dad-10670	62	11	on	on	ADP
dad-10670	62	12	large	large	ADJ
dad-10670	62	13	instances	instance	NOUN
dad-10670	62	14	(	(	PUNCT
dad-10670	62	15	marcu	marcu	NOUN
dad-10670	62	16	,	,	PUNCT
dad-10670	62	17	2000	2000	NUM
dad-10670	62	18	)	)	PUNCT
dad-10670	62	19	are	be	AUX
dad-10670	62	20	overcome	overcome	VERB
dad-10670	62	21	by	by	ADP
dad-10670	62	22	a	a	DET
dad-10670	62	23	dynamic	dynamic	ADJ
dad-10670	62	24	programming	programming	NOUN
dad-10670	62	25	approach	approach	NOUN
dad-10670	62	26	.	.	PUNCT
dad-10670	63	1	although	although	SCONJ
dad-10670	63	2	spade	spade	NOUN
dad-10670	63	3	does	do	AUX
dad-10670	63	4	not	not	PART
dad-10670	63	5	support	support	VERB
dad-10670	63	6	full	full	ADJ
dad-10670	63	7	text	text	NOUN
dad-10670	63	8	discourse	discourse	NOUN
dad-10670	63	9	parsing	parsing	NOUN
dad-10670	63	10	,	,	PUNCT
dad-10670	63	11	it	it	PRON
dad-10670	63	12	demonstrates	demonstrate	VERB
dad-10670	63	13	the	the	DET
dad-10670	63	14	correlation	correlation	NOUN
dad-10670	63	15	between	between	ADP
dad-10670	63	16	syntactic	syntactic	ADJ
dad-10670	63	17	and	and	CCONJ
dad-10670	63	18	discourse	discourse	NOUN
dad-10670	63	19	information	information	NOUN
dad-10670	63	20	and	and	CCONJ
dad-10670	63	21	their	their	PRON
dad-10670	63	22	capability	capability	NOUN
dad-10670	63	23	to	to	ADP
dad-10670	63	24	identifying	identify	VERB
dad-10670	63	25	structure	structure	NOUN
dad-10670	63	26	and	and	CCONJ
dad-10670	63	27	relations	relation	NOUN
dad-10670	63	28	empirically	empirically	ADV
dad-10670	63	29	,	,	PUNCT
dad-10670	63	30	particularly	particularly	ADV
dad-10670	63	31	when	when	SCONJ
dad-10670	63	32	no	no	DET
dad-10670	63	33	signaling	signal	VERB
dad-10670	63	34	cue	cue	NOUN
dad-10670	63	35	-	-	PUNCT
dad-10670	63	36	word	word	NOUN
dad-10670	63	37	is	be	AUX
dad-10670	63	38	present	present	ADJ
dad-10670	63	39	(	(	PUNCT
dad-10670	63	40	between	between	ADP
dad-10670	63	41	60	60	NUM
dad-10670	63	42	%	%	NOUN
dad-10670	63	43	and	and	CCONJ
dad-10670	63	44	70	70	NUM
dad-10670	63	45	%	%	NOUN
dad-10670	63	46	of	of	ADP
dad-10670	63	47	the	the	DET
dad-10670	63	48	total	total	NOUN
dad-10670	63	49	(	(	PUNCT
dad-10670	63	50	taboada	taboada	NOUN
dad-10670	63	51	,	,	PUNCT
dad-10670	63	52	2006	2006	NUM
dad-10670	63	53	)	)	PUNCT
dad-10670	63	54	)	)	PUNCT
dad-10670	63	55	.	.	PUNCT
dad-10670	64	1	the	the	DET
dad-10670	64	2	authors	author	NOUN
dad-10670	64	3	’	’	PART
dad-10670	64	4	exploration	exploration	NOUN
dad-10670	64	5	of	of	ADP
dad-10670	64	6	‘	'	PUNCT
dad-10670	64	7	dominance	dominance	NOUN
dad-10670	64	8	sets	set	NOUN
dad-10670	64	9	’	'	PUNCT
dad-10670	64	10	in	in	ADP
dad-10670	64	11	lexicalized	lexicalized	ADJ
dad-10670	64	12	syntax	syntax	NOUN
dad-10670	64	13	trees	tree	NOUN
dad-10670	64	14	has	have	AUX
dad-10670	64	15	provided	provide	VERB
dad-10670	64	16	the	the	DET
dad-10670	64	17	basis	basis	NOUN
dad-10670	64	18	for	for	ADP
dad-10670	64	19	a	a	DET
dad-10670	64	20	set	set	NOUN
dad-10670	64	21	of	of	ADP
dad-10670	64	22	lexico	lexico	NOUN
dad-10670	64	23	-	-	PUNCT
dad-10670	64	24	syntactic	syntactic	ADJ
dad-10670	64	25	features	feature	NOUN
dad-10670	64	26	in	in	ADP
dad-10670	64	27	our	our	PRON
dad-10670	64	28	own	own	ADJ
dad-10670	64	29	algorithm	algorithm	NOUN
dad-10670	64	30	.	.	PUNCT
dad-10670	65	1	a	a	DET
dad-10670	65	2	key	key	ADJ
dad-10670	65	3	difference	difference	NOUN
dad-10670	65	4	to	to	ADP
dad-10670	65	5	our	our	PRON
dad-10670	65	6	approach	approach	NOUN
dad-10670	65	7	is	be	AUX
dad-10670	65	8	we	we	PRON
dad-10670	65	9	are	be	AUX
dad-10670	65	10	able	able	ADJ
dad-10670	65	11	to	to	PART
dad-10670	65	12	process	process	VERB
dad-10670	65	13	full	full	ADJ
dad-10670	65	14	text	text	NOUN
dad-10670	65	15	input	input	NOUN
dad-10670	65	16	,	,	PUNCT
dad-10670	65	17	rather	rather	ADV
dad-10670	65	18	than	than	ADP
dad-10670	65	19	individual	individual	ADJ
dad-10670	65	20	sentences	sentence	NOUN
dad-10670	65	21	.	.	PUNCT
dad-10670	66	1	to	to	ADP
dad-10670	66	2	the	the	DET
dad-10670	66	3	best	good	ADJ
dad-10670	66	4	of	of	ADP
dad-10670	66	5	our	our	PRON
dad-10670	66	6	knowledge	knowledge	NOUN
dad-10670	66	7	,	,	PUNCT
dad-10670	66	8	reitter	reitter	NOUN
dad-10670	66	9	(	(	PUNCT
dad-10670	66	10	2003b	2003b	NUM
dad-10670	66	11	)	)	PUNCT
dad-10670	66	12	presents	present	VERB
dad-10670	66	13	the	the	DET
dad-10670	66	14	only	only	ADJ
dad-10670	66	15	method	method	NOUN
dad-10670	66	16	based	base	VERB
dad-10670	66	17	exclusively	exclusively	ADV
dad-10670	66	18	on	on	ADP
dad-10670	66	19	feature	feature	NOUN
dad-10670	66	20	-	-	PUNCT
dad-10670	66	21	rich	rich	ADJ
dad-10670	66	22	supervised	supervised	ADJ
dad-10670	66	23	learning	learn	VERB
dad-10670	66	24	to	to	PART
dad-10670	66	25	produce	produce	VERB
dad-10670	66	26	text	text	NOUN
dad-10670	66	27	level	level	NOUN
dad-10670	66	28	discourse	discourse	NOUN
dad-10670	66	29	parse	parse	NOUN
dad-10670	66	30	trees	tree	NOUN
dad-10670	66	31	.	.	PUNCT
dad-10670	67	1	his	his	PRON
dad-10670	67	2	algorithm	algorithm	NOUN
dad-10670	67	3	relies	rely	VERB
dad-10670	67	4	on	on	ADP
dad-10670	67	5	training	train	VERB
dad-10670	67	6	a	a	DET
dad-10670	67	7	set	set	NOUN
dad-10670	67	8	of	of	ADP
dad-10670	67	9	support	support	NOUN
dad-10670	67	10	vector	vector	NOUN
dad-10670	67	11	machine	machine	NOUN
dad-10670	67	12	classifiers	classifier	NOUN
dad-10670	67	13	to	to	PART
dad-10670	67	14	score	score	VERB
dad-10670	67	15	and	and	CCONJ
dad-10670	67	16	to	to	PART
dad-10670	67	17	label	label	VERB
dad-10670	67	18	relations	relation	NOUN
dad-10670	67	19	between	between	ADP
dad-10670	67	20	spans	span	NOUN
dad-10670	67	21	.	.	PUNCT
dad-10670	68	1	although	although	SCONJ
dad-10670	68	2	the	the	DET
dad-10670	68	3	author	author	NOUN
dad-10670	68	4	’s	’s	PART
dad-10670	68	5	suggestion	suggestion	NOUN
dad-10670	68	6	of	of	ADP
dad-10670	68	7	a	a	DET
dad-10670	68	8	bottom	bottom	ADJ
dad-10670	68	9	-	-	PUNCT
dad-10670	68	10	up	up	ADP
dad-10670	68	11	tree	tree	NOUN
dad-10670	68	12	building	building	NOUN
dad-10670	68	13	algorithm	algorithm	NOUN
dad-10670	68	14	using	use	VERB
dad-10670	68	15	chart	chart	NOUN
dad-10670	68	16	parsing	parse	VERB
dad-10670	68	17	style	style	NOUN
dad-10670	68	18	techniques	technique	NOUN
dad-10670	68	19	has	have	AUX
dad-10670	68	20	not	not	PART
dad-10670	68	21	been	be	AUX
dad-10670	68	22	implemented	implement	VERB
dad-10670	68	23	yet	yet	ADV
dad-10670	68	24	,	,	PUNCT
dad-10670	68	25	the	the	DET
dad-10670	68	26	results	result	NOUN
dad-10670	68	27	for	for	ADP
dad-10670	68	28	raw	raw	ADJ
dad-10670	68	29	instance	instance	NOUN
dad-10670	68	30	classification	classification	NOUN
dad-10670	68	31	provides	provide	VERB
dad-10670	68	32	a	a	DET
dad-10670	68	33	good	good	ADJ
dad-10670	68	34	intermediate	intermediate	ADJ
dad-10670	68	35	benchmark	benchmark	NOUN
dad-10670	68	36	for	for	ADP
dad-10670	68	37	the	the	DET
dad-10670	68	38	evaluation	evaluation	NOUN
dad-10670	68	39	of	of	ADP
dad-10670	68	40	our	our	PRON
dad-10670	68	41	own	own	ADJ
dad-10670	68	42	instance	instance	NOUN
dad-10670	68	43	classifier	classifier	NOUN
dad-10670	68	44	.	.	PUNCT
dad-10670	69	1	more	more	ADV
dad-10670	69	2	recently	recently	ADV
dad-10670	69	3	,	,	PUNCT
dad-10670	69	4	baldridge	baldridge	NOUN
dad-10670	69	5	and	and	CCONJ
dad-10670	69	6	lascarides	lascaride	NOUN
dad-10670	69	7	(	(	PUNCT
dad-10670	69	8	2005	2005	NUM
dad-10670	69	9	)	)	PUNCT
dad-10670	69	10	successfully	successfully	ADV
dad-10670	69	11	implemented	implement	VERB
dad-10670	69	12	a	a	DET
dad-10670	69	13	probabilistic	probabilistic	ADJ
dad-10670	69	14	parser	parser	NOUN
dad-10670	69	15	that	that	PRON
dad-10670	69	16	uses	uses	AUX
dad-10670	69	17	headed	head	VERB
dad-10670	69	18	trees	tree	NOUN
dad-10670	69	19	to	to	PART
dad-10670	69	20	label	label	NOUN
dad-10670	69	21	discourse	discourse	NOUN
dad-10670	69	22	relations	relation	NOUN
dad-10670	69	23	.	.	PUNCT
dad-10670	70	1	the	the	DET
dad-10670	70	2	authors	author	NOUN
dad-10670	70	3	employed	employ	VERB
dad-10670	70	4	the	the	DET
dad-10670	70	5	more	more	ADV
dad-10670	70	6	specific	specific	ADJ
dad-10670	70	7	framework	framework	NOUN
dad-10670	70	8	of	of	ADP
dad-10670	70	9	segmented	segment	VERB
dad-10670	70	10	discourse	discourse	NOUN
dad-10670	70	11	representation	representation	NOUN
dad-10670	70	12	theory	theory	NOUN
dad-10670	70	13	(	(	PUNCT
dad-10670	70	14	asher	asher	NOUN
dad-10670	70	15	and	and	CCONJ
dad-10670	70	16	lascarides	lascaride	NOUN
dad-10670	70	17	,	,	PUNCT
dad-10670	70	18	2003	2003	NUM
dad-10670	70	19	)	)	PUNCT
dad-10670	70	20	rather	rather	ADV
dad-10670	70	21	than	than	ADP
dad-10670	70	22	rst	rst	PROPN
dad-10670	70	23	,	,	PUNCT
dad-10670	70	24	which	which	PRON
dad-10670	70	25	they	they	PRON
dad-10670	70	26	applied	apply	VERB
dad-10670	70	27	to	to	ADP
dad-10670	70	28	texts	text	NOUN
dad-10670	70	29	in	in	ADP
dad-10670	70	30	dialogue	dialogue	NOUN
dad-10670	70	31	form	form	NOUN
dad-10670	70	32	.	.	PUNCT
dad-10670	71	1	in	in	ADP
dad-10670	71	2	sagae	sagae	NOUN
dad-10670	71	3	(	(	PUNCT
dad-10670	71	4	2009	2009	NUM
dad-10670	71	5	)	)	PUNCT
dad-10670	71	6	,	,	PUNCT
dad-10670	71	7	a	a	DET
dad-10670	71	8	discourse	discourse	NOUN
dad-10670	71	9	parser	parser	NOUN
dad-10670	71	10	based	base	VERB
dad-10670	71	11	on	on	ADP
dad-10670	71	12	shift	shift	NOUN
dad-10670	71	13	-	-	PUNCT
dad-10670	71	14	reduced	reduce	VERB
dad-10670	71	15	parsing	parsing	NOUN
dad-10670	71	16	is	be	AUX
dad-10670	71	17	presented	present	VERB
dad-10670	71	18	.	.	PUNCT
dad-10670	72	1	the	the	DET
dad-10670	72	2	parser	parser	NOUN
dad-10670	72	3	is	be	AUX
dad-10670	72	4	trained	train	VERB
dad-10670	72	5	on	on	ADP
dad-10670	72	6	the	the	DET
dad-10670	72	7	rst	rst	PROPN
dad-10670	72	8	-	-	PUNCT
dad-10670	72	9	dt	dt	PROPN
dad-10670	72	10	,	,	PUNCT
dad-10670	72	11	employs	employ	VERB
dad-10670	72	12	transition	transition	NOUN
dad-10670	72	13	algorithms	algorithm	NOUN
dad-10670	72	14	for	for	ADP
dad-10670	72	15	dependency	dependency	NOUN
dad-10670	72	16	and	and	CCONJ
dad-10670	72	17	constituent	constituent	NOUN
dad-10670	72	18	trees	tree	NOUN
dad-10670	72	19	,	,	PUNCT
dad-10670	72	20	and	and	CCONJ
dad-10670	72	21	uses	use	VERB
dad-10670	72	22	its	its	PRON
dad-10670	72	23	own	own	ADJ
dad-10670	72	24	syntax	syntax	NOUN
dad-10670	72	25	and	and	CCONJ
dad-10670	72	26	dependency	dependency	NOUN
dad-10670	72	27	tagger	tagger	NOUN
dad-10670	72	28	.	.	PUNCT
dad-10670	73	1	it	it	PRON
dad-10670	73	2	brings	bring	VERB
dad-10670	73	3	noticeable	noticeable	ADJ
dad-10670	73	4	improvements	improvement	NOUN
dad-10670	73	5	in	in	ADP
dad-10670	73	6	accuracy	accuracy	NOUN
dad-10670	73	7	and	and	CCONJ
dad-10670	73	8	speed	speed	NOUN
dad-10670	73	9	against	against	ADP
dad-10670	73	10	marcu	marcu	NOUN
dad-10670	73	11	’s	’s	PART
dad-10670	73	12	original	original	ADJ
dad-10670	73	13	chart	chart	NOUN
dad-10670	73	14	parsing	parse	VERB
dad-10670	73	15	approach	approach	NOUN
dad-10670	73	16	(	(	PUNCT
dad-10670	73	17	marcu	marcu	NOUN
dad-10670	73	18	,	,	PUNCT
dad-10670	73	19	2000	2000	NUM
dad-10670	73	20	)	)	PUNCT
dad-10670	73	21	.	.	PUNCT
dad-10670	74	1	the	the	DET
dad-10670	74	2	parser	parser	NOUN
dad-10670	74	3	also	also	ADV
dad-10670	74	4	includes	include	VERB
dad-10670	74	5	a	a	DET
dad-10670	74	6	discourse	discourse	NOUN
dad-10670	74	7	segmenter	segmenter	NOUN
dad-10670	74	8	based	base	VERB
dad-10670	74	9	on	on	ADP
dad-10670	74	10	a	a	DET
dad-10670	74	11	binary	binary	ADJ
dad-10670	74	12	classifier	classifier	NOUN
dad-10670	74	13	trained	train	VERB
dad-10670	74	14	on	on	ADP
dad-10670	74	15	lexico	lexico	NOUN
dad-10670	74	16	-	-	PUNCT
dad-10670	74	17	syntactic	syntactic	ADJ
dad-10670	74	18	features	feature	NOUN
dad-10670	74	19	.	.	PUNCT
dad-10670	75	1	the	the	DET
dad-10670	75	2	discourse	discourse	NOUN
dad-10670	75	3	segmenter	segmenter	NOUN
dad-10670	75	4	performs	perform	VERB
dad-10670	75	5	better	well	ADJ
dad-10670	75	6	than	than	ADP
dad-10670	75	7	the	the	DET
dad-10670	75	8	one	one	NOUN
dad-10670	75	9	presented	present	VERB
dad-10670	75	10	in	in	ADP
dad-10670	75	11	reitter	reitter	PROPN
dad-10670	75	12	(	(	PUNCT
dad-10670	75	13	2003b	2003b	NUM
dad-10670	75	14	)	)	PUNCT
dad-10670	75	15	and	and	CCONJ
dad-10670	75	16	is	be	AUX
dad-10670	75	17	significantly	significantly	ADV
dad-10670	75	18	faster	fast	ADV
dad-10670	75	19	.	.	PUNCT
dad-10670	76	1	importantly	importantly	ADV
dad-10670	76	2	,	,	PUNCT
dad-10670	76	3	compared	compare	VERB
dad-10670	76	4	to	to	PART
dad-10670	76	5	soricut	soricut	VERB
dad-10670	76	6	and	and	CCONJ
dad-10670	76	7	marcu	marcu	NOUN
dad-10670	76	8	(	(	PUNCT
dad-10670	76	9	2003	2003	NUM
dad-10670	76	10	)	)	PUNCT
dad-10670	76	11	,	,	PUNCT
dad-10670	76	12	the	the	DET
dad-10670	76	13	proposed	propose	VERB
dad-10670	76	14	system	system	NOUN
dad-10670	76	15	is	be	AUX
dad-10670	76	16	able	able	ADJ
dad-10670	76	17	to	to	PART
dad-10670	76	18	create	create	VERB
dad-10670	76	19	discourse	discourse	NOUN
dad-10670	76	20	structures	structure	NOUN
dad-10670	76	21	for	for	ADP
dad-10670	76	22	full	full	ADJ
dad-10670	76	23	texts	text	NOUN
dad-10670	76	24	,	,	PUNCT
dad-10670	76	25	not	not	PART
dad-10670	76	26	only	only	ADV
dad-10670	76	27	sentences	sentence	NOUN
dad-10670	76	28	.	.	PUNCT
dad-10670	77	1	for	for	ADP
dad-10670	77	2	discourse	discourse	NOUN
dad-10670	77	3	segmentation	segmentation	NOUN
dad-10670	77	4	,	,	PUNCT
dad-10670	77	5	the	the	DET
dad-10670	77	6	author	author	NOUN
dad-10670	77	7	reports	report	VERB
dad-10670	77	8	an	an	DET
dad-10670	77	9	f	f	NOUN
dad-10670	77	10	-	-	PUNCT
dad-10670	77	11	score	score	NOUN
dad-10670	77	12	of	of	ADP
dad-10670	77	13	86.7	86.7	NUM
dad-10670	77	14	%	%	NOUN
dad-10670	77	15	and	and	CCONJ
dad-10670	77	16	an	an	DET
dad-10670	77	17	f	f	NOUN
dad-10670	77	18	-	-	PUNCT
dad-10670	77	19	score	score	NOUN
dad-10670	77	20	of	of	ADP
dad-10670	77	21	44.5	44.5	NUM
dad-10670	77	22	%	%	NOUN
dad-10670	77	23	for	for	ADP
dad-10670	77	24	creating	create	VERB
dad-10670	77	25	text	text	NOUN
dad-10670	77	26	level	level	NOUN
dad-10670	77	27	discourse	discourse	NOUN
dad-10670	77	28	trees	tree	NOUN
dad-10670	77	29	.	.	PUNCT
dad-10670	78	1	subba	subba	NOUN
dad-10670	78	2	and	and	CCONJ
dad-10670	78	3	di	di	PROPN
dad-10670	78	4	eugenio	eugenio	PROPN
dad-10670	78	5	(	(	PUNCT
dad-10670	78	6	2009	2009	NUM
dad-10670	78	7	)	)	PUNCT
dad-10670	78	8	propose	propose	VERB
dad-10670	78	9	a	a	DET
dad-10670	78	10	discourse	discourse	NOUN
dad-10670	78	11	parser	parser	NOUN
dad-10670	78	12	based	base	VERB
dad-10670	78	13	on	on	ADP
dad-10670	78	14	inductive	inductive	ADJ
dad-10670	78	15	logic	logic	NOUN
dad-10670	78	16	programming	programming	NOUN
dad-10670	78	17	,	,	PUNCT
dad-10670	78	18	a	a	DET
dad-10670	78	19	first	first	ADJ
dad-10670	78	20	-	-	PUNCT
dad-10670	78	21	order	order	NOUN
dad-10670	78	22	logic	logic	NOUN
dad-10670	78	23	approach	approach	NOUN
dad-10670	78	24	for	for	ADP
dad-10670	78	25	learning	learn	VERB
dad-10670	78	26	discourse	discourse	NOUN
dad-10670	78	27	relations	relation	NOUN
dad-10670	78	28	.	.	PUNCT
dad-10670	79	1	in	in	ADP
dad-10670	79	2	their	their	PRON
dad-10670	79	3	system	system	NOUN
dad-10670	79	4	,	,	PUNCT
dad-10670	79	5	besides	besides	SCONJ
dad-10670	79	6	traditional	traditional	ADJ
dad-10670	79	7	syntactic	syntactic	ADJ
dad-10670	79	8	and	and	CCONJ
dad-10670	79	9	discourse	discourse	VERB
dad-10670	79	10	cue	cue	NOUN
dad-10670	79	11	information	information	NOUN
dad-10670	79	12	,	,	PUNCT
dad-10670	79	13	rules	rule	NOUN
dad-10670	79	14	are	be	AUX
dad-10670	79	15	learnt	learn	VERB
dad-10670	79	16	based	base	VERB
dad-10670	79	17	on	on	ADP
dad-10670	79	18	semantic	semantic	ADJ
dad-10670	79	19	information	information	NOUN
dad-10670	79	20	from	from	ADP
dad-10670	79	21	wordnet	wordnet	NOUN
dad-10670	79	22	(	(	PUNCT
dad-10670	79	23	fellbaum	fellbaum	NOUN
dad-10670	79	24	,	,	PUNCT
dad-10670	79	25	1998	1998	NUM
dad-10670	79	26	)	)	PUNCT
dad-10670	79	27	,	,	PUNCT
dad-10670	79	28	similarity	similarity	NOUN
dad-10670	79	29	measures	measure	NOUN
dad-10670	79	30	,	,	PUNCT
dad-10670	79	31	structural	structural	ADJ
dad-10670	79	32	properties	property	NOUN
dad-10670	79	33	between	between	ADP
dad-10670	79	34	text	text	NOUN
dad-10670	79	35	segments	segment	NOUN
dad-10670	79	36	,	,	PUNCT
dad-10670	79	37	as	as	ADV
dad-10670	79	38	well	well	ADV
dad-10670	79	39	as	as	ADP
dad-10670	79	40	compositional	compositional	ADJ
dad-10670	79	41	semantics	semantic	NOUN
dad-10670	79	42	.	.	PUNCT
dad-10670	80	1	then	then	ADV
dad-10670	80	2	,	,	PUNCT
dad-10670	80	3	a	a	DET
dad-10670	80	4	shift	shift	NOUN
dad-10670	80	5	-	-	PUNCT
dad-10670	80	6	reduce	reduce	NOUN
dad-10670	80	7	parsing	parsing	NOUN
dad-10670	80	8	model	model	NOUN
dad-10670	80	9	is	be	AUX
dad-10670	80	10	employed	employ	VERB
dad-10670	80	11	to	to	PART
dad-10670	80	12	produce	produce	VERB
dad-10670	80	13	discourse	discourse	NOUN
dad-10670	80	14	structures	structure	NOUN
dad-10670	80	15	.	.	PUNCT
dad-10670	81	1	the	the	DET
dad-10670	81	2	authors	author	NOUN
dad-10670	81	3	employ	employ	VERB
dad-10670	81	4	a	a	DET
dad-10670	81	5	custom	custom	NOUN
dad-10670	81	6	corpus	corpus	NOUN
dad-10670	81	7	of	of	ADP
dad-10670	81	8	instructional	instructional	ADJ
dad-10670	81	9	texts	text	NOUN
dad-10670	81	10	,	,	PUNCT
dad-10670	81	11	manually	manually	ADV
dad-10670	81	12	segmented	segment	VERB
dad-10670	81	13	into	into	ADP
dad-10670	81	14	discourse	discourse	NOUN
dad-10670	81	15	units	unit	NOUN
dad-10670	81	16	,	,	PUNCT
dad-10670	81	17	and	and	CCONJ
dad-10670	81	18	annotated	annotate	VERB
dad-10670	81	19	with	with	ADP
dad-10670	81	20	a	a	DET
dad-10670	81	21	custom	custom	NOUN
dad-10670	81	22	set	set	NOUN
dad-10670	81	23	of	of	ADP
dad-10670	81	24	26	26	NUM
dad-10670	81	25	discourse	discourse	NOUN
dad-10670	81	26	relations	relation	NOUN
dad-10670	81	27	.	.	PUNCT
dad-10670	82	1	for	for	ADP
dad-10670	82	2	sentence	sentence	NOUN
dad-10670	82	3	-	-	PUNCT
dad-10670	82	4	level	level	NOUN
dad-10670	82	5	relation	relation	NOUN
dad-10670	82	6	labeling	labeling	NOUN
dad-10670	82	7	,	,	PUNCT
dad-10670	82	8	their	their	PRON
dad-10670	82	9	system	system	NOUN
dad-10670	82	10	reaches	reach	VERB
dad-10670	82	11	an	an	DET
dad-10670	82	12	f	f	NOUN
dad-10670	82	13	-	-	PUNCT
dad-10670	82	14	score	score	NOUN
dad-10670	82	15	of	of	ADP
dad-10670	82	16	63	63	NUM
dad-10670	82	17	%	%	NOUN
dad-10670	82	18	,	,	PUNCT
dad-10670	82	19	while	while	SCONJ
dad-10670	82	20	the	the	DET
dad-10670	82	21	f	f	NOUN
dad-10670	82	22	-	-	PUNCT
dad-10670	82	23	score	score	NOUN
dad-10670	82	24	for	for	ADP
dad-10670	82	25	text	text	NOUN
dad-10670	82	26	level	level	NOUN
dad-10670	82	27	discourse	discourse	NOUN
dad-10670	82	28	trees	tree	NOUN
dad-10670	82	29	is	be	AUX
dad-10670	82	30	35.44	35.44	NUM
dad-10670	82	31	%	%	NOUN
dad-10670	82	32	.	.	PUNCT
dad-10670	83	1	a	a	DET
dad-10670	83	2	statistical	statistical	ADJ
dad-10670	83	3	discourse	discourse	NOUN
dad-10670	83	4	segmenter	segmenter	NOUN
dad-10670	83	5	based	base	VERB
dad-10670	83	6	on	on	ADP
dad-10670	83	7	artificial	artificial	ADJ
dad-10670	83	8	neural	neural	ADJ
dad-10670	83	9	networks	network	NOUN
dad-10670	83	10	is	be	AUX
dad-10670	83	11	presented	present	VERB
dad-10670	83	12	in	in	ADP
dad-10670	83	13	subba	subba	NOUN
dad-10670	83	14	and	and	CCONJ
dad-10670	83	15	di	di	NOUN
dad-10670	83	16	eugenio	eugenio	PROPN
dad-10670	83	17	(	(	PUNCT
dad-10670	83	18	2007	2007	NUM
dad-10670	83	19	)	)	PUNCT
dad-10670	83	20	.	.	PUNCT
dad-10670	84	1	the	the	DET
dad-10670	84	2	system	system	NOUN
dad-10670	84	3	employs	employ	VERB
dad-10670	84	4	a	a	DET
dad-10670	84	5	multi	multi	ADJ
dad-10670	84	6	-	-	ADJ
dad-10670	84	7	layer	layer	ADJ
dad-10670	84	8	perceptron	perceptron	PROPN
dad-10670	84	9	model	model	NOUN
dad-10670	84	10	and	and	CCONJ
dad-10670	84	11	is	be	AUX
dad-10670	84	12	trained	train	VERB
dad-10670	84	13	on	on	ADP
dad-10670	84	14	the	the	DET
dad-10670	84	15	rst	rst	NOUN
dad-10670	84	16	-	-	PUNCT
dad-10670	84	17	dt	dt	NOUN
dad-10670	84	18	using	use	VERB
dad-10670	84	19	syntactic	syntactic	ADJ
dad-10670	84	20	and	and	CCONJ
dad-10670	84	21	lexical	lexical	ADJ
dad-10670	84	22	information	information	NOUN
dad-10670	84	23	,	,	PUNCT
dad-10670	84	24	in	in	ADP
dad-10670	84	25	particular	particular	ADJ
dad-10670	84	26	discourse	discourse	NOUN
dad-10670	84	27	cues	cue	NOUN
dad-10670	84	28	.	.	PUNCT
dad-10670	85	1	bagging	bagging	NOUN
dad-10670	85	2	(	(	PUNCT
dad-10670	85	3	breiman	breiman	NOUN
dad-10670	85	4	,	,	PUNCT
dad-10670	85	5	1996	1996	NUM
dad-10670	85	6	)	)	PUNCT
dad-10670	85	7	is	be	AUX
dad-10670	85	8	applied	apply	VERB
dad-10670	85	9	to	to	PART
dad-10670	85	10	reduce	reduce	VERB
dad-10670	85	11	over	over	ADV
dad-10670	85	12	-	-	PUNCT
dad-10670	85	13	fitting	fitting	ADJ
dad-10670	85	14	.	.	PUNCT
dad-10670	86	1	the	the	DET
dad-10670	86	2	performance	performance	NOUN
dad-10670	86	3	of	of	ADP
dad-10670	86	4	this	this	DET
dad-10670	86	5	segmenter	segmenter	NOUN
dad-10670	86	6	is	be	AUX
dad-10670	86	7	comparable	comparable	ADJ
dad-10670	86	8	to	to	ADP
dad-10670	86	9	the	the	DET
dad-10670	86	10	one	one	NUM
dad-10670	86	11	4	4	NUM
dad-10670	86	12	hilda	hilda	VERB
dad-10670	86	13	:	:	PUNCT
dad-10670	86	14	a	a	DET
dad-10670	86	15	discourse	discourse	NOUN
dad-10670	86	16	parser	parser	NOUN
dad-10670	86	17	using	use	VERB
dad-10670	86	18	support	support	NOUN
dad-10670	86	19	vector	vector	NOUN
dad-10670	86	20	machine	machine	NOUN
dad-10670	86	21	classification	classification	NOUN
dad-10670	86	22	employed	employ	VERB
dad-10670	86	23	in	in	ADP
dad-10670	86	24	soricut	soricut	NOUN
dad-10670	86	25	and	and	CCONJ
dad-10670	86	26	marcu	marcu	NOUN
dad-10670	86	27	(	(	PUNCT
dad-10670	86	28	2003	2003	NUM
dad-10670	86	29	)	)	PUNCT
dad-10670	86	30	,	,	PUNCT
dad-10670	86	31	with	with	ADP
dad-10670	86	32	an	an	DET
dad-10670	86	33	f	f	NOUN
dad-10670	86	34	-	-	PUNCT
dad-10670	86	35	score	score	NOUN
dad-10670	86	36	of	of	ADP
dad-10670	86	37	84.4	84.4	NUM
dad-10670	86	38	%	%	NOUN
dad-10670	86	39	(	(	PUNCT
dad-10670	86	40	86	86	NUM
dad-10670	86	41	%	%	NOUN
dad-10670	86	42	when	when	SCONJ
dad-10670	86	43	using	use	VERB
dad-10670	86	44	perfect	perfect	ADJ
dad-10670	86	45	parse	parse	NOUN
dad-10670	86	46	trees	tree	NOUN
dad-10670	86	47	)	)	PUNCT
dad-10670	86	48	.	.	PUNCT
dad-10670	87	1	next	next	ADV
dad-10670	87	2	we	we	PRON
dad-10670	87	3	turn	turn	VERB
dad-10670	87	4	to	to	ADP
dad-10670	87	5	rule	rule	NOUN
dad-10670	87	6	-	-	PUNCT
dad-10670	87	7	based	base	VERB
dad-10670	87	8	approaches	approach	NOUN
dad-10670	87	9	to	to	ADP
dad-10670	87	10	segmentation	segmentation	NOUN
dad-10670	87	11	and	and	CCONJ
dad-10670	87	12	discourse	discourse	NOUN
dad-10670	87	13	parsing	parsing	NOUN
dad-10670	87	14	.	.	PUNCT
dad-10670	88	1	le	le	PROPN
dad-10670	88	2	thanh	thanh	VERB
dad-10670	88	3	et	et	PROPN
dad-10670	88	4	al	al	PROPN
dad-10670	88	5	.	.	PROPN
dad-10670	89	1	(	(	PUNCT
dad-10670	89	2	2004b	2004b	NUM
dad-10670	89	3	)	)	PUNCT
dad-10670	89	4	propose	propose	VERB
dad-10670	89	5	a	a	DET
dad-10670	89	6	multi	multi	ADJ
dad-10670	89	7	-	-	ADJ
dad-10670	89	8	step	step	ADJ
dad-10670	89	9	algorithm	algorithm	NOUN
dad-10670	89	10	to	to	ADP
dad-10670	89	11	segment	segment	NOUN
dad-10670	89	12	text	text	NOUN
dad-10670	89	13	and	and	CCONJ
dad-10670	89	14	organize	organize	VERB
dad-10670	89	15	its	its	PRON
dad-10670	89	16	spans	span	NOUN
dad-10670	89	17	into	into	ADP
dad-10670	89	18	trees	tree	NOUN
dad-10670	89	19	for	for	ADP
dad-10670	89	20	each	each	DET
dad-10670	89	21	successive	successive	ADJ
dad-10670	89	22	level	level	NOUN
dad-10670	89	23	of	of	ADP
dad-10670	89	24	text	text	NOUN
dad-10670	89	25	organization	organization	NOUN
dad-10670	89	26	:	:	PUNCT
dad-10670	89	27	sentence	sentence	NOUN
dad-10670	89	28	level	level	NOUN
dad-10670	89	29	,	,	PUNCT
dad-10670	89	30	paragraph	paragraph	NOUN
dad-10670	89	31	level	level	NOUN
dad-10670	89	32	,	,	PUNCT
dad-10670	89	33	and	and	CCONJ
dad-10670	89	34	text	text	NOUN
dad-10670	89	35	level	level	NOUN
dad-10670	89	36	.	.	PUNCT
dad-10670	90	1	within	within	ADP
dad-10670	90	2	each	each	DET
dad-10670	90	3	level	level	NOUN
dad-10670	90	4	,	,	PUNCT
dad-10670	90	5	the	the	DET
dad-10670	90	6	authors	author	NOUN
dad-10670	90	7	explore	explore	VERB
dad-10670	90	8	a	a	DET
dad-10670	90	9	search	search	NOUN
dad-10670	90	10	space	space	NOUN
dad-10670	90	11	consisting	consist	VERB
dad-10670	90	12	of	of	ADP
dad-10670	90	13	all	all	DET
dad-10670	90	14	valid	valid	ADJ
dad-10670	90	15	trees	tree	NOUN
dad-10670	90	16	fitting	fit	VERB
dad-10670	90	17	the	the	DET
dad-10670	90	18	‘	'	PUNCT
dad-10670	90	19	sequentiality	sequentiality	NOUN
dad-10670	90	20	principle	principle	NOUN
dad-10670	90	21	’	'	PUNCT
dad-10670	90	22	(	(	PUNCT
dad-10670	90	23	i.e.	i.e.	X
dad-10670	90	24	,	,	PUNCT
dad-10670	90	25	two	two	NUM
dad-10670	90	26	spans	span	NOUN
dad-10670	90	27	connected	connect	VERB
dad-10670	90	28	in	in	ADP
dad-10670	90	29	the	the	DET
dad-10670	90	30	tree	tree	NOUN
dad-10670	90	31	are	be	AUX
dad-10670	90	32	adjacent	adjacent	ADJ
dad-10670	90	33	in	in	ADP
dad-10670	90	34	the	the	DET
dad-10670	90	35	original	original	ADJ
dad-10670	90	36	text	text	NOUN
dad-10670	90	37	,	,	PUNCT
dad-10670	90	38	see	see	VERB
dad-10670	90	39	section	section	NOUN
dad-10670	90	40	3.2	3.2	NUM
dad-10670	90	41	)	)	PUNCT
dad-10670	90	42	.	.	PUNCT
dad-10670	91	1	the	the	DET
dad-10670	91	2	multi	multi	ADJ
dad-10670	91	3	-	-	ADJ
dad-10670	91	4	level	level	ADJ
dad-10670	91	5	nature	nature	NOUN
dad-10670	91	6	of	of	ADP
dad-10670	91	7	their	their	PRON
dad-10670	91	8	algorithm	algorithm	NOUN
dad-10670	91	9	mitigates	mitigate	VERB
dad-10670	91	10	the	the	DET
dad-10670	91	11	combinatorial	combinatorial	ADJ
dad-10670	91	12	explosion	explosion	NOUN
dad-10670	91	13	effect	effect	NOUN
dad-10670	91	14	.	.	PUNCT
dad-10670	92	1	however	however	ADV
dad-10670	92	2	,	,	PUNCT
dad-10670	92	3	at	at	ADP
dad-10670	92	4	the	the	DET
dad-10670	92	5	text	text	NOUN
dad-10670	92	6	level	level	NOUN
dad-10670	92	7	,	,	PUNCT
dad-10670	92	8	the	the	DET
dad-10670	92	9	algorithm	algorithm	NOUN
dad-10670	92	10	has	have	VERB
dad-10670	92	11	to	to	PART
dad-10670	92	12	score	score	VERB
dad-10670	92	13	a	a	DET
dad-10670	92	14	large	large	ADJ
dad-10670	92	15	number	number	NOUN
dad-10670	92	16	of	of	ADP
dad-10670	92	17	trees	tree	NOUN
dad-10670	92	18	to	to	PART
dad-10670	92	19	extract	extract	VERB
dad-10670	92	20	the	the	DET
dad-10670	92	21	best	good	ADJ
dad-10670	92	22	candidate	candidate	NOUN
dad-10670	92	23	—	—	PUNCT
dad-10670	92	24	despite	despite	SCONJ
dad-10670	92	25	the	the	DET
dad-10670	92	26	use	use	NOUN
dad-10670	92	27	of	of	ADP
dad-10670	92	28	beam	beam	NOUN
dad-10670	92	29	search	search	NOUN
dad-10670	92	30	as	as	ADP
dad-10670	92	31	a	a	DET
dad-10670	92	32	method	method	NOUN
dad-10670	92	33	to	to	PART
dad-10670	92	34	explore	explore	VERB
dad-10670	92	35	the	the	DET
dad-10670	92	36	solution	solution	NOUN
dad-10670	92	37	space	space	NOUN
dad-10670	92	38	.	.	PUNCT
dad-10670	93	1	hence	hence	ADV
dad-10670	93	2	,	,	PUNCT
dad-10670	93	3	this	this	DET
dad-10670	93	4	approach	approach	NOUN
dad-10670	93	5	is	be	AUX
dad-10670	93	6	impractical	impractical	ADJ
dad-10670	93	7	for	for	ADP
dad-10670	93	8	large	large	ADJ
dad-10670	93	9	input	input	NOUN
dad-10670	93	10	text	text	NOUN
dad-10670	93	11	.	.	PUNCT
dad-10670	94	1	furthermore	furthermore	ADV
dad-10670	94	2	,	,	PUNCT
dad-10670	94	3	the	the	DET
dad-10670	94	4	assumption	assumption	NOUN
dad-10670	94	5	of	of	ADP
dad-10670	94	6	strong	strong	ADJ
dad-10670	94	7	tree	tree	NOUN
dad-10670	94	8	-	-	PUNCT
dad-10670	94	9	consistency	consistency	NOUN
dad-10670	94	10	(	(	PUNCT
dad-10670	94	11	the	the	DET
dad-10670	94	12	adjacency	adjacency	PROPN
dad-10670	94	13	hypothesis	hypothesis	NOUN
dad-10670	94	14	)	)	PUNCT
dad-10670	94	15	at	at	ADP
dad-10670	94	16	each	each	DET
dad-10670	94	17	organizational	organizational	ADJ
dad-10670	94	18	level	level	NOUN
dad-10670	94	19	is	be	AUX
dad-10670	94	20	not	not	PART
dad-10670	94	21	supported	support	VERB
dad-10670	94	22	in	in	ADP
dad-10670	94	23	the	the	DET
dad-10670	94	24	rst	rst	PROPN
dad-10670	94	25	corpus	corpus	NOUN
dad-10670	94	26	,	,	PUNCT
dad-10670	94	27	where	where	SCONJ
dad-10670	94	28	a	a	DET
dad-10670	94	29	majority	majority	NOUN
dad-10670	94	30	of	of	ADP
dad-10670	94	31	documents	document	NOUN
dad-10670	94	32	contain	contain	VERB
dad-10670	94	33	organizational	organizational	ADJ
dad-10670	94	34	blocks	block	NOUN
dad-10670	94	35	that	that	PRON
dad-10670	94	36	do	do	AUX
dad-10670	94	37	not	not	PART
dad-10670	94	38	map	map	VERB
dad-10670	94	39	to	to	ADP
dad-10670	94	40	valid	valid	ADJ
dad-10670	94	41	subtrees	subtree	NOUN
dad-10670	94	42	.	.	PUNCT
dad-10670	95	1	tofiloski	tofiloski	VERB
dad-10670	95	2	et	et	PROPN
dad-10670	95	3	al	al	PROPN
dad-10670	95	4	.	.	PROPN
dad-10670	96	1	(	(	PUNCT
dad-10670	96	2	2009	2009	NUM
dad-10670	96	3	)	)	PUNCT
dad-10670	96	4	argue	argue	VERB
dad-10670	96	5	that	that	SCONJ
dad-10670	96	6	using	use	VERB
dad-10670	96	7	rules	rule	NOUN
dad-10670	96	8	has	have	VERB
dad-10670	96	9	certain	certain	ADJ
dad-10670	96	10	advantages	advantage	NOUN
dad-10670	96	11	over	over	ADP
dad-10670	96	12	automatic	automatic	ADJ
dad-10670	96	13	learning	learning	NOUN
dad-10670	96	14	methods	method	NOUN
dad-10670	96	15	and	and	CCONJ
dad-10670	96	16	present	present	VERB
dad-10670	96	17	a	a	DET
dad-10670	96	18	rule	rule	NOUN
dad-10670	96	19	-	-	PUNCT
dad-10670	96	20	based	base	VERB
dad-10670	96	21	discourse	discourse	NOUN
dad-10670	96	22	segmenter	segmenter	NOUN
dad-10670	96	23	.	.	PUNCT
dad-10670	97	1	their	their	PRON
dad-10670	97	2	proposed	propose	VERB
dad-10670	97	3	model	model	NOUN
dad-10670	97	4	does	do	AUX
dad-10670	97	5	not	not	PART
dad-10670	97	6	depend	depend	VERB
dad-10670	97	7	on	on	ADP
dad-10670	97	8	a	a	DET
dad-10670	97	9	specific	specific	ADJ
dad-10670	97	10	training	training	NOUN
dad-10670	97	11	corpus	corpus	NOUN
dad-10670	97	12	and	and	CCONJ
dad-10670	97	13	supports	support	VERB
dad-10670	97	14	high	high	ADJ
dad-10670	97	15	-	-	PUNCT
dad-10670	97	16	precision	precision	NOUN
dad-10670	97	17	segmentation	segmentation	NOUN
dad-10670	97	18	by	by	ADP
dad-10670	97	19	inserting	insert	VERB
dad-10670	97	20	fewer	few	ADJ
dad-10670	97	21	but	but	CCONJ
dad-10670	97	22	‘	'	PUNCT
dad-10670	97	23	quality	quality	NOUN
dad-10670	97	24	’	'	PUNCT
dad-10670	97	25	boundaries	boundary	NOUN
dad-10670	97	26	.	.	PUNCT
dad-10670	98	1	however	however	ADV
dad-10670	98	2	,	,	PUNCT
dad-10670	98	3	segmentation	segmentation	NOUN
dad-10670	98	4	is	be	AUX
dad-10670	98	5	done	do	VERB
dad-10670	98	6	in	in	ADP
dad-10670	98	7	a	a	DET
dad-10670	98	8	manner	manner	NOUN
dad-10670	98	9	different	different	ADJ
dad-10670	98	10	from	from	ADP
dad-10670	98	11	the	the	DET
dad-10670	98	12	segmentation	segmentation	NOUN
dad-10670	98	13	guidelines	guideline	NOUN
dad-10670	98	14	used	use	VERB
dad-10670	98	15	in	in	ADP
dad-10670	98	16	the	the	DET
dad-10670	98	17	rst	rst	PROPN
dad-10670	98	18	-	-	PUNCT
dad-10670	98	19	dt	dt	NOUN
dad-10670	98	20	corpus	corpus	PROPN
dad-10670	98	21	(	(	PUNCT
dad-10670	98	22	carlson	carlson	PROPN
dad-10670	98	23	et	et	PROPN
dad-10670	98	24	al	al	PROPN
dad-10670	98	25	.	.	PROPN
dad-10670	98	26	,	,	PUNCT
dad-10670	98	27	2001	2001	NUM
dad-10670	98	28	)	)	PUNCT
dad-10670	98	29	.	.	PUNCT
dad-10670	99	1	first	first	ADV
dad-10670	99	2	,	,	PUNCT
dad-10670	99	3	the	the	DET
dad-10670	99	4	authors	author	NOUN
dad-10670	99	5	only	only	ADV
dad-10670	99	6	create	create	VERB
dad-10670	99	7	edus	edus	NOUN
dad-10670	99	8	that	that	PRON
dad-10670	99	9	contain	contain	VERB
dad-10670	99	10	verbs	verb	NOUN
dad-10670	99	11	.	.	PUNCT
dad-10670	100	1	they	they	PRON
dad-10670	100	2	try	try	VERB
dad-10670	100	3	to	to	PART
dad-10670	100	4	capture	capture	VERB
dad-10670	100	5	specific	specific	ADJ
dad-10670	100	6	relations	relation	NOUN
dad-10670	100	7	(	(	PUNCT
dad-10670	100	8	e.g.	e.g.	ADV
dad-10670	100	9	,	,	PUNCT
dad-10670	100	10	condition	condition	NOUN
dad-10670	100	11	,	,	PUNCT
dad-10670	100	12	purpose	purpose	NOUN
dad-10670	100	13	)	)	PUNCT
dad-10670	100	14	and	and	CCONJ
dad-10670	100	15	avoid	avoid	VERB
dad-10670	100	16	less	less	ADV
dad-10670	100	17	informative	informative	ADJ
dad-10670	100	18	relations	relation	NOUN
dad-10670	100	19	,	,	PUNCT
dad-10670	100	20	such	such	ADJ
dad-10670	100	21	as	as	ADP
dad-10670	100	22	elaboration	elaboration	NOUN
dad-10670	100	23	.	.	PUNCT
dad-10670	101	1	further	far	ADV
dad-10670	101	2	,	,	PUNCT
dad-10670	101	3	complement	complement	NOUN
dad-10670	101	4	clauses	clause	NOUN
dad-10670	101	5	are	be	AUX
dad-10670	101	6	not	not	PART
dad-10670	101	7	placed	place	VERB
dad-10670	101	8	in	in	ADP
dad-10670	101	9	independent	independent	ADJ
dad-10670	101	10	units	unit	NOUN
dad-10670	101	11	in	in	ADP
dad-10670	101	12	order	order	NOUN
dad-10670	101	13	not	not	PART
dad-10670	101	14	to	to	PART
dad-10670	101	15	break	break	VERB
dad-10670	101	16	np	np	DET
dad-10670	101	17	constituency	constituency	NOUN
dad-10670	101	18	,	,	PUNCT
dad-10670	101	19	and	and	CCONJ
dad-10670	101	20	avoid	avoid	VERB
dad-10670	101	21	same	same	ADJ
dad-10670	101	22	-	-	PUNCT
dad-10670	101	23	unit	unit	NOUN
dad-10670	101	24	relations	relation	NOUN
dad-10670	101	25	.	.	PUNCT
dad-10670	102	1	for	for	ADP
dad-10670	102	2	instance	instance	NOUN
dad-10670	102	3	,	,	PUNCT
dad-10670	102	4	“	"	PUNCT
dad-10670	102	5	he	he	PRON
dad-10670	102	6	said	say	VERB
dad-10670	102	7	that	that	SCONJ
dad-10670	102	8	”	"	PUNCT
dad-10670	102	9	is	be	AUX
dad-10670	102	10	not	not	PART
dad-10670	102	11	considered	consider	VERB
dad-10670	102	12	an	an	DET
dad-10670	102	13	independent	independent	ADJ
dad-10670	102	14	edu	edu	PROPN
dad-10670	102	15	.	.	PUNCT
dad-10670	103	1	in	in	ADP
dad-10670	103	2	this	this	DET
dad-10670	103	3	article	article	NOUN
dad-10670	103	4	,	,	PUNCT
dad-10670	103	5	we	we	PRON
dad-10670	103	6	will	will	AUX
dad-10670	103	7	not	not	PART
dad-10670	103	8	enter	enter	VERB
dad-10670	103	9	the	the	DET
dad-10670	103	10	discussion	discussion	NOUN
dad-10670	103	11	of	of	ADP
dad-10670	103	12	what	what	PRON
dad-10670	103	13	constitutes	constitute	VERB
dad-10670	103	14	the	the	DET
dad-10670	103	15	best	good	ADJ
dad-10670	103	16	segmentation	segmentation	NOUN
dad-10670	103	17	guidelines	guideline	NOUN
dad-10670	103	18	,	,	PUNCT
dad-10670	103	19	and	and	CCONJ
dad-10670	103	20	simply	simply	ADV
dad-10670	103	21	take	take	VERB
dad-10670	103	22	the	the	DET
dad-10670	103	23	rst	rst	NOUN
dad-10670	103	24	-	-	PUNCT
dad-10670	103	25	dt	dt	NOUN
dad-10670	103	26	corpus	corpus	NOUN
dad-10670	103	27	as	as	ADP
dad-10670	103	28	the	the	DET
dad-10670	103	29	basis	basis	NOUN
dad-10670	103	30	for	for	ADP
dad-10670	103	31	segmentation	segmentation	NOUN
dad-10670	103	32	by	by	ADP
dad-10670	103	33	supervised	supervised	ADJ
dad-10670	103	34	methods	method	NOUN
dad-10670	103	35	.	.	PUNCT
dad-10670	104	1	3	3	X
dad-10670	104	2	.	.	X
dad-10670	104	3	approach	approach	NOUN
dad-10670	104	4	3.1	3.1	NUM
dad-10670	104	5	the	the	DET
dad-10670	104	6	rst	rst	PROPN
dad-10670	104	7	discourse	discourse	NOUN
dad-10670	104	8	treebank	treebank	NOUN
dad-10670	104	9	as	as	SCONJ
dad-10670	104	10	previously	previously	ADV
dad-10670	104	11	mentioned	mention	VERB
dad-10670	104	12	,	,	PUNCT
dad-10670	104	13	rst	rst	PROPN
dad-10670	104	14	-	-	PUNCT
dad-10670	104	15	dt	dt	PROPN
dad-10670	104	16	is	be	AUX
dad-10670	104	17	a	a	DET
dad-10670	104	18	large	large	ADJ
dad-10670	104	19	corpus	corpus	NOUN
dad-10670	104	20	of	of	ADP
dad-10670	104	21	documents	document	NOUN
dad-10670	104	22	annotated	annotate	VERB
dad-10670	104	23	with	with	ADP
dad-10670	104	24	edu	edu	PROPN
dad-10670	104	25	segmentation	segmentation	NOUN
dad-10670	104	26	and	and	CCONJ
dad-10670	104	27	full	full	ADJ
dad-10670	104	28	text	text	NOUN
dad-10670	104	29	level	level	NOUN
dad-10670	104	30	rhetorical	rhetorical	ADJ
dad-10670	104	31	structure	structure	NOUN
dad-10670	104	32	(	(	PUNCT
dad-10670	104	33	carlson	carlson	PROPN
dad-10670	104	34	et	et	PROPN
dad-10670	104	35	al	al	PROPN
dad-10670	104	36	.	.	PROPN
dad-10670	104	37	,	,	PUNCT
dad-10670	104	38	2001	2001	NUM
dad-10670	104	39	)	)	PUNCT
dad-10670	104	40	.	.	PUNCT
dad-10670	105	1	the	the	DET
dad-10670	105	2	corpus	corpus	NOUN
dad-10670	105	3	was	be	AUX
dad-10670	105	4	released	release	VERB
dad-10670	105	5	in	in	ADP
dad-10670	105	6	2002	2002	NUM
dad-10670	105	7	and	and	CCONJ
dad-10670	105	8	contains	contain	VERB
dad-10670	105	9	385	385	NUM
dad-10670	105	10	articles	article	NOUN
dad-10670	105	11	transcribed	transcribe	VERB
dad-10670	105	12	from	from	ADP
dad-10670	105	13	the	the	DET
dad-10670	105	14	wall	wall	PROPN
dad-10670	105	15	street	street	PROPN
dad-10670	105	16	journal	journal	PROPN
dad-10670	105	17	.	.	PUNCT
dad-10670	106	1	these	these	DET
dad-10670	106	2	articles	article	NOUN
dad-10670	106	3	are	be	AUX
dad-10670	106	4	a	a	DET
dad-10670	106	5	subset	subset	NOUN
dad-10670	106	6	of	of	ADP
dad-10670	106	7	the	the	DET
dad-10670	106	8	penn	penn	PROPN
dad-10670	106	9	treebank	treebank	PROPN
dad-10670	106	10	corpus	corpus	PROPN
dad-10670	106	11	of	of	ADP
dad-10670	106	12	english	english	PROPN
dad-10670	106	13	(	(	PUNCT
dad-10670	106	14	marcus	marcus	PROPN
dad-10670	106	15	et	et	PROPN
dad-10670	106	16	al	al	PROPN
dad-10670	106	17	.	.	PROPN
dad-10670	106	18	,	,	PUNCT
dad-10670	106	19	1993	1993	NUM
dad-10670	106	20	)	)	PUNCT
dad-10670	106	21	.	.	PUNCT
dad-10670	107	1	the	the	DET
dad-10670	107	2	corpus	corpus	NOUN
dad-10670	107	3	allows	allow	VERB
dad-10670	107	4	us	we	PRON
dad-10670	107	5	to	to	AUX
dad-10670	107	6	both	both	PRON
dad-10670	107	7	train	train	VERB
dad-10670	107	8	and	and	CCONJ
dad-10670	107	9	evaluate	evaluate	VERB
dad-10670	107	10	the	the	DET
dad-10670	107	11	performance	performance	NOUN
dad-10670	107	12	of	of	ADP
dad-10670	107	13	our	our	PRON
dad-10670	107	14	algorithm	algorithm	NOUN
dad-10670	107	15	on	on	ADP
dad-10670	107	16	a	a	DET
dad-10670	107	17	large	large	ADJ
dad-10670	107	18	number	number	NOUN
dad-10670	107	19	of	of	ADP
dad-10670	107	20	documents	document	NOUN
dad-10670	107	21	of	of	ADP
dad-10670	107	22	varied	varied	ADJ
dad-10670	107	23	lengths	length	NOUN
dad-10670	107	24	.	.	PUNCT
dad-10670	108	1	although	although	SCONJ
dad-10670	108	2	the	the	DET
dad-10670	108	3	original	original	ADJ
dad-10670	108	4	rst	rst	NOUN
dad-10670	108	5	set	set	NOUN
dad-10670	108	6	is	be	AUX
dad-10670	108	7	composed	compose	VERB
dad-10670	108	8	of	of	ADP
dad-10670	108	9	24	24	NUM
dad-10670	108	10	discourse	discourse	NOUN
dad-10670	108	11	relations	relation	NOUN
dad-10670	108	12	(	(	PUNCT
dad-10670	108	13	mann	mann	PROPN
dad-10670	108	14	and	and	CCONJ
dad-10670	108	15	thompson	thompson	PROPN
dad-10670	108	16	,	,	PUNCT
dad-10670	108	17	1988	1988	NUM
dad-10670	108	18	)	)	PUNCT
dad-10670	108	19	,	,	PUNCT
dad-10670	108	20	this	this	DET
dad-10670	108	21	list	list	NOUN
dad-10670	108	22	has	have	AUX
dad-10670	108	23	evolved	evolve	VERB
dad-10670	108	24	to	to	PART
dad-10670	108	25	fit	fit	VERB
dad-10670	108	26	the	the	DET
dad-10670	108	27	type	type	NOUN
dad-10670	108	28	of	of	ADP
dad-10670	108	29	application	application	NOUN
dad-10670	108	30	and	and	CCONJ
dad-10670	108	31	expressive	expressive	ADJ
dad-10670	108	32	power	power	NOUN
dad-10670	108	33	required	require	VERB
dad-10670	108	34	by	by	ADP
dad-10670	108	35	each	each	DET
dad-10670	108	36	researcher	researcher	NOUN
dad-10670	108	37	.	.	PUNCT
dad-10670	109	1	in	in	ADP
dad-10670	109	2	the	the	DET
dad-10670	109	3	rst	rst	PROPN
dad-10670	109	4	-	-	PUNCT
dad-10670	109	5	dt	dt	PROPN
dad-10670	109	6	,	,	PUNCT
dad-10670	109	7	relation	relation	NOUN
dad-10670	109	8	annotation	annotation	NOUN
dad-10670	109	9	is	be	AUX
dad-10670	109	10	done	do	VERB
dad-10670	109	11	using	use	VERB
dad-10670	109	12	a	a	DET
dad-10670	109	13	set	set	NOUN
dad-10670	109	14	of	of	ADP
dad-10670	109	15	78	78	NUM
dad-10670	109	16	rhetorical	rhetorical	ADJ
dad-10670	109	17	relations	relation	NOUN
dad-10670	109	18	,	,	PUNCT
dad-10670	109	19	which	which	PRON
dad-10670	109	20	enables	enable	VERB
dad-10670	109	21	a	a	DET
dad-10670	109	22	high	high	ADJ
dad-10670	109	23	level	level	NOUN
dad-10670	109	24	of	of	ADP
dad-10670	109	25	expressivity	expressivity	NOUN
dad-10670	109	26	.	.	PUNCT
dad-10670	110	1	these	these	DET
dad-10670	110	2	relations	relation	NOUN
dad-10670	110	3	are	be	AUX
dad-10670	110	4	divided	divide	VERB
dad-10670	110	5	in	in	ADP
dad-10670	110	6	two	two	NUM
dad-10670	110	7	categories	category	NOUN
dad-10670	110	8	:	:	PUNCT
dad-10670	110	9	53	53	NUM
dad-10670	110	10	mononuclear	mononuclear	ADJ
dad-10670	110	11	relations	relation	NOUN
dad-10670	110	12	and	and	CCONJ
dad-10670	110	13	25	25	NUM
dad-10670	110	14	multi	multi	ADJ
dad-10670	110	15	-	-	ADJ
dad-10670	110	16	nuclear	nuclear	ADJ
dad-10670	110	17	relations	relation	NOUN
dad-10670	110	18	.	.	PUNCT
dad-10670	111	1	3.2	3.2	NUM
dad-10670	111	2	working	working	NOUN
dad-10670	111	3	assumptions	assumption	NOUN
dad-10670	111	4	an	an	DET
dad-10670	111	5	essential	essential	ADJ
dad-10670	111	6	characteristic	characteristic	NOUN
dad-10670	111	7	of	of	ADP
dad-10670	111	8	rst	rst	PROPN
dad-10670	111	9	is	be	AUX
dad-10670	111	10	that	that	SCONJ
dad-10670	111	11	“	"	PUNCT
dad-10670	111	12	a	a	DET
dad-10670	111	13	left	leave	VERB
dad-10670	111	14	-	-	PUNCT
dad-10670	111	15	to	to	ADP
dad-10670	111	16	-	-	PUNCT
dad-10670	111	17	right	right	ADJ
dad-10670	111	18	reading	reading	NOUN
dad-10670	111	19	of	of	ADP
dad-10670	111	20	the	the	DET
dad-10670	111	21	terminal	terminal	ADJ
dad-10670	111	22	frontier	frontier	NOUN
dad-10670	111	23	of	of	ADP
dad-10670	111	24	a	a	DET
dad-10670	111	25	discourse	discourse	NOUN
dad-10670	111	26	tree	tree	NOUN
dad-10670	111	27	associated	associate	VERB
dad-10670	111	28	with	with	ADP
dad-10670	111	29	a	a	DET
dad-10670	111	30	text	text	NOUN
dad-10670	111	31	corresponds	correspond	VERB
dad-10670	111	32	to	to	ADP
dad-10670	111	33	the	the	DET
dad-10670	111	34	span	span	NOUN
dad-10670	111	35	of	of	ADP
dad-10670	111	36	text	text	NOUN
dad-10670	111	37	analyzed	analyze	VERB
dad-10670	111	38	,	,	PUNCT
dad-10670	111	39	in	in	ADP
dad-10670	111	40	the	the	DET
dad-10670	111	41	same	same	ADJ
dad-10670	111	42	linear	linear	ADJ
dad-10670	111	43	order	order	NOUN
dad-10670	111	44	.	.	PUNCT
dad-10670	111	45	”	"	PUNCT
dad-10670	112	1	(	(	PUNCT
dad-10670	112	2	marcu	marcu	NOUN
dad-10670	112	3	,	,	PUNCT
dad-10670	112	4	2000	2000	NUM
dad-10670	112	5	)	)	PUNCT
dad-10670	112	6	.	.	PUNCT
dad-10670	113	1	this	this	DET
dad-10670	113	2	‘	'	PUNCT
dad-10670	113	3	sequentiality	sequentiality	NOUN
dad-10670	113	4	principle	principle	NOUN
dad-10670	113	5	’	'	PUNCT
dad-10670	113	6	guarantees	guarantee	VERB
dad-10670	113	7	that	that	SCONJ
dad-10670	113	8	only	only	ADV
dad-10670	113	9	consecutive	consecutive	ADJ
dad-10670	113	10	spans	span	NOUN
dad-10670	113	11	of	of	ADP
dad-10670	113	12	a	a	DET
dad-10670	113	13	text	text	NOUN
dad-10670	113	14	can	can	AUX
dad-10670	113	15	be	be	AUX
dad-10670	113	16	put	put	VERB
dad-10670	113	17	into	into	ADP
dad-10670	113	18	relation	relation	NOUN
dad-10670	113	19	,	,	PUNCT
dad-10670	113	20	which	which	PRON
dad-10670	113	21	dramatically	dramatically	ADV
dad-10670	113	22	reduces	reduce	VERB
dad-10670	113	23	the	the	DET
dad-10670	113	24	size	size	NOUN
dad-10670	113	25	of	of	ADP
dad-10670	113	26	the	the	DET
dad-10670	113	27	solution	solution	NOUN
dad-10670	113	28	space	space	NOUN
dad-10670	113	29	(	(	PUNCT
dad-10670	113	30	see	see	VERB
dad-10670	113	31	section	section	NOUN
dad-10670	113	32	3.3	3.3	NUM
dad-10670	113	33	)	)	PUNCT
dad-10670	113	34	.	.	PUNCT
dad-10670	114	1	5	5	NUM
dad-10670	114	2	hernault	hernault	ADJ
dad-10670	114	3	,	,	PUNCT
dad-10670	114	4	prendinger	prendinger	NOUN
dad-10670	114	5	,	,	PUNCT
dad-10670	114	6	duverle	duverle	NOUN
dad-10670	114	7	,	,	PUNCT
dad-10670	114	8	and	and	CCONJ
dad-10670	114	9	ishizuka	ishizuka	VERB
dad-10670	114	10	when	when	SCONJ
dad-10670	114	11	building	build	VERB
dad-10670	114	12	discourse	discourse	NOUN
dad-10670	114	13	parsers	parser	NOUN
dad-10670	114	14	,	,	PUNCT
dad-10670	114	15	researchers	researcher	NOUN
dad-10670	114	16	have	have	AUX
dad-10670	114	17	opted	opt	VERB
dad-10670	114	18	to	to	PART
dad-10670	114	19	use	use	VERB
dad-10670	114	20	small	small	ADJ
dad-10670	114	21	relation	relation	NOUN
dad-10670	114	22	sets	set	NOUN
dad-10670	114	23	,	,	PUNCT
dad-10670	114	24	which	which	PRON
dad-10670	114	25	makes	make	VERB
dad-10670	114	26	the	the	DET
dad-10670	114	27	problem	problem	NOUN
dad-10670	114	28	of	of	ADP
dad-10670	114	29	assigning	assign	VERB
dad-10670	114	30	relations	relation	NOUN
dad-10670	114	31	easier	easy	ADJ
dad-10670	114	32	.	.	PUNCT
dad-10670	115	1	for	for	ADP
dad-10670	115	2	instance	instance	NOUN
dad-10670	115	3	,	,	PUNCT
dad-10670	115	4	marcu	marcu	NOUN
dad-10670	115	5	(	(	PUNCT
dad-10670	115	6	2000	2000	NUM
dad-10670	115	7	)	)	PUNCT
dad-10670	115	8	used	use	VERB
dad-10670	115	9	15	15	NUM
dad-10670	115	10	relations	relation	NOUN
dad-10670	115	11	,	,	PUNCT
dad-10670	115	12	while	while	SCONJ
dad-10670	115	13	le	le	PROPN
dad-10670	115	14	thanh	thanh	VERB
dad-10670	115	15	et	et	PROPN
dad-10670	115	16	al	al	PROPN
dad-10670	115	17	.	.	PROPN
dad-10670	116	1	(	(	PUNCT
dad-10670	116	2	2004b	2004b	NUM
dad-10670	116	3	)	)	PUNCT
dad-10670	116	4	used	use	VERB
dad-10670	116	5	14	14	NUM
dad-10670	116	6	relations	relation	NOUN
dad-10670	116	7	.	.	PUNCT
dad-10670	117	1	although	although	SCONJ
dad-10670	117	2	the	the	DET
dad-10670	117	3	rich	rich	ADJ
dad-10670	117	4	78	78	NUM
dad-10670	117	5	-	-	PUNCT
dad-10670	117	6	relations	relation	NOUN
dad-10670	117	7	set	set	NOUN
dad-10670	117	8	of	of	ADP
dad-10670	117	9	the	the	DET
dad-10670	117	10	rst	rst	NOUN
dad-10670	117	11	-	-	PUNCT
dad-10670	117	12	dt	dt	NOUN
dad-10670	117	13	enables	enable	VERB
dad-10670	117	14	a	a	DET
dad-10670	117	15	high	high	ADJ
dad-10670	117	16	level	level	NOUN
dad-10670	117	17	of	of	ADP
dad-10670	117	18	expressivity	expressivity	NOUN
dad-10670	117	19	,	,	PUNCT
dad-10670	117	20	which	which	PRON
dad-10670	117	21	is	be	AUX
dad-10670	117	22	useful	useful	ADJ
dad-10670	117	23	in	in	ADP
dad-10670	117	24	the	the	DET
dad-10670	117	25	case	case	NOUN
dad-10670	117	26	of	of	ADP
dad-10670	117	27	linguistic	linguistic	ADJ
dad-10670	117	28	studies	study	NOUN
dad-10670	117	29	,	,	PUNCT
dad-10670	117	30	it	it	PRON
dad-10670	117	31	has	have	VERB
dad-10670	117	32	unnecessary	unnecessary	ADJ
dad-10670	117	33	precision	precision	NOUN
dad-10670	117	34	for	for	ADP
dad-10670	117	35	most	most	ADJ
dad-10670	117	36	text	text	NOUN
dad-10670	117	37	analysis	analysis	NOUN
dad-10670	117	38	applications	application	NOUN
dad-10670	117	39	.	.	PUNCT
dad-10670	118	1	furthermore	furthermore	ADV
dad-10670	118	2	,	,	PUNCT
dad-10670	118	3	from	from	ADP
dad-10670	118	4	a	a	DET
dad-10670	118	5	machine	machine	NOUN
dad-10670	118	6	learning	learning	NOUN
dad-10670	118	7	perspective	perspective	NOUN
dad-10670	118	8	,	,	PUNCT
dad-10670	118	9	working	work	VERB
dad-10670	118	10	with	with	ADP
dad-10670	118	11	a	a	DET
dad-10670	118	12	smaller	small	ADJ
dad-10670	118	13	set	set	NOUN
dad-10670	118	14	of	of	ADP
dad-10670	118	15	relations	relation	NOUN
dad-10670	118	16	improves	improve	VERB
dad-10670	118	17	the	the	DET
dad-10670	118	18	computational	computational	ADJ
dad-10670	118	19	properties	property	NOUN
dad-10670	118	20	of	of	ADP
dad-10670	118	21	the	the	DET
dad-10670	118	22	problem	problem	NOUN
dad-10670	118	23	.	.	PUNCT
dad-10670	119	1	a	a	DET
dad-10670	119	2	problem	problem	NOUN
dad-10670	119	3	with	with	ADP
dad-10670	119	4	fewer	few	ADJ
dad-10670	119	5	classes	class	NOUN
dad-10670	119	6	guarantees	guarantee	VERB
dad-10670	119	7	a	a	DET
dad-10670	119	8	better	well	ADJ
dad-10670	119	9	separability	separability	NOUN
dad-10670	119	10	between	between	ADP
dad-10670	119	11	the	the	DET
dad-10670	119	12	different	different	ADJ
dad-10670	119	13	classes	class	NOUN
dad-10670	119	14	,	,	PUNCT
dad-10670	119	15	as	as	SCONJ
dad-10670	119	16	it	it	PRON
dad-10670	119	17	abstracts	abstract	VERB
dad-10670	119	18	from	from	ADP
dad-10670	119	19	the	the	DET
dad-10670	119	20	ambiguities	ambiguity	NOUN
dad-10670	119	21	inherent	inherent	ADJ
dad-10670	119	22	to	to	ADP
dad-10670	119	23	fine	fine	ADV
dad-10670	119	24	-	-	PUNCT
dad-10670	119	25	grained	grain	VERB
dad-10670	119	26	relations	relation	NOUN
dad-10670	119	27	.	.	PUNCT
dad-10670	120	1	we	we	PRON
dad-10670	120	2	adopt	adopt	VERB
dad-10670	120	3	a	a	DET
dad-10670	120	4	similar	similar	ADJ
dad-10670	120	5	strategy	strategy	NOUN
dad-10670	120	6	and	and	CCONJ
dad-10670	120	7	work	work	VERB
dad-10670	120	8	with	with	ADP
dad-10670	120	9	the	the	DET
dad-10670	120	10	18	18	NUM
dad-10670	120	11	relations	relation	NOUN
dad-10670	120	12	defined	define	VERB
dad-10670	120	13	in	in	ADP
dad-10670	120	14	carlson	carlson	PROPN
dad-10670	120	15	et	et	PROPN
dad-10670	120	16	al	al	PROPN
dad-10670	120	17	.	.	PROPN
dad-10670	121	1	(	(	PUNCT
dad-10670	121	2	2001	2001	NUM
dad-10670	121	3	)	)	PUNCT
dad-10670	121	4	,	,	PUNCT
dad-10670	121	5	and	and	CCONJ
dad-10670	121	6	previously	previously	ADV
dad-10670	121	7	used	use	VERB
dad-10670	121	8	by	by	ADP
dad-10670	121	9	soricut	soricut	NOUN
dad-10670	121	10	and	and	CCONJ
dad-10670	121	11	marcu	marcu	NOUN
dad-10670	121	12	(	(	PUNCT
dad-10670	121	13	2003	2003	NUM
dad-10670	121	14	)	)	PUNCT
dad-10670	121	15	.	.	PUNCT
dad-10670	122	1	in	in	ADP
dad-10670	122	2	this	this	DET
dad-10670	122	3	reduced	reduce	VERB
dad-10670	122	4	set	set	NOUN
dad-10670	122	5	,	,	PUNCT
dad-10670	122	6	the	the	DET
dad-10670	122	7	original	original	ADJ
dad-10670	122	8	rst	rst	PROPN
dad-10670	122	9	-	-	PUNCT
dad-10670	122	10	dt	dt	NOUN
dad-10670	122	11	relations	relation	NOUN
dad-10670	122	12	are	be	AUX
dad-10670	122	13	partitioned	partition	VERB
dad-10670	122	14	into	into	ADP
dad-10670	122	15	16	16	NUM
dad-10670	122	16	categories	category	NOUN
dad-10670	122	17	,	,	PUNCT
dad-10670	122	18	which	which	PRON
dad-10670	122	19	correspond	correspond	VERB
dad-10670	122	20	to	to	ADP
dad-10670	122	21	(	(	PUNCT
dad-10670	122	22	merged	merged	ADJ
dad-10670	122	23	)	)	PUNCT
dad-10670	122	24	relations	relation	NOUN
dad-10670	122	25	(	(	PUNCT
dad-10670	122	26	called	call	VERB
dad-10670	122	27	“	"	PUNCT
dad-10670	122	28	classes	class	NOUN
dad-10670	122	29	”	"	PUNCT
dad-10670	122	30	in	in	ADP
dad-10670	122	31	carlson	carlson	PROPN
dad-10670	122	32	et	et	PROPN
dad-10670	122	33	al	al	PROPN
dad-10670	122	34	.	.	PROPN
dad-10670	123	1	(	(	PUNCT
dad-10670	123	2	2001	2001	NUM
dad-10670	123	3	)	)	PUNCT
dad-10670	123	4	)	)	PUNCT
dad-10670	123	5	,	,	PUNCT
dad-10670	123	6	depending	depend	VERB
dad-10670	123	7	on	on	ADP
dad-10670	123	8	their	their	PRON
dad-10670	123	9	rhetorical	rhetorical	ADJ
dad-10670	123	10	similarity.1	similarity.1	NOUN
dad-10670	123	11	for	for	ADP
dad-10670	123	12	instance	instance	NOUN
dad-10670	123	13	,	,	PUNCT
dad-10670	123	14	the	the	DET
dad-10670	123	15	semantically	semantically	ADV
dad-10670	123	16	similar	similar	ADJ
dad-10670	123	17	problem	problem	NOUN
dad-10670	123	18	-	-	PUNCT
dad-10670	123	19	solution	solution	NOUN
dad-10670	123	20	,	,	PUNCT
dad-10670	123	21	question	question	NOUN
dad-10670	123	22	-	-	PUNCT
dad-10670	123	23	answer	answer	NOUN
dad-10670	123	24	,	,	PUNCT
dad-10670	123	25	statement	statement	NOUN
dad-10670	123	26	-	-	PUNCT
dad-10670	123	27	response	response	NOUN
dad-10670	123	28	,	,	PUNCT
dad-10670	123	29	topic	topic	NOUN
dad-10670	123	30	-	-	PUNCT
dad-10670	123	31	comment	comment	NOUN
dad-10670	123	32	and	and	CCONJ
dad-10670	123	33	comment	comment	NOUN
dad-10670	123	34	-	-	PUNCT
dad-10670	123	35	topic	topic	NOUN
dad-10670	123	36	relations	relation	NOUN
dad-10670	123	37	,	,	PUNCT
dad-10670	123	38	are	be	AUX
dad-10670	123	39	merged	merge	VERB
dad-10670	123	40	to	to	PART
dad-10670	123	41	topiccomment	topiccomment	VERB
dad-10670	123	42	.	.	PUNCT
dad-10670	124	1	two	two	NUM
dad-10670	124	2	extra	extra	ADJ
dad-10670	124	3	relations	relation	NOUN
dad-10670	124	4	are	be	AUX
dad-10670	124	5	used	use	VERB
dad-10670	124	6	for	for	ADP
dad-10670	124	7	helping	help	VERB
dad-10670	124	8	the	the	DET
dad-10670	124	9	structuring	structuring	NOUN
dad-10670	124	10	of	of	ADP
dad-10670	124	11	the	the	DET
dad-10670	124	12	text	text	NOUN
dad-10670	124	13	,	,	PUNCT
dad-10670	124	14	textual	textual	ADJ
dad-10670	124	15	-	-	PUNCT
dad-10670	124	16	organization	organization	NOUN
dad-10670	124	17	and	and	CCONJ
dad-10670	124	18	same	same	ADJ
dad-10670	124	19	-	-	PUNCT
dad-10670	124	20	unit	unit	NOUN
dad-10670	124	21	.	.	PUNCT
dad-10670	125	1	in	in	ADP
dad-10670	125	2	rst	rst	PROPN
dad-10670	125	3	,	,	PUNCT
dad-10670	125	4	multi	multi	ADJ
dad-10670	125	5	-	-	ADJ
dad-10670	125	6	nuclear	nuclear	ADJ
dad-10670	125	7	relations	relation	NOUN
dad-10670	125	8	can	can	AUX
dad-10670	125	9	take	take	VERB
dad-10670	125	10	an	an	DET
dad-10670	125	11	arbitrary	arbitrary	ADJ
dad-10670	125	12	number	number	NOUN
dad-10670	125	13	of	of	ADP
dad-10670	125	14	arguments	argument	NOUN
dad-10670	125	15	.	.	PUNCT
dad-10670	126	1	for	for	ADP
dad-10670	126	2	instance	instance	NOUN
dad-10670	126	3	,	,	PUNCT
dad-10670	126	4	the	the	DET
dad-10670	126	5	list	list	NOUN
dad-10670	126	6	relation	relation	NOUN
dad-10670	126	7	can	can	AUX
dad-10670	126	8	connect	connect	VERB
dad-10670	126	9	any	any	DET
dad-10670	126	10	number	number	NOUN
dad-10670	126	11	of	of	ADP
dad-10670	126	12	spans	span	NOUN
dad-10670	126	13	,	,	PUNCT
dad-10670	126	14	leading	lead	VERB
dad-10670	126	15	to	to	ADP
dad-10670	126	16	n	n	CCONJ
dad-10670	126	17	-	-	PUNCT
dad-10670	126	18	ary	ary	PROPN
dad-10670	126	19	tree	tree	NOUN
dad-10670	126	20	representations	representation	NOUN
dad-10670	126	21	.	.	PUNCT
dad-10670	127	1	in	in	ADP
dad-10670	127	2	order	order	NOUN
dad-10670	127	3	to	to	PART
dad-10670	127	4	maintain	maintain	VERB
dad-10670	127	5	compatibility	compatibility	NOUN
dad-10670	127	6	with	with	ADP
dad-10670	127	7	our	our	PRON
dad-10670	127	8	svm	svm	ADJ
dad-10670	127	9	classification	classification	NOUN
dad-10670	127	10	approach	approach	NOUN
dad-10670	127	11	,	,	PUNCT
dad-10670	127	12	we	we	PRON
dad-10670	127	13	have	have	VERB
dad-10670	127	14	to	to	PART
dad-10670	127	15	convert	convert	VERB
dad-10670	127	16	n	n	CCONJ
dad-10670	127	17	-	-	PUNCT
dad-10670	127	18	ary	ary	PROPN
dad-10670	127	19	trees	tree	NOUN
dad-10670	127	20	to	to	ADP
dad-10670	127	21	binary	binary	NOUN
dad-10670	127	22	trees	tree	NOUN
dad-10670	127	23	.	.	PUNCT
dad-10670	128	1	this	this	DET
dad-10670	128	2	conversion	conversion	NOUN
dad-10670	128	3	can	can	AUX
dad-10670	128	4	be	be	AUX
dad-10670	128	5	done	do	VERB
dad-10670	128	6	trivially	trivially	ADV
dad-10670	128	7	,	,	PUNCT
dad-10670	128	8	by	by	ADP
dad-10670	128	9	consecutively	consecutively	ADV
dad-10670	128	10	nesting	nest	VERB
dad-10670	128	11	the	the	DET
dad-10670	128	12	arguments	argument	NOUN
dad-10670	128	13	of	of	ADP
dad-10670	128	14	multi	multi	ADJ
dad-10670	128	15	-	-	ADJ
dad-10670	128	16	nuclear	nuclear	ADJ
dad-10670	128	17	relations	relation	NOUN
dad-10670	128	18	,	,	PUNCT
dad-10670	128	19	in	in	ADP
dad-10670	128	20	the	the	DET
dad-10670	128	21	fashion	fashion	NOUN
dad-10670	128	22	shown	show	VERB
dad-10670	128	23	in	in	ADP
dad-10670	128	24	figure	figure	NOUN
dad-10670	128	25	4	4	NUM
dad-10670	128	26	.	.	PUNCT
dad-10670	128	27	list	list	NOUN
dad-10670	129	1	[	[	X
dad-10670	129	2	1	1	X
dad-10670	129	3	]	]	X
dad-10670	129	4	[	[	X
dad-10670	129	5	2	2	X
dad-10670	129	6	]	]	X
dad-10670	129	7	[	[	X
dad-10670	129	8	3	3	X
dad-10670	129	9	]	]	X
dad-10670	129	10	[	[	X
dad-10670	129	11	4	4	NUM
dad-10670	129	12	]	]	SYM
dad-10670	129	13	1	1	NUM
dad-10670	129	14	list	list	NOUN
dad-10670	129	15	[	[	X
dad-10670	129	16	1	1	NUM
dad-10670	129	17	]	]	X
dad-10670	129	18	list	list	NOUN
dad-10670	129	19	[	[	X
dad-10670	129	20	2	2	NUM
dad-10670	129	21	]	]	X
dad-10670	129	22	list	list	NOUN
dad-10670	129	23	[	[	X
dad-10670	129	24	3	3	X
dad-10670	129	25	]	]	X
dad-10670	129	26	[	[	X
dad-10670	129	27	4	4	NUM
dad-10670	129	28	]	]	SYM
dad-10670	129	29	1	1	NUM
dad-10670	129	30	figure	figure	NOUN
dad-10670	129	31	4	4	NUM
dad-10670	129	32	:	:	PUNCT
dad-10670	129	33	binarization	binarization	NOUN
dad-10670	129	34	of	of	ADP
dad-10670	129	35	multi	multi	ADJ
dad-10670	129	36	-	-	ADJ
dad-10670	129	37	nuclear	nuclear	ADJ
dad-10670	129	38	relations	relation	NOUN
dad-10670	129	39	it	it	PRON
dad-10670	129	40	is	be	AUX
dad-10670	129	41	interesting	interesting	ADJ
dad-10670	129	42	to	to	PART
dad-10670	129	43	note	note	VERB
dad-10670	129	44	that	that	SCONJ
dad-10670	129	45	the	the	DET
dad-10670	129	46	reverse	reverse	ADJ
dad-10670	129	47	transformation	transformation	NOUN
dad-10670	129	48	is	be	AUX
dad-10670	129	49	possible	possible	ADJ
dad-10670	129	50	most	most	ADJ
dad-10670	129	51	of	of	ADP
dad-10670	129	52	the	the	DET
dad-10670	129	53	time	time	NOUN
dad-10670	129	54	.	.	PUNCT
dad-10670	130	1	however	however	ADV
dad-10670	130	2	,	,	PUNCT
dad-10670	130	3	a	a	DET
dad-10670	130	4	loss	loss	NOUN
dad-10670	130	5	of	of	ADP
dad-10670	130	6	reversibility	reversibility	NOUN
dad-10670	130	7	occurs	occur	VERB
dad-10670	130	8	when	when	SCONJ
dad-10670	130	9	two	two	NUM
dad-10670	130	10	similar	similar	ADJ
dad-10670	130	11	multi	multi	ADJ
dad-10670	130	12	-	-	ADJ
dad-10670	130	13	nuclear	nuclear	ADJ
dad-10670	130	14	relations	relation	NOUN
dad-10670	130	15	are	be	AUX
dad-10670	130	16	nested	nest	VERB
dad-10670	130	17	consecutively	consecutively	ADV
dad-10670	130	18	(	(	PUNCT
dad-10670	130	19	for	for	ADP
dad-10670	130	20	instance	instance	NOUN
dad-10670	130	21	,	,	PUNCT
dad-10670	130	22	two	two	NUM
dad-10670	130	23	lists	list	NOUN
dad-10670	130	24	)	)	PUNCT
dad-10670	130	25	.	.	PUNCT
dad-10670	131	1	however	however	ADV
dad-10670	131	2	,	,	PUNCT
dad-10670	131	3	these	these	DET
dad-10670	131	4	cases	case	NOUN
dad-10670	131	5	occur	occur	VERB
dad-10670	131	6	rarely	rarely	ADV
dad-10670	131	7	in	in	ADP
dad-10670	131	8	practice	practice	NOUN
dad-10670	131	9	—	—	PUNCT
dad-10670	131	10	only	only	ADV
dad-10670	131	11	15	15	NUM
dad-10670	131	12	times	time	NOUN
dad-10670	131	13	in	in	ADP
dad-10670	131	14	the	the	DET
dad-10670	131	15	380	380	NUM
dad-10670	131	16	documents	document	NOUN
dad-10670	131	17	of	of	ADP
dad-10670	131	18	our	our	PRON
dad-10670	131	19	corpus	corpus	NOUN
dad-10670	131	20	.	.	PUNCT
dad-10670	132	1	3.3	3.3	NUM
dad-10670	132	2	outline	outline	NOUN
dad-10670	132	3	of	of	ADP
dad-10670	132	4	discourse	discourse	NOUN
dad-10670	132	5	segmentation	segmentation	NOUN
dad-10670	132	6	and	and	CCONJ
dad-10670	132	7	discourse	discourse	NOUN
dad-10670	132	8	relation	relation	NOUN
dad-10670	132	9	labeling	labeling	NOUN
dad-10670	132	10	figure	figure	NOUN
dad-10670	132	11	5	5	NUM
dad-10670	132	12	shows	show	VERB
dad-10670	132	13	the	the	DET
dad-10670	132	14	basic	basic	ADJ
dad-10670	132	15	workflow	workflow	NOUN
dad-10670	132	16	of	of	ADP
dad-10670	132	17	the	the	DET
dad-10670	132	18	hilda	hilda	PROPN
dad-10670	132	19	parser	parser	NOUN
dad-10670	132	20	(	(	PUNCT
dad-10670	132	21	details	detail	NOUN
dad-10670	132	22	of	of	ADP
dad-10670	132	23	the	the	DET
dad-10670	132	24	workflow	workflow	NOUN
dad-10670	132	25	are	be	AUX
dad-10670	132	26	provided	provide	VERB
dad-10670	132	27	in	in	ADP
dad-10670	132	28	appendix	appendix	NOUN
dad-10670	132	29	a	a	PRON
dad-10670	132	30	)	)	PUNCT
dad-10670	132	31	.	.	PUNCT
dad-10670	133	1	1	1	X
dad-10670	133	2	.	.	X
dad-10670	133	3	for	for	ADP
dad-10670	133	4	a	a	DET
dad-10670	133	5	complete	complete	ADJ
dad-10670	133	6	list	list	NOUN
dad-10670	133	7	of	of	ADP
dad-10670	133	8	the	the	DET
dad-10670	133	9	relations	relation	NOUN
dad-10670	133	10	used	use	VERB
dad-10670	133	11	,	,	PUNCT
dad-10670	133	12	see	see	VERB
dad-10670	133	13	appendix	appendix	PROPN
dad-10670	133	14	b.	b.	PROPN
dad-10670	133	15	6	6	NUM
dad-10670	133	16	hilda	hilda	VERB
dad-10670	133	17	:	:	PUNCT
dad-10670	134	1	a	a	DET
dad-10670	134	2	discourse	discourse	NOUN
dad-10670	134	3	parser	parser	NOUN
dad-10670	134	4	using	use	VERB
dad-10670	134	5	support	support	NOUN
dad-10670	134	6	vector	vector	NOUN
dad-10670	134	7	machine	machine	NOUN
dad-10670	134	8	classification	classification	NOUN
dad-10670	134	9	•	•	NOUN
dad-10670	134	10	a	a	DET
dad-10670	134	11	text	text	NOUN
dad-10670	134	12	is	be	AUX
dad-10670	134	13	first	first	ADV
dad-10670	134	14	segmented	segment	VERB
dad-10670	134	15	into	into	ADP
dad-10670	134	16	edus	edus	NOUN
dad-10670	134	17	.	.	PUNCT
dad-10670	135	1	•	•	NUM
dad-10670	135	2	then	then	ADV
dad-10670	135	3	the	the	DET
dad-10670	135	4	relation	relation	NOUN
dad-10670	135	5	labeling	labeling	NOUN
dad-10670	135	6	step	step	NOUN
dad-10670	135	7	evaluates	evaluate	NOUN
dad-10670	135	8	which	which	PRON
dad-10670	135	9	relations	relation	NOUN
dad-10670	135	10	are	be	AUX
dad-10670	135	11	likely	likely	ADJ
dad-10670	135	12	to	to	PART
dad-10670	135	13	hold	hold	VERB
dad-10670	135	14	between	between	ADP
dad-10670	135	15	consecutive	consecutive	ADJ
dad-10670	135	16	edus	edus	NOUN
dad-10670	135	17	.	.	PUNCT
dad-10670	136	1	the	the	DET
dad-10670	136	2	two	two	NUM
dad-10670	136	3	edus	edus	NOUN
dad-10670	136	4	which	which	PRON
dad-10670	136	5	are	be	AUX
dad-10670	136	6	most	most	ADV
dad-10670	136	7	probably	probably	ADV
dad-10670	136	8	connected	connect	VERB
dad-10670	136	9	by	by	ADP
dad-10670	136	10	a	a	DET
dad-10670	136	11	rhetorical	rhetorical	ADJ
dad-10670	136	12	relation	relation	NOUN
dad-10670	136	13	are	be	AUX
dad-10670	136	14	merged	merge	VERB
dad-10670	136	15	into	into	ADP
dad-10670	136	16	a	a	DET
dad-10670	136	17	rhetorical	rhetorical	ADJ
dad-10670	136	18	structure	structure	NOUN
dad-10670	136	19	tree	tree	NOUN
dad-10670	136	20	of	of	ADP
dad-10670	136	21	two	two	NUM
dad-10670	136	22	edus	edus	NOUN
dad-10670	136	23	.	.	PUNCT
dad-10670	137	1	•	•	NUM
dad-10670	137	2	next	next	ADV
dad-10670	137	3	,	,	PUNCT
dad-10670	137	4	we	we	PRON
dad-10670	137	5	go	go	VERB
dad-10670	137	6	back	back	ADV
dad-10670	137	7	to	to	ADP
dad-10670	137	8	the	the	DET
dad-10670	137	9	labeling	labeling	NOUN
dad-10670	137	10	step	step	NOUN
dad-10670	137	11	to	to	PART
dad-10670	137	12	re	re	VERB
dad-10670	137	13	-	-	VERB
dad-10670	137	14	evaluate	evaluate	VERB
dad-10670	137	15	which	which	PRON
dad-10670	137	16	relations	relation	NOUN
dad-10670	137	17	are	be	AUX
dad-10670	137	18	the	the	DET
dad-10670	137	19	most	most	ADV
dad-10670	137	20	likely	likely	ADJ
dad-10670	137	21	to	to	PART
dad-10670	137	22	hold	hold	VERB
dad-10670	137	23	between	between	ADP
dad-10670	137	24	spans	span	NOUN
dad-10670	137	25	(	(	PUNCT
dad-10670	137	26	rhetorical	rhetorical	ADJ
dad-10670	137	27	structure	structure	NOUN
dad-10670	137	28	trees	tree	NOUN
dad-10670	137	29	of	of	ADP
dad-10670	137	30	any	any	DET
dad-10670	137	31	size	size	NOUN
dad-10670	137	32	,	,	PUNCT
dad-10670	137	33	including	include	VERB
dad-10670	137	34	atomical	atomical	ADJ
dad-10670	137	35	edus	edus	NOUN
dad-10670	137	36	)	)	PUNCT
dad-10670	137	37	.	.	PUNCT
dad-10670	138	1	•	•	NOUN
dad-10670	138	2	this	this	DET
dad-10670	138	3	procedure	procedure	NOUN
dad-10670	138	4	is	be	AUX
dad-10670	138	5	repeated	repeat	VERB
dad-10670	138	6	until	until	SCONJ
dad-10670	138	7	all	all	DET
dad-10670	138	8	spans	span	NOUN
dad-10670	138	9	are	be	AUX
dad-10670	138	10	merged	merge	VERB
dad-10670	138	11	.	.	PUNCT
dad-10670	139	1	the	the	DET
dad-10670	139	2	outlined	outline	VERB
dad-10670	139	3	method	method	NOUN
dad-10670	139	4	enables	enable	VERB
dad-10670	139	5	us	we	PRON
dad-10670	139	6	to	to	PART
dad-10670	139	7	find	find	VERB
dad-10670	139	8	a	a	DET
dad-10670	139	9	good	good	ADJ
dad-10670	139	10	approximation	approximation	NOUN
dad-10670	139	11	of	of	ADP
dad-10670	139	12	‘	'	PUNCT
dad-10670	139	13	perfect	perfect	ADJ
dad-10670	139	14	’	'	PUNCT
dad-10670	139	15	discourse	discourse	NOUN
dad-10670	139	16	trees	tree	NOUN
dad-10670	139	17	(	(	PUNCT
dad-10670	139	18	i.e.	i.e.	X
dad-10670	139	19	,	,	PUNCT
dad-10670	139	20	the	the	DET
dad-10670	139	21	ones	one	NOUN
dad-10670	139	22	created	create	VERB
dad-10670	139	23	by	by	ADP
dad-10670	139	24	human	human	ADJ
dad-10670	139	25	annotators	annotator	NOUN
dad-10670	139	26	)	)	PUNCT
dad-10670	139	27	in	in	ADP
dad-10670	139	28	linear	linear	ADJ
dad-10670	139	29	time	time	NOUN
dad-10670	139	30	complexity	complexity	NOUN
dad-10670	139	31	with	with	ADP
dad-10670	139	32	respect	respect	NOUN
dad-10670	139	33	to	to	ADP
dad-10670	139	34	the	the	DET
dad-10670	139	35	length	length	NOUN
dad-10670	139	36	of	of	ADP
dad-10670	139	37	the	the	DET
dad-10670	139	38	input	input	NOUN
dad-10670	139	39	text	text	NOUN
dad-10670	139	40	.	.	PUNCT
dad-10670	140	1	input	input	NOUN
dad-10670	140	2	text	text	NOUN
dad-10670	140	3	bdiscourse	bdiscourse	NOUN
dad-10670	140	4	segmentation	segmentation	NOUN
dad-10670	140	5	relation	relation	NOUN
dad-10670	140	6	labeling	labeling	NOUN
dad-10670	140	7	tree	tree	NOUN
dad-10670	140	8	construction	construction	NOUN
dad-10670	140	9	discourse	discourse	NOUN
dad-10670	140	10	tree	tree	NOUN
dad-10670	140	11	figure	figure	NOUN
dad-10670	140	12	5	5	NUM
dad-10670	140	13	:	:	PUNCT
dad-10670	140	14	basic	basic	ADJ
dad-10670	140	15	parser	parser	NOUN
dad-10670	140	16	workflow	workflow	NOUN
dad-10670	140	17	the	the	DET
dad-10670	140	18	tasks	task	NOUN
dad-10670	140	19	of	of	ADP
dad-10670	140	20	discourse	discourse	NOUN
dad-10670	140	21	segmentation	segmentation	NOUN
dad-10670	140	22	and	and	CCONJ
dad-10670	140	23	relation	relation	NOUN
dad-10670	140	24	labeling	labeling	NOUN
dad-10670	140	25	are	be	AUX
dad-10670	140	26	modeled	model	VERB
dad-10670	140	27	as	as	ADP
dad-10670	140	28	supervised	supervise	VERB
dad-10670	140	29	classification	classification	NOUN
dad-10670	140	30	tasks	task	NOUN
dad-10670	140	31	.	.	PUNCT
dad-10670	141	1	we	we	PRON
dad-10670	141	2	chose	choose	VERB
dad-10670	141	3	to	to	PART
dad-10670	141	4	use	use	VERB
dad-10670	141	5	support	support	NOUN
dad-10670	141	6	vector	vector	NOUN
dad-10670	141	7	machines	machine	NOUN
dad-10670	141	8	(	(	PUNCT
dad-10670	141	9	svm	svm	PROPN
dad-10670	141	10	)	)	PUNCT
dad-10670	141	11	(	(	PUNCT
dad-10670	141	12	vapnik	vapnik	X
dad-10670	141	13	,	,	PUNCT
dad-10670	141	14	1995	1995	NUM
dad-10670	141	15	)	)	PUNCT
dad-10670	141	16	.	.	PUNCT
dad-10670	142	1	svms	svms	NOUN
dad-10670	142	2	are	be	AUX
dad-10670	142	3	a	a	DET
dad-10670	142	4	set	set	NOUN
dad-10670	142	5	of	of	ADP
dad-10670	142	6	maximum	maximum	ADJ
dad-10670	142	7	-	-	PUNCT
dad-10670	142	8	margin	margin	NOUN
dad-10670	142	9	classifiers	classifier	NOUN
dad-10670	142	10	,	,	PUNCT
dad-10670	142	11	i.e.	i.e.	X
dad-10670	142	12	,	,	PUNCT
dad-10670	142	13	they	they	PRON
dad-10670	142	14	minimize	minimize	VERB
dad-10670	142	15	the	the	DET
dad-10670	142	16	classification	classification	NOUN
dad-10670	142	17	error	error	NOUN
dad-10670	142	18	and	and	CCONJ
dad-10670	142	19	maximize	maximize	VERB
dad-10670	142	20	the	the	DET
dad-10670	142	21	geometric	geometric	ADJ
dad-10670	142	22	margin	margin	NOUN
dad-10670	142	23	.	.	PUNCT
dad-10670	143	1	they	they	PRON
dad-10670	143	2	have	have	AUX
dad-10670	143	3	been	be	AUX
dad-10670	143	4	used	use	VERB
dad-10670	143	5	extensively	extensively	ADV
dad-10670	143	6	in	in	ADP
dad-10670	143	7	domains	domain	NOUN
dad-10670	143	8	ranging	range	VERB
dad-10670	143	9	from	from	ADP
dad-10670	143	10	bio	bio	NOUN
dad-10670	143	11	-	-	NOUN
dad-10670	143	12	informatics	informatic	NOUN
dad-10670	143	13	to	to	ADP
dad-10670	143	14	natural	natural	ADJ
dad-10670	143	15	language	language	NOUN
dad-10670	143	16	processing	processing	NOUN
dad-10670	143	17	,	,	PUNCT
dad-10670	143	18	and	and	CCONJ
dad-10670	143	19	are	be	AUX
dad-10670	143	20	considered	consider	VERB
dad-10670	143	21	as	as	ADP
dad-10670	143	22	a	a	DET
dad-10670	143	23	state	state	NOUN
dad-10670	143	24	-	-	PUNCT
dad-10670	143	25	of	of	ADP
dad-10670	143	26	-	-	PUNCT
dad-10670	143	27	the	the	DET
dad-10670	143	28	-	-	PUNCT
dad-10670	143	29	art	art	NOUN
dad-10670	143	30	classifier	classifier	NOUN
dad-10670	143	31	for	for	ADP
dad-10670	143	32	many	many	ADJ
dad-10670	143	33	practical	practical	ADJ
dad-10670	143	34	tasks	task	NOUN
dad-10670	143	35	.	.	PUNCT
dad-10670	144	1	in	in	ADP
dad-10670	144	2	discourse	discourse	NOUN
dad-10670	144	3	segmentation	segmentation	NOUN
dad-10670	144	4	,	,	PUNCT
dad-10670	144	5	the	the	DET
dad-10670	144	6	problem	problem	NOUN
dad-10670	144	7	is	be	AUX
dad-10670	144	8	to	to	PART
dad-10670	144	9	assign	assign	VERB
dad-10670	144	10	each	each	DET
dad-10670	144	11	word	word	NOUN
dad-10670	144	12	of	of	ADP
dad-10670	144	13	the	the	DET
dad-10670	144	14	input	input	NOUN
dad-10670	144	15	text	text	NOUN
dad-10670	144	16	an	an	DET
dad-10670	144	17	observation	observation	NOUN
dad-10670	144	18	category	category	NOUN
dad-10670	144	19	c	c	NOUN
dad-10670	144	20	∈	∈	PROPN
dad-10670	144	21	{	{	PUNCT
dad-10670	144	22	0	0	NUM
dad-10670	144	23	,	,	PUNCT
dad-10670	144	24	1	1	NUM
dad-10670	144	25	}	}	PUNCT
dad-10670	144	26	,	,	PUNCT
dad-10670	144	27	where	where	SCONJ
dad-10670	144	28	0	0	NUM
dad-10670	144	29	indicates	indicate	VERB
dad-10670	144	30	that	that	SCONJ
dad-10670	144	31	a	a	DET
dad-10670	144	32	word	word	NOUN
dad-10670	144	33	is	be	AUX
dad-10670	144	34	not	not	PART
dad-10670	144	35	a	a	DET
dad-10670	144	36	boundary	boundary	NOUN
dad-10670	144	37	,	,	PUNCT
dad-10670	144	38	and	and	CCONJ
dad-10670	144	39	1	1	NUM
dad-10670	144	40	indicates	indicate	VERB
dad-10670	144	41	that	that	SCONJ
dad-10670	144	42	it	it	PRON
dad-10670	144	43	is	be	AUX
dad-10670	144	44	a	a	DET
dad-10670	144	45	boundary	boundary	NOUN
dad-10670	144	46	.	.	PUNCT
dad-10670	145	1	for	for	ADP
dad-10670	145	2	example	example	NOUN
dad-10670	145	3	,	,	PUNCT
dad-10670	145	4	in	in	ADP
dad-10670	145	5	the	the	DET
dad-10670	145	6	snippet	snippet	NOUN
dad-10670	145	7	from	from	ADP
dad-10670	145	8	figure	figure	NOUN
dad-10670	145	9	1	1	NUM
dad-10670	145	10	,	,	PUNCT
dad-10670	145	11	the	the	DET
dad-10670	145	12	words	word	NOUN
dad-10670	145	13	said	say	VERB
dad-10670	145	14	,	,	PUNCT
dad-10670	145	15	quarter	quarter	NOUN
dad-10670	145	16	and	and	CCONJ
dad-10670	145	17	yesterday	yesterday	NOUN
dad-10670	145	18	are	be	AUX
dad-10670	145	19	assigned	assign	VERB
dad-10670	145	20	1	1	NUM
dad-10670	145	21	,	,	PUNCT
dad-10670	145	22	whereas	whereas	SCONJ
dad-10670	145	23	all	all	DET
dad-10670	145	24	other	other	ADJ
dad-10670	145	25	words	word	NOUN
dad-10670	145	26	belong	belong	VERB
dad-10670	145	27	to	to	ADP
dad-10670	145	28	category	category	NOUN
dad-10670	145	29	0	0	NUM
dad-10670	145	30	.	.	PUNCT
dad-10670	146	1	hence	hence	ADV
dad-10670	146	2	,	,	PUNCT
dad-10670	146	3	we	we	PRON
dad-10670	146	4	train	train	VERB
dad-10670	146	5	a	a	DET
dad-10670	146	6	binary	binary	PROPN
dad-10670	146	7	classifier	classifier	NOUN
dad-10670	146	8	seg	seg	PROPN
dad-10670	146	9	:	:	PUNCT
dad-10670	146	10	w	w	X
dad-10670	146	11	→	→	PUNCT
dad-10670	146	12	{	{	PUNCT
dad-10670	146	13	0	0	NUM
dad-10670	146	14	,	,	PUNCT
dad-10670	146	15	1	1	NUM
dad-10670	146	16	}	}	PUNCT
dad-10670	146	17	,	,	PUNCT
dad-10670	146	18	wherew	wherew	NOUN
dad-10670	146	19	is	be	AUX
dad-10670	146	20	the	the	DET
dad-10670	146	21	set	set	NOUN
dad-10670	146	22	of	of	ADP
dad-10670	146	23	all	all	DET
dad-10670	146	24	words	word	NOUN
dad-10670	146	25	occurring	occur	VERB
dad-10670	146	26	in	in	ADP
dad-10670	146	27	the	the	DET
dad-10670	146	28	input	input	NOUN
dad-10670	146	29	text	text	NOUN
dad-10670	146	30	,	,	PUNCT
dad-10670	146	31	denoted	denote	VERB
dad-10670	146	32	by	by	ADP
dad-10670	146	33	inputtext	inputtext	ADJ
dad-10670	146	34	.	.	PUNCT
dad-10670	147	1	after	after	ADP
dad-10670	147	2	the	the	DET
dad-10670	147	3	segmentation	segmentation	NOUN
dad-10670	147	4	step	step	NOUN
dad-10670	147	5	,	,	PUNCT
dad-10670	147	6	we	we	PRON
dad-10670	147	7	obtain	obtain	VERB
dad-10670	147	8	a	a	DET
dad-10670	147	9	list	list	NOUN
dad-10670	147	10	of	of	ADP
dad-10670	147	11	edus	edus	NOUN
dad-10670	147	12	,	,	PUNCT
dad-10670	147	13	e	e	PROPN
dad-10670	147	14	=	=	SYM
dad-10670	147	15	〈	〈	PROPN
dad-10670	147	16	e1	e1	PROPN
dad-10670	147	17	,	,	PUNCT
dad-10670	147	18	e2	e2	PROPN
dad-10670	147	19	,	,	PUNCT
dad-10670	147	20	.	.	PUNCT
dad-10670	147	21	.	.	PUNCT
dad-10670	148	1	.	.	PUNCT
dad-10670	149	1	〉	〉	NOUN
dad-10670	149	2	.	.	PUNCT
dad-10670	150	1	the	the	DET
dad-10670	150	2	concatenation	concatenation	NOUN
dad-10670	150	3	of	of	ADP
dad-10670	150	4	these	these	DET
dad-10670	150	5	edus	edus	NOUN
dad-10670	150	6	,	,	PUNCT
dad-10670	150	7	from	from	ADP
dad-10670	150	8	left	leave	VERB
dad-10670	150	9	to	to	ADP
dad-10670	150	10	right	right	NOUN
dad-10670	150	11	,	,	PUNCT
dad-10670	150	12	gives	give	VERB
dad-10670	150	13	back	back	ADP
dad-10670	150	14	inputtext	inputtext	ADJ
dad-10670	150	15	.	.	PUNCT
dad-10670	151	1	for	for	ADP
dad-10670	151	2	relation	relation	NOUN
dad-10670	151	3	labeling	labeling	NOUN
dad-10670	151	4	and	and	CCONJ
dad-10670	151	5	tree	tree	NOUN
dad-10670	151	6	construction	construction	NOUN
dad-10670	151	7	,	,	PUNCT
dad-10670	151	8	we	we	PRON
dad-10670	151	9	apply	apply	VERB
dad-10670	151	10	the	the	DET
dad-10670	151	11	following	following	ADJ
dad-10670	151	12	method	method	NOUN
dad-10670	151	13	.	.	PUNCT
dad-10670	152	1	given	give	VERB
dad-10670	152	2	two	two	NUM
dad-10670	152	3	consecutive	consecutive	ADJ
dad-10670	152	4	spans	span	NOUN
dad-10670	152	5	(	(	PUNCT
dad-10670	152	6	edus	edus	NOUN
dad-10670	152	7	or	or	CCONJ
dad-10670	152	8	subtrees	subtree	NOUN
dad-10670	152	9	)	)	PUNCT
dad-10670	152	10	,	,	PUNCT
dad-10670	152	11	we	we	PRON
dad-10670	152	12	determine	determine	VERB
dad-10670	152	13	(	(	PUNCT
dad-10670	152	14	1	1	X
dad-10670	152	15	)	)	PUNCT
dad-10670	152	16	the	the	DET
dad-10670	152	17	likelihood	likelihood	NOUN
dad-10670	152	18	of	of	ADP
dad-10670	152	19	a	a	DET
dad-10670	152	20	direct	direct	ADJ
dad-10670	152	21	structural	structural	ADJ
dad-10670	152	22	relation	relation	NOUN
dad-10670	152	23	,	,	PUNCT
dad-10670	152	24	(	(	PUNCT
dad-10670	152	25	2	2	X
dad-10670	152	26	)	)	PUNCT
dad-10670	152	27	the	the	DET
dad-10670	152	28	probabilities	probability	NOUN
dad-10670	152	29	for	for	ADP
dad-10670	152	30	the	the	DET
dad-10670	152	31	label	label	NOUN
dad-10670	152	32	of	of	ADP
dad-10670	152	33	the	the	DET
dad-10670	152	34	relation	relation	NOUN
dad-10670	152	35	with	with	ADP
dad-10670	152	36	nuclearity	nuclearity	NOUN
dad-10670	152	37	of	of	ADP
dad-10670	152	38	the	the	DET
dad-10670	152	39	spans	span	NOUN
dad-10670	152	40	.	.	PUNCT
dad-10670	153	1	a	a	DET
dad-10670	153	2	full	full	ADJ
dad-10670	153	3	rhetorical	rhetorical	ADJ
dad-10670	153	4	structure	structure	NOUN
dad-10670	153	5	tree	tree	NOUN
dad-10670	153	6	is	be	AUX
dad-10670	153	7	produced	produce	VERB
dad-10670	153	8	by	by	ADP
dad-10670	153	9	applying	apply	VERB
dad-10670	153	10	this	this	DET
dad-10670	153	11	mechanism	mechanism	NOUN
dad-10670	153	12	repeatedly	repeatedly	ADV
dad-10670	153	13	in	in	ADP
dad-10670	153	14	a	a	DET
dad-10670	153	15	straightforward	straightforward	ADJ
dad-10670	153	16	bottom	bottom	ADJ
dad-10670	153	17	-	-	PUNCT
dad-10670	153	18	up	up	ADP
dad-10670	153	19	fashion	fashion	NOUN
dad-10670	153	20	.	.	PUNCT
dad-10670	154	1	definition	definition	NOUN
dad-10670	154	2	1	1	NUM
dad-10670	154	3	(	(	PUNCT
dad-10670	154	4	discourse	discourse	NOUN
dad-10670	154	5	relation	relation	NOUN
dad-10670	154	6	set	set	NOUN
dad-10670	154	7	)	)	PUNCT
dad-10670	154	8	a	a	DET
dad-10670	154	9	discourse	discourse	NOUN
dad-10670	154	10	relation	relation	NOUN
dad-10670	154	11	setr	setr	NOUN
dad-10670	154	12	is	be	AUX
dad-10670	154	13	defined	define	VERB
dad-10670	154	14	as	as	ADP
dad-10670	154	15	a	a	DET
dad-10670	154	16	set	set	NOUN
dad-10670	154	17	r	r	NOUN
dad-10670	154	18	=	=	SYM
dad-10670	154	19	{	{	PUNCT
dad-10670	154	20	r1	r1	PROPN
dad-10670	154	21	,	,	PUNCT
dad-10670	154	22	r2	r2	PROPN
dad-10670	154	23	,	,	PUNCT
dad-10670	154	24	.	.	PUNCT
dad-10670	154	25	.	.	PUNCT
dad-10670	154	26	.	.	PUNCT
dad-10670	155	1	ri	ri	PROPN
dad-10670	155	2	,	,	PUNCT
dad-10670	155	3	.	.	PUNCT
dad-10670	155	4	.	.	PUNCT
dad-10670	156	1	.	.	PUNCT
dad-10670	157	1	,	,	PUNCT
dad-10670	157	2	rn	rn	PROPN
dad-10670	157	3	}	}	PUNCT
dad-10670	157	4	,	,	PUNCT
dad-10670	157	5	such	such	ADJ
dad-10670	157	6	that	that	SCONJ
dad-10670	157	7	∀i	∀i	NOUN
dad-10670	157	8	,	,	PUNCT
dad-10670	157	9	ri	ri	PROPN
dad-10670	157	10	=	=	SYM
dad-10670	157	11	〈	〈	PROPN
dad-10670	157	12	rri	rri	PROPN
dad-10670	157	13	,	,	PUNCT
dad-10670	157	14	lefti	lefti	NOUN
dad-10670	157	15	,	,	PUNCT
dad-10670	157	16	righti	righti	PROPN
dad-10670	157	17	〉	〉	PROPN
dad-10670	157	18	,	,	PUNCT
dad-10670	157	19	whereby	whereby	ADV
dad-10670	157	20	:	:	PUNCT
dad-10670	157	21	•	•	NUM
dad-10670	157	22	rri	rri	PROPN
dad-10670	157	23	∈	∈	PROPN
dad-10670	157	24	{	{	PUNCT
dad-10670	157	25	attribution	attribution	NOUN
dad-10670	157	26	,	,	PUNCT
dad-10670	157	27	cause	cause	NOUN
dad-10670	157	28	,	,	PUNCT
dad-10670	157	29	elaboration	elaboration	NOUN
dad-10670	157	30	,	,	PUNCT
dad-10670	157	31	list	list	NOUN
dad-10670	157	32	,	,	PUNCT
dad-10670	157	33	.	.	PUNCT
dad-10670	157	34	.	.	PUNCT
dad-10670	157	35	.	.	PUNCT
dad-10670	158	1	}	}	PUNCT
dad-10670	158	2	(	(	PUNCT
dad-10670	158	3	see	see	VERB
dad-10670	158	4	appendix	appendix	NOUN
dad-10670	158	5	b	b	NOUN
dad-10670	158	6	for	for	ADP
dad-10670	158	7	a	a	DET
dad-10670	158	8	full	full	ADJ
dad-10670	158	9	list	list	NOUN
dad-10670	158	10	of	of	ADP
dad-10670	158	11	relations	relation	NOUN
dad-10670	158	12	considered	consider	VERB
dad-10670	158	13	in	in	ADP
dad-10670	158	14	hilda	hilda	PROPN
dad-10670	158	15	)	)	PUNCT
dad-10670	158	16	;	;	PUNCT
dad-10670	158	17	•	•	NUM
dad-10670	158	18	lefti	lefti	NOUN
dad-10670	158	19	,	,	PUNCT
dad-10670	158	20	righti	righti	PROPN
dad-10670	158	21	∈	∈	PROPN
dad-10670	158	22	{	{	PUNCT
dad-10670	158	23	nucleus	nucleus	PROPN
dad-10670	158	24	,	,	PUNCT
dad-10670	158	25	satellite	satellite	NOUN
dad-10670	158	26	}	}	PUNCT
dad-10670	158	27	;	;	PUNCT
dad-10670	158	28	7	7	NUM
dad-10670	158	29	hernault	hernault	NOUN
dad-10670	158	30	,	,	PUNCT
dad-10670	158	31	prendinger	prendinger	NOUN
dad-10670	158	32	,	,	PUNCT
dad-10670	158	33	duverle	duverle	NOUN
dad-10670	158	34	,	,	PUNCT
dad-10670	158	35	and	and	CCONJ
dad-10670	158	36	ishizuka	ishizuka	VERB
dad-10670	158	37	•	•	NOUN
dad-10670	158	38	all	all	DET
dad-10670	158	39	ri	ri	NOUN
dad-10670	158	40	triplets	triplet	NOUN
dad-10670	158	41	are	be	AUX
dad-10670	158	42	valid	valid	ADJ
dad-10670	158	43	,	,	PUNCT
dad-10670	158	44	i.e.	i.e.	X
dad-10670	158	45	,	,	PUNCT
dad-10670	158	46	given	give	VERB
dad-10670	158	47	a	a	DET
dad-10670	158	48	rhetorical	rhetorical	ADJ
dad-10670	158	49	relation	relation	NOUN
dad-10670	158	50	rri	rri	PROPN
dad-10670	158	51	,	,	PUNCT
dad-10670	158	52	the	the	DET
dad-10670	158	53	left	left	ADJ
dad-10670	158	54	and	and	CCONJ
dad-10670	158	55	right	right	ADJ
dad-10670	158	56	nuclearities	nuclearitie	NOUN
dad-10670	158	57	lefti	lefti	VERB
dad-10670	158	58	and	and	CCONJ
dad-10670	158	59	righti	righti	PROPN
dad-10670	158	60	are	be	AUX
dad-10670	158	61	allowed	allow	VERB
dad-10670	158	62	nuclearity	nuclearity	NOUN
dad-10670	158	63	options	option	NOUN
dad-10670	158	64	.	.	PUNCT
dad-10670	159	1	our	our	PRON
dad-10670	159	2	working	work	VERB
dad-10670	159	3	rhetorical	rhetorical	ADJ
dad-10670	159	4	relation	relation	NOUN
dad-10670	159	5	set	set	VERB
dad-10670	159	6	and	and	CCONJ
dad-10670	159	7	allowed	allow	VERB
dad-10670	159	8	nuclearity	nuclearity	NOUN
dad-10670	159	9	options	option	NOUN
dad-10670	159	10	are	be	AUX
dad-10670	159	11	defined	define	VERB
dad-10670	159	12	in	in	ADP
dad-10670	159	13	appendix	appendix	PROPN
dad-10670	159	14	b.	b.	PROPN
dad-10670	160	1	we	we	PRON
dad-10670	160	2	employ	employ	VERB
dad-10670	160	3	18	18	NUM
dad-10670	160	4	rhetorical	rhetorical	ADJ
dad-10670	160	5	relations	relation	NOUN
dad-10670	160	6	which	which	PRON
dad-10670	160	7	,	,	PUNCT
dad-10670	160	8	when	when	SCONJ
dad-10670	160	9	nuclearized	nuclearize	VERB
dad-10670	160	10	,	,	PUNCT
dad-10670	160	11	give	give	VERB
dad-10670	160	12	a	a	DET
dad-10670	160	13	total	total	NOUN
dad-10670	160	14	of	of	ADP
dad-10670	160	15	41	41	NUM
dad-10670	160	16	possible	possible	ADJ
dad-10670	160	17	ri	ri	NOUN
dad-10670	160	18	triplets	triplet	NOUN
dad-10670	160	19	(	(	PUNCT
dad-10670	160	20	the	the	DET
dad-10670	160	21	cardinal	cardinal	ADJ
dad-10670	160	22	number	number	NOUN
dad-10670	160	23	ofr	ofr	PROPN
dad-10670	160	24	)	)	PUNCT
dad-10670	160	25	.	.	PUNCT
dad-10670	161	1	definition	definition	NOUN
dad-10670	161	2	2	2	NUM
dad-10670	161	3	(	(	PUNCT
dad-10670	161	4	valid	valid	ADJ
dad-10670	161	5	rs	rs	NOUN
dad-10670	161	6	-	-	PUNCT
dad-10670	161	7	tree	tree	NOUN
dad-10670	161	8	)	)	PUNCT
dad-10670	161	9	a	a	DET
dad-10670	161	10	rs	rs	NOUN
dad-10670	161	11	-	-	PUNCT
dad-10670	161	12	tree	tree	NOUN
dad-10670	161	13	t	t	NOUN
dad-10670	161	14	is	be	AUX
dad-10670	161	15	valid	valid	ADJ
dad-10670	161	16	if	if	SCONJ
dad-10670	161	17	it	it	PRON
dad-10670	161	18	satisfies	satisfy	VERB
dad-10670	161	19	the	the	DET
dad-10670	161	20	following	follow	VERB
dad-10670	161	21	properties	property	NOUN
dad-10670	161	22	:	:	PUNCT
dad-10670	162	1	1	1	X
dad-10670	162	2	.	.	X
dad-10670	162	3	all	all	DET
dad-10670	162	4	leaf	leaf	NOUN
dad-10670	162	5	nodes	node	NOUN
dad-10670	162	6	of	of	ADP
dad-10670	162	7	t	t	PROPN
dad-10670	162	8	are	be	AUX
dad-10670	162	9	edus;2	edus;2	PROPN
dad-10670	162	10	2	2	NUM
dad-10670	162	11	.	.	PUNCT
dad-10670	163	1	all	all	DET
dad-10670	163	2	non	non	ADJ
dad-10670	163	3	-	-	ADJ
dad-10670	163	4	leaf	leaf	ADJ
dad-10670	163	5	nodes	node	NOUN
dad-10670	163	6	of	of	ADP
dad-10670	163	7	t	t	PROPN
dad-10670	163	8	are	be	AUX
dad-10670	163	9	tagged	tag	VERB
dad-10670	163	10	with	with	ADP
dad-10670	163	11	a	a	DET
dad-10670	163	12	discourse	discourse	NOUN
dad-10670	163	13	relation	relation	NOUN
dad-10670	163	14	ri	ri	PROPN
dad-10670	163	15	∈	∈	PROPN
dad-10670	163	16	r.	r.	PROPN
dad-10670	163	17	t	t	PROPN
dad-10670	164	1	=	=	PUNCT
dad-10670	164	2	{	{	PUNCT
dad-10670	164	3	t1	t1	NOUN
dad-10670	164	4	,	,	PUNCT
dad-10670	164	5	t2	t2	NOUN
dad-10670	164	6	,	,	PUNCT
dad-10670	164	7	.	.	PUNCT
dad-10670	164	8	.	.	PUNCT
dad-10670	164	9	.	.	PUNCT
dad-10670	165	1	ti	ti	X
dad-10670	165	2	,	,	PUNCT
dad-10670	165	3	.	.	PUNCT
dad-10670	165	4	.	.	PUNCT
dad-10670	165	5	.	.	PUNCT
dad-10670	166	1	}	}	PUNCT
dad-10670	166	2	denotes	denote	VERB
dad-10670	166	3	the	the	DET
dad-10670	166	4	set	set	NOUN
dad-10670	166	5	of	of	ADP
dad-10670	166	6	all	all	DET
dad-10670	166	7	valid	valid	ADJ
dad-10670	166	8	rs	r	NOUN
dad-10670	166	9	-	-	PUNCT
dad-10670	166	10	trees	tree	NOUN
dad-10670	166	11	.	.	PUNCT
dad-10670	167	1	in	in	ADP
dad-10670	167	2	our	our	PRON
dad-10670	167	3	algorithm	algorithm	NOUN
dad-10670	167	4	,	,	PUNCT
dad-10670	167	5	rs	r	NOUN
dad-10670	167	6	-	-	PUNCT
dad-10670	167	7	trees	tree	NOUN
dad-10670	167	8	are	be	AUX
dad-10670	167	9	implemented	implement	VERB
dad-10670	167	10	using	use	VERB
dad-10670	167	11	binary	binary	ADJ
dad-10670	167	12	tree	tree	NOUN
dad-10670	167	13	structures	structure	NOUN
dad-10670	167	14	.	.	PUNCT
dad-10670	168	1	to	to	PART
dad-10670	168	2	obtain	obtain	VERB
dad-10670	168	3	a	a	DET
dad-10670	168	4	good	good	ADJ
dad-10670	168	5	classification	classification	NOUN
dad-10670	168	6	accuracy	accuracy	NOUN
dad-10670	168	7	,	,	PUNCT
dad-10670	168	8	we	we	PRON
dad-10670	168	9	train	train	VERB
dad-10670	168	10	two	two	NUM
dad-10670	168	11	separate	separate	ADJ
dad-10670	168	12	classifiers	classifier	NOUN
dad-10670	168	13	:	:	PUNCT
dad-10670	168	14	•	•	ADP
dad-10670	168	15	a	a	DET
dad-10670	168	16	binary	binary	NOUN
dad-10670	168	17	classifier	classifier	NOUN
dad-10670	168	18	for	for	ADP
dad-10670	168	19	structure	structure	NOUN
dad-10670	168	20	labeling	labeling	NOUN
dad-10670	168	21	,	,	PUNCT
dad-10670	168	22	i.e.	i.e.	X
dad-10670	168	23	,	,	PUNCT
dad-10670	168	24	existence	existence	NOUN
dad-10670	168	25	of	of	ADP
dad-10670	168	26	a	a	DET
dad-10670	168	27	rhetorical	rhetorical	ADJ
dad-10670	168	28	relation	relation	NOUN
dad-10670	168	29	between	between	ADP
dad-10670	168	30	two	two	NUM
dad-10670	168	31	valid	valid	ADJ
dad-10670	168	32	rs	r	NOUN
dad-10670	168	33	-	-	PUNCT
dad-10670	168	34	trees	tree	NOUN
dad-10670	168	35	:	:	PUNCT
dad-10670	168	36	struct	struct	NOUN
dad-10670	168	37	:	:	PUNCT
dad-10670	168	38	t	t	PROPN
dad-10670	168	39	×	×	PROPN
dad-10670	168	40	t	t	PROPN
dad-10670	168	41	→	→	PUNCT
dad-10670	168	42	{	{	PUNCT
dad-10670	168	43	0	0	NUM
dad-10670	168	44	,	,	PUNCT
dad-10670	168	45	1	1	NUM
dad-10670	168	46	}	}	PUNCT
dad-10670	168	47	.	.	PUNCT
dad-10670	169	1	•	•	NOUN
dad-10670	169	2	a	a	DET
dad-10670	169	3	multi	multi	ADJ
dad-10670	169	4	-	-	ADJ
dad-10670	169	5	class	class	ADJ
dad-10670	169	6	classifier	classifier	NOUN
dad-10670	169	7	for	for	ADP
dad-10670	169	8	rhetorical	rhetorical	ADJ
dad-10670	169	9	relation	relation	NOUN
dad-10670	169	10	and	and	CCONJ
dad-10670	169	11	nuclearity	nuclearity	NOUN
dad-10670	169	12	labeling	labeling	NOUN
dad-10670	169	13	:	:	PUNCT
dad-10670	169	14	label	label	NOUN
dad-10670	169	15	:	:	PUNCT
dad-10670	169	16	t	t	PROPN
dad-10670	169	17	×	×	PROPN
dad-10670	169	18	t	t	PROPN
dad-10670	169	19	→	→	SYM
dad-10670	169	20	r.	r.	PROPN
dad-10670	169	21	label	label	NOUN
dad-10670	169	22	returns	return	NOUN
dad-10670	169	23	predicted	predict	VERB
dad-10670	169	24	relation	relation	NOUN
dad-10670	169	25	label	label	NOUN
dad-10670	169	26	and	and	CCONJ
dad-10670	169	27	nuclearities	nuclearitie	NOUN
dad-10670	169	28	.	.	PUNCT
dad-10670	170	1	the	the	DET
dad-10670	170	2	detailed	detailed	ADJ
dad-10670	170	3	algorithm	algorithm	NOUN
dad-10670	170	4	for	for	ADP
dad-10670	170	5	tree	tree	NOUN
dad-10670	170	6	construction	construction	NOUN
dad-10670	170	7	is	be	AUX
dad-10670	170	8	provided	provide	VERB
dad-10670	170	9	in	in	ADP
dad-10670	170	10	figure	figure	NOUN
dad-10670	170	11	6	6	NUM
dad-10670	170	12	.	.	PUNCT
dad-10670	171	1	we	we	PRON
dad-10670	171	2	first	first	ADV
dad-10670	171	3	create	create	VERB
dad-10670	171	4	a	a	DET
dad-10670	171	5	list	list	NOUN
dad-10670	171	6	containing	contain	VERB
dad-10670	171	7	all	all	DET
dad-10670	171	8	edus	edus	NOUN
dad-10670	171	9	of	of	ADP
dad-10670	171	10	the	the	DET
dad-10670	171	11	input	input	NOUN
dad-10670	171	12	text	text	NOUN
dad-10670	171	13	,	,	PUNCT
dad-10670	171	14	in	in	ADP
dad-10670	171	15	left	left	ADJ
dad-10670	171	16	-	-	PUNCT
dad-10670	171	17	to	to	ADP
dad-10670	171	18	-	-	PUNCT
dad-10670	171	19	right	right	ADJ
dad-10670	171	20	reading	reading	NOUN
dad-10670	171	21	order	order	NOUN
dad-10670	171	22	.	.	PUNCT
dad-10670	172	1	the	the	DET
dad-10670	172	2	binary	binary	PROPN
dad-10670	172	3	classifier	classifier	PROPN
dad-10670	172	4	struct	struct	NOUN
dad-10670	172	5	is	be	AUX
dad-10670	172	6	then	then	ADV
dad-10670	172	7	applied	apply	VERB
dad-10670	172	8	to	to	ADP
dad-10670	172	9	all	all	DET
dad-10670	172	10	pairs	pair	NOUN
dad-10670	172	11	of	of	ADP
dad-10670	172	12	consecutive	consecutive	ADJ
dad-10670	172	13	elements	element	NOUN
dad-10670	172	14	from	from	ADP
dad-10670	172	15	this	this	DET
dad-10670	172	16	list	list	NOUN
dad-10670	172	17	,	,	PUNCT
dad-10670	172	18	in	in	ADP
dad-10670	172	19	order	order	NOUN
dad-10670	172	20	to	to	PART
dad-10670	172	21	determine	determine	VERB
dad-10670	172	22	the	the	DET
dad-10670	172	23	probability	probability	NOUN
dad-10670	172	24	of	of	ADP
dad-10670	172	25	a	a	DET
dad-10670	172	26	structural	structural	ADJ
dad-10670	172	27	relation	relation	NOUN
dad-10670	172	28	between	between	ADP
dad-10670	172	29	consecutive	consecutive	ADJ
dad-10670	172	30	edus	edus	NOUN
dad-10670	172	31	.	.	PUNCT
dad-10670	173	1	next	next	ADV
dad-10670	173	2	we	we	PRON
dad-10670	173	3	select	select	VERB
dad-10670	173	4	the	the	DET
dad-10670	173	5	pair	pair	NOUN
dad-10670	173	6	with	with	ADP
dad-10670	173	7	the	the	DET
dad-10670	173	8	highest	high	ADJ
dad-10670	173	9	probability	probability	NOUN
dad-10670	173	10	and	and	CCONJ
dad-10670	173	11	apply	apply	VERB
dad-10670	173	12	label	label	NOUN
dad-10670	173	13	to	to	ADP
dad-10670	173	14	it	it	PRON
dad-10670	173	15	,	,	PUNCT
dad-10670	173	16	in	in	ADP
dad-10670	173	17	order	order	NOUN
dad-10670	173	18	to	to	PART
dad-10670	173	19	determine	determine	VERB
dad-10670	173	20	the	the	DET
dad-10670	173	21	relation	relation	NOUN
dad-10670	173	22	’s	’s	PART
dad-10670	173	23	label	label	NOUN
dad-10670	173	24	and	and	CCONJ
dad-10670	173	25	nuclearity	nuclearity	NOUN
dad-10670	173	26	.	.	PUNCT
dad-10670	174	1	we	we	PRON
dad-10670	174	2	can	can	AUX
dad-10670	174	3	now	now	ADV
dad-10670	174	4	remove	remove	VERB
dad-10670	174	5	the	the	DET
dad-10670	174	6	two	two	NUM
dad-10670	174	7	selected	select	VERB
dad-10670	174	8	edus	edus	NOUN
dad-10670	174	9	from	from	ADP
dad-10670	174	10	the	the	DET
dad-10670	174	11	list	list	NOUN
dad-10670	174	12	,	,	PUNCT
dad-10670	174	13	and	and	CCONJ
dad-10670	174	14	replace	replace	VERB
dad-10670	174	15	them	they	PRON
dad-10670	174	16	by	by	ADP
dad-10670	174	17	a	a	DET
dad-10670	174	18	newly	newly	ADV
dad-10670	174	19	created	create	VERB
dad-10670	174	20	rs	rs	NOUN
dad-10670	174	21	-	-	PUNCT
dad-10670	174	22	tree	tree	NOUN
dad-10670	174	23	labeled	label	VERB
dad-10670	174	24	by	by	ADP
dad-10670	174	25	the	the	DET
dad-10670	174	26	estimated	estimate	VERB
dad-10670	174	27	relation	relation	NOUN
dad-10670	174	28	label	label	NOUN
dad-10670	174	29	and	and	CCONJ
dad-10670	174	30	nuclearity	nuclearity	NOUN
dad-10670	174	31	,	,	PUNCT
dad-10670	174	32	whose	whose	DET
dad-10670	174	33	left	left	ADJ
dad-10670	174	34	and	and	CCONJ
dad-10670	174	35	right	right	ADJ
dad-10670	174	36	children	child	NOUN
dad-10670	174	37	are	be	AUX
dad-10670	174	38	the	the	DET
dad-10670	174	39	selected	select	VERB
dad-10670	174	40	edus	edus	NOUN
dad-10670	174	41	.	.	PUNCT
dad-10670	175	1	we	we	PRON
dad-10670	175	2	update	update	VERB
dad-10670	175	3	the	the	DET
dad-10670	175	4	probabilities	probability	NOUN
dad-10670	175	5	of	of	ADP
dad-10670	175	6	structural	structural	ADJ
dad-10670	175	7	relations	relation	NOUN
dad-10670	175	8	holding	hold	VERB
dad-10670	175	9	between	between	ADP
dad-10670	175	10	the	the	DET
dad-10670	175	11	newly	newly	ADV
dad-10670	175	12	created	create	VERB
dad-10670	175	13	tree	tree	NOUN
dad-10670	175	14	and	and	CCONJ
dad-10670	175	15	adjacent	adjacent	ADJ
dad-10670	175	16	elements	element	NOUN
dad-10670	175	17	in	in	ADP
dad-10670	175	18	the	the	DET
dad-10670	175	19	list	list	NOUN
dad-10670	175	20	,	,	PUNCT
dad-10670	175	21	using	use	VERB
dad-10670	175	22	struct	struct	NOUN
dad-10670	175	23	.	.	PUNCT
dad-10670	176	1	subsequently	subsequently	ADV
dad-10670	176	2	,	,	PUNCT
dad-10670	176	3	we	we	PRON
dad-10670	176	4	repeat	repeat	VERB
dad-10670	176	5	the	the	DET
dad-10670	176	6	process	process	NOUN
dad-10670	176	7	until	until	SCONJ
dad-10670	176	8	the	the	DET
dad-10670	176	9	list	list	NOUN
dad-10670	176	10	contains	contain	VERB
dad-10670	176	11	only	only	ADV
dad-10670	176	12	one	one	NUM
dad-10670	176	13	element	element	NOUN
dad-10670	176	14	,	,	PUNCT
dad-10670	176	15	which	which	PRON
dad-10670	176	16	is	be	AUX
dad-10670	176	17	our	our	PRON
dad-10670	176	18	final	final	ADJ
dad-10670	176	19	discourse	discourse	NOUN
dad-10670	176	20	tree	tree	NOUN
dad-10670	176	21	,	,	PUNCT
dad-10670	176	22	built	build	VERB
dad-10670	176	23	bottom	bottom	NOUN
dad-10670	176	24	-	-	PUNCT
dad-10670	176	25	to	to	ADP
dad-10670	176	26	-	-	PUNCT
dad-10670	176	27	top	top	NOUN
dad-10670	176	28	.	.	PUNCT
dad-10670	177	1	in	in	ADP
dad-10670	177	2	doing	do	VERB
dad-10670	177	3	so	so	ADV
dad-10670	177	4	,	,	PUNCT
dad-10670	177	5	we	we	PRON
dad-10670	177	6	have	have	AUX
dad-10670	177	7	adopted	adopt	VERB
dad-10670	177	8	a	a	DET
dad-10670	177	9	‘	'	PUNCT
dad-10670	177	10	greedy	greedy	ADJ
dad-10670	177	11	’	'	PUNCT
dad-10670	177	12	approach	approach	NOUN
dad-10670	177	13	,	,	PUNCT
dad-10670	177	14	i.e.	i.e.	X
dad-10670	177	15	,	,	PUNCT
dad-10670	177	16	at	at	ADP
dad-10670	177	17	each	each	DET
dad-10670	177	18	iteration	iteration	NOUN
dad-10670	177	19	the	the	DET
dad-10670	177	20	two	two	NUM
dad-10670	177	21	spans	span	NOUN
dad-10670	177	22	most	most	ADV
dad-10670	177	23	likely	likely	ADV
dad-10670	177	24	connected	connect	VERB
dad-10670	177	25	by	by	ADP
dad-10670	177	26	a	a	DET
dad-10670	177	27	relation	relation	NOUN
dad-10670	177	28	have	have	AUX
dad-10670	177	29	been	be	AUX
dad-10670	177	30	merged	merge	VERB
dad-10670	177	31	.	.	PUNCT
dad-10670	178	1	hence	hence	ADV
dad-10670	178	2	,	,	PUNCT
dad-10670	178	3	our	our	PRON
dad-10670	178	4	algorithm	algorithm	NOUN
dad-10670	178	5	,	,	PUNCT
dad-10670	178	6	which	which	PRON
dad-10670	178	7	makes	make	VERB
dad-10670	178	8	locally	locally	ADV
dad-10670	178	9	-	-	PUNCT
dad-10670	178	10	optimal	optimal	ADJ
dad-10670	178	11	choices	choice	NOUN
dad-10670	178	12	,	,	PUNCT
dad-10670	178	13	performs	perform	VERB
dad-10670	178	14	in	in	ADP
dad-10670	178	15	time	time	NOUN
dad-10670	178	16	complexity	complexity	NOUN
dad-10670	178	17	of	of	ADP
dad-10670	178	18	o(n	o(n	PROPN
dad-10670	178	19	)	)	PUNCT
dad-10670	178	20	.	.	PUNCT
dad-10670	179	1	3.4	3.4	NUM
dad-10670	179	2	features	feature	VERB
dad-10670	179	3	3.4.1	3.4.1	NUM
dad-10670	179	4	discourse	discourse	NOUN
dad-10670	179	5	segmentation	segmentation	NOUN
dad-10670	179	6	the	the	DET
dad-10670	179	7	first	first	ADJ
dad-10670	179	8	step	step	NOUN
dad-10670	179	9	of	of	ADP
dad-10670	179	10	the	the	DET
dad-10670	179	11	parser	parser	NOUN
dad-10670	179	12	is	be	AUX
dad-10670	179	13	to	to	PART
dad-10670	179	14	segment	segment	VERB
dad-10670	179	15	the	the	DET
dad-10670	179	16	text	text	NOUN
dad-10670	179	17	into	into	ADP
dad-10670	179	18	units	unit	NOUN
dad-10670	179	19	.	.	PUNCT
dad-10670	180	1	for	for	ADP
dad-10670	180	2	this	this	DET
dad-10670	180	3	task	task	NOUN
dad-10670	180	4	,	,	PUNCT
dad-10670	180	5	we	we	PRON
dad-10670	180	6	use	use	VERB
dad-10670	180	7	a	a	DET
dad-10670	180	8	combination	combination	NOUN
dad-10670	180	9	of	of	ADP
dad-10670	180	10	syntactic	syntactic	ADJ
dad-10670	180	11	and	and	CCONJ
dad-10670	180	12	lexical	lexical	ADJ
dad-10670	180	13	features	feature	NOUN
dad-10670	180	14	:	:	PUNCT
dad-10670	180	15	words	word	NOUN
dad-10670	180	16	,	,	PUNCT
dad-10670	180	17	pos	pos	NOUN
dad-10670	180	18	tags	tag	NOUN
dad-10670	180	19	,	,	PUNCT
dad-10670	180	20	and	and	CCONJ
dad-10670	180	21	lexical	lexical	ADJ
dad-10670	180	22	heads	head	NOUN
dad-10670	180	23	.	.	PUNCT
dad-10670	181	1	in	in	ADP
dad-10670	181	2	particular	particular	ADJ
dad-10670	181	3	,	,	PUNCT
dad-10670	181	4	we	we	PRON
dad-10670	181	5	use	use	VERB
dad-10670	181	6	the	the	DET
dad-10670	181	7	lexicosyntactic	lexicosyntactic	ADJ
dad-10670	181	8	features	feature	NOUN
dad-10670	181	9	of	of	ADP
dad-10670	181	10	soricut	soricut	NOUN
dad-10670	181	11	and	and	CCONJ
dad-10670	181	12	marcu	marcu	NOUN
dad-10670	181	13	(	(	PUNCT
dad-10670	181	14	2003	2003	NUM
dad-10670	181	15	)	)	PUNCT
dad-10670	181	16	,	,	PUNCT
dad-10670	181	17	which	which	PRON
dad-10670	181	18	were	be	AUX
dad-10670	181	19	found	find	VERB
dad-10670	181	20	good	good	ADJ
dad-10670	181	21	indicators	indicator	NOUN
dad-10670	181	22	for	for	ADP
dad-10670	181	23	the	the	DET
dad-10670	181	24	presence	presence	NOUN
dad-10670	181	25	of	of	ADP
dad-10670	181	26	edu	edu	PROPN
dad-10670	181	27	boundaries	boundaries	PROPN
dad-10670	181	28	.	.	PUNCT
dad-10670	182	1	figure	figure	NOUN
dad-10670	182	2	9	9	NUM
dad-10670	182	3	shows	show	VERB
dad-10670	182	4	the	the	DET
dad-10670	182	5	parse	parse	NOUN
dad-10670	182	6	tree	tree	NOUN
dad-10670	182	7	of	of	ADP
dad-10670	182	8	the	the	DET
dad-10670	182	9	sentence	sentence	NOUN
dad-10670	182	10	“	"	PUNCT
dad-10670	182	11	farm	farm	NOUN
dad-10670	182	12	lending	lending	NOUN
dad-10670	182	13	was	be	AUX
dad-10670	182	14	enacted	enact	VERB
dad-10670	182	15	to	to	PART
dad-10670	182	16	correct	correct	VERB
dad-10670	182	17	this	this	DET
dad-10670	182	18	problem	problem	NOUN
dad-10670	182	19	by	by	ADP
dad-10670	182	20	providing	provide	VERB
dad-10670	182	21	a	a	DET
dad-10670	182	22	reliable	reliable	ADJ
dad-10670	182	23	flow	flow	NOUN
dad-10670	182	24	of	of	ADP
dad-10670	182	25	lendable	lendable	ADJ
dad-10670	182	26	funds	fund	NOUN
dad-10670	182	27	.	.	PUNCT
dad-10670	182	28	”	"	PUNCT
dad-10670	183	1	(	(	PUNCT
dad-10670	183	2	see	see	VERB
dad-10670	183	3	also	also	ADV
dad-10670	183	4	figure	figure	VERB
dad-10670	183	5	1	1	NUM
dad-10670	183	6	)	)	PUNCT
dad-10670	183	7	.	.	PUNCT
dad-10670	184	1	here	here	ADV
dad-10670	184	2	,	,	PUNCT
dad-10670	184	3	lexical	lexical	ADJ
dad-10670	184	4	heads	head	NOUN
dad-10670	184	5	are	be	AUX
dad-10670	184	6	generated	generate	VERB
dad-10670	184	7	using	use	VERB
dad-10670	184	8	projection	projection	NOUN
dad-10670	184	9	rules	rule	NOUN
dad-10670	184	10	from	from	ADP
dad-10670	184	11	magerman	magerman	NOUN
dad-10670	184	12	(	(	PUNCT
dad-10670	184	13	1995	1995	NUM
dad-10670	184	14	)	)	PUNCT
dad-10670	184	15	and	and	CCONJ
dad-10670	184	16	indicated	indicate	VERB
dad-10670	184	17	between	between	ADP
dad-10670	184	18	brackets	bracket	NOUN
dad-10670	184	19	.	.	PUNCT
dad-10670	185	1	for	for	ADP
dad-10670	185	2	a	a	DET
dad-10670	185	3	word	word	NOUN
dad-10670	185	4	w	w	NOUN
dad-10670	185	5	,	,	PUNCT
dad-10670	185	6	we	we	PRON
dad-10670	185	7	look	look	VERB
dad-10670	185	8	at	at	ADP
dad-10670	185	9	its	its	PRON
dad-10670	185	10	highest	high	ADJ
dad-10670	185	11	ancestor	ancestor	NOUN
dad-10670	185	12	in	in	ADP
dad-10670	185	13	the	the	DET
dad-10670	185	14	parse	parse	NOUN
dad-10670	185	15	tree	tree	NOUN
dad-10670	185	16	with	with	ADP
dad-10670	185	17	a	a	DET
dad-10670	185	18	lexical	lexical	ADJ
dad-10670	185	19	head	head	NOUN
dad-10670	185	20	equal	equal	ADJ
dad-10670	185	21	to	to	ADP
dad-10670	185	22	w	w	PROPN
dad-10670	185	23	,	,	PUNCT
dad-10670	185	24	and	and	CCONJ
dad-10670	185	25	with	with	ADP
dad-10670	185	26	a	a	DET
dad-10670	185	27	right	right	ADV
dad-10670	185	28	-	-	PUNCT
dad-10670	185	29	sibling	sible	VERB
dad-10670	185	30	.	.	PUNCT
dad-10670	186	1	2	2	X
dad-10670	186	2	.	.	X
dad-10670	186	3	note	note	VERB
dad-10670	186	4	that	that	SCONJ
dad-10670	186	5	a	a	DET
dad-10670	186	6	single	single	ADJ
dad-10670	186	7	edu	edu	PROPN
dad-10670	186	8	also	also	ADV
dad-10670	186	9	constitutes	constitute	VERB
dad-10670	186	10	a	a	DET
dad-10670	186	11	valid	valid	ADJ
dad-10670	186	12	rs	rs	NOUN
dad-10670	186	13	-	-	PUNCT
dad-10670	186	14	tree	tree	NOUN
dad-10670	186	15	.	.	PUNCT
dad-10670	187	1	8	8	NUM
dad-10670	187	2	hilda	hilda	VERB
dad-10670	187	3	:	:	PUNCT
dad-10670	187	4	a	a	DET
dad-10670	187	5	discourse	discourse	NOUN
dad-10670	187	6	parser	parser	NOUN
dad-10670	187	7	using	use	VERB
dad-10670	187	8	support	support	NOUN
dad-10670	187	9	vector	vector	NOUN
dad-10670	187	10	machine	machine	NOUN
dad-10670	187	11	classification	classification	NOUN
dad-10670	187	12	require	require	NOUN
dad-10670	187	13	:	:	PUNCT
dad-10670	187	14	e	e	X
dad-10670	187	15	=	=	SYM
dad-10670	187	16	〈	〈	PROPN
dad-10670	187	17	e1	e1	PROPN
dad-10670	187	18	,	,	PUNCT
dad-10670	187	19	e2	e2	PROPN
dad-10670	187	20	.	.	PUNCT
dad-10670	187	21	.	.	PUNCT
dad-10670	187	22	.	.	PUNCT
dad-10670	188	1	〉	〉	NOUN
dad-10670	188	2	,	,	PUNCT
dad-10670	188	3	list	list	NOUN
dad-10670	188	4	of	of	ADP
dad-10670	188	5	all	all	DET
dad-10670	188	6	the	the	DET
dad-10670	188	7	text	text	NOUN
dad-10670	188	8	’s	’s	PART
dad-10670	188	9	edu	edu	PROPN
dad-10670	188	10	ensure	ensure	VERB
dad-10670	188	11	:	:	PUNCT
dad-10670	188	12	finaltree	finaltree	NOUN
dad-10670	188	13	is	be	AUX
dad-10670	188	14	a	a	DET
dad-10670	188	15	valid	valid	ADJ
dad-10670	188	16	rs	rs	NOUN
dad-10670	188	17	-	-	PUNCT
dad-10670	188	18	tree	tree	NOUN
dad-10670	188	19	for	for	ADP
dad-10670	188	20	inputtext	inputtext	ADJ
dad-10670	188	21	l←	l←	X
dad-10670	188	22	e	e	NOUN
dad-10670	188	23	for	for	ADP
dad-10670	188	24	all	all	PRON
dad-10670	188	25	(	(	PUNCT
dad-10670	188	26	li	li	PROPN
dad-10670	188	27	,	,	PUNCT
dad-10670	188	28	li+1	li+1	NOUN
dad-10670	188	29	)	)	PUNCT
dad-10670	188	30	in	in	ADP
dad-10670	188	31	l	l	NOUN
dad-10670	188	32	do	do	AUX
dad-10670	188	33	scores[i]←	scores[i]←	PROPN
dad-10670	188	34	struct(li	struct(li	PROPN
dad-10670	188	35	,	,	PUNCT
dad-10670	188	36	li+1	li+1	PROPN
dad-10670	188	37	)	)	PUNCT
dad-10670	188	38	a	a	DET
dad-10670	188	39	end	end	NOUN
dad-10670	188	40	for	for	ADP
dad-10670	188	41	while	while	SCONJ
dad-10670	188	42	|l|	|l|	VERB
dad-10670	188	43	>	>	X
dad-10670	188	44	1	1	NUM
dad-10670	188	45	do	do	AUX
dad-10670	188	46	i←	i←	ADV
dad-10670	188	47	argmax(scores	argmax(score	NOUN
dad-10670	188	48	)	)	PUNCT
dad-10670	189	1	newlabel←	newlabel←	PROPN
dad-10670	189	2	label(li	label(li	PROPN
dad-10670	189	3	,	,	PUNCT
dad-10670	189	4	li+1	li+1	NOUN
dad-10670	189	5	)	)	PUNCT
dad-10670	189	6	newsubtree←	newsubtree←	PROPN
dad-10670	189	7	createtree(li	createtree(li	PROPN
dad-10670	189	8	,	,	PUNCT
dad-10670	189	9	li+1	li+1	PROPN
dad-10670	189	10	,	,	PUNCT
dad-10670	189	11	newlabel)b	newlabel)b	PUNCT
dad-10670	189	12	scores[i−	scores[i−	PROPN
dad-10670	189	13	1]←	1]←	NUM
dad-10670	189	14	struct(li−1	struct(li−1	PROPN
dad-10670	189	15	,	,	PUNCT
dad-10670	189	16	newsubtree	newsubtree	PROPN
dad-10670	189	17	)	)	PUNCT
dad-10670	189	18	scores[i+	scores[i+	NOUN
dad-10670	189	19	2]←	2]←	NUM
dad-10670	189	20	struct(newsubtree	struct(newsubtree	PROPN
dad-10670	189	21	,	,	PUNCT
dad-10670	189	22	li+2	li+2	X
dad-10670	189	23	)	)	PUNCT
dad-10670	189	24	delete(scores[i	delete(scores[i	ADJ
dad-10670	189	25	]	]	PUNCT
dad-10670	189	26	)	)	PUNCT
dad-10670	189	27	delete(scores[i+	delete(scores[i+	NOUN
dad-10670	189	28	1	1	NUM
dad-10670	189	29	]	]	PUNCT
dad-10670	189	30	)	)	PUNCT
dad-10670	189	31	l←	l←	PUNCT
dad-10670	190	1	[	[	X
dad-10670	190	2	l0	l0	X
dad-10670	190	3	,	,	PUNCT
dad-10670	190	4	.	.	PUNCT
dad-10670	190	5	.	.	PUNCT
dad-10670	191	1	.	.	PUNCT
dad-10670	192	1	,	,	PUNCT
dad-10670	192	2	li−1	li−1	PROPN
dad-10670	192	3	,	,	PUNCT
dad-10670	192	4	newsubtree	newsubtree	NOUN
dad-10670	192	5	,	,	PUNCT
dad-10670	192	6	li+2	li+2	PROPN
dad-10670	192	7	,	,	PUNCT
dad-10670	192	8	.	.	PUNCT
dad-10670	192	9	.	.	PUNCT
dad-10670	192	10	.	.	PUNCT
dad-10670	193	1	]	]	PUNCT
dad-10670	193	2	end	end	VERB
dad-10670	193	3	while	while	SCONJ
dad-10670	193	4	finaltree←	finaltree←	PROPN
dad-10670	193	5	l0	l0	PROPN
dad-10670	193	6	return	return	VERB
dad-10670	193	7	finaltree	finaltree	NOUN
dad-10670	193	8	a.	a.	NOUN
dad-10670	193	9	li	li	PROPN
dad-10670	193	10	denotes	denote	VERB
dad-10670	193	11	the	the	DET
dad-10670	193	12	i	i	PROPN
dad-10670	193	13	-	-	PUNCT
dad-10670	193	14	th	th	X
dad-10670	193	15	element	element	NOUN
dad-10670	193	16	of	of	ADP
dad-10670	193	17	list	list	NOUN
dad-10670	193	18	l.	l.	PROPN
dad-10670	193	19	b.	b.	PROPN
dad-10670	193	20	createtree	createtree	PROPN
dad-10670	193	21	is	be	AUX
dad-10670	193	22	a	a	DET
dad-10670	193	23	function	function	NOUN
dad-10670	193	24	that	that	PRON
dad-10670	193	25	takes	take	VERB
dad-10670	193	26	two	two	NUM
dad-10670	193	27	rs	r	NOUN
dad-10670	193	28	-	-	PUNCT
dad-10670	193	29	trees	tree	NOUN
dad-10670	193	30	t	t	NOUN
dad-10670	193	31	and	and	CCONJ
dad-10670	193	32	t	t	PROPN
dad-10670	193	33	′	′	PROPN
dad-10670	193	34	,	,	PUNCT
dad-10670	193	35	a	a	DET
dad-10670	193	36	relation	relation	NOUN
dad-10670	193	37	r	r	NOUN
dad-10670	193	38	∈	∈	NOUN
dad-10670	193	39	r	r	NOUN
dad-10670	193	40	,	,	PUNCT
dad-10670	193	41	and	and	CCONJ
dad-10670	193	42	returns	return	VERB
dad-10670	193	43	the	the	DET
dad-10670	193	44	rs	rs	NOUN
dad-10670	193	45	-	-	PUNCT
dad-10670	193	46	tree	tree	NOUN
dad-10670	193	47	whose	whose	DET
dad-10670	193	48	left	left	ADJ
dad-10670	193	49	child	child	NOUN
dad-10670	193	50	is	be	AUX
dad-10670	193	51	t	t	PROPN
dad-10670	193	52	,	,	PUNCT
dad-10670	193	53	whose	whose	DET
dad-10670	193	54	right	right	ADJ
dad-10670	193	55	child	child	NOUN
dad-10670	193	56	is	be	AUX
dad-10670	193	57	t	t	NOUN
dad-10670	193	58	′	′	NUM
dad-10670	193	59	,	,	PUNCT
dad-10670	193	60	and	and	CCONJ
dad-10670	193	61	which	which	PRON
dad-10670	193	62	is	be	AUX
dad-10670	193	63	tagged	tag	VERB
dad-10670	193	64	with	with	ADP
dad-10670	193	65	relation	relation	NOUN
dad-10670	193	66	r.	r.	PROPN
dad-10670	193	67	figure	figure	VERB
dad-10670	193	68	6	6	NUM
dad-10670	193	69	:	:	PUNCT
dad-10670	193	70	bottom	bottom	ADJ
dad-10670	193	71	-	-	PUNCT
dad-10670	193	72	up	up	ADP
dad-10670	193	73	tree	tree	NOUN
dad-10670	193	74	construction	construction	NOUN
dad-10670	193	75	this	this	DET
dad-10670	193	76	highest	high	ADJ
dad-10670	193	77	-	-	PUNCT
dad-10670	193	78	ancestor	ancestor	NOUN
dad-10670	193	79	node	node	NOUN
dad-10670	193	80	is	be	AUX
dad-10670	193	81	called	call	VERB
dad-10670	193	82	nw	nw	PROPN
dad-10670	193	83	.	.	PUNCT
dad-10670	194	1	then	then	ADV
dad-10670	194	2	,	,	PUNCT
dad-10670	194	3	we	we	PRON
dad-10670	194	4	call	call	VERB
dad-10670	194	5	its	its	PRON
dad-10670	194	6	parent	parent	NOUN
dad-10670	194	7	np	np	NOUN
dad-10670	194	8	,	,	PUNCT
dad-10670	194	9	and	and	CCONJ
dad-10670	194	10	its	its	PRON
dad-10670	194	11	right	right	ADV
dad-10670	194	12	-	-	PUNCT
dad-10670	194	13	sibling	sible	VERB
dad-10670	194	14	nr	nr	NOUN
dad-10670	194	15	.	.	PROPN
dad-10670	195	1	for	for	ADP
dad-10670	195	2	instance	instance	NOUN
dad-10670	195	3	,	,	PUNCT
dad-10670	195	4	when	when	SCONJ
dad-10670	195	5	applying	apply	VERB
dad-10670	195	6	this	this	DET
dad-10670	195	7	process	process	NOUN
dad-10670	195	8	to	to	ADP
dad-10670	195	9	the	the	DET
dad-10670	195	10	word	word	NOUN
dad-10670	195	11	“	"	PUNCT
dad-10670	195	12	enacted	enact	VERB
dad-10670	195	13	”	"	PUNCT
dad-10670	195	14	,	,	PUNCT
dad-10670	195	15	we	we	PRON
dad-10670	195	16	obtain	obtain	VERB
dad-10670	195	17	nw	nw	X
dad-10670	195	18	=	=	PUNCT
dad-10670	195	19	vbn(enacted	vbn(enacte	VERB
dad-10670	195	20	)	)	PUNCT
dad-10670	195	21	,	,	PUNCT
dad-10670	195	22	np	np	NOUN
dad-10670	195	23	=	=	PUNCT
dad-10670	195	24	vp(enacted	vp(enacted	PROPN
dad-10670	195	25	)	)	PUNCT
dad-10670	195	26	,	,	PUNCT
dad-10670	195	27	nr	nr	NOUN
dad-10670	195	28	=	=	NOUN
dad-10670	195	29	s(correct	s(correct	ADJ
dad-10670	195	30	)	)	PUNCT
dad-10670	195	31	.	.	PUNCT
dad-10670	196	1	we	we	PRON
dad-10670	196	2	can	can	AUX
dad-10670	196	3	now	now	ADV
dad-10670	196	4	define	define	VERB
dad-10670	196	5	the	the	DET
dad-10670	196	6	contextual	contextual	ADJ
dad-10670	196	7	features	feature	NOUN
dad-10670	196	8	of	of	ADP
dad-10670	196	9	the	the	DET
dad-10670	196	10	word	word	NOUN
dad-10670	196	11	at	at	ADP
dad-10670	196	12	position	position	NOUN
dad-10670	196	13	i	i	PRON
dad-10670	196	14	in	in	ADP
dad-10670	196	15	the	the	DET
dad-10670	196	16	text	text	NOUN
dad-10670	196	17	.	.	PUNCT
dad-10670	197	1	it	it	PRON
dad-10670	197	2	is	be	AUX
dad-10670	197	3	the	the	DET
dad-10670	197	4	set	set	NOUN
dad-10670	197	5	composed	compose	VERB
dad-10670	197	6	of	of	ADP
dad-10670	197	7	the	the	DET
dad-10670	197	8	word	word	NOUN
dad-10670	197	9	wi	wi	PROPN
dad-10670	197	10	,	,	PUNCT
dad-10670	197	11	its	its	PRON
dad-10670	197	12	pos	pos	NOUN
dad-10670	197	13	,	,	PUNCT
dad-10670	197	14	as	as	ADV
dad-10670	197	15	well	well	ADV
dad-10670	197	16	as	as	ADP
dad-10670	197	17	the	the	DET
dad-10670	197	18	pos	pos	NOUN
dad-10670	197	19	and	and	CCONJ
dad-10670	197	20	lexical	lexical	ADJ
dad-10670	197	21	heads	head	NOUN
dad-10670	197	22	of	of	ADP
dad-10670	197	23	nwi	nwi	PROPN
dad-10670	197	24	,	,	PUNCT
dad-10670	197	25	npi	npi	PROPN
dad-10670	197	26	,	,	PUNCT
dad-10670	197	27	and	and	CCONJ
dad-10670	197	28	nri	nri	PROPN
dad-10670	197	29	.	.	PUNCT
dad-10670	198	1	next	next	ADV
dad-10670	198	2	,	,	PUNCT
dad-10670	198	3	the	the	DET
dad-10670	198	4	features	feature	NOUN
dad-10670	198	5	for	for	ADP
dad-10670	198	6	position	position	NOUN
dad-10670	198	7	i	i	PRON
dad-10670	198	8	in	in	ADP
dad-10670	198	9	the	the	DET
dad-10670	198	10	text	text	NOUN
dad-10670	198	11	are	be	AUX
dad-10670	198	12	created	create	VERB
dad-10670	198	13	by	by	ADP
dad-10670	198	14	concatenating	concatenate	VERB
dad-10670	198	15	the	the	DET
dad-10670	198	16	contextual	contextual	ADJ
dad-10670	198	17	features	feature	NOUN
dad-10670	198	18	at	at	ADP
dad-10670	198	19	positions	position	NOUN
dad-10670	198	20	i	i	PRON
dad-10670	198	21	−	−	PROPN
dad-10670	198	22	2	2	NUM
dad-10670	198	23	,	,	PUNCT
dad-10670	198	24	i	i	PRON
dad-10670	198	25	−	−	PROPN
dad-10670	198	26	1	1	NUM
dad-10670	198	27	,	,	PUNCT
dad-10670	198	28	and	and	CCONJ
dad-10670	198	29	i.	i.	NOUN
dad-10670	198	30	those	those	DET
dad-10670	198	31	positions	position	NOUN
dad-10670	198	32	were	be	AUX
dad-10670	198	33	determined	determine	VERB
dad-10670	198	34	empirically	empirically	ADV
dad-10670	198	35	to	to	PART
dad-10670	198	36	give	give	VERB
dad-10670	198	37	the	the	DET
dad-10670	198	38	maximum	maximum	ADJ
dad-10670	198	39	contextual	contextual	ADJ
dad-10670	198	40	coverage	coverage	NOUN
dad-10670	198	41	.	.	PUNCT
dad-10670	199	1	3.4.2	3.4.2	NUM
dad-10670	199	2	relation	relation	NOUN
dad-10670	199	3	labeling	labeling	NOUN
dad-10670	199	4	for	for	ADP
dad-10670	199	5	relation	relation	NOUN
dad-10670	199	6	labeling	labeling	NOUN
dad-10670	199	7	,	,	PUNCT
dad-10670	199	8	several	several	ADJ
dad-10670	199	9	shallow	shallow	ADJ
dad-10670	199	10	lexical	lexical	ADJ
dad-10670	199	11	and	and	CCONJ
dad-10670	199	12	syntactic	syntactic	ADJ
dad-10670	199	13	features	feature	NOUN
dad-10670	199	14	are	be	AUX
dad-10670	199	15	taken	take	VERB
dad-10670	199	16	into	into	ADP
dad-10670	199	17	account	account	NOUN
dad-10670	199	18	,	,	PUNCT
dad-10670	199	19	including	include	VERB
dad-10670	199	20	features	feature	NOUN
dad-10670	199	21	of	of	ADP
dad-10670	199	22	textual	textual	ADJ
dad-10670	199	23	organization	organization	NOUN
dad-10670	199	24	,	,	PUNCT
dad-10670	199	25	lexical	lexical	ADJ
dad-10670	199	26	features	feature	NOUN
dad-10670	199	27	,	,	PUNCT
dad-10670	199	28	‘	'	PUNCT
dad-10670	199	29	dominance	dominance	NOUN
dad-10670	199	30	sets	set	NOUN
dad-10670	199	31	’	'	PUNCT
dad-10670	199	32	(	(	PUNCT
dad-10670	199	33	soricut	soricut	VERB
dad-10670	199	34	and	and	CCONJ
dad-10670	199	35	marcu	marcu	NOUN
dad-10670	199	36	,	,	PUNCT
dad-10670	199	37	2003	2003	NUM
dad-10670	199	38	)	)	PUNCT
dad-10670	199	39	,	,	PUNCT
dad-10670	199	40	and	and	CCONJ
dad-10670	199	41	structural	structural	ADJ
dad-10670	199	42	features	feature	NOUN
dad-10670	199	43	.	.	PUNCT
dad-10670	200	1	the	the	DET
dad-10670	200	2	first	first	ADJ
dad-10670	200	3	type	type	NOUN
dad-10670	200	4	of	of	ADP
dad-10670	200	5	features	feature	NOUN
dad-10670	200	6	we	we	PRON
dad-10670	200	7	incorporate	incorporate	VERB
dad-10670	200	8	is	be	AUX
dad-10670	200	9	related	relate	VERB
dad-10670	200	10	to	to	ADP
dad-10670	200	11	textual	textual	ADJ
dad-10670	200	12	organization	organization	NOUN
dad-10670	200	13	.	.	PUNCT
dad-10670	201	1	a	a	DET
dad-10670	201	2	number	number	NOUN
dad-10670	201	3	of	of	ADP
dad-10670	201	4	previous	previous	ADJ
dad-10670	201	5	efforts	effort	NOUN
dad-10670	201	6	(	(	PUNCT
dad-10670	201	7	marcu	marcu	NOUN
dad-10670	201	8	,	,	PUNCT
dad-10670	201	9	1996	1996	NUM
dad-10670	201	10	;	;	PUNCT
dad-10670	201	11	soricut	soricut	VERB
dad-10670	201	12	and	and	CCONJ
dad-10670	201	13	marcu	marcu	NOUN
dad-10670	201	14	,	,	PUNCT
dad-10670	201	15	2003	2003	NUM
dad-10670	201	16	)	)	PUNCT
dad-10670	201	17	have	have	AUX
dad-10670	201	18	shown	show	VERB
dad-10670	201	19	that	that	SCONJ
dad-10670	201	20	there	there	PRON
dad-10670	201	21	is	be	VERB
dad-10670	201	22	a	a	DET
dad-10670	201	23	strong	strong	ADJ
dad-10670	201	24	correlation	correlation	NOUN
dad-10670	201	25	between	between	ADP
dad-10670	201	26	different	different	ADJ
dad-10670	201	27	organizational	organizational	ADJ
dad-10670	201	28	levels	level	NOUN
dad-10670	201	29	of	of	ADP
dad-10670	201	30	textual	textual	ADJ
dad-10670	201	31	units	unit	NOUN
dad-10670	201	32	and	and	CCONJ
dad-10670	201	33	subtrees	subtree	NOUN
dad-10670	201	34	of	of	ADP
dad-10670	201	35	the	the	DET
dad-10670	201	36	rs	rs	NOUN
dad-10670	201	37	-	-	PUNCT
dad-10670	201	38	tree	tree	NOUN
dad-10670	201	39	,	,	PUNCT
dad-10670	201	40	both	both	CCONJ
dad-10670	201	41	at	at	ADP
dad-10670	201	42	the	the	DET
dad-10670	201	43	sentence	sentence	NOUN
dad-10670	201	44	level	level	NOUN
dad-10670	201	45	and	and	CCONJ
dad-10670	201	46	at	at	ADP
dad-10670	201	47	the	the	DET
dad-10670	201	48	paragraph	paragraph	NOUN
dad-10670	201	49	level	level	NOUN
dad-10670	201	50	.	.	PUNCT
dad-10670	202	1	hence	hence	ADV
dad-10670	202	2	,	,	PUNCT
dad-10670	202	3	the	the	DET
dad-10670	202	4	following	follow	VERB
dad-10670	202	5	features	feature	NOUN
dad-10670	202	6	provide	provide	VERB
dad-10670	202	7	valuable	valuable	ADJ
dad-10670	202	8	high	high	ADJ
dad-10670	202	9	-	-	PUNCT
dad-10670	202	10	level	level	NOUN
dad-10670	202	11	cues	cue	NOUN
dad-10670	202	12	for	for	ADP
dad-10670	202	13	the	the	DET
dad-10670	202	14	task	task	NOUN
dad-10670	202	15	of	of	ADP
dad-10670	202	16	scoring	scoring	NOUN
dad-10670	202	17	span	span	NOUN
dad-10670	202	18	relation	relation	NOUN
dad-10670	202	19	priority	priority	NOUN
dad-10670	202	20	(	(	PUNCT
dad-10670	202	21	classifier	classifier	NOUN
dad-10670	202	22	struct	struct	NOUN
dad-10670	202	23	)	)	PUNCT
dad-10670	202	24	.	.	PUNCT
dad-10670	203	1	as	as	SCONJ
dad-10670	203	2	hypothesized	hypothesize	VERB
dad-10670	203	3	by	by	ADP
dad-10670	203	4	reitter	reitter	PROPN
dad-10670	203	5	(	(	PUNCT
dad-10670	203	6	2003b	2003b	NUM
dad-10670	203	7	)	)	PUNCT
dad-10670	203	8	,	,	PUNCT
dad-10670	203	9	there	there	PRON
dad-10670	203	10	is	be	VERB
dad-10670	203	11	a	a	DET
dad-10670	203	12	correlation	correlation	NOUN
dad-10670	203	13	between	between	ADP
dad-10670	203	14	span	span	NOUN
dad-10670	203	15	length	length	NOUN
dad-10670	203	16	and	and	CCONJ
dad-10670	203	17	type	type	NOUN
dad-10670	203	18	of	of	ADP
dad-10670	203	19	rhetorical	rhetorical	ADJ
dad-10670	203	20	relations	relation	NOUN
dad-10670	203	21	.	.	PUNCT
dad-10670	204	1	for	for	ADP
dad-10670	204	2	instance	instance	NOUN
dad-10670	204	3	,	,	PUNCT
dad-10670	204	4	the	the	DET
dad-10670	204	5	satellite	satellite	NOUN
dad-10670	204	6	of	of	ADP
dad-10670	204	7	a	a	DET
dad-10670	204	8	contrast	contrast	NOUN
dad-10670	204	9	relation	relation	NOUN
dad-10670	204	10	tends	tend	VERB
dad-10670	204	11	to	to	PART
dad-10670	204	12	be	be	AUX
dad-10670	204	13	shorter	short	ADJ
dad-10670	204	14	than	than	ADP
dad-10670	204	15	its	its	PRON
dad-10670	204	16	nucleus	nucleus	NOUN
dad-10670	204	17	.	.	PUNCT
dad-10670	205	1	9	9	NUM
dad-10670	205	2	hernault	hernault	ADJ
dad-10670	205	3	,	,	PUNCT
dad-10670	205	4	prendinger	prendinger	NOUN
dad-10670	205	5	,	,	PUNCT
dad-10670	205	6	duverle	duverle	NOUN
dad-10670	205	7	,	,	PUNCT
dad-10670	205	8	and	and	CCONJ
dad-10670	205	9	ishizuka	ishizuka	VERB
dad-10670	205	10	the	the	DET
dad-10670	205	11	organizational	organizational	ADJ
dad-10670	205	12	features	feature	NOUN
dad-10670	205	13	taken	take	VERB
dad-10670	205	14	into	into	ADP
dad-10670	205	15	account	account	NOUN
dad-10670	205	16	in	in	ADP
dad-10670	205	17	our	our	PRON
dad-10670	205	18	system	system	NOUN
dad-10670	205	19	are	be	AUX
dad-10670	205	20	presented	present	VERB
dad-10670	205	21	in	in	ADP
dad-10670	205	22	table	table	NOUN
dad-10670	205	23	1	1	NUM
dad-10670	205	24	.	.	PUNCT
dad-10670	206	1	in	in	ADP
dad-10670	206	2	this	this	DET
dad-10670	206	3	table	table	NOUN
dad-10670	206	4	,	,	PUNCT
dad-10670	206	5	all	all	DET
dad-10670	206	6	subtree	subtree	NOUN
dad-10670	206	7	-	-	PUNCT
dad-10670	206	8	specific	specific	ADJ
dad-10670	206	9	features	feature	NOUN
dad-10670	206	10	,	,	PUNCT
dad-10670	206	11	which	which	PRON
dad-10670	206	12	are	be	AUX
dad-10670	206	13	symmetrically	symmetrically	ADV
dad-10670	206	14	extracted	extract	VERB
dad-10670	206	15	from	from	ADP
dad-10670	206	16	both	both	PRON
dad-10670	206	17	left	left	ADJ
dad-10670	206	18	and	and	CCONJ
dad-10670	206	19	right	right	ADJ
dad-10670	206	20	candidate	candidate	NOUN
dad-10670	206	21	spans	span	NOUN
dad-10670	206	22	,	,	PUNCT
dad-10670	206	23	are	be	AUX
dad-10670	206	24	suffixed	suffix	VERB
dad-10670	206	25	by	by	ADP
dad-10670	206	26	“	"	PUNCT
dad-10670	206	27	s[pan	s[pan	NOUN
dad-10670	206	28	]	]	PUNCT
dad-10670	206	29	”	"	PUNCT
dad-10670	206	30	.	.	PUNCT
dad-10670	207	1	the	the	DET
dad-10670	207	2	other	other	ADJ
dad-10670	207	3	features	feature	NOUN
dad-10670	207	4	,	,	PUNCT
dad-10670	207	5	calculated	calculate	VERB
dad-10670	207	6	as	as	ADP
dad-10670	207	7	a	a	DET
dad-10670	207	8	function	function	NOUN
dad-10670	207	9	of	of	ADP
dad-10670	207	10	the	the	DET
dad-10670	207	11	two	two	NUM
dad-10670	207	12	subtrees	subtree	NOUN
dad-10670	207	13	considered	consider	VERB
dad-10670	207	14	as	as	ADP
dad-10670	207	15	a	a	DET
dad-10670	207	16	pair	pair	NOUN
dad-10670	207	17	,	,	PUNCT
dad-10670	207	18	are	be	AUX
dad-10670	207	19	indicated	indicate	VERB
dad-10670	207	20	by	by	ADP
dad-10670	207	21	“	"	PUNCT
dad-10670	207	22	f[ull	f[ull	NOUN
dad-10670	207	23	]	]	PUNCT
dad-10670	207	24	”	"	PUNCT
dad-10670	207	25	.	.	PUNCT
dad-10670	208	1	table	table	NOUN
dad-10670	208	2	1	1	NUM
dad-10670	208	3	:	:	PUNCT
dad-10670	208	4	features	feature	VERB
dad-10670	208	5	encoding	encode	VERB
dad-10670	208	6	textual	textual	ADJ
dad-10670	208	7	organization	organization	NOUN
dad-10670	208	8	feature	feature	NOUN
dad-10670	208	9	name	name	NOUN
dad-10670	208	10	scope	scope	NOUN
dad-10670	208	11	belong	belong	VERB
dad-10670	208	12	to	to	ADP
dad-10670	208	13	same	same	ADJ
dad-10670	208	14	sentence	sentence	NOUN
dad-10670	208	15	f	f	PROPN
dad-10670	208	16	belong	belong	VERB
dad-10670	208	17	to	to	ADP
dad-10670	208	18	same	same	ADJ
dad-10670	208	19	paragraph	paragraph	NOUN
dad-10670	208	20	f	f	NOUN
dad-10670	208	21	number	number	NOUN
dad-10670	208	22	of	of	ADP
dad-10670	208	23	paragraph	paragraph	NOUN
dad-10670	208	24	boundaries	boundary	NOUN
dad-10670	208	25	s	s	PART
dad-10670	208	26	number	number	NOUN
dad-10670	208	27	of	of	ADP
dad-10670	208	28	sentence	sentence	NOUN
dad-10670	208	29	boundaries	boundary	NOUN
dad-10670	208	30	s	s	PART
dad-10670	208	31	length	length	NOUN
dad-10670	208	32	in	in	ADP
dad-10670	208	33	tokens	token	NOUN
dad-10670	208	34	s	s	PART
dad-10670	208	35	length	length	NOUN
dad-10670	208	36	in	in	ADP
dad-10670	208	37	edus	edus	NOUN
dad-10670	208	38	s	s	PART
dad-10670	208	39	distance	distance	NOUN
dad-10670	208	40	to	to	ADP
dad-10670	208	41	beginning	beginning	NOUN
dad-10670	208	42	of	of	ADP
dad-10670	208	43	sentence	sentence	NOUN
dad-10670	208	44	in	in	ADP
dad-10670	208	45	tokens	token	NOUN
dad-10670	208	46	s	s	PART
dad-10670	208	47	size	size	NOUN
dad-10670	208	48	of	of	ADP
dad-10670	208	49	span	span	NOUN
dad-10670	208	50	over	over	ADP
dad-10670	208	51	sentence	sentence	NOUN
dad-10670	208	52	in	in	ADP
dad-10670	208	53	edus	edus	NOUN
dad-10670	208	54	s	s	PART
dad-10670	208	55	size	size	NOUN
dad-10670	208	56	of	of	ADP
dad-10670	208	57	span	span	NOUN
dad-10670	208	58	over	over	ADP
dad-10670	208	59	sentence	sentence	NOUN
dad-10670	208	60	in	in	ADP
dad-10670	208	61	tokens	token	NOUN
dad-10670	208	62	s	s	PART
dad-10670	208	63	size	size	NOUN
dad-10670	208	64	of	of	ADP
dad-10670	208	65	both	both	DET
dad-10670	208	66	spans	span	NOUN
dad-10670	208	67	over	over	ADP
dad-10670	208	68	sentence	sentence	NOUN
dad-10670	208	69	in	in	ADP
dad-10670	208	70	tokens	tokens	PROPN
dad-10670	208	71	f	f	PROPN
dad-10670	208	72	distance	distance	NOUN
dad-10670	208	73	to	to	ADP
dad-10670	208	74	beginning	beginning	NOUN
dad-10670	208	75	of	of	ADP
dad-10670	208	76	sentence	sentence	NOUN
dad-10670	208	77	in	in	ADP
dad-10670	208	78	edus	edus	NOUN
dad-10670	208	79	s	s	PART
dad-10670	208	80	distance	distance	NOUN
dad-10670	208	81	to	to	ADP
dad-10670	208	82	beginning	beginning	NOUN
dad-10670	208	83	of	of	ADP
dad-10670	208	84	text	text	NOUN
dad-10670	208	85	in	in	ADP
dad-10670	208	86	tokens	token	NOUN
dad-10670	208	87	s	s	PART
dad-10670	208	88	distance	distance	NOUN
dad-10670	208	89	to	to	PART
dad-10670	208	90	end	end	VERB
dad-10670	208	91	of	of	ADP
dad-10670	208	92	sentence	sentence	NOUN
dad-10670	208	93	in	in	ADP
dad-10670	208	94	tokens	token	NOUN
dad-10670	208	95	s	s	PART
dad-10670	208	96	discourse	discourse	NOUN
dad-10670	208	97	cues	cue	NOUN
dad-10670	208	98	,	,	PUNCT
dad-10670	208	99	such	such	ADJ
dad-10670	208	100	as	as	ADP
dad-10670	208	101	because	because	ADV
dad-10670	208	102	,	,	PUNCT
dad-10670	208	103	however	however	ADV
dad-10670	208	104	,	,	PUNCT
dad-10670	208	105	etc	etc	X
dad-10670	208	106	are	be	AUX
dad-10670	208	107	another	another	DET
dad-10670	208	108	type	type	NOUN
dad-10670	208	109	of	of	ADP
dad-10670	208	110	feature	feature	NOUN
dad-10670	208	111	that	that	PRON
dad-10670	208	112	frequently	frequently	ADV
dad-10670	208	113	indicate	indicate	VERB
dad-10670	208	114	the	the	DET
dad-10670	208	115	presence	presence	NOUN
dad-10670	208	116	of	of	ADP
dad-10670	208	117	discourse	discourse	NOUN
dad-10670	208	118	relations	relation	NOUN
dad-10670	208	119	.	.	PUNCT
dad-10670	209	1	instead	instead	ADV
dad-10670	209	2	of	of	ADP
dad-10670	209	3	working	work	VERB
dad-10670	209	4	with	with	ADP
dad-10670	209	5	a	a	DET
dad-10670	209	6	pre	pre	ADJ
dad-10670	209	7	-	-	ADJ
dad-10670	209	8	defined	define	VERB
dad-10670	209	9	dictionary	dictionary	NOUN
dad-10670	209	10	of	of	ADP
dad-10670	209	11	discourse	discourse	NOUN
dad-10670	209	12	cues	cue	NOUN
dad-10670	209	13	,	,	PUNCT
dad-10670	209	14	we	we	PRON
dad-10670	209	15	chose	choose	VERB
dad-10670	209	16	to	to	PART
dad-10670	209	17	build	build	VERB
dad-10670	209	18	an	an	DET
dad-10670	209	19	empirical	empirical	ADJ
dad-10670	209	20	n	n	CCONJ
dad-10670	209	21	-	-	PUNCT
dad-10670	209	22	gram	gram	NOUN
dad-10670	209	23	dictionary	dictionary	NOUN
dad-10670	209	24	from	from	ADP
dad-10670	209	25	the	the	DET
dad-10670	209	26	training	training	NOUN
dad-10670	209	27	corpus	corpus	NOUN
dad-10670	209	28	,	,	PUNCT
dad-10670	209	29	ranked	rank	VERB
dad-10670	209	30	by	by	ADP
dad-10670	209	31	frequency	frequency	NOUN
dad-10670	209	32	,	,	PUNCT
dad-10670	209	33	in	in	ADP
dad-10670	209	34	order	order	NOUN
dad-10670	209	35	to	to	PART
dad-10670	209	36	keep	keep	VERB
dad-10670	209	37	it	it	PRON
dad-10670	209	38	to	to	ADP
dad-10670	209	39	a	a	DET
dad-10670	209	40	reasonable	reasonable	ADJ
dad-10670	209	41	size	size	NOUN
dad-10670	209	42	.	.	PUNCT
dad-10670	210	1	this	this	DET
dad-10670	210	2	method	method	NOUN
dad-10670	210	3	has	have	VERB
dad-10670	210	4	the	the	DET
dad-10670	210	5	advantage	advantage	NOUN
dad-10670	210	6	of	of	ADP
dad-10670	210	7	capturing	capture	VERB
dad-10670	210	8	non	non	ADJ
dad-10670	210	9	-	-	ADJ
dad-10670	210	10	lexical	lexical	ADJ
dad-10670	210	11	signals	signal	NOUN
dad-10670	210	12	,	,	PUNCT
dad-10670	210	13	such	such	ADJ
dad-10670	210	14	as	as	ADP
dad-10670	210	15	punctuation	punctuation	NOUN
dad-10670	210	16	and	and	CCONJ
dad-10670	210	17	paragraph	paragraph	NOUN
dad-10670	210	18	boundaries	boundary	NOUN
dad-10670	210	19	.	.	PUNCT
dad-10670	211	1	we	we	PRON
dad-10670	211	2	encode	encode	VERB
dad-10670	211	3	the	the	DET
dad-10670	211	4	prefix	prefix	NOUN
dad-10670	211	5	and	and	CCONJ
dad-10670	211	6	suffix	suffix	NOUN
dad-10670	211	7	of	of	ADP
dad-10670	211	8	each	each	DET
dad-10670	211	9	span	span	NOUN
dad-10670	211	10	as	as	SCONJ
dad-10670	211	11	ordered	order	VERB
dad-10670	211	12	3	3	NUM
dad-10670	211	13	-	-	PUNCT
dad-10670	211	14	grams	gram	NOUN
dad-10670	211	15	,	,	PUNCT
dad-10670	211	16	which	which	PRON
dad-10670	211	17	were	be	AUX
dad-10670	211	18	found	find	VERB
dad-10670	211	19	to	to	PART
dad-10670	211	20	give	give	VERB
dad-10670	211	21	the	the	DET
dad-10670	211	22	best	good	ADJ
dad-10670	211	23	signal	signal	NOUN
dad-10670	211	24	-	-	PUNCT
dad-10670	211	25	to	to	ADP
dad-10670	211	26	-	-	PUNCT
dad-10670	211	27	noise	noise	NOUN
dad-10670	211	28	ratio	ratio	NOUN
dad-10670	211	29	(	(	PUNCT
dad-10670	211	30	see	see	VERB
dad-10670	211	31	manning	manning	NOUN
dad-10670	211	32	and	and	CCONJ
dad-10670	211	33	schütze	schütze	NOUN
dad-10670	211	34	,	,	PUNCT
dad-10670	211	35	1999	1999	NUM
dad-10670	211	36	)	)	PUNCT
dad-10670	211	37	.	.	PUNCT
dad-10670	212	1	encoding	encode	VERB
dad-10670	212	2	the	the	DET
dad-10670	212	3	beginning	beginning	NOUN
dad-10670	212	4	and	and	CCONJ
dad-10670	212	5	end	end	NOUN
dad-10670	212	6	of	of	ADP
dad-10670	212	7	a	a	DET
dad-10670	212	8	span	span	NOUN
dad-10670	212	9	is	be	AUX
dad-10670	212	10	motivated	motivate	VERB
dad-10670	212	11	by	by	ADP
dad-10670	212	12	the	the	DET
dad-10670	212	13	hypothesis	hypothesis	NOUN
dad-10670	212	14	that	that	PRON
dad-10670	212	15	the	the	DET
dad-10670	212	16	most	most	ADV
dad-10670	212	17	meaningful	meaningful	ADJ
dad-10670	212	18	rhetorical	rhetorical	ADJ
dad-10670	212	19	signals	signal	NOUN
dad-10670	212	20	are	be	AUX
dad-10670	212	21	found	find	VERB
dad-10670	212	22	at	at	ADP
dad-10670	212	23	the	the	DET
dad-10670	212	24	edges	edge	NOUN
dad-10670	212	25	of	of	ADP
dad-10670	212	26	the	the	DET
dad-10670	212	27	span	span	NOUN
dad-10670	212	28	(	(	PUNCT
dad-10670	212	29	schilder	schilder	PROPN
dad-10670	212	30	,	,	PUNCT
dad-10670	212	31	2002	2002	NUM
dad-10670	212	32	)	)	PUNCT
dad-10670	212	33	.	.	PUNCT
dad-10670	213	1	for	for	ADP
dad-10670	213	2	instance	instance	NOUN
dad-10670	213	3	,	,	PUNCT
dad-10670	213	4	for	for	ADP
dad-10670	213	5	the	the	DET
dad-10670	213	6	span	span	NOUN
dad-10670	213	7	in	in	ADP
dad-10670	213	8	figure	figure	NOUN
dad-10670	213	9	3	3	NUM
dad-10670	213	10	consisting	consist	VERB
dad-10670	213	11	of	of	ADP
dad-10670	213	12	units	unit	NOUN
dad-10670	213	13	1b	1b	NUM
dad-10670	213	14	and	and	CCONJ
dad-10670	213	15	1c	1c	NOUN
dad-10670	213	16	connected	connect	VERB
dad-10670	213	17	by	by	ADP
dad-10670	213	18	a	a	DET
dad-10670	213	19	manner	manner	NOUN
dad-10670	213	20	-	-	PUNCT
dad-10670	213	21	means	means	NOUN
dad-10670	213	22	relation	relation	NOUN
dad-10670	213	23	,	,	PUNCT
dad-10670	213	24	the	the	DET
dad-10670	213	25	beginning	beginning	NOUN
dad-10670	213	26	and	and	CCONJ
dad-10670	213	27	end	end	NOUN
dad-10670	213	28	-	-	PUNCT
dad-10670	213	29	of	of	ADP
dad-10670	213	30	-	-	PUNCT
dad-10670	213	31	span	span	NOUN
dad-10670	213	32	3	3	NUM
dad-10670	213	33	-	-	PUNCT
dad-10670	213	34	grams	gram	NOUN
dad-10670	213	35	we	we	PRON
dad-10670	213	36	encode	encode	VERB
dad-10670	213	37	in	in	ADP
dad-10670	213	38	our	our	PRON
dad-10670	213	39	features	feature	NOUN
dad-10670	213	40	(	(	PUNCT
dad-10670	213	41	to	to	PART
dad-10670	213	42	,	,	PUNCT
dad-10670	213	43	correct	correct	ADJ
dad-10670	213	44	,	,	PUNCT
dad-10670	213	45	this	this	PRON
dad-10670	213	46	)	)	PUNCT
dad-10670	213	47	and	and	CCONJ
dad-10670	213	48	(	(	PUNCT
dad-10670	213	49	of	of	ADP
dad-10670	213	50	,	,	PUNCT
dad-10670	213	51	lendable	lendable	ADJ
dad-10670	213	52	,	,	PUNCT
dad-10670	213	53	funds	fund	NOUN
dad-10670	213	54	)	)	PUNCT
dad-10670	213	55	,	,	PUNCT
dad-10670	213	56	respectively	respectively	ADV
dad-10670	213	57	.	.	PUNCT
dad-10670	214	1	in	in	ADP
dad-10670	214	2	total	total	NOUN
dad-10670	214	3	,	,	PUNCT
dad-10670	214	4	our	our	PRON
dad-10670	214	5	3	3	NUM
dad-10670	214	6	-	-	PUNCT
dad-10670	214	7	gram	gram	NOUN
dad-10670	214	8	dictionary	dictionary	NOUN
dad-10670	214	9	is	be	AUX
dad-10670	214	10	of	of	ADP
dad-10670	214	11	size	size	NOUN
dad-10670	214	12	12,000	12,000	NUM
dad-10670	214	13	,	,	PUNCT
dad-10670	214	14	and	and	CCONJ
dad-10670	214	15	encoding	encode	VERB
dad-10670	214	16	prefix	prefix	NOUN
dad-10670	214	17	and	and	CCONJ
dad-10670	214	18	suffix	suffix	NOUN
dad-10670	214	19	raise	raise	VERB
dad-10670	214	20	the	the	DET
dad-10670	214	21	encoding	encoding	NOUN
dad-10670	214	22	size	size	NOUN
dad-10670	214	23	to	to	ADP
dad-10670	214	24	22	22	NUM
dad-10670	214	25	×	×	NOUN
dad-10670	214	26	|3	|3	NOUN
dad-10670	214	27	-	-	PUNCT
dad-10670	214	28	gram	gram	NOUN
dad-10670	214	29	dictionary|	dictionary|	NUM
dad-10670	214	30	=	=	SYM
dad-10670	214	31	48	48	NUM
dad-10670	214	32	,	,	PUNCT
dad-10670	214	33	000	000	NUM
dad-10670	214	34	.	.	PUNCT
dad-10670	215	1	this	this	DET
dad-10670	215	2	approach	approach	NOUN
dad-10670	215	3	was	be	AUX
dad-10670	215	4	validated	validate	VERB
dad-10670	215	5	by	by	ADP
dad-10670	215	6	comparing	compare	VERB
dad-10670	215	7	it	it	PRON
dad-10670	215	8	to	to	ADP
dad-10670	215	9	results	result	NOUN
dad-10670	215	10	obtained	obtain	VERB
dad-10670	215	11	from	from	ADP
dad-10670	215	12	using	use	VERB
dad-10670	215	13	a	a	DET
dad-10670	215	14	fixed	fix	VERB
dad-10670	215	15	dictionary	dictionary	NOUN
dad-10670	215	16	of	of	ADP
dad-10670	215	17	300	300	NUM
dad-10670	215	18	cue	cue	NOUN
dad-10670	215	19	-	-	PUNCT
dad-10670	215	20	phrases	phrase	NOUN
dad-10670	215	21	instead	instead	ADV
dad-10670	215	22	(	(	PUNCT
dad-10670	215	23	oberlander	oberlander	NOUN
dad-10670	215	24	and	and	CCONJ
dad-10670	215	25	moore	moore	PROPN
dad-10670	215	26	,	,	PUNCT
dad-10670	215	27	1999	1999	NUM
dad-10670	215	28	)	)	PUNCT
dad-10670	215	29	.	.	PUNCT
dad-10670	216	1	performance	performance	NOUN
dad-10670	216	2	was	be	AUX
dad-10670	216	3	found	find	VERB
dad-10670	216	4	to	to	PART
dad-10670	216	5	be	be	AUX
dad-10670	216	6	lower	low	ADJ
dad-10670	216	7	when	when	SCONJ
dad-10670	216	8	using	use	VERB
dad-10670	216	9	only	only	ADV
dad-10670	216	10	the	the	DET
dad-10670	216	11	cue	cue	ADJ
dad-10670	216	12	-	-	PUNCT
dad-10670	216	13	phrase	phrase	NOUN
dad-10670	216	14	dictionary	dictionary	NOUN
dad-10670	216	15	,	,	PUNCT
dad-10670	216	16	and	and	CCONJ
dad-10670	216	17	marginally	marginally	ADV
dad-10670	216	18	lower	low	ADJ
dad-10670	216	19	when	when	SCONJ
dad-10670	216	20	using	use	VERB
dad-10670	216	21	both	both	PRON
dad-10670	216	22	together	together	ADV
dad-10670	216	23	.	.	PUNCT
dad-10670	217	1	to	to	PART
dad-10670	217	2	improve	improve	VERB
dad-10670	217	3	discourse	discourse	NOUN
dad-10670	217	4	cue	cue	NOUN
dad-10670	217	5	detection	detection	NOUN
dad-10670	217	6	and	and	CCONJ
dad-10670	217	7	to	to	PART
dad-10670	217	8	make	make	VERB
dad-10670	217	9	it	it	PRON
dad-10670	217	10	less	less	ADV
dad-10670	217	11	dependent	dependent	ADJ
dad-10670	217	12	on	on	ADP
dad-10670	217	13	lexical	lexical	ADJ
dad-10670	217	14	information	information	NOUN
dad-10670	217	15	,	,	PUNCT
dad-10670	217	16	we	we	PRON
dad-10670	217	17	complement	complement	VERB
dad-10670	217	18	it	it	PRON
dad-10670	217	19	with	with	ADP
dad-10670	217	20	shallow	shallow	ADJ
dad-10670	217	21	syntactic	syntactic	ADJ
dad-10670	217	22	information	information	NOUN
dad-10670	217	23	,	,	PUNCT
dad-10670	217	24	by	by	ADP
dad-10670	217	25	encoding	encode	VERB
dad-10670	217	26	pos	pos	NOUN
dad-10670	217	27	tags	tag	NOUN
dad-10670	217	28	for	for	ADP
dad-10670	217	29	the	the	DET
dad-10670	217	30	prefix	prefix	NOUN
dad-10670	217	31	and	and	CCONJ
dad-10670	217	32	suffix	suffix	NOUN
dad-10670	217	33	of	of	ADP
dad-10670	217	34	each	each	DET
dad-10670	217	35	span	span	NOUN
dad-10670	217	36	.	.	PUNCT
dad-10670	218	1	the	the	DET
dad-10670	218	2	penn	penn	PROPN
dad-10670	218	3	treebank	treebank	PROPN
dad-10670	218	4	contains	contain	VERB
dad-10670	218	5	npos	npos	NOUN
dad-10670	218	6	=	=	NUM
dad-10670	218	7	384	384	NUM
dad-10670	218	8	tags	tag	NOUN
dad-10670	218	9	,	,	PUNCT
dad-10670	218	10	hence	hence	ADV
dad-10670	218	11	it	it	PRON
dad-10670	218	12	requires	require	VERB
dad-10670	218	13	2	2	NUM
dad-10670	218	14	×	×	NOUN
dad-10670	218	15	3	3	NUM
dad-10670	218	16	×	×	NOUN
dad-10670	218	17	npos	npos	NOUN
dad-10670	218	18	=	=	NOUN
dad-10670	218	19	2304	2304	NUM
dad-10670	218	20	additional	additional	ADJ
dad-10670	218	21	dimensions	dimension	NOUN
dad-10670	218	22	.	.	PUNCT
dad-10670	219	1	to	to	PART
dad-10670	219	2	reflect	reflect	VERB
dad-10670	219	3	the	the	DET
dad-10670	219	4	way	way	NOUN
dad-10670	219	5	relations	relation	NOUN
dad-10670	219	6	attach	attach	VERB
dad-10670	219	7	in	in	ADP
dad-10670	219	8	rs	r	NOUN
dad-10670	219	9	-	-	PUNCT
dad-10670	219	10	trees	tree	NOUN
dad-10670	219	11	,	,	PUNCT
dad-10670	219	12	we	we	PRON
dad-10670	219	13	use	use	VERB
dad-10670	219	14	the	the	DET
dad-10670	219	15	notion	notion	NOUN
dad-10670	219	16	of	of	ADP
dad-10670	219	17	dominance	dominance	NOUN
dad-10670	219	18	sets	set	NOUN
dad-10670	219	19	as	as	SCONJ
dad-10670	219	20	defined	define	VERB
dad-10670	219	21	in	in	ADP
dad-10670	219	22	soricut	soricut	NOUN
dad-10670	219	23	and	and	CCONJ
dad-10670	219	24	marcu	marcu	NOUN
dad-10670	219	25	(	(	PUNCT
dad-10670	219	26	2003	2003	NUM
dad-10670	219	27	)	)	PUNCT
dad-10670	219	28	.	.	PUNCT
dad-10670	220	1	a	a	DET
dad-10670	220	2	dominance	dominance	NOUN
dad-10670	220	3	set	set	NOUN
dad-10670	220	4	is	be	AUX
dad-10670	220	5	formed	form	VERB
dad-10670	220	6	by	by	ADP
dad-10670	220	7	consecutive	consecutive	ADJ
dad-10670	220	8	edus	edus	NOUN
dad-10670	220	9	of	of	ADP
dad-10670	220	10	a	a	DET
dad-10670	220	11	sentence	sentence	NOUN
dad-10670	220	12	that	that	PRON
dad-10670	220	13	are	be	AUX
dad-10670	220	14	connected	connect	VERB
dad-10670	220	15	in	in	ADP
dad-10670	220	16	the	the	DET
dad-10670	220	17	lexicalized	lexicalize	VERB
dad-10670	220	18	syntax	syntax	NOUN
dad-10670	220	19	tree	tree	NOUN
dad-10670	220	20	by	by	ADP
dad-10670	220	21	a	a	DET
dad-10670	220	22	head	head	NOUN
dad-10670	220	23	node	node	NOUN
dad-10670	220	24	(	(	PUNCT
dad-10670	220	25	the	the	DET
dad-10670	220	26	dominating	dominating	NOUN
dad-10670	220	27	node	node	NOUN
dad-10670	220	28	)	)	PUNCT
dad-10670	220	29	and	and	CCONJ
dad-10670	220	30	an	an	DET
dad-10670	220	31	attachment	attachment	NOUN
dad-10670	220	32	node	node	NOUN
dad-10670	220	33	(	(	PUNCT
dad-10670	220	34	the	the	DET
dad-10670	220	35	dominated	dominate	VERB
dad-10670	220	36	node	node	NOUN
dad-10670	220	37	)	)	PUNCT
dad-10670	220	38	.	.	PUNCT
dad-10670	221	1	for	for	ADP
dad-10670	221	2	instance	instance	NOUN
dad-10670	221	3	,	,	PUNCT
dad-10670	221	4	the	the	DET
dad-10670	221	5	sentence	sentence	NOUN
dad-10670	221	6	of	of	ADP
dad-10670	221	7	figure	figure	NOUN
dad-10670	221	8	1	1	NUM
dad-10670	221	9	contains	contain	VERB
dad-10670	221	10	three	three	NUM
dad-10670	221	11	edus	edus	NOUN
dad-10670	221	12	,	,	PUNCT
dad-10670	221	13	which	which	PRON
dad-10670	221	14	10	10	NUM
dad-10670	221	15	hilda	hilda	VERB
dad-10670	221	16	:	:	PUNCT
dad-10670	221	17	a	a	DET
dad-10670	221	18	discourse	discourse	NOUN
dad-10670	221	19	parser	parser	NOUN
dad-10670	221	20	using	use	VERB
dad-10670	221	21	support	support	NOUN
dad-10670	221	22	vector	vector	NOUN
dad-10670	221	23	machine	machine	NOUN
dad-10670	221	24	classification	classification	NOUN
dad-10670	221	25	are	be	AUX
dad-10670	221	26	connected	connect	VERB
dad-10670	221	27	by	by	ADP
dad-10670	221	28	two	two	NUM
dad-10670	221	29	rhetorical	rhetorical	ADJ
dad-10670	221	30	relations	relation	NOUN
dad-10670	221	31	,	,	PUNCT
dad-10670	221	32	enablement	enablement	NOUN
dad-10670	221	33	and	and	CCONJ
dad-10670	221	34	manner	manner	NOUN
dad-10670	221	35	-	-	PUNCT
dad-10670	221	36	means	mean	NOUN
dad-10670	221	37	.	.	PUNCT
dad-10670	222	1	this	this	PRON
dad-10670	222	2	leads	lead	VERB
dad-10670	222	3	to	to	ADP
dad-10670	222	4	two	two	NUM
dad-10670	222	5	possible	possible	ADJ
dad-10670	222	6	tree	tree	NOUN
dad-10670	222	7	constructions	construction	NOUN
dad-10670	222	8	(	(	PUNCT
dad-10670	222	9	see	see	VERB
dad-10670	222	10	figure	figure	NOUN
dad-10670	222	11	8)	8)	NUM
dad-10670	222	12	,	,	PUNCT
dad-10670	222	13	in	in	ADP
dad-10670	222	14	which	which	PRON
dad-10670	222	15	edu	edu	PROPN
dad-10670	222	16	1b	1b	PROPN
dad-10670	222	17	is	be	AUX
dad-10670	222	18	connected	connect	VERB
dad-10670	222	19	either	either	ADV
dad-10670	222	20	to	to	ADP
dad-10670	222	21	1a	1a	PROPN
dad-10670	222	22	or	or	CCONJ
dad-10670	222	23	to	to	ADP
dad-10670	222	24	1c	1c	NUM
dad-10670	222	25	.	.	PUNCT
dad-10670	223	1	in	in	ADP
dad-10670	223	2	the	the	DET
dad-10670	223	3	left	left	ADJ
dad-10670	223	4	-	-	PUNCT
dad-10670	223	5	hand	hand	NOUN
dad-10670	223	6	construction	construction	NOUN
dad-10670	223	7	,	,	PUNCT
dad-10670	223	8	1a	1a	PROPN
dad-10670	223	9	and	and	CCONJ
dad-10670	223	10	1b	1b	NUM
dad-10670	223	11	are	be	AUX
dad-10670	223	12	connected	connect	VERB
dad-10670	223	13	first	first	ADV
dad-10670	223	14	,	,	PUNCT
dad-10670	223	15	then	then	ADV
dad-10670	223	16	the	the	DET
dad-10670	223	17	subtree	subtree	NOUN
dad-10670	223	18	they	they	PRON
dad-10670	223	19	span	span	VERB
dad-10670	223	20	is	be	AUX
dad-10670	223	21	connected	connect	VERB
dad-10670	223	22	to	to	ADP
dad-10670	223	23	1c	1c	NUM
dad-10670	223	24	.	.	PUNCT
dad-10670	224	1	in	in	ADP
dad-10670	224	2	the	the	DET
dad-10670	224	3	right	right	ADJ
dad-10670	224	4	-	-	PUNCT
dad-10670	224	5	hand	hand	NOUN
dad-10670	224	6	construction	construction	NOUN
dad-10670	224	7	,	,	PUNCT
dad-10670	224	8	1b	1b	NUM
dad-10670	224	9	and	and	CCONJ
dad-10670	224	10	1c	1c	NOUN
dad-10670	224	11	are	be	AUX
dad-10670	224	12	connected	connect	VERB
dad-10670	224	13	first	first	ADV
dad-10670	224	14	,	,	PUNCT
dad-10670	224	15	then	then	ADV
dad-10670	224	16	the	the	DET
dad-10670	224	17	subtree	subtree	NOUN
dad-10670	224	18	they	they	PRON
dad-10670	224	19	span	span	VERB
dad-10670	224	20	is	be	AUX
dad-10670	224	21	connected	connect	VERB
dad-10670	224	22	to	to	ADP
dad-10670	224	23	1a	1a	NUM
dad-10670	224	24	.	.	PUNCT
dad-10670	225	1	it	it	PRON
dad-10670	225	2	is	be	AUX
dad-10670	225	3	possible	possible	ADJ
dad-10670	225	4	to	to	PART
dad-10670	225	5	infer	infer	VERB
dad-10670	225	6	the	the	DET
dad-10670	225	7	correct	correct	ADJ
dad-10670	225	8	logical	logical	ADJ
dad-10670	225	9	nesting	nesting	ADJ
dad-10670	225	10	order	order	NOUN
dad-10670	225	11	by	by	ADP
dad-10670	225	12	studying	study	VERB
dad-10670	225	13	the	the	DET
dad-10670	225	14	associated	associated	ADJ
dad-10670	225	15	syntax	syntax	NOUN
dad-10670	225	16	tree	tree	NOUN
dad-10670	225	17	,	,	PUNCT
dad-10670	225	18	and	and	CCONJ
dad-10670	225	19	observing	observe	VERB
dad-10670	225	20	the	the	DET
dad-10670	225	21	sub	sub	NOUN
dad-10670	225	22	-	-	NOUN
dad-10670	225	23	trees	tree	NOUN
dad-10670	225	24	spanned	span	VERB
dad-10670	225	25	by	by	ADP
dad-10670	225	26	each	each	DET
dad-10670	225	27	edu	edu	PROPN
dad-10670	225	28	.	.	PUNCT
dad-10670	226	1	in	in	ADP
dad-10670	226	2	figure	figure	NOUN
dad-10670	226	3	9	9	NUM
dad-10670	226	4	,	,	PUNCT
dad-10670	226	5	looking	look	VERB
dad-10670	226	6	at	at	ADP
dad-10670	226	7	the	the	DET
dad-10670	226	8	frontiers	frontier	NOUN
dad-10670	226	9	between	between	ADP
dad-10670	226	10	dominating	dominating	NOUN
dad-10670	226	11	nodes	node	NOUN
dad-10670	226	12	(	(	PUNCT
dad-10670	226	13	diamond	diamond	NOUN
dad-10670	226	14	-	-	PUNCT
dad-10670	226	15	shaped	shape	VERB
dad-10670	226	16	)	)	PUNCT
dad-10670	226	17	and	and	CCONJ
dad-10670	226	18	dominated	dominate	VERB
dad-10670	226	19	nodes	node	NOUN
dad-10670	226	20	(	(	PUNCT
dad-10670	226	21	oval	oval	NOUN
dad-10670	226	22	-	-	PUNCT
dad-10670	226	23	shaped	shape	VERB
dad-10670	226	24	)	)	PUNCT
dad-10670	226	25	indicates	indicate	VERB
dad-10670	226	26	the	the	DET
dad-10670	226	27	correct	correct	ADJ
dad-10670	226	28	dominance	dominance	NOUN
dad-10670	226	29	order	order	NOUN
dad-10670	226	30	,	,	PUNCT
dad-10670	226	31	1a	1a	X
dad-10670	226	32	>	>	X
dad-10670	226	33	1b	1b	PROPN
dad-10670	226	34	>	>	X
dad-10670	226	35	1c	1c	NOUN
dad-10670	226	36	.	.	PUNCT
dad-10670	227	1	thus	thus	ADV
dad-10670	227	2	,	,	PUNCT
dad-10670	227	3	1b	1b	NUM
dad-10670	227	4	should	should	AUX
dad-10670	227	5	be	be	AUX
dad-10670	227	6	attached	attach	VERB
dad-10670	227	7	to	to	ADP
dad-10670	227	8	1c	1c	NUM
dad-10670	227	9	,	,	PUNCT
dad-10670	227	10	and	and	CCONJ
dad-10670	227	11	the	the	DET
dad-10670	227	12	correct	correct	ADJ
dad-10670	227	13	tree	tree	NOUN
dad-10670	227	14	in	in	ADP
dad-10670	227	15	figure	figure	NOUN
dad-10670	227	16	8	8	NUM
dad-10670	227	17	is	be	AUX
dad-10670	227	18	the	the	DET
dad-10670	227	19	right	right	ADJ
dad-10670	227	20	-	-	PUNCT
dad-10670	227	21	hand	hand	NOUN
dad-10670	227	22	one	one	NUM
dad-10670	227	23	.	.	PUNCT
dad-10670	228	1	soricut	soricut	VERB
dad-10670	228	2	and	and	CCONJ
dad-10670	228	3	marcu	marcu	NOUN
dad-10670	228	4	(	(	PUNCT
dad-10670	228	5	2003	2003	NUM
dad-10670	228	6	)	)	PUNCT
dad-10670	228	7	also	also	ADV
dad-10670	228	8	note	note	VERB
dad-10670	228	9	that	that	SCONJ
dad-10670	228	10	only	only	ADV
dad-10670	228	11	taking	take	VERB
dad-10670	228	12	the	the	DET
dad-10670	228	13	pos	pos	NOUN
dad-10670	228	14	tags	tag	NOUN
dad-10670	228	15	of	of	ADP
dad-10670	228	16	the	the	DET
dad-10670	228	17	dominance	dominance	NOUN
dad-10670	228	18	nodes	node	NOUN
dad-10670	228	19	into	into	ADP
dad-10670	228	20	account	account	NOUN
dad-10670	228	21	is	be	AUX
dad-10670	228	22	too	too	ADV
dad-10670	228	23	general	general	ADJ
dad-10670	228	24	.	.	PUNCT
dad-10670	229	1	then	then	ADV
dad-10670	229	2	,	,	PUNCT
dad-10670	229	3	in	in	ADP
dad-10670	229	4	order	order	NOUN
dad-10670	229	5	to	to	PART
dad-10670	229	6	augment	augment	VERB
dad-10670	229	7	information	information	NOUN
dad-10670	229	8	about	about	ADP
dad-10670	229	9	the	the	DET
dad-10670	229	10	lexico	lexico	NOUN
dad-10670	229	11	-	-	PUNCT
dad-10670	229	12	syntactic	syntactic	ADJ
dad-10670	229	13	context	context	NOUN
dad-10670	229	14	,	,	PUNCT
dad-10670	229	15	lexical	lexical	ADJ
dad-10670	229	16	heads	head	NOUN
dad-10670	229	17	of	of	ADP
dad-10670	229	18	those	those	DET
dad-10670	229	19	nodes	node	NOUN
dad-10670	229	20	are	be	AUX
dad-10670	229	21	also	also	ADV
dad-10670	229	22	taken	take	VERB
dad-10670	229	23	into	into	ADP
dad-10670	229	24	account	account	NOUN
dad-10670	229	25	.	.	PUNCT
dad-10670	230	1	for	for	ADP
dad-10670	230	2	approximating	approximate	VERB
dad-10670	230	3	soricut	soricut	NOUN
dad-10670	230	4	and	and	CCONJ
dad-10670	230	5	marcu	marcu	NOUN
dad-10670	230	6	’s	’s	PART
dad-10670	230	7	definition	definition	NOUN
dad-10670	230	8	of	of	ADP
dad-10670	230	9	dominance	dominance	NOUN
dad-10670	230	10	sets	set	NOUN
dad-10670	230	11	,	,	PUNCT
dad-10670	230	12	we	we	PRON
dad-10670	230	13	incorporate	incorporate	VERB
dad-10670	230	14	the	the	DET
dad-10670	230	15	lexicosyntactic	lexicosyntactic	ADJ
dad-10670	230	16	and	and	CCONJ
dad-10670	230	17	structural	structural	ADJ
dad-10670	230	18	features	feature	NOUN
dad-10670	230	19	presented	present	VERB
dad-10670	230	20	in	in	ADP
dad-10670	230	21	table	table	NOUN
dad-10670	230	22	2	2	NUM
dad-10670	230	23	into	into	ADP
dad-10670	230	24	our	our	PRON
dad-10670	230	25	model	model	NOUN
dad-10670	230	26	.	.	PUNCT
dad-10670	231	1	it	it	PRON
dad-10670	231	2	is	be	AUX
dad-10670	231	3	worth	worth	ADJ
dad-10670	231	4	noting	note	VERB
dad-10670	231	5	that	that	SCONJ
dad-10670	231	6	most	most	ADJ
dad-10670	231	7	of	of	ADP
dad-10670	231	8	these	these	DET
dad-10670	231	9	features	feature	NOUN
dad-10670	231	10	apply	apply	VERB
dad-10670	231	11	only	only	ADV
dad-10670	231	12	when	when	SCONJ
dad-10670	231	13	both	both	DET
dad-10670	231	14	spans	span	NOUN
dad-10670	231	15	belong	belong	VERB
dad-10670	231	16	to	to	ADP
dad-10670	231	17	the	the	DET
dad-10670	231	18	same	same	ADJ
dad-10670	231	19	sentence	sentence	NOUN
dad-10670	231	20	.	.	PUNCT
dad-10670	232	1	otherwise	otherwise	ADV
dad-10670	232	2	,	,	PUNCT
dad-10670	232	3	the	the	DET
dad-10670	232	4	parse	parse	NOUN
dad-10670	232	5	trees	tree	NOUN
dad-10670	232	6	of	of	ADP
dad-10670	232	7	spans	span	NOUN
dad-10670	232	8	are	be	AUX
dad-10670	232	9	disjunct	disjunct	ADJ
dad-10670	232	10	,	,	PUNCT
dad-10670	232	11	and	and	CCONJ
dad-10670	232	12	the	the	DET
dad-10670	232	13	notion	notion	NOUN
dad-10670	232	14	of	of	ADP
dad-10670	232	15	dominance	dominance	NOUN
dad-10670	232	16	does	do	AUX
dad-10670	232	17	not	not	PART
dad-10670	232	18	apply	apply	VERB
dad-10670	232	19	.	.	PUNCT
dad-10670	233	1	in	in	ADP
dad-10670	233	2	this	this	DET
dad-10670	233	3	case	case	NOUN
dad-10670	233	4	,	,	PUNCT
dad-10670	233	5	the	the	DET
dad-10670	233	6	features	feature	NOUN
dad-10670	233	7	receive	receive	VERB
dad-10670	233	8	default	default	NOUN
dad-10670	233	9	values	value	NOUN
dad-10670	233	10	during	during	ADP
dad-10670	233	11	feature	feature	NOUN
dad-10670	233	12	extraction	extraction	NOUN
dad-10670	233	13	.	.	PUNCT
dad-10670	234	1	table	table	NOUN
dad-10670	234	2	2	2	NUM
dad-10670	234	3	:	:	PUNCT
dad-10670	234	4	features	feature	VERB
dad-10670	234	5	encoding	encode	VERB
dad-10670	234	6	dominance	dominance	NOUN
dad-10670	234	7	sets	set	VERB
dad-10670	234	8	feature	feature	NOUN
dad-10670	234	9	name	name	NOUN
dad-10670	234	10	scope	scope	NOUN
dad-10670	234	11	distance	distance	NOUN
dad-10670	234	12	to	to	PART
dad-10670	234	13	root	root	VERB
dad-10670	234	14	of	of	ADP
dad-10670	234	15	the	the	DET
dad-10670	234	16	syntax	syntax	NOUN
dad-10670	234	17	tree	tree	NOUN
dad-10670	234	18	s	s	PART
dad-10670	234	19	distance	distance	NOUN
dad-10670	234	20	to	to	ADP
dad-10670	234	21	common	common	ADJ
dad-10670	234	22	ancestor	ancestor	NOUN
dad-10670	234	23	in	in	ADP
dad-10670	234	24	the	the	DET
dad-10670	234	25	syntax	syntax	NOUN
dad-10670	234	26	tree	tree	NOUN
dad-10670	234	27	s	s	PART
dad-10670	234	28	delta	delta	NOUN
dad-10670	234	29	of	of	ADP
dad-10670	234	30	distances	distance	NOUN
dad-10670	234	31	to	to	ADP
dad-10670	234	32	common	common	ADJ
dad-10670	234	33	ancestor	ancestor	NOUN
dad-10670	234	34	f	f	PROPN
dad-10670	234	35	dominating	dominate	VERB
dad-10670	234	36	node	node	NOUN
dad-10670	234	37	’s	’s	PART
dad-10670	234	38	lexical	lexical	ADJ
dad-10670	234	39	head	head	NOUN
dad-10670	234	40	in	in	ADP
dad-10670	234	41	span	span	NOUN
dad-10670	234	42	s	s	PART
dad-10670	234	43	common	common	ADJ
dad-10670	234	44	ancestor	ancestor	NOUN
dad-10670	234	45	’s	’s	PART
dad-10670	234	46	pos	pos	NOUN
dad-10670	234	47	tag	tag	NOUN
dad-10670	234	48	f	f	PROPN
dad-10670	234	49	common	common	ADJ
dad-10670	234	50	ancestor	ancestor	NOUN
dad-10670	234	51	’s	’s	PART
dad-10670	234	52	lexical	lexical	ADJ
dad-10670	234	53	head	head	NOUN
dad-10670	234	54	f	f	PROPN
dad-10670	234	55	dominating	dominate	VERB
dad-10670	234	56	node	node	NOUN
dad-10670	234	57	’s	’s	PART
dad-10670	234	58	pos	pos	NOUN
dad-10670	234	59	tag	tag	NOUN
dad-10670	234	60	f	f	PROPN
dad-10670	234	61	dominating	dominate	VERB
dad-10670	234	62	node	node	NOUN
dad-10670	234	63	’s	’s	PART
dad-10670	234	64	lexical	lexical	ADJ
dad-10670	234	65	head	head	NOUN
dad-10670	234	66	f	f	PROPN
dad-10670	234	67	dominated	dominate	VERB
dad-10670	234	68	node	node	PROPN
dad-10670	234	69	’s	’s	PART
dad-10670	234	70	pos	pos	NOUN
dad-10670	234	71	tag	tag	NOUN
dad-10670	234	72	f	f	PROPN
dad-10670	234	73	dominated	dominate	VERB
dad-10670	234	74	node	node	PROPN
dad-10670	234	75	’s	’s	PART
dad-10670	234	76	lexical	lexical	ADJ
dad-10670	234	77	head	head	NOUN
dad-10670	234	78	f	f	PROPN
dad-10670	234	79	dominated	dominate	VERB
dad-10670	234	80	node	node	PROPN
dad-10670	234	81	’s	’s	PART
dad-10670	234	82	sibling	sible	VERB
dad-10670	234	83	’s	’s	PART
dad-10670	234	84	pos	pos	NOUN
dad-10670	234	85	tag	tag	NOUN
dad-10670	234	86	f	f	PROPN
dad-10670	234	87	dominated	dominate	VERB
dad-10670	234	88	node	node	PROPN
dad-10670	234	89	’s	’s	PART
dad-10670	234	90	sibling	sible	VERB
dad-10670	234	91	’s	’s	PART
dad-10670	234	92	lexical	lexical	ADJ
dad-10670	234	93	head	head	NOUN
dad-10670	234	94	f	f	PROPN
dad-10670	234	95	relative	relative	ADJ
dad-10670	234	96	position	position	NOUN
dad-10670	234	97	of	of	ADP
dad-10670	234	98	lexical	lexical	ADJ
dad-10670	234	99	head	head	NOUN
dad-10670	234	100	in	in	ADP
dad-10670	234	101	sentence	sentence	NOUN
dad-10670	234	102	s	s	VERB
dad-10670	234	103	the	the	DET
dad-10670	234	104	strong	strong	ADJ
dad-10670	234	105	compositionality	compositionality	NOUN
dad-10670	234	106	criterion	criterion	NOUN
dad-10670	234	107	was	be	AUX
dad-10670	234	108	found	find	VERB
dad-10670	234	109	by	by	ADP
dad-10670	234	110	marcu	marcu	NOUN
dad-10670	234	111	through	through	ADP
dad-10670	234	112	empirical	empirical	ADJ
dad-10670	234	113	analysis	analysis	NOUN
dad-10670	234	114	of	of	ADP
dad-10670	234	115	large	large	ADJ
dad-10670	234	116	rst	rst	ADJ
dad-10670	234	117	trees	tree	NOUN
dad-10670	234	118	:	:	PUNCT
dad-10670	234	119	“	"	PUNCT
dad-10670	234	120	[	[	X
dad-10670	234	121	.	.	PUNCT
dad-10670	234	122	.	.	PUNCT
dad-10670	234	123	.	.	PUNCT
dad-10670	235	1	]	]	PUNCT
dad-10670	236	1	whenever	whenever	SCONJ
dad-10670	236	2	two	two	NUM
dad-10670	236	3	large	large	ADJ
dad-10670	236	4	text	text	NOUN
dad-10670	236	5	spans	span	NOUN
dad-10670	236	6	are	be	AUX
dad-10670	236	7	connected	connect	VERB
dad-10670	236	8	through	through	ADP
dad-10670	236	9	a	a	DET
dad-10670	236	10	rhetorical	rhetorical	ADJ
dad-10670	236	11	relation	relation	NOUN
dad-10670	236	12	,	,	PUNCT
dad-10670	236	13	that	that	SCONJ
dad-10670	236	14	rhetorical	rhetorical	ADJ
dad-10670	236	15	relation	relation	NOUN
dad-10670	236	16	holds	hold	VERB
dad-10670	236	17	also	also	ADV
dad-10670	236	18	between	between	ADP
dad-10670	236	19	the	the	DET
dad-10670	236	20	most	most	ADV
dad-10670	236	21	important	important	ADJ
dad-10670	236	22	parts	part	NOUN
dad-10670	236	23	(	(	PUNCT
dad-10670	236	24	i.e.	i.e.	X
dad-10670	236	25	the	the	DET
dad-10670	236	26	nuclei	nucleus	NOUN
dad-10670	236	27	)	)	PUNCT
dad-10670	236	28	of	of	ADP
dad-10670	236	29	the	the	DET
dad-10670	236	30	constituent	constituent	NOUN
dad-10670	236	31	spans	span	NOUN
dad-10670	236	32	.	.	PUNCT
dad-10670	236	33	”	"	PUNCT
dad-10670	237	1	(	(	PUNCT
dad-10670	237	2	see	see	VERB
dad-10670	237	3	marcu	marcu	NOUN
dad-10670	237	4	(	(	PUNCT
dad-10670	237	5	1996	1996	NUM
dad-10670	237	6	,	,	PUNCT
dad-10670	237	7	2	2	NUM
dad-10670	237	8	)	)	PUNCT
dad-10670	237	9	)	)	PUNCT
dad-10670	237	10	.	.	PUNCT
dad-10670	238	1	this	this	DET
dad-10670	238	2	criterion	criterion	NOUN
dad-10670	238	3	,	,	PUNCT
dad-10670	238	4	he	he	PRON
dad-10670	238	5	argues	argue	VERB
dad-10670	238	6	,	,	PUNCT
dad-10670	238	7	provides	provide	VERB
dad-10670	238	8	a	a	DET
dad-10670	238	9	good	good	ADJ
dad-10670	238	10	indicator	indicator	NOUN
dad-10670	238	11	of	of	ADP
dad-10670	238	12	validity	validity	NOUN
dad-10670	238	13	in	in	ADP
dad-10670	238	14	the	the	DET
dad-10670	238	15	case	case	NOUN
dad-10670	238	16	of	of	ADP
dad-10670	238	17	text	text	NOUN
dad-10670	238	18	level	level	NOUN
dad-10670	238	19	trees	tree	NOUN
dad-10670	238	20	where	where	SCONJ
dad-10670	238	21	relations	relation	NOUN
dad-10670	238	22	can	can	AUX
dad-10670	238	23	connect	connect	VERB
dad-10670	238	24	large	large	ADJ
dad-10670	238	25	spans	span	NOUN
dad-10670	238	26	of	of	ADP
dad-10670	238	27	text	text	NOUN
dad-10670	238	28	beyond	beyond	ADP
dad-10670	238	29	the	the	DET
dad-10670	238	30	sentence	sentence	NOUN
dad-10670	238	31	level	level	NOUN
dad-10670	238	32	.	.	PUNCT
dad-10670	239	1	the	the	DET
dad-10670	239	2	idea	idea	NOUN
dad-10670	239	3	of	of	ADP
dad-10670	239	4	strong	strong	ADJ
dad-10670	239	5	compositionality	compositionality	NOUN
dad-10670	239	6	is	be	AUX
dad-10670	239	7	implemented	implement	VERB
dad-10670	239	8	by	by	ADP
dad-10670	239	9	replicating	replicate	VERB
dad-10670	239	10	the	the	DET
dad-10670	239	11	set	set	NOUN
dad-10670	239	12	of	of	ADP
dad-10670	239	13	lexical	lexical	ADJ
dad-10670	239	14	and	and	CCONJ
dad-10670	239	15	syntactic	syntactic	ADJ
dad-10670	239	16	features	feature	NOUN
dad-10670	239	17	of	of	ADP
dad-10670	239	18	table	table	NOUN
dad-10670	239	19	2	2	NUM
dad-10670	239	20	,	,	PUNCT
dad-10670	239	21	which	which	PRON
dad-10670	239	22	are	be	AUX
dad-10670	239	23	extracted	extract	VERB
dad-10670	239	24	from	from	ADP
dad-10670	239	25	the	the	DET
dad-10670	239	26	‘	'	PUNCT
dad-10670	239	27	most	most	ADV
dad-10670	239	28	important	important	ADJ
dad-10670	239	29	’	'	PUNCT
dad-10670	239	30	edu	edu	PROPN
dad-10670	239	31	,	,	PUNCT
dad-10670	239	32	called	call	VERB
dad-10670	239	33	main	main	ADJ
dad-10670	239	34	constituent	constituent	NOUN
dad-10670	239	35	,	,	PUNCT
dad-10670	239	36	of	of	ADP
dad-10670	239	37	each	each	DET
dad-10670	239	38	span	span	NOUN
dad-10670	239	39	.	.	PUNCT
dad-10670	240	1	the	the	DET
dad-10670	240	2	main	main	ADJ
dad-10670	240	3	constituent	constituent	NOUN
dad-10670	240	4	of	of	ADP
dad-10670	240	5	the	the	DET
dad-10670	240	6	left	left	ADJ
dad-10670	240	7	span	span	NOUN
dad-10670	240	8	is	be	AUX
dad-10670	240	9	found	find	VERB
dad-10670	240	10	by	by	ADP
dad-10670	240	11	recursively	recursively	ADV
dad-10670	240	12	selecting	select	VERB
dad-10670	240	13	the	the	DET
dad-10670	240	14	right	right	ADV
dad-10670	240	15	-	-	PUNCT
dad-10670	240	16	most	most	ADV
dad-10670	240	17	nucleus	nucleus	NOUN
dad-10670	240	18	,	,	PUNCT
dad-10670	240	19	starting	start	VERB
dad-10670	240	20	from	from	ADP
dad-10670	240	21	the	the	DET
dad-10670	240	22	top	top	ADJ
dad-10670	240	23	relation	relation	NOUN
dad-10670	240	24	.	.	PUNCT
dad-10670	241	1	for	for	ADP
dad-10670	241	2	the	the	DET
dad-10670	241	3	right	right	ADJ
dad-10670	241	4	span	span	NOUN
dad-10670	241	5	,	,	PUNCT
dad-10670	241	6	the	the	DET
dad-10670	241	7	left	leave	VERB
dad-10670	241	8	-	-	PUNCT
dad-10670	241	9	most	most	ADJ
dad-10670	241	10	nucleus	nucleus	NOUN
dad-10670	241	11	is	be	AUX
dad-10670	241	12	selected	select	VERB
dad-10670	241	13	instead	instead	ADV
dad-10670	241	14	.	.	PUNCT
dad-10670	242	1	we	we	PRON
dad-10670	242	2	chose	choose	VERB
dad-10670	242	3	the	the	DET
dad-10670	242	4	leftmost	leftmost	ADJ
dad-10670	242	5	and	and	CCONJ
dad-10670	242	6	rightmost	rightmost	PROPN
dad-10670	242	7	nuclei	nucleus	NOUN
dad-10670	242	8	,	,	PUNCT
dad-10670	242	9	respectively	respectively	ADV
dad-10670	242	10	,	,	PUNCT
dad-10670	242	11	so	so	SCONJ
dad-10670	242	12	that	that	SCONJ
dad-10670	242	13	the	the	DET
dad-10670	242	14	edus	edus	NOUN
dad-10670	242	15	considered	consider	VERB
dad-10670	242	16	are	be	AUX
dad-10670	242	17	as	as	ADV
dad-10670	242	18	close	close	ADJ
dad-10670	242	19	11	11	NUM
dad-10670	242	20	hernault	hernault	NOUN
dad-10670	242	21	,	,	PUNCT
dad-10670	242	22	prendinger	prendinger	NOUN
dad-10670	242	23	,	,	PUNCT
dad-10670	242	24	duverle	duverle	NOUN
dad-10670	242	25	,	,	PUNCT
dad-10670	242	26	and	and	CCONJ
dad-10670	242	27	ishizuka	ishizuka	VERB
dad-10670	242	28	as	as	ADP
dad-10670	242	29	possible	possible	ADJ
dad-10670	242	30	to	to	ADP
dad-10670	242	31	each	each	DET
dad-10670	242	32	other	other	ADJ
dad-10670	242	33	.	.	PUNCT
dad-10670	243	1	marcu	marcu	NOUN
dad-10670	243	2	(	(	PUNCT
dad-10670	243	3	1996	1996	NUM
dad-10670	243	4	)	)	PUNCT
dad-10670	243	5	recommends	recommend	VERB
dad-10670	243	6	selecting	select	VERB
dad-10670	243	7	the	the	DET
dad-10670	243	8	union	union	NOUN
dad-10670	243	9	of	of	ADP
dad-10670	243	10	all	all	DET
dad-10670	243	11	nuclei	nucleus	NOUN
dad-10670	243	12	in	in	ADP
dad-10670	243	13	each	each	DET
dad-10670	243	14	span	span	NOUN
dad-10670	243	15	.	.	PUNCT
dad-10670	244	1	however	however	ADV
dad-10670	244	2	,	,	PUNCT
dad-10670	244	3	the	the	DET
dad-10670	244	4	restriction	restriction	NOUN
dad-10670	244	5	we	we	PRON
dad-10670	244	6	impose	impose	VERB
dad-10670	244	7	is	be	AUX
dad-10670	244	8	necessary	necessary	ADJ
dad-10670	244	9	to	to	PART
dad-10670	244	10	accommodate	accommodate	VERB
dad-10670	244	11	the	the	DET
dad-10670	244	12	fixed	fix	VERB
dad-10670	244	13	length	length	NOUN
dad-10670	244	14	nature	nature	NOUN
dad-10670	244	15	of	of	ADP
dad-10670	244	16	our	our	PRON
dad-10670	244	17	feature	feature	NOUN
dad-10670	244	18	vectors	vector	NOUN
dad-10670	244	19	.	.	PUNCT
dad-10670	245	1	finally	finally	ADV
dad-10670	245	2	,	,	PUNCT
dad-10670	245	3	to	to	PART
dad-10670	245	4	guide	guide	VERB
dad-10670	245	5	decisions	decision	NOUN
dad-10670	245	6	regarding	regard	VERB
dad-10670	245	7	high	high	ADJ
dad-10670	245	8	-	-	PUNCT
dad-10670	245	9	level	level	NOUN
dad-10670	245	10	relation	relation	NOUN
dad-10670	245	11	classification	classification	NOUN
dad-10670	245	12	on	on	ADP
dad-10670	245	13	large	large	ADJ
dad-10670	245	14	spans	span	NOUN
dad-10670	245	15	,	,	PUNCT
dad-10670	245	16	we	we	PRON
dad-10670	245	17	encode	encode	VERB
dad-10670	245	18	the	the	DET
dad-10670	245	19	rhetorical	rhetorical	ADJ
dad-10670	245	20	structure	structure	NOUN
dad-10670	245	21	of	of	ADP
dad-10670	245	22	a	a	DET
dad-10670	245	23	span	span	NOUN
dad-10670	245	24	into	into	ADP
dad-10670	245	25	the	the	DET
dad-10670	245	26	feature	feature	NOUN
dad-10670	245	27	vector	vector	NOUN
dad-10670	245	28	.	.	PUNCT
dad-10670	246	1	we	we	PRON
dad-10670	246	2	(	(	PUNCT
dad-10670	246	3	informally	informally	ADV
dad-10670	246	4	)	)	PUNCT
dad-10670	246	5	observed	observe	VERB
dad-10670	246	6	some	some	DET
dad-10670	246	7	level	level	NOUN
dad-10670	246	8	of	of	ADP
dad-10670	246	9	correlation	correlation	NOUN
dad-10670	246	10	between	between	ADP
dad-10670	246	11	relations	relation	NOUN
dad-10670	246	12	at	at	ADP
dad-10670	246	13	different	different	ADJ
dad-10670	246	14	levels	level	NOUN
dad-10670	246	15	of	of	ADP
dad-10670	246	16	the	the	DET
dad-10670	246	17	tree	tree	NOUN
dad-10670	246	18	throughout	throughout	ADP
dad-10670	246	19	the	the	DET
dad-10670	246	20	corpus	corpus	NOUN
dad-10670	246	21	.	.	PUNCT
dad-10670	247	1	this	this	PRON
dad-10670	247	2	is	be	AUX
dad-10670	247	3	trivially	trivially	ADV
dad-10670	247	4	the	the	DET
dad-10670	247	5	case	case	NOUN
dad-10670	247	6	for	for	ADP
dad-10670	247	7	n	n	CCONJ
dad-10670	247	8	-	-	PUNCT
dad-10670	247	9	ary	ary	PROPN
dad-10670	247	10	relations	relation	NOUN
dad-10670	247	11	such	such	ADJ
dad-10670	247	12	list	list	NOUN
dad-10670	247	13	which	which	PRON
dad-10670	247	14	have	have	AUX
dad-10670	247	15	been	be	AUX
dad-10670	247	16	binarized	binarize	VERB
dad-10670	247	17	in	in	ADP
dad-10670	247	18	our	our	PRON
dad-10670	247	19	representation	representation	NOUN
dad-10670	247	20	;	;	PUNCT
dad-10670	247	21	the	the	DET
dad-10670	247	22	presence	presence	NOUN
dad-10670	247	23	of	of	ADP
dad-10670	247	24	several	several	ADJ
dad-10670	247	25	list	list	NOUN
dad-10670	247	26	relations	relation	NOUN
dad-10670	247	27	in	in	ADP
dad-10670	247	28	right	right	ADV
dad-10670	247	29	-	-	PUNCT
dad-10670	247	30	most	most	ADJ
dad-10670	247	31	nodes	node	NOUN
dad-10670	247	32	of	of	ADP
dad-10670	247	33	a	a	DET
dad-10670	247	34	subtree	subtree	NOUN
dad-10670	247	35	greatly	greatly	ADV
dad-10670	247	36	increases	increase	VERB
dad-10670	247	37	the	the	DET
dad-10670	247	38	chance	chance	NOUN
dad-10670	247	39	that	that	SCONJ
dad-10670	247	40	the	the	DET
dad-10670	247	41	parent	parent	NOUN
dad-10670	247	42	relation	relation	NOUN
dad-10670	247	43	might	might	AUX
dad-10670	247	44	be	be	AUX
dad-10670	247	45	a	a	DET
dad-10670	247	46	list	list	NOUN
dad-10670	247	47	itself	itself	PRON
dad-10670	247	48	.	.	PUNCT
dad-10670	248	1	we	we	PRON
dad-10670	248	2	encode	encode	VERB
dad-10670	248	3	this	this	DET
dad-10670	248	4	structure	structure	NOUN
dad-10670	248	5	as	as	ADP
dad-10670	248	6	a	a	DET
dad-10670	248	7	breadth	breadth	NOUN
dad-10670	248	8	-	-	PUNCT
dad-10670	248	9	first	first	ADJ
dad-10670	248	10	,	,	PUNCT
dad-10670	248	11	flat	flat	ADJ
dad-10670	248	12	list	list	NOUN
dad-10670	248	13	representation	representation	NOUN
dad-10670	248	14	of	of	ADP
dad-10670	248	15	the	the	DET
dad-10670	248	16	binary	binary	ADJ
dad-10670	248	17	tree	tree	NOUN
dad-10670	248	18	,	,	PUNCT
dad-10670	248	19	whereby	whereby	SCONJ
dad-10670	248	20	each	each	DET
dad-10670	248	21	element	element	NOUN
dad-10670	248	22	of	of	ADP
dad-10670	248	23	the	the	DET
dad-10670	248	24	list	list	NOUN
dad-10670	248	25	is	be	AUX
dad-10670	248	26	split	split	VERB
dad-10670	248	27	over	over	ADP
dad-10670	248	28	one	one	NUM
dad-10670	248	29	of	of	ADP
dad-10670	248	30	the	the	DET
dad-10670	248	31	41	41	NUM
dad-10670	248	32	potential	potential	ADJ
dad-10670	248	33	class	class	NOUN
dad-10670	248	34	labels	label	NOUN
dad-10670	248	35	.	.	PUNCT
dad-10670	249	1	as	as	SCONJ
dad-10670	249	2	the	the	DET
dad-10670	249	3	feature	feature	NOUN
dad-10670	249	4	vectors	vector	NOUN
dad-10670	249	5	have	have	VERB
dad-10670	249	6	a	a	DET
dad-10670	249	7	fixed	fix	VERB
dad-10670	249	8	length	length	NOUN
dad-10670	249	9	,	,	PUNCT
dad-10670	249	10	we	we	PRON
dad-10670	249	11	set	set	VERB
dad-10670	249	12	an	an	DET
dad-10670	249	13	arbitrary	arbitrary	ADJ
dad-10670	249	14	limitation	limitation	NOUN
dad-10670	249	15	on	on	ADP
dad-10670	249	16	the	the	DET
dad-10670	249	17	depth	depth	NOUN
dad-10670	249	18	encoded	encode	VERB
dad-10670	249	19	.	.	PUNCT
dad-10670	250	1	this	this	PRON
dad-10670	250	2	increases	increase	VERB
dad-10670	250	3	the	the	DET
dad-10670	250	4	encoding	encoding	NOUN
dad-10670	250	5	size	size	NOUN
dad-10670	250	6	by	by	ADP
dad-10670	250	7	2×|r|×2d	2×|r|×2d	NUM
dad-10670	250	8	=	=	SYM
dad-10670	250	9	41×2d+1	41×2d+1	NUM
dad-10670	250	10	,	,	PUNCT
dad-10670	250	11	where	where	SCONJ
dad-10670	250	12	d	d	NOUN
dad-10670	250	13	is	be	AUX
dad-10670	250	14	the	the	DET
dad-10670	250	15	maximal	maximal	ADJ
dad-10670	250	16	depth	depth	NOUN
dad-10670	250	17	considered	consider	VERB
dad-10670	250	18	.	.	PUNCT
dad-10670	251	1	experiments	experiment	NOUN
dad-10670	251	2	have	have	AUX
dad-10670	251	3	shown	show	VERB
dad-10670	251	4	optimal	optimal	ADJ
dad-10670	251	5	performances	performance	NOUN
dad-10670	251	6	for	for	ADP
dad-10670	251	7	2	2	NUM
dad-10670	251	8	≤	≤	NUM
dad-10670	251	9	d	d	NOUN
dad-10670	251	10	≤	≤	NUM
dad-10670	251	11	3	3	NUM
dad-10670	251	12	(	(	PUNCT
dad-10670	251	13	we	we	PRON
dad-10670	251	14	selected	select	VERB
dad-10670	251	15	d	d	NOUN
dad-10670	251	16	=	=	NOUN
dad-10670	251	17	3	3	NUM
dad-10670	251	18	)	)	PUNCT
dad-10670	251	19	with	with	ADP
dad-10670	251	20	sharp	sharp	ADJ
dad-10670	251	21	decreases	decrease	NOUN
dad-10670	251	22	for	for	ADP
dad-10670	251	23	values	value	NOUN
dad-10670	251	24	of	of	ADP
dad-10670	251	25	d	d	PROPN
dad-10670	251	26	>	>	X
dad-10670	251	27	4	4	NUM
dad-10670	251	28	.	.	PUNCT
dad-10670	252	1	this	this	DET
dad-10670	252	2	result	result	NOUN
dad-10670	252	3	can	can	AUX
dad-10670	252	4	be	be	AUX
dad-10670	252	5	explained	explain	VERB
dad-10670	252	6	by	by	ADP
dad-10670	252	7	the	the	DET
dad-10670	252	8	excessively	excessively	ADV
dad-10670	252	9	noisy	noisy	ADJ
dad-10670	252	10	data	datum	NOUN
dad-10670	252	11	brought	bring	VERB
dad-10670	252	12	about	about	ADP
dad-10670	252	13	by	by	ADP
dad-10670	252	14	an	an	DET
dad-10670	252	15	exponentially	exponentially	ADV
dad-10670	252	16	growing	grow	VERB
dad-10670	252	17	set	set	NOUN
dad-10670	252	18	of	of	ADP
dad-10670	252	19	features	feature	NOUN
dad-10670	252	20	.	.	PUNCT
dad-10670	253	1	4	4	X
dad-10670	253	2	.	.	X
dad-10670	253	3	experiments	experiment	NOUN
dad-10670	253	4	4.1	4.1	NUM
dad-10670	253	5	discourse	discourse	NOUN
dad-10670	253	6	segmentation	segmentation	NOUN
dad-10670	253	7	we	we	PRON
dad-10670	253	8	first	first	ADV
dad-10670	253	9	measure	measure	VERB
dad-10670	253	10	the	the	DET
dad-10670	253	11	performance	performance	NOUN
dad-10670	253	12	of	of	ADP
dad-10670	253	13	the	the	DET
dad-10670	253	14	seg	seg	PROPN
dad-10670	253	15	classifier	classifier	NOUN
dad-10670	253	16	.	.	PUNCT
dad-10670	254	1	after	after	ADP
dad-10670	254	2	feature	feature	NOUN
dad-10670	254	3	extraction	extraction	NOUN
dad-10670	254	4	from	from	ADP
dad-10670	254	5	the	the	DET
dad-10670	254	6	341	341	NUM
dad-10670	254	7	training	training	NOUN
dad-10670	254	8	documents	document	NOUN
dad-10670	254	9	and	and	CCONJ
dad-10670	254	10	38	38	NUM
dad-10670	254	11	test	test	NOUN
dad-10670	254	12	documents	document	NOUN
dad-10670	254	13	of	of	ADP
dad-10670	254	14	the	the	DET
dad-10670	254	15	rst	rst	NOUN
dad-10670	254	16	-	-	PUNCT
dad-10670	254	17	dt	dt	NOUN
dad-10670	254	18	,	,	PUNCT
dad-10670	254	19	we	we	PRON
dad-10670	254	20	obtain	obtain	VERB
dad-10670	254	21	177,633	177,633	NUM
dad-10670	254	22	training	training	NOUN
dad-10670	254	23	vectors	vector	NOUN
dad-10670	254	24	and	and	CCONJ
dad-10670	254	25	21,667	21,667	NUM
dad-10670	254	26	test	test	NOUN
dad-10670	254	27	vectors	vector	NOUN
dad-10670	254	28	for	for	ADP
dad-10670	254	29	this	this	DET
dad-10670	254	30	task	task	NOUN
dad-10670	254	31	.	.	PUNCT
dad-10670	255	1	because	because	SCONJ
dad-10670	255	2	a	a	DET
dad-10670	255	3	sentence	sentence	NOUN
dad-10670	255	4	typically	typically	ADV
dad-10670	255	5	contains	contain	VERB
dad-10670	255	6	few	few	ADJ
dad-10670	255	7	edu	edu	PROPN
dad-10670	255	8	boundaries	boundary	NOUN
dad-10670	255	9	,	,	PUNCT
dad-10670	255	10	our	our	PRON
dad-10670	255	11	dataset	dataset	NOUN
dad-10670	255	12	is	be	AUX
dad-10670	255	13	skewed	skew	VERB
dad-10670	255	14	,	,	PUNCT
dad-10670	255	15	with	with	ADP
dad-10670	255	16	approximately	approximately	ADV
dad-10670	255	17	89	89	NUM
dad-10670	255	18	%	%	NOUN
dad-10670	255	19	negative	negative	ADJ
dad-10670	255	20	training	training	NOUN
dad-10670	255	21	examples	example	NOUN
dad-10670	255	22	and	and	CCONJ
dad-10670	255	23	11	11	NUM
dad-10670	255	24	%	%	NOUN
dad-10670	255	25	positive	positive	ADJ
dad-10670	255	26	examples	example	NOUN
dad-10670	255	27	.	.	PUNCT
dad-10670	256	1	our	our	PRON
dad-10670	256	2	segmenter	segmenter	NOUN
dad-10670	256	3	is	be	AUX
dad-10670	256	4	evaluated	evaluate	VERB
dad-10670	256	5	with	with	ADP
dad-10670	256	6	respect	respect	NOUN
dad-10670	256	7	to	to	ADP
dad-10670	256	8	two	two	NUM
dad-10670	256	9	types	type	NOUN
dad-10670	256	10	of	of	ADP
dad-10670	256	11	competing	compete	VERB
dad-10670	256	12	systems	system	NOUN
dad-10670	256	13	.	.	PUNCT
dad-10670	257	1	first	first	ADV
dad-10670	257	2	,	,	PUNCT
dad-10670	257	3	we	we	PRON
dad-10670	257	4	measure	measure	VERB
dad-10670	257	5	the	the	DET
dad-10670	257	6	segmentation	segmentation	NOUN
dad-10670	257	7	result	result	NOUN
dad-10670	257	8	when	when	SCONJ
dad-10670	257	9	using	use	VERB
dad-10670	257	10	parse	parse	NOUN
dad-10670	257	11	trees	tree	NOUN
dad-10670	257	12	from	from	ADP
dad-10670	257	13	the	the	DET
dad-10670	257	14	penn	penn	PROPN
dad-10670	257	15	treebank	treebank	PROPN
dad-10670	257	16	(	(	PUNCT
dad-10670	257	17	marcus	marcus	PROPN
dad-10670	257	18	et	et	PROPN
dad-10670	257	19	al	al	PROPN
dad-10670	257	20	.	.	PROPN
dad-10670	257	21	,	,	PUNCT
dad-10670	257	22	1993	1993	NUM
dad-10670	257	23	)	)	PUNCT
dad-10670	257	24	as	as	ADP
dad-10670	257	25	our	our	PRON
dad-10670	257	26	gold	gold	NOUN
dad-10670	257	27	standard	standard	NOUN
dad-10670	257	28	.	.	PUNCT
dad-10670	258	1	second	second	ADJ
dad-10670	258	2	,	,	PUNCT
dad-10670	258	3	as	as	ADP
dad-10670	258	4	a	a	DET
dad-10670	258	5	practical	practical	ADJ
dad-10670	258	6	evaluation	evaluation	NOUN
dad-10670	258	7	,	,	PUNCT
dad-10670	258	8	we	we	PRON
dad-10670	258	9	compare	compare	VERB
dad-10670	258	10	the	the	DET
dad-10670	258	11	performance	performance	NOUN
dad-10670	258	12	when	when	SCONJ
dad-10670	258	13	using	use	VERB
dad-10670	258	14	parse	parse	NOUN
dad-10670	258	15	trees	tree	NOUN
dad-10670	258	16	generated	generate	VERB
dad-10670	258	17	respectively	respectively	ADV
dad-10670	258	18	by	by	ADP
dad-10670	258	19	the	the	DET
dad-10670	258	20	stanford	stanford	PROPN
dad-10670	258	21	parser3	parser3	PROPN
dad-10670	258	22	(	(	PUNCT
dad-10670	258	23	klein	klein	PROPN
dad-10670	258	24	and	and	CCONJ
dad-10670	258	25	manning	manning	NOUN
dad-10670	258	26	,	,	PUNCT
dad-10670	258	27	2003	2003	NUM
dad-10670	258	28	)	)	PUNCT
dad-10670	258	29	and	and	CCONJ
dad-10670	258	30	by	by	ADP
dad-10670	258	31	the	the	DET
dad-10670	258	32	charniak	charniak	PROPN
dad-10670	258	33	parser4	parser4	NOUN
dad-10670	258	34	(	(	PUNCT
dad-10670	258	35	charniak	charniak	PROPN
dad-10670	258	36	,	,	PUNCT
dad-10670	258	37	2000	2000	NUM
dad-10670	258	38	)	)	PUNCT
dad-10670	258	39	.	.	PUNCT
dad-10670	259	1	evaluation	evaluation	NOUN
dad-10670	259	2	is	be	AUX
dad-10670	259	3	done	do	VERB
dad-10670	259	4	on	on	ADP
dad-10670	259	5	the	the	DET
dad-10670	259	6	test	test	NOUN
dad-10670	259	7	subset	subset	NOUN
dad-10670	259	8	of	of	ADP
dad-10670	259	9	the	the	DET
dad-10670	259	10	rst	rst	NOUN
dad-10670	259	11	-	-	PUNCT
dad-10670	259	12	dt	dt	NOUN
dad-10670	259	13	.	.	PUNCT
dad-10670	260	1	in	in	ADP
dad-10670	260	2	our	our	PRON
dad-10670	260	3	experiments	experiment	NOUN
dad-10670	260	4	,	,	PUNCT
dad-10670	260	5	we	we	PRON
dad-10670	260	6	use	use	VERB
dad-10670	260	7	the	the	DET
dad-10670	260	8	same	same	ADJ
dad-10670	260	9	measure	measure	NOUN
dad-10670	260	10	as	as	SCONJ
dad-10670	260	11	soricut	soricut	VERB
dad-10670	260	12	and	and	CCONJ
dad-10670	260	13	marcu	marcu	NOUN
dad-10670	260	14	(	(	PUNCT
dad-10670	260	15	2003	2003	NUM
dad-10670	260	16	)	)	PUNCT
dad-10670	260	17	and	and	CCONJ
dad-10670	260	18	tofiloski	tofiloski	VERB
dad-10670	260	19	et	et	PROPN
dad-10670	260	20	al	al	PROPN
dad-10670	260	21	.	.	PUNCT
dad-10670	261	1	(	(	PUNCT
dad-10670	261	2	2009	2009	NUM
dad-10670	261	3	)	)	PUNCT
dad-10670	261	4	,	,	PUNCT
dad-10670	261	5	i.e.	i.e.	X
dad-10670	261	6	,	,	PUNCT
dad-10670	261	7	we	we	PRON
dad-10670	261	8	only	only	ADV
dad-10670	261	9	measure	measure	VERB
dad-10670	261	10	the	the	DET
dad-10670	261	11	score	score	NOUN
dad-10670	261	12	on	on	ADP
dad-10670	261	13	boundaries	boundary	NOUN
dad-10670	261	14	inside	inside	ADP
dad-10670	261	15	sentences	sentence	NOUN
dad-10670	261	16	.	.	PUNCT
dad-10670	262	1	this	this	PRON
dad-10670	262	2	avoids	avoid	VERB
dad-10670	262	3	artificially	artificially	ADV
dad-10670	262	4	boosting	boost	VERB
dad-10670	262	5	the	the	DET
dad-10670	262	6	performance	performance	NOUN
dad-10670	262	7	by	by	ADP
dad-10670	262	8	including	include	VERB
dad-10670	262	9	obvious	obvious	ADJ
dad-10670	262	10	end	end	NOUN
dad-10670	262	11	-	-	PUNCT
dad-10670	262	12	of	of	ADP
dad-10670	262	13	-	-	PUNCT
dad-10670	262	14	sentence	sentence	NOUN
dad-10670	262	15	or	or	CCONJ
dad-10670	262	16	end	end	NOUN
dad-10670	262	17	-	-	PUNCT
dad-10670	262	18	of	of	ADP
dad-10670	262	19	-	-	PUNCT
dad-10670	262	20	paragraph	paragraph	NOUN
dad-10670	262	21	boundaries	boundary	NOUN
dad-10670	262	22	.	.	PUNCT
dad-10670	263	1	the	the	DET
dad-10670	263	2	radial	radial	ADJ
dad-10670	263	3	basis	basis	NOUN
dad-10670	263	4	function	function	NOUN
dad-10670	263	5	(	(	PUNCT
dad-10670	263	6	rbf	rbf	PROPN
dad-10670	263	7	)	)	PUNCT
dad-10670	263	8	(	(	PUNCT
dad-10670	263	9	vapnik	vapnik	X
dad-10670	263	10	,	,	PUNCT
dad-10670	263	11	1995	1995	NUM
dad-10670	263	12	)	)	PUNCT
dad-10670	263	13	kernel	kernel	NOUN
dad-10670	263	14	is	be	AUX
dad-10670	263	15	selected	select	VERB
dad-10670	263	16	,	,	PUNCT
dad-10670	263	17	and	and	CCONJ
dad-10670	263	18	parameter	parameter	NOUN
dad-10670	263	19	estimation	estimation	NOUN
dad-10670	263	20	is	be	AUX
dad-10670	263	21	done	do	VERB
dad-10670	263	22	using	use	VERB
dad-10670	263	23	grid	grid	NOUN
dad-10670	263	24	search	search	NOUN
dad-10670	263	25	with	with	ADP
dad-10670	263	26	5	5	NUM
dad-10670	263	27	-	-	ADJ
dad-10670	263	28	fold	fold	ADJ
dad-10670	263	29	cross	cross	NOUN
dad-10670	263	30	validation	validation	NOUN
dad-10670	263	31	(	(	PUNCT
dad-10670	263	32	staelin	staelin	NOUN
dad-10670	263	33	,	,	PUNCT
dad-10670	263	34	2003	2003	NUM
dad-10670	263	35	)	)	PUNCT
dad-10670	263	36	.	.	PUNCT
dad-10670	264	1	in	in	ADP
dad-10670	264	2	practice	practice	NOUN
dad-10670	264	3	,	,	PUNCT
dad-10670	264	4	we	we	PRON
dad-10670	264	5	did	do	AUX
dad-10670	264	6	not	not	PART
dad-10670	264	7	observe	observe	VERB
dad-10670	264	8	any	any	DET
dad-10670	264	9	problem	problem	NOUN
dad-10670	264	10	related	relate	VERB
dad-10670	264	11	to	to	ADP
dad-10670	264	12	the	the	DET
dad-10670	264	13	class	class	NOUN
dad-10670	264	14	imbalance	imbalance	NOUN
dad-10670	264	15	in	in	ADP
dad-10670	264	16	the	the	DET
dad-10670	264	17	training	training	NOUN
dad-10670	264	18	set	set	VERB
dad-10670	264	19	with	with	ADP
dad-10670	264	20	these	these	DET
dad-10670	264	21	parameters	parameter	NOUN
dad-10670	264	22	.	.	PUNCT
dad-10670	265	1	the	the	DET
dad-10670	265	2	performance	performance	NOUN
dad-10670	265	3	of	of	ADP
dad-10670	265	4	our	our	PRON
dad-10670	265	5	and	and	CCONJ
dad-10670	265	6	other	other	ADJ
dad-10670	265	7	available	available	ADJ
dad-10670	265	8	segmenters	segmenter	NOUN
dad-10670	265	9	is	be	AUX
dad-10670	265	10	shown	show	VERB
dad-10670	265	11	in	in	ADP
dad-10670	265	12	table	table	NOUN
dad-10670	265	13	3	3	NUM
dad-10670	265	14	.	.	PUNCT
dad-10670	265	15	nnds	nnd	NOUN
dad-10670	265	16	refers	refer	VERB
dad-10670	265	17	to	to	ADP
dad-10670	265	18	the	the	DET
dad-10670	265	19	neural	neural	ADJ
dad-10670	265	20	-	-	PUNCT
dad-10670	265	21	networks	network	NOUN
dad-10670	265	22	discourse	discourse	NOUN
dad-10670	265	23	segmenter	segmenter	NOUN
dad-10670	265	24	(	(	PUNCT
dad-10670	265	25	subba	subba	NOUN
dad-10670	265	26	and	and	CCONJ
dad-10670	265	27	di	di	NOUN
dad-10670	265	28	eugenio	eugenio	PROPN
dad-10670	265	29	,	,	PUNCT
dad-10670	265	30	2007	2007	NUM
dad-10670	265	31	)	)	PUNCT
dad-10670	265	32	,	,	PUNCT
dad-10670	265	33	spade	spade	NOUN
dad-10670	265	34	-	-	PUNCT
dad-10670	265	35	seg	seg	PROPN
dad-10670	265	36	is	be	AUX
dad-10670	265	37	the	the	DET
dad-10670	265	38	segmentation	segmentation	NOUN
dad-10670	265	39	module	module	NOUN
dad-10670	265	40	of	of	ADP
dad-10670	265	41	spade	spade	NOUN
dad-10670	265	42	,	,	PUNCT
dad-10670	265	43	described	describe	VERB
dad-10670	265	44	in	in	ADP
dad-10670	265	45	soricut	soricut	NOUN
dad-10670	265	46	and	and	CCONJ
dad-10670	265	47	marcu	marcu	NOUN
dad-10670	265	48	(	(	PUNCT
dad-10670	265	49	2003	2003	NUM
dad-10670	265	50	)	)	PUNCT
dad-10670	265	51	,	,	PUNCT
dad-10670	265	52	and	and	CCONJ
dad-10670	265	53	sag	sag	NOUN
dad-10670	265	54	refers	refer	VERB
dad-10670	265	55	to	to	ADP
dad-10670	265	56	the	the	DET
dad-10670	265	57	segmenter	segmenter	NOUN
dad-10670	265	58	included	include	VERB
dad-10670	265	59	in	in	ADP
dad-10670	265	60	the	the	DET
dad-10670	265	61	discourse	discourse	NOUN
dad-10670	265	62	parser	parser	NOUN
dad-10670	265	63	of	of	ADP
dad-10670	265	64	sagae	sagae	NOUN
dad-10670	265	65	(	(	PUNCT
dad-10670	265	66	2009	2009	NUM
dad-10670	265	67	)	)	PUNCT
dad-10670	265	68	.	.	PUNCT
dad-10670	266	1	table	table	NOUN
dad-10670	266	2	3	3	NUM
dad-10670	266	3	(	(	PUNCT
dad-10670	266	4	top	top	ADJ
dad-10670	266	5	row	row	NOUN
dad-10670	266	6	)	)	PUNCT
dad-10670	266	7	shows	show	VERB
dad-10670	266	8	that	that	SCONJ
dad-10670	266	9	hilda	hilda	PROPN
dad-10670	266	10	-	-	PUNCT
dad-10670	266	11	seg	seg	PROPN
dad-10670	266	12	significantly	significantly	ADV
dad-10670	266	13	outperforms	outperform	VERB
dad-10670	266	14	other	other	ADJ
dad-10670	266	15	discourse	discourse	NOUN
dad-10670	266	16	segmenters	segmenter	NOUN
dad-10670	266	17	.	.	PUNCT
dad-10670	267	1	with	with	ADP
dad-10670	267	2	gold	gold	NOUN
dad-10670	267	3	standard	standard	ADJ
dad-10670	267	4	trees	tree	NOUN
dad-10670	267	5	,	,	PUNCT
dad-10670	267	6	spade	spade	NOUN
dad-10670	267	7	-	-	PUNCT
dad-10670	267	8	seg	seg	NOUN
dad-10670	267	9	and	and	CCONJ
dad-10670	267	10	nnds	nnd	VERB
dad-10670	267	11	yield	yield	VERB
dad-10670	267	12	f	f	NOUN
dad-10670	267	13	-	-	PUNCT
dad-10670	267	14	scores	score	NOUN
dad-10670	267	15	of	of	ADP
dad-10670	267	16	84.7	84.7	NUM
dad-10670	267	17	%	%	NOUN
dad-10670	267	18	and	and	CCONJ
dad-10670	267	19	86	86	NUM
dad-10670	267	20	%	%	NOUN
dad-10670	267	21	,	,	PUNCT
dad-10670	267	22	respectively	respectively	ADV
dad-10670	267	23	,	,	PUNCT
dad-10670	267	24	versus	versus	ADP
dad-10670	267	25	95	95	NUM
dad-10670	267	26	%	%	NOUN
dad-10670	267	27	for	for	ADP
dad-10670	267	28	hilda	hilda	NOUN
dad-10670	267	29	-	-	PUNCT
dad-10670	267	30	seg	seg	PROPN
dad-10670	267	31	.	.	PUNCT
dad-10670	268	1	the	the	DET
dad-10670	268	2	measure	measure	NOUN
dad-10670	268	3	of	of	ADP
dad-10670	268	4	the	the	DET
dad-10670	268	5	human	human	ADJ
dad-10670	268	6	annotators	annotator	NOUN
dad-10670	268	7	’	'	PUNCT
dad-10670	268	8	agreement	agreement	NOUN
dad-10670	268	9	for	for	ADP
dad-10670	268	10	the	the	DET
dad-10670	268	11	segmentation	segmentation	NOUN
dad-10670	268	12	task	task	NOUN
dad-10670	268	13	has	have	AUX
dad-10670	268	14	been	be	AUX
dad-10670	268	15	3	3	NUM
dad-10670	268	16	.	.	PUNCT
dad-10670	268	17	http://nlp.stanford.edu/software/lex-parser.shtml	http://nlp.stanford.edu/software/lex-parser.shtml	PROPN
dad-10670	268	18	4	4	NUM
dad-10670	268	19	.	.	PUNCT
dad-10670	268	20	ftp://ftp.cs.brown.edu/pub/nlparser/	ftp://ftp.cs.brown.edu/pub/nlparser/	PROPN
dad-10670	268	21	12	12	NUM
dad-10670	268	22	hilda	hilda	VERB
dad-10670	268	23	:	:	PUNCT
dad-10670	268	24	a	a	DET
dad-10670	268	25	discourse	discourse	NOUN
dad-10670	268	26	parser	parser	NOUN
dad-10670	268	27	using	use	VERB
dad-10670	268	28	support	support	NOUN
dad-10670	268	29	vector	vector	NOUN
dad-10670	268	30	machine	machine	NOUN
dad-10670	268	31	classification	classification	NOUN
dad-10670	268	32	table	table	NOUN
dad-10670	268	33	3	3	NUM
dad-10670	268	34	:	:	PUNCT
dad-10670	268	35	performance	performance	NOUN
dad-10670	268	36	comparison	comparison	NOUN
dad-10670	268	37	with	with	ADP
dad-10670	268	38	other	other	ADJ
dad-10670	268	39	segmenters	segmenter	NOUN
dad-10670	269	1	system	system	VERB
dad-10670	269	2	trees	tree	NOUN
dad-10670	269	3	precision	precision	NOUN
dad-10670	269	4	recall	recall	NOUN
dad-10670	269	5	f	f	NOUN
dad-10670	269	6	-	-	PUNCT
dad-10670	269	7	score	score	NOUN
dad-10670	269	8	spade	spade	NOUN
dad-10670	269	9	-	-	PUNCT
dad-10670	269	10	seg	seg	PROPN
dad-10670	269	11	penn	penn	PROPN
dad-10670	269	12	84.1	84.1	NUM
dad-10670	269	13	85.4	85.4	NUM
dad-10670	269	14	84.7	84.7	NUM
dad-10670	269	15	nnds	nnd	NOUN
dad-10670	269	16	penn	penn	PROPN
dad-10670	269	17	85.5	85.5	NUM
dad-10670	269	18	86.6	86.6	NUM
dad-10670	269	19	86.0	86.0	NUM
dad-10670	269	20	hilda	hilda	PROPN
dad-10670	269	21	-	-	PUNCT
dad-10670	269	22	seg	seg	PROPN
dad-10670	269	23	penn	penn	PROPN
dad-10670	269	24	95.5	95.5	NUM
dad-10670	269	25	94.5	94.5	NUM
dad-10670	269	26	95.0	95.0	NUM
dad-10670	269	27	spade	spade	NOUN
dad-10670	269	28	-	-	PUNCT
dad-10670	269	29	seg	seg	NOUN
dad-10670	269	30	charniak	charniak	NOUN
dad-10670	269	31	83.5	83.5	NUM
dad-10670	269	32	82.7	82.7	NUM
dad-10670	269	33	83.1	83.1	NUM
dad-10670	269	34	nnds	nnd	NOUN
dad-10670	269	35	charniak	charniak	NOUN
dad-10670	269	36	83.9	83.9	NUM
dad-10670	269	37	84.8	84.8	NUM
dad-10670	269	38	84.4	84.4	NUM
dad-10670	269	39	hilda	hilda	PROPN
dad-10670	269	40	-	-	PUNCT
dad-10670	269	41	seg	seg	PROPN
dad-10670	269	42	charniak	charniak	PROPN
dad-10670	269	43	94.7	94.7	NUM
dad-10670	269	44	93.4	93.4	NUM
dad-10670	269	45	94.0	94.0	NUM
dad-10670	269	46	sag	sag	NOUN
dad-10670	269	47	sag	sag	NOUN
dad-10670	269	48	87.4	87.4	NUM
dad-10670	269	49	86.0	86.0	NUM
dad-10670	269	50	86.7	86.7	NUM
dad-10670	269	51	hilda	hilda	PROPN
dad-10670	269	52	-	-	PUNCT
dad-10670	269	53	seg	seg	PROPN
dad-10670	269	54	stanford	stanford	PROPN
dad-10670	269	55	94.5	94.5	NUM
dad-10670	269	56	93.1	93.1	NUM
dad-10670	269	57	93.8	93.8	NUM
dad-10670	269	58	human	human	ADJ
dad-10670	269	59	agreement	agreement	NOUN
dad-10670	269	60	–	–	PUNCT
dad-10670	269	61	98.5	98.5	NUM
dad-10670	269	62	98.2	98.2	NUM
dad-10670	269	63	98.3	98.3	NUM
dad-10670	269	64	calculated	calculate	VERB
dad-10670	269	65	in	in	ADP
dad-10670	269	66	soricut	soricut	NOUN
dad-10670	269	67	and	and	CCONJ
dad-10670	269	68	marcu	marcu	NOUN
dad-10670	269	69	(	(	PUNCT
dad-10670	269	70	2003	2003	NUM
dad-10670	269	71	)	)	PUNCT
dad-10670	269	72	,	,	PUNCT
dad-10670	269	73	with	with	ADP
dad-10670	269	74	a	a	DET
dad-10670	269	75	f	f	NOUN
dad-10670	269	76	-	-	PUNCT
dad-10670	269	77	score	score	NOUN
dad-10670	269	78	of	of	ADP
dad-10670	269	79	98.3	98.3	NUM
dad-10670	269	80	%	%	NOUN
dad-10670	269	81	.	.	PUNCT
dad-10670	270	1	using	use	VERB
dad-10670	270	2	perfect	perfect	ADJ
dad-10670	270	3	parse	parse	NOUN
dad-10670	270	4	trees	tree	NOUN
dad-10670	270	5	,	,	PUNCT
dad-10670	270	6	hildaseg	hildaseg	PROPN
dad-10670	270	7	reaches	reach	VERB
dad-10670	270	8	96.6	96.6	NUM
dad-10670	270	9	%	%	NOUN
dad-10670	270	10	of	of	ADP
dad-10670	270	11	the	the	DET
dad-10670	270	12	annotators	annotator	NOUN
dad-10670	270	13	’	'	PUNCT
dad-10670	270	14	level	level	NOUN
dad-10670	270	15	.	.	PUNCT
dad-10670	271	1	when	when	SCONJ
dad-10670	271	2	the	the	DET
dad-10670	271	3	stanford	stanford	PROPN
dad-10670	271	4	parse	parse	NOUN
dad-10670	271	5	trees	tree	NOUN
dad-10670	271	6	are	be	AUX
dad-10670	271	7	selected	select	VERB
dad-10670	271	8	,	,	PUNCT
dad-10670	271	9	95.4	95.4	NUM
dad-10670	271	10	%	%	NOUN
dad-10670	271	11	of	of	ADP
dad-10670	271	12	this	this	DET
dad-10670	271	13	level	level	NOUN
dad-10670	271	14	is	be	AUX
dad-10670	271	15	reached	reach	VERB
dad-10670	271	16	.	.	PUNCT
dad-10670	272	1	the	the	DET
dad-10670	272	2	current	current	ADJ
dad-10670	272	3	state	state	NOUN
dad-10670	272	4	-	-	PUNCT
dad-10670	272	5	of	of	ADP
dad-10670	272	6	-	-	PUNCT
dad-10670	272	7	the	the	DET
dad-10670	272	8	-	-	PUNCT
dad-10670	272	9	art	art	NOUN
dad-10670	272	10	discourse	discourse	NOUN
dad-10670	272	11	segmenter	segmenter	NOUN
dad-10670	272	12	,	,	PUNCT
dad-10670	272	13	sag	sag	NOUN
dad-10670	272	14	,	,	PUNCT
dad-10670	272	15	in	in	ADP
dad-10670	272	16	which	which	PRON
dad-10670	272	17	a	a	DET
dad-10670	272	18	custom	custom	NOUN
dad-10670	272	19	syntax	syntax	NOUN
dad-10670	272	20	parser	parser	NOUN
dad-10670	272	21	is	be	AUX
dad-10670	272	22	employed	employ	VERB
dad-10670	272	23	,	,	PUNCT
dad-10670	272	24	has	have	VERB
dad-10670	272	25	an	an	DET
dad-10670	272	26	f	f	NOUN
dad-10670	272	27	-	-	PUNCT
dad-10670	272	28	score	score	NOUN
dad-10670	272	29	of	of	ADP
dad-10670	272	30	86.7	86.7	NUM
dad-10670	272	31	%	%	NOUN
dad-10670	272	32	for	for	ADP
dad-10670	272	33	this	this	DET
dad-10670	272	34	task	task	NOUN
dad-10670	272	35	.	.	PUNCT
dad-10670	273	1	when	when	SCONJ
dad-10670	273	2	using	use	VERB
dad-10670	273	3	stanford	stanford	PROPN
dad-10670	273	4	parse	parse	NOUN
dad-10670	273	5	trees	tree	NOUN
dad-10670	273	6	,	,	PUNCT
dad-10670	273	7	our	our	PRON
dad-10670	273	8	segmenter	segmenter	NOUN
dad-10670	273	9	thus	thus	ADV
dad-10670	273	10	provides	provide	VERB
dad-10670	273	11	a	a	DET
dad-10670	273	12	8.2	8.2	NUM
dad-10670	273	13	%	%	NOUN
dad-10670	273	14	performance	performance	NOUN
dad-10670	273	15	increase	increase	NOUN
dad-10670	273	16	compared	compare	VERB
dad-10670	273	17	to	to	ADP
dad-10670	273	18	sag	sag	NOUN
dad-10670	273	19	(	(	PUNCT
dad-10670	273	20	9.6	9.6	NUM
dad-10670	273	21	%	%	NOUN
dad-10670	273	22	when	when	SCONJ
dad-10670	273	23	using	use	VERB
dad-10670	273	24	perfect	perfect	ADJ
dad-10670	273	25	trees	tree	NOUN
dad-10670	273	26	)	)	PUNCT
dad-10670	273	27	.	.	PUNCT
dad-10670	274	1	in	in	ADP
dad-10670	274	2	table	table	NOUN
dad-10670	274	3	3	3	NUM
dad-10670	274	4	,	,	PUNCT
dad-10670	274	5	we	we	PRON
dad-10670	274	6	decided	decide	VERB
dad-10670	274	7	not	not	PART
dad-10670	274	8	to	to	PART
dad-10670	274	9	include	include	VERB
dad-10670	274	10	the	the	DET
dad-10670	274	11	results	result	NOUN
dad-10670	274	12	of	of	ADP
dad-10670	274	13	the	the	DET
dad-10670	274	14	rule	rule	NOUN
dad-10670	274	15	-	-	PUNCT
dad-10670	274	16	based	base	VERB
dad-10670	274	17	segmenters	segmenter	NOUN
dad-10670	274	18	of	of	ADP
dad-10670	274	19	le	le	X
dad-10670	274	20	thanh	thanh	PROPN
dad-10670	274	21	et	et	PROPN
dad-10670	274	22	al	al	PROPN
dad-10670	274	23	.	.	PROPN
dad-10670	274	24	(	(	PUNCT
dad-10670	274	25	2004a	2004a	NUM
dad-10670	274	26	)	)	PUNCT
dad-10670	274	27	and	and	CCONJ
dad-10670	274	28	tofiloski	tofiloski	VERB
dad-10670	274	29	et	et	PROPN
dad-10670	274	30	al	al	PROPN
dad-10670	274	31	.	.	PUNCT
dad-10670	275	1	(	(	PUNCT
dad-10670	275	2	2009	2009	NUM
dad-10670	275	3	)	)	PUNCT
dad-10670	275	4	,	,	PUNCT
dad-10670	275	5	for	for	ADP
dad-10670	275	6	several	several	ADJ
dad-10670	275	7	reasons	reason	NOUN
dad-10670	275	8	.	.	PUNCT
dad-10670	276	1	first	first	ADV
dad-10670	276	2	,	,	PUNCT
dad-10670	276	3	le	le	AUX
dad-10670	276	4	thanh	thanh	VERB
dad-10670	276	5	et	et	PROPN
dad-10670	276	6	al	al	PROPN
dad-10670	276	7	.	.	PROPN
dad-10670	277	1	(	(	PUNCT
dad-10670	277	2	2004a	2004a	NUM
dad-10670	277	3	)	)	PUNCT
dad-10670	277	4	report	report	VERB
dad-10670	277	5	their	their	PRON
dad-10670	277	6	results	result	NOUN
dad-10670	277	7	using	use	VERB
dad-10670	277	8	a	a	DET
dad-10670	277	9	‘	'	PUNCT
dad-10670	277	10	softer	soft	ADJ
dad-10670	277	11	’	'	PUNCT
dad-10670	277	12	metric	metric	NOUN
dad-10670	277	13	,	,	PUNCT
dad-10670	277	14	in	in	ADP
dad-10670	277	15	which	which	PRON
dad-10670	277	16	end	end	NOUN
dad-10670	277	17	-	-	PUNCT
dad-10670	277	18	of	of	ADP
dad-10670	277	19	-	-	PUNCT
dad-10670	277	20	sentence	sentence	NOUN
dad-10670	277	21	boundaries	boundary	NOUN
dad-10670	277	22	are	be	AUX
dad-10670	277	23	taken	take	VERB
dad-10670	277	24	into	into	ADP
dad-10670	277	25	account	account	NOUN
dad-10670	277	26	.	.	PUNCT
dad-10670	278	1	the	the	DET
dad-10670	278	2	authors	author	NOUN
dad-10670	278	3	used	use	VERB
dad-10670	278	4	penn	penn	PROPN
dad-10670	278	5	treebank	treebank	PROPN
dad-10670	278	6	parse	parse	NOUN
dad-10670	278	7	trees	tree	NOUN
dad-10670	278	8	,	,	PUNCT
dad-10670	278	9	and	and	CCONJ
dad-10670	278	10	after	after	ADP
dad-10670	278	11	evaluation	evaluation	NOUN
dad-10670	278	12	on	on	ADP
dad-10670	278	13	8	8	NUM
dad-10670	278	14	texts	text	NOUN
dad-10670	278	15	of	of	ADP
dad-10670	278	16	the	the	DET
dad-10670	278	17	rst	rst	NOUN
dad-10670	278	18	-	-	PUNCT
dad-10670	278	19	dt	dt	PROPN
dad-10670	278	20	,	,	PUNCT
dad-10670	278	21	obtain	obtain	VERB
dad-10670	278	22	a	a	DET
dad-10670	278	23	precision	precision	NOUN
dad-10670	278	24	value	value	NOUN
dad-10670	278	25	of	of	ADP
dad-10670	278	26	81.4	81.4	NUM
dad-10670	278	27	%	%	NOUN
dad-10670	278	28	and	and	CCONJ
dad-10670	278	29	a	a	DET
dad-10670	278	30	recall	recall	NOUN
dad-10670	278	31	value	value	NOUN
dad-10670	278	32	of	of	ADP
dad-10670	278	33	79.2	79.2	NUM
dad-10670	278	34	%	%	NOUN
dad-10670	278	35	.	.	PUNCT
dad-10670	279	1	finally	finally	ADV
dad-10670	279	2	,	,	PUNCT
dad-10670	279	3	the	the	DET
dad-10670	279	4	results	result	NOUN
dad-10670	279	5	of	of	ADP
dad-10670	279	6	tofiloski	tofiloski	NOUN
dad-10670	279	7	et	et	PROPN
dad-10670	279	8	al	al	PROPN
dad-10670	279	9	.	.	PROPN
dad-10670	280	1	(	(	PUNCT
dad-10670	280	2	2009	2009	NUM
dad-10670	280	3	)	)	PUNCT
dad-10670	280	4	can	can	AUX
dad-10670	280	5	not	not	PART
dad-10670	280	6	be	be	AUX
dad-10670	280	7	compared	compare	VERB
dad-10670	280	8	directly	directly	ADV
dad-10670	280	9	to	to	ADP
dad-10670	280	10	ours	ours	PRON
dad-10670	280	11	,	,	PUNCT
dad-10670	280	12	as	as	SCONJ
dad-10670	280	13	different	different	ADJ
dad-10670	280	14	segmentation	segmentation	NOUN
dad-10670	280	15	guidelines	guideline	NOUN
dad-10670	280	16	were	be	AUX
dad-10670	280	17	used	use	VERB
dad-10670	280	18	.	.	PUNCT
dad-10670	281	1	the	the	DET
dad-10670	281	2	authors	author	NOUN
dad-10670	281	3	report	report	VERB
dad-10670	281	4	a	a	DET
dad-10670	281	5	precision	precision	NOUN
dad-10670	281	6	value	value	NOUN
dad-10670	281	7	of	of	ADP
dad-10670	281	8	89	89	NUM
dad-10670	281	9	%	%	NOUN
dad-10670	281	10	and	and	CCONJ
dad-10670	281	11	a	a	DET
dad-10670	281	12	recall	recall	NOUN
dad-10670	281	13	value	value	NOUN
dad-10670	281	14	of	of	ADP
dad-10670	281	15	86	86	NUM
dad-10670	281	16	%	%	NOUN
dad-10670	281	17	when	when	SCONJ
dad-10670	281	18	using	use	VERB
dad-10670	281	19	charniak	charniak	NOUN
dad-10670	281	20	parse	parse	NOUN
dad-10670	281	21	trees	tree	NOUN
dad-10670	281	22	,	,	PUNCT
dad-10670	281	23	and	and	CCONJ
dad-10670	281	24	a	a	DET
dad-10670	281	25	precision	precision	NOUN
dad-10670	281	26	value	value	NOUN
dad-10670	281	27	of	of	ADP
dad-10670	281	28	82	82	NUM
dad-10670	281	29	%	%	NOUN
dad-10670	281	30	and	and	CCONJ
dad-10670	281	31	a	a	DET
dad-10670	281	32	recall	recall	NOUN
dad-10670	281	33	value	value	NOUN
dad-10670	281	34	of	of	ADP
dad-10670	281	35	86	86	NUM
dad-10670	281	36	%	%	NOUN
dad-10670	281	37	when	when	SCONJ
dad-10670	281	38	using	use	VERB
dad-10670	281	39	stanford	stanford	PROPN
dad-10670	281	40	trees	tree	NOUN
dad-10670	281	41	.	.	PUNCT
dad-10670	282	1	moreover	moreover	ADV
dad-10670	282	2	,	,	PUNCT
dad-10670	282	3	these	these	DET
dad-10670	282	4	scores	score	NOUN
dad-10670	282	5	were	be	AUX
dad-10670	282	6	measured	measure	VERB
dad-10670	282	7	on	on	ADP
dad-10670	282	8	three	three	NUM
dad-10670	282	9	texts	text	NOUN
dad-10670	282	10	of	of	ADP
dad-10670	282	11	the	the	DET
dad-10670	282	12	rst	rst	NOUN
dad-10670	282	13	-	-	PUNCT
dad-10670	282	14	dt	dt	NOUN
dad-10670	282	15	only	only	ADV
dad-10670	282	16	,	,	PUNCT
dad-10670	282	17	which	which	PRON
dad-10670	282	18	makes	make	VERB
dad-10670	282	19	comparison	comparison	NOUN
dad-10670	282	20	even	even	ADV
dad-10670	282	21	more	more	ADV
dad-10670	282	22	difficult	difficult	ADJ
dad-10670	282	23	.	.	PUNCT
dad-10670	283	1	to	to	PART
dad-10670	283	2	investigate	investigate	VERB
dad-10670	283	3	which	which	DET
dad-10670	283	4	features	feature	NOUN
dad-10670	283	5	contributed	contribute	VERB
dad-10670	283	6	to	to	ADP
dad-10670	283	7	the	the	DET
dad-10670	283	8	performance	performance	NOUN
dad-10670	283	9	of	of	ADP
dad-10670	283	10	the	the	DET
dad-10670	283	11	discourse	discourse	NOUN
dad-10670	283	12	segmenter	segmenter	NOUN
dad-10670	283	13	,	,	PUNCT
dad-10670	283	14	we	we	PRON
dad-10670	283	15	run	run	VERB
dad-10670	283	16	several	several	ADJ
dad-10670	283	17	experiments	experiment	NOUN
dad-10670	283	18	that	that	PRON
dad-10670	283	19	measure	measure	VERB
dad-10670	283	20	the	the	DET
dad-10670	283	21	segmentation	segmentation	NOUN
dad-10670	283	22	performance	performance	NOUN
dad-10670	283	23	when	when	SCONJ
dad-10670	283	24	training	train	VERB
dad-10670	283	25	with	with	ADP
dad-10670	283	26	different	different	ADJ
dad-10670	283	27	combinations	combination	NOUN
dad-10670	283	28	of	of	ADP
dad-10670	283	29	features	feature	NOUN
dad-10670	283	30	,	,	PUNCT
dad-10670	283	31	taken	take	VERB
dad-10670	283	32	at	at	ADP
dad-10670	283	33	different	different	ADJ
dad-10670	283	34	relative	relative	ADJ
dad-10670	283	35	positions	position	NOUN
dad-10670	283	36	.	.	PUNCT
dad-10670	284	1	for	for	ADP
dad-10670	284	2	each	each	DET
dad-10670	284	3	experiment	experiment	NOUN
dad-10670	284	4	,	,	PUNCT
dad-10670	284	5	we	we	PRON
dad-10670	284	6	use	use	VERB
dad-10670	284	7	parse	parse	NOUN
dad-10670	284	8	trees	tree	NOUN
dad-10670	284	9	from	from	ADP
dad-10670	284	10	the	the	DET
dad-10670	284	11	stanford	stanford	PROPN
dad-10670	284	12	parser	parser	NOUN
dad-10670	284	13	.	.	PUNCT
dad-10670	285	1	the	the	DET
dad-10670	285	2	results	result	NOUN
dad-10670	285	3	are	be	AUX
dad-10670	285	4	indicated	indicate	VERB
dad-10670	285	5	in	in	ADP
dad-10670	285	6	table	table	NOUN
dad-10670	285	7	4	4	NUM
dad-10670	285	8	.	.	PUNCT
dad-10670	286	1	position	position	NOUN
dad-10670	286	2	(	(	PUNCT
dad-10670	286	3	0	0	NUM
dad-10670	286	4	)	)	PUNCT
dad-10670	286	5	and	and	CCONJ
dad-10670	286	6	(	(	PUNCT
dad-10670	286	7	−1	−1	NOUN
dad-10670	286	8	,	,	PUNCT
dad-10670	286	9	0	0	NUM
dad-10670	286	10	)	)	PUNCT
dad-10670	286	11	indicate	indicate	VERB
dad-10670	286	12	that	that	SCONJ
dad-10670	286	13	,	,	PUNCT
dad-10670	286	14	when	when	SCONJ
dad-10670	286	15	extracting	extract	VERB
dad-10670	286	16	features	feature	NOUN
dad-10670	286	17	for	for	ADP
dad-10670	286	18	the	the	DET
dad-10670	286	19	word	word	NOUN
dad-10670	286	20	at	at	ADP
dad-10670	286	21	position	position	NOUN
dad-10670	286	22	i	i	PRON
dad-10670	286	23	in	in	ADP
dad-10670	286	24	the	the	DET
dad-10670	286	25	text	text	NOUN
dad-10670	286	26	,	,	PUNCT
dad-10670	286	27	we	we	PRON
dad-10670	286	28	only	only	ADV
dad-10670	286	29	used	use	VERB
dad-10670	286	30	contextual	contextual	ADJ
dad-10670	286	31	features	feature	NOUN
dad-10670	286	32	from	from	ADP
dad-10670	286	33	position	position	NOUN
dad-10670	286	34	i	i	PRON
dad-10670	286	35	,	,	PUNCT
dad-10670	286	36	and	and	CCONJ
dad-10670	286	37	the	the	DET
dad-10670	286	38	concatenation	concatenation	NOUN
dad-10670	286	39	of	of	ADP
dad-10670	286	40	contextual	contextual	ADJ
dad-10670	286	41	features	feature	NOUN
dad-10670	286	42	from	from	ADP
dad-10670	286	43	positions	position	NOUN
dad-10670	286	44	i−1	i−1	PROPN
dad-10670	286	45	and	and	CCONJ
dad-10670	286	46	i	i	PROPN
dad-10670	286	47	,	,	PUNCT
dad-10670	286	48	respectively	respectively	ADV
dad-10670	286	49	.	.	PUNCT
dad-10670	287	1	we	we	PRON
dad-10670	287	2	observe	observe	VERB
dad-10670	287	3	that	that	SCONJ
dad-10670	287	4	,	,	PUNCT
dad-10670	287	5	for	for	ADP
dad-10670	287	6	position	position	NOUN
dad-10670	287	7	(	(	PUNCT
dad-10670	287	8	0	0	NUM
dad-10670	287	9	)	)	PUNCT
dad-10670	287	10	,	,	PUNCT
dad-10670	287	11	using	use	VERB
dad-10670	287	12	as	as	SCONJ
dad-10670	287	13	features	feature	VERB
dad-10670	287	14	the	the	DET
dad-10670	287	15	current	current	ADJ
dad-10670	287	16	word	word	NOUN
dad-10670	287	17	w	w	NOUN
dad-10670	287	18	and	and	CCONJ
dad-10670	287	19	its	its	PRON
dad-10670	287	20	part	part	NOUN
dad-10670	287	21	of	of	ADP
dad-10670	287	22	speech	speech	NOUN
dad-10670	287	23	pos(w	pos(w	ADJ
dad-10670	287	24	)	)	PUNCT
dad-10670	287	25	only	only	ADV
dad-10670	287	26	,	,	PUNCT
dad-10670	287	27	we	we	PRON
dad-10670	287	28	obtain	obtain	VERB
dad-10670	287	29	a	a	DET
dad-10670	287	30	precision	precision	NOUN
dad-10670	287	31	of	of	ADP
dad-10670	287	32	82.7	82.7	NUM
dad-10670	287	33	%	%	NOUN
dad-10670	287	34	,	,	PUNCT
dad-10670	287	35	but	but	CCONJ
dad-10670	287	36	a	a	DET
dad-10670	287	37	comparatively	comparatively	ADV
dad-10670	287	38	lower	low	ADJ
dad-10670	287	39	recall	recall	NOUN
dad-10670	287	40	of	of	ADP
dad-10670	287	41	70.8	70.8	NUM
dad-10670	287	42	%	%	NOUN
dad-10670	287	43	.	.	PUNCT
dad-10670	288	1	however	however	ADV
dad-10670	288	2	,	,	PUNCT
dad-10670	288	3	when	when	SCONJ
dad-10670	288	4	using	use	VERB
dad-10670	288	5	as	as	SCONJ
dad-10670	288	6	features	feature	NOUN
dad-10670	288	7	the	the	DET
dad-10670	288	8	three	three	NUM
dad-10670	288	9	nodes	node	NOUN
dad-10670	288	10	nw	nw	PROPN
dad-10670	288	11	,	,	PUNCT
dad-10670	288	12	np	np	INTJ
dad-10670	288	13	,	,	PUNCT
dad-10670	288	14	nr	nr	PROPN
dad-10670	288	15	of	of	ADP
dad-10670	288	16	the	the	DET
dad-10670	288	17	parse	parse	NOUN
dad-10670	288	18	tree	tree	NOUN
dad-10670	288	19	only	only	ADV
dad-10670	288	20	(	(	PUNCT
dad-10670	288	21	see	see	VERB
dad-10670	288	22	3.4.1	3.4.1	NUM
dad-10670	288	23	)	)	PUNCT
dad-10670	288	24	,	,	PUNCT
dad-10670	288	25	the	the	DET
dad-10670	288	26	precision	precision	NOUN
dad-10670	288	27	is	be	AUX
dad-10670	288	28	slightly	slightly	ADV
dad-10670	288	29	lower	low	ADJ
dad-10670	288	30	(	(	PUNCT
dad-10670	288	31	82	82	NUM
dad-10670	288	32	%	%	NOUN
dad-10670	288	33	)	)	PUNCT
dad-10670	288	34	but	but	CCONJ
dad-10670	288	35	the	the	DET
dad-10670	288	36	recall	recall	NOUN
dad-10670	288	37	higher	high	ADJ
dad-10670	288	38	than	than	ADP
dad-10670	288	39	the	the	DET
dad-10670	288	40	previous	previous	ADJ
dad-10670	288	41	case	case	NOUN
dad-10670	288	42	,	,	PUNCT
dad-10670	288	43	at	at	ADP
dad-10670	288	44	78.4	78.4	NUM
dad-10670	288	45	%	%	NOUN
dad-10670	288	46	.	.	PUNCT
dad-10670	289	1	when	when	SCONJ
dad-10670	289	2	using	use	VERB
dad-10670	289	3	the	the	DET
dad-10670	289	4	combination	combination	NOUN
dad-10670	289	5	of	of	ADP
dad-10670	289	6	all	all	DET
dad-10670	289	7	these	these	DET
dad-10670	289	8	features	feature	NOUN
dad-10670	289	9	,	,	PUNCT
dad-10670	289	10	we	we	PRON
dad-10670	289	11	obtain	obtain	VERB
dad-10670	289	12	a	a	DET
dad-10670	289	13	precision	precision	NOUN
dad-10670	289	14	of	of	ADP
dad-10670	289	15	81.9	81.9	NUM
dad-10670	289	16	%	%	NOUN
dad-10670	289	17	and	and	CCONJ
dad-10670	289	18	a	a	DET
dad-10670	289	19	recall	recall	NOUN
dad-10670	289	20	of	of	ADP
dad-10670	289	21	79.4	79.4	NUM
dad-10670	289	22	%	%	NOUN
dad-10670	289	23	.	.	PUNCT
dad-10670	290	1	with	with	ADP
dad-10670	290	2	this	this	DET
dad-10670	290	3	set	set	NOUN
dad-10670	290	4	of	of	ADP
dad-10670	290	5	relevant	relevant	ADJ
dad-10670	290	6	features	feature	NOUN
dad-10670	290	7	defined	define	VERB
dad-10670	290	8	,	,	PUNCT
dad-10670	290	9	we	we	PRON
dad-10670	290	10	now	now	ADV
dad-10670	290	11	increase	increase	VERB
dad-10670	290	12	the	the	DET
dad-10670	290	13	coverage	coverage	NOUN
dad-10670	290	14	by	by	ADP
dad-10670	290	15	also	also	ADV
dad-10670	290	16	including	include	VERB
dad-10670	290	17	features	feature	NOUN
dad-10670	290	18	extracted	extract	VERB
dad-10670	290	19	from	from	ADP
dad-10670	290	20	the	the	DET
dad-10670	290	21	previous	previous	ADJ
dad-10670	290	22	word	word	NOUN
dad-10670	290	23	in	in	ADP
dad-10670	290	24	our	our	PRON
dad-10670	290	25	feature	feature	NOUN
dad-10670	290	26	set	set	VERB
dad-10670	290	27	.	.	PUNCT
dad-10670	291	1	with	with	ADP
dad-10670	291	2	these	these	DET
dad-10670	291	3	features	feature	NOUN
dad-10670	291	4	,	,	PUNCT
dad-10670	291	5	now	now	ADV
dad-10670	291	6	taken	take	VERB
dad-10670	291	7	from	from	ADP
dad-10670	291	8	positions	position	NOUN
dad-10670	291	9	13	13	NUM
dad-10670	291	10	hernault	hernault	NOUN
dad-10670	291	11	,	,	PUNCT
dad-10670	291	12	prendinger	prendinger	NOUN
dad-10670	291	13	,	,	PUNCT
dad-10670	291	14	duverle	duverle	NOUN
dad-10670	291	15	,	,	PUNCT
dad-10670	291	16	and	and	CCONJ
dad-10670	291	17	ishizuka	ishizuka	PROPN
dad-10670	291	18	(	(	PUNCT
dad-10670	291	19	−1	−1	NOUN
dad-10670	291	20	,	,	PUNCT
dad-10670	291	21	0	0	NUM
dad-10670	291	22	)	)	PUNCT
dad-10670	291	23	,	,	PUNCT
dad-10670	291	24	we	we	PRON
dad-10670	291	25	notice	notice	VERB
dad-10670	291	26	a	a	DET
dad-10670	291	27	sharp	sharp	ADJ
dad-10670	291	28	f	f	ADJ
dad-10670	291	29	-	-	PUNCT
dad-10670	291	30	score	score	NOUN
dad-10670	291	31	increase	increase	NOUN
dad-10670	291	32	of	of	ADP
dad-10670	291	33	14.5	14.5	NUM
dad-10670	291	34	%	%	NOUN
dad-10670	291	35	.	.	PUNCT
dad-10670	292	1	in	in	ADP
dad-10670	292	2	particular	particular	ADJ
dad-10670	292	3	,	,	PUNCT
dad-10670	292	4	precision	precision	NOUN
dad-10670	292	5	reaches	reach	VERB
dad-10670	292	6	93.3	93.3	NUM
dad-10670	292	7	%	%	NOUN
dad-10670	292	8	and	and	CCONJ
dad-10670	292	9	recall	recall	VERB
dad-10670	292	10	91.5	91.5	NUM
dad-10670	292	11	%	%	NOUN
dad-10670	292	12	.	.	PUNCT
dad-10670	293	1	finally	finally	ADV
dad-10670	293	2	,	,	PUNCT
dad-10670	293	3	we	we	PRON
dad-10670	293	4	add	add	VERB
dad-10670	293	5	the	the	DET
dad-10670	293	6	contextual	contextual	ADJ
dad-10670	293	7	features	feature	NOUN
dad-10670	293	8	for	for	ADP
dad-10670	293	9	the	the	DET
dad-10670	293	10	word	word	NOUN
dad-10670	293	11	at	at	ADP
dad-10670	293	12	position	position	NOUN
dad-10670	293	13	i	i	PRON
dad-10670	293	14	−	−	VERB
dad-10670	293	15	2	2	NUM
dad-10670	293	16	to	to	ADP
dad-10670	293	17	the	the	DET
dad-10670	293	18	feature	feature	NOUN
dad-10670	293	19	set	set	VERB
dad-10670	293	20	.	.	PUNCT
dad-10670	294	1	the	the	DET
dad-10670	294	2	f	f	NOUN
dad-10670	294	3	-	-	PUNCT
dad-10670	294	4	score	score	NOUN
dad-10670	294	5	increase	increase	NOUN
dad-10670	294	6	in	in	ADP
dad-10670	294	7	this	this	DET
dad-10670	294	8	case	case	NOUN
dad-10670	294	9	is	be	AUX
dad-10670	294	10	much	much	ADV
dad-10670	294	11	smaller	small	ADJ
dad-10670	294	12	(	(	PUNCT
dad-10670	294	13	1.5	1.5	NUM
dad-10670	294	14	%	%	NOUN
dad-10670	294	15	)	)	PUNCT
dad-10670	294	16	.	.	PUNCT
dad-10670	295	1	we	we	PRON
dad-10670	295	2	now	now	ADV
dad-10670	295	3	attain	attain	VERB
dad-10670	295	4	a	a	DET
dad-10670	295	5	precision	precision	NOUN
dad-10670	295	6	of	of	ADP
dad-10670	295	7	94.5	94.5	NUM
dad-10670	295	8	%	%	NOUN
dad-10670	295	9	and	and	CCONJ
dad-10670	295	10	a	a	DET
dad-10670	295	11	recall	recall	NOUN
dad-10670	295	12	of	of	ADP
dad-10670	295	13	93.1	93.1	NUM
dad-10670	295	14	%	%	NOUN
dad-10670	295	15	.	.	PUNCT
dad-10670	296	1	from	from	ADP
dad-10670	296	2	this	this	DET
dad-10670	296	3	point	point	NOUN
dad-10670	296	4	forward	forward	ADV
dad-10670	296	5	,	,	PUNCT
dad-10670	296	6	adding	add	VERB
dad-10670	296	7	features	feature	NOUN
dad-10670	296	8	from	from	ADP
dad-10670	296	9	anterior	anterior	ADJ
dad-10670	296	10	words	word	NOUN
dad-10670	296	11	did	do	AUX
dad-10670	296	12	not	not	PART
dad-10670	296	13	improve	improve	VERB
dad-10670	296	14	performance	performance	NOUN
dad-10670	296	15	,	,	PUNCT
dad-10670	296	16	although	although	SCONJ
dad-10670	296	17	it	it	PRON
dad-10670	296	18	significantly	significantly	ADV
dad-10670	296	19	increased	increase	VERB
dad-10670	296	20	training	training	NOUN
dad-10670	296	21	time	time	NOUN
dad-10670	296	22	.	.	PUNCT
dad-10670	297	1	finally	finally	ADV
dad-10670	297	2	,	,	PUNCT
dad-10670	297	3	we	we	PRON
dad-10670	297	4	did	do	AUX
dad-10670	297	5	not	not	PART
dad-10670	297	6	notice	notice	VERB
dad-10670	297	7	any	any	DET
dad-10670	297	8	significant	significant	ADJ
dad-10670	297	9	performance	performance	NOUN
dad-10670	297	10	increase	increase	NOUN
dad-10670	297	11	when	when	SCONJ
dad-10670	297	12	including	include	VERB
dad-10670	297	13	contextual	contextual	ADJ
dad-10670	297	14	features	feature	NOUN
dad-10670	297	15	for	for	ADP
dad-10670	297	16	words	word	NOUN
dad-10670	297	17	located	locate	VERB
dad-10670	297	18	ahead	ahead	ADV
dad-10670	297	19	of	of	ADP
dad-10670	297	20	the	the	DET
dad-10670	297	21	current	current	ADJ
dad-10670	297	22	position	position	NOUN
dad-10670	297	23	.	.	PUNCT
dad-10670	298	1	table	table	NOUN
dad-10670	298	2	4	4	NUM
dad-10670	298	3	:	:	PUNCT
dad-10670	298	4	segmentation	segmentation	NOUN
dad-10670	298	5	performance	performance	NOUN
dad-10670	298	6	with	with	ADP
dad-10670	298	7	different	different	ADJ
dad-10670	298	8	sets	set	NOUN
dad-10670	298	9	of	of	ADP
dad-10670	298	10	features	feature	NOUN
dad-10670	298	11	positions	position	NOUN
dad-10670	298	12	features	feature	VERB
dad-10670	298	13	p	p	NOUN
dad-10670	298	14	r	r	NOUN
dad-10670	298	15	f	f	X
dad-10670	298	16	(	(	PUNCT
dad-10670	298	17	0	0	NUM
dad-10670	298	18	)	)	PUNCT
dad-10670	298	19	w	w	NOUN
dad-10670	298	20	80.7	80.7	NUM
dad-10670	298	21	69.4	69.4	NUM
dad-10670	298	22	74.6	74.6	NUM
dad-10670	298	23	(	(	PUNCT
dad-10670	298	24	0	0	NUM
dad-10670	298	25	)	)	PUNCT
dad-10670	298	26	w	w	NOUN
dad-10670	298	27	,	,	PUNCT
dad-10670	298	28	pos(w	pos(w	ADJ
dad-10670	298	29	)	)	PUNCT
dad-10670	298	30	82.7	82.7	NUM
dad-10670	298	31	70.8	70.8	NUM
dad-10670	298	32	76.3	76.3	NUM
dad-10670	298	33	(	(	PUNCT
dad-10670	298	34	0	0	NUM
dad-10670	298	35	)	)	PUNCT
dad-10670	298	36	nw	nw	PROPN
dad-10670	298	37	,	,	PUNCT
dad-10670	298	38	np	np	INTJ
dad-10670	298	39	,	,	PUNCT
dad-10670	298	40	nr	nr	PROPN
dad-10670	298	41	82.0	82.0	NUM
dad-10670	298	42	78.4	78.4	NUM
dad-10670	298	43	80.2	80.2	NUM
dad-10670	298	44	(	(	PUNCT
dad-10670	298	45	0	0	NUM
dad-10670	298	46	)	)	PUNCT
dad-10670	298	47	w	w	NOUN
dad-10670	298	48	,	,	PUNCT
dad-10670	298	49	pos(w	pos(w	PROPN
dad-10670	298	50	)	)	PUNCT
dad-10670	298	51	,	,	PUNCT
dad-10670	298	52	nw	nw	PROPN
dad-10670	298	53	,	,	PUNCT
dad-10670	298	54	np	np	INTJ
dad-10670	298	55	,	,	PUNCT
dad-10670	298	56	nr	nr	PROPN
dad-10670	298	57	81.9	81.9	NUM
dad-10670	298	58	79.4	79.4	NUM
dad-10670	298	59	80.7	80.7	NUM
dad-10670	298	60	(	(	PUNCT
dad-10670	298	61	−1	−1	NOUN
dad-10670	298	62	,	,	PUNCT
dad-10670	298	63	0	0	NUM
dad-10670	298	64	)	)	PUNCT
dad-10670	298	65	w	w	NOUN
dad-10670	298	66	,	,	PUNCT
dad-10670	298	67	pos(w	pos(w	PROPN
dad-10670	298	68	)	)	PUNCT
dad-10670	298	69	,	,	PUNCT
dad-10670	298	70	nw	nw	PROPN
dad-10670	298	71	,	,	PUNCT
dad-10670	298	72	np	np	INTJ
dad-10670	298	73	,	,	PUNCT
dad-10670	298	74	nr	nr	PROPN
dad-10670	298	75	93.3	93.3	NUM
dad-10670	298	76	91.5	91.5	NUM
dad-10670	298	77	92.4	92.4	NUM
dad-10670	298	78	(	(	PUNCT
dad-10670	298	79	−2,−1	−2,−1	ADJ
dad-10670	298	80	,	,	PUNCT
dad-10670	298	81	0	0	NUM
dad-10670	298	82	)	)	PUNCT
dad-10670	298	83	w	w	NOUN
dad-10670	298	84	,	,	PUNCT
dad-10670	298	85	pos(w	pos(w	PROPN
dad-10670	298	86	)	)	PUNCT
dad-10670	298	87	,	,	PUNCT
dad-10670	298	88	nw	nw	PROPN
dad-10670	298	89	,	,	PUNCT
dad-10670	298	90	np	np	INTJ
dad-10670	298	91	,	,	PUNCT
dad-10670	298	92	nr	nr	PROPN
dad-10670	298	93	94.5	94.5	NUM
dad-10670	298	94	93.1	93.1	NUM
dad-10670	298	95	93.8	93.8	NUM
dad-10670	298	96	4.2	4.2	NUM
dad-10670	298	97	relation	relation	NOUN
dad-10670	298	98	labeling	labeling	NOUN
dad-10670	298	99	although	although	SCONJ
dad-10670	298	100	our	our	PRON
dad-10670	298	101	final	final	ADJ
dad-10670	298	102	goal	goal	NOUN
dad-10670	298	103	is	be	AUX
dad-10670	298	104	to	to	PART
dad-10670	298	105	achieve	achieve	VERB
dad-10670	298	106	good	good	ADJ
dad-10670	298	107	performance	performance	NOUN
dad-10670	298	108	on	on	ADP
dad-10670	298	109	the	the	DET
dad-10670	298	110	entire	entire	ADJ
dad-10670	298	111	tree	tree	NOUN
dad-10670	298	112	-	-	PUNCT
dad-10670	298	113	building	building	NOUN
dad-10670	298	114	task	task	NOUN
dad-10670	298	115	,	,	PUNCT
dad-10670	298	116	a	a	DET
dad-10670	298	117	useful	useful	ADJ
dad-10670	298	118	intermediate	intermediate	ADJ
dad-10670	298	119	evaluation	evaluation	NOUN
dad-10670	298	120	of	of	ADP
dad-10670	298	121	our	our	PRON
dad-10670	298	122	system	system	NOUN
dad-10670	298	123	can	can	AUX
dad-10670	298	124	be	be	AUX
dad-10670	298	125	conducted	conduct	VERB
dad-10670	298	126	by	by	ADP
dad-10670	298	127	measuring	measure	VERB
dad-10670	298	128	the	the	DET
dad-10670	298	129	raw	raw	ADJ
dad-10670	298	130	performance	performance	NOUN
dad-10670	298	131	of	of	ADP
dad-10670	298	132	our	our	PRON
dad-10670	298	133	svm	svm	ADJ
dad-10670	298	134	classifiers	classifier	NOUN
dad-10670	298	135	.	.	PUNCT
dad-10670	299	1	the	the	DET
dad-10670	299	2	binary	binary	PROPN
dad-10670	299	3	classifier	classifier	PROPN
dad-10670	299	4	struct	struct	NOUN
dad-10670	299	5	is	be	AUX
dad-10670	299	6	trained	train	VERB
dad-10670	299	7	on	on	ADP
dad-10670	299	8	52,683	52,683	NUM
dad-10670	299	9	instances	instance	NOUN
dad-10670	299	10	(	(	PUNCT
dad-10670	299	11	split	split	VERB
dad-10670	299	12	approximately	approximately	ADV
dad-10670	299	13	between	between	ADP
dad-10670	299	14	1/3	1/3	NUM
dad-10670	299	15	positive	positive	ADJ
dad-10670	299	16	and	and	CCONJ
dad-10670	299	17	2/3	2/3	NUM
dad-10670	299	18	negative	negative	ADJ
dad-10670	299	19	examples	example	NOUN
dad-10670	299	20	)	)	PUNCT
dad-10670	299	21	,	,	PUNCT
dad-10670	299	22	extracted	extract	VERB
dad-10670	299	23	from	from	ADP
dad-10670	299	24	341	341	NUM
dad-10670	299	25	documents	document	NOUN
dad-10670	299	26	,	,	PUNCT
dad-10670	299	27	and	and	CCONJ
dad-10670	299	28	tested	test	VERB
dad-10670	299	29	on	on	ADP
dad-10670	299	30	8,558	8,558	NUM
dad-10670	299	31	instances	instance	NOUN
dad-10670	299	32	extracted	extract	VERB
dad-10670	299	33	from	from	ADP
dad-10670	299	34	38	38	NUM
dad-10670	299	35	documents	document	NOUN
dad-10670	299	36	.	.	PUNCT
dad-10670	300	1	the	the	DET
dad-10670	300	2	feature	feature	NOUN
dad-10670	300	3	space	space	NOUN
dad-10670	300	4	has	have	VERB
dad-10670	300	5	a	a	DET
dad-10670	300	6	dimension	dimension	NOUN
dad-10670	300	7	of	of	ADP
dad-10670	300	8	136,987	136,987	NUM
dad-10670	300	9	.	.	PUNCT
dad-10670	301	1	the	the	DET
dad-10670	301	2	other	other	ADJ
dad-10670	301	3	classifier	classifier	NOUN
dad-10670	301	4	,	,	PUNCT
dad-10670	301	5	label	label	NOUN
dad-10670	301	6	,	,	PUNCT
dad-10670	301	7	is	be	AUX
dad-10670	301	8	trained	train	VERB
dad-10670	301	9	on	on	ADP
dad-10670	301	10	17,742	17,742	NUM
dad-10670	301	11	instances	instance	NOUN
dad-10670	301	12	(	(	PUNCT
dad-10670	301	13	labeled	label	VERB
dad-10670	301	14	across	across	ADP
dad-10670	301	15	41	41	NUM
dad-10670	301	16	classes	class	NOUN
dad-10670	301	17	)	)	PUNCT
dad-10670	301	18	and	and	CCONJ
dad-10670	301	19	tested	test	VERB
dad-10670	301	20	on	on	ADP
dad-10670	301	21	2,887	2,887	NUM
dad-10670	301	22	instances	instance	NOUN
dad-10670	301	23	,	,	PUNCT
dad-10670	301	24	the	the	DET
dad-10670	301	25	same	same	ADJ
dad-10670	301	26	feature	feature	NOUN
dad-10670	301	27	vector	vector	NOUN
dad-10670	301	28	dimension	dimension	NOUN
dad-10670	301	29	as	as	ADP
dad-10670	301	30	struct	struct	NOUN
dad-10670	301	31	.	.	PUNCT
dad-10670	302	1	the	the	DET
dad-10670	302	2	optimal	optimal	ADJ
dad-10670	302	3	training	training	NOUN
dad-10670	302	4	parameters	parameter	NOUN
dad-10670	302	5	for	for	ADP
dad-10670	302	6	each	each	DET
dad-10670	302	7	kernel	kernel	PROPN
dad-10670	302	8	function	function	NOUN
dad-10670	302	9	are	be	AUX
dad-10670	302	10	obtained	obtain	VERB
dad-10670	302	11	through	through	ADP
dad-10670	302	12	automated	automate	VERB
dad-10670	302	13	grid	grid	NOUN
dad-10670	302	14	search	search	NOUN
dad-10670	302	15	with	with	ADP
dad-10670	302	16	5	5	NUM
dad-10670	302	17	-	-	ADJ
dad-10670	302	18	fold	fold	ADJ
dad-10670	302	19	cross	cross	NOUN
dad-10670	302	20	-	-	NOUN
dad-10670	302	21	validation	validation	ADJ
dad-10670	302	22	(	(	PUNCT
dad-10670	302	23	staelin	staelin	NOUN
dad-10670	302	24	,	,	PUNCT
dad-10670	302	25	2003	2003	NUM
dad-10670	302	26	)	)	PUNCT
dad-10670	302	27	.	.	PUNCT
dad-10670	303	1	table	table	NOUN
dad-10670	303	2	5	5	NUM
dad-10670	303	3	shows	show	VERB
dad-10670	303	4	the	the	DET
dad-10670	303	5	training	training	NOUN
dad-10670	303	6	time	time	NOUN
dad-10670	303	7	and	and	CCONJ
dad-10670	303	8	accuracy	accuracy	NOUN
dad-10670	303	9	for	for	ADP
dad-10670	303	10	these	these	DET
dad-10670	303	11	experiments	experiment	NOUN
dad-10670	303	12	,	,	PUNCT
dad-10670	303	13	using	use	VERB
dad-10670	303	14	various	various	ADJ
dad-10670	303	15	software	software	NOUN
dad-10670	303	16	and	and	CCONJ
dad-10670	303	17	selected	select	VERB
dad-10670	303	18	different	different	ADJ
dad-10670	303	19	kernels	kernel	NOUN
dad-10670	303	20	,	,	PUNCT
dad-10670	303	21	for	for	ADP
dad-10670	303	22	both	both	DET
dad-10670	303	23	struct	struct	NOUN
dad-10670	303	24	and	and	CCONJ
dad-10670	303	25	label	label	NOUN
dad-10670	303	26	,	,	PUNCT
dad-10670	303	27	as	as	ADV
dad-10670	303	28	well	well	ADV
dad-10670	303	29	as	as	ADP
dad-10670	303	30	a	a	DET
dad-10670	303	31	comparison	comparison	NOUN
dad-10670	303	32	to	to	ADP
dad-10670	303	33	the	the	DET
dad-10670	303	34	results	result	NOUN
dad-10670	303	35	of	of	ADP
dad-10670	303	36	reitter	reitter	NOUN
dad-10670	303	37	(	(	PUNCT
dad-10670	303	38	2003a	2003a	NUM
dad-10670	303	39	)	)	PUNCT
dad-10670	303	40	.	.	PUNCT
dad-10670	304	1	for	for	ADP
dad-10670	304	2	struct	struct	NOUN
dad-10670	304	3	we	we	PRON
dad-10670	304	4	used	use	VERB
dad-10670	304	5	the	the	DET
dad-10670	304	6	regression	regression	NOUN
dad-10670	304	7	mode	mode	NOUN
dad-10670	304	8	of	of	ADP
dad-10670	304	9	the	the	DET
dad-10670	304	10	svm	svm	ADJ
dad-10670	304	11	software	software	NOUN
dad-10670	304	12	to	to	PART
dad-10670	304	13	obtain	obtain	VERB
dad-10670	304	14	probabilistic	probabilistic	ADJ
dad-10670	304	15	output	output	NOUN
dad-10670	304	16	.	.	PUNCT
dad-10670	305	1	note	note	VERB
dad-10670	305	2	that	that	SCONJ
dad-10670	305	3	in	in	ADP
dad-10670	305	4	table	table	NOUN
dad-10670	305	5	5	5	NUM
dad-10670	305	6	,	,	PUNCT
dad-10670	305	7	the	the	DET
dad-10670	305	8	results	result	NOUN
dad-10670	305	9	for	for	ADP
dad-10670	305	10	struct	struct	NOUN
dad-10670	305	11	and	and	CCONJ
dad-10670	305	12	label	label	NOUN
dad-10670	305	13	are	be	AUX
dad-10670	305	14	independent	independent	ADJ
dad-10670	305	15	,	,	PUNCT
dad-10670	305	16	as	as	SCONJ
dad-10670	305	17	they	they	PRON
dad-10670	305	18	indicate	indicate	VERB
dad-10670	305	19	performance	performance	NOUN
dad-10670	305	20	for	for	ADP
dad-10670	305	21	cross	cross	NOUN
dad-10670	305	22	-	-	NOUN
dad-10670	305	23	validation	validation	NOUN
dad-10670	305	24	.	.	PUNCT
dad-10670	306	1	14	14	NUM
dad-10670	306	2	hilda	hilda	VERB
dad-10670	306	3	:	:	PUNCT
dad-10670	306	4	a	a	DET
dad-10670	306	5	discourse	discourse	NOUN
dad-10670	306	6	parser	parser	NOUN
dad-10670	306	7	using	use	VERB
dad-10670	306	8	support	support	NOUN
dad-10670	306	9	vector	vector	NOUN
dad-10670	306	10	machine	machine	NOUN
dad-10670	306	11	classification	classification	NOUN
dad-10670	306	12	table	table	NOUN
dad-10670	306	13	5	5	NUM
dad-10670	306	14	:	:	PUNCT
dad-10670	306	15	svm	svm	PROPN
dad-10670	306	16	classifiers	classifier	NOUN
dad-10670	306	17	performances	performance	NOUN
dad-10670	306	18	–	–	PUNCT
dad-10670	306	19	ll	ll	NOUN
dad-10670	306	20	:	:	PUNCT
dad-10670	306	21	liblineara	liblineara	VERB
dad-10670	306	22	,	,	PUNCT
dad-10670	306	23	svml	svml	ADJ
dad-10670	306	24	:	:	PUNCT
dad-10670	306	25	svm	svm	VERB
dad-10670	306	26	lightb	lightb	NOUN
dad-10670	306	27	,	,	PUNCT
dad-10670	306	28	svmm	svmm	NOUN
dad-10670	306	29	:	:	PUNCT
dad-10670	306	30	svm	svm	PROPN
dad-10670	306	31	multiclassc	multiclassc	PROPN
dad-10670	306	32	,	,	PUNCT
dad-10670	306	33	ls	ls	PROPN
dad-10670	306	34	:	:	PUNCT
dad-10670	306	35	libsvmd	libsvmd	PROPN
dad-10670	306	36	classifier	classifier	PROPN
dad-10670	306	37	struct	struct	VERB
dad-10670	306	38	label	label	NOUN
dad-10670	306	39	reitter	reitter	NOUN
dad-10670	306	40	kernel	kernel	PROPN
dad-10670	306	41	linear	linear	PROPN
dad-10670	306	42	polyn	polyn	PROPN
dad-10670	306	43	.	.	PUNCT
dad-10670	307	1	rbf	rbf	PROPN
dad-10670	307	2	linear	linear	PROPN
dad-10670	307	3	rbf	rbf	PROPN
dad-10670	307	4	rbf	rbf	PROPN
dad-10670	307	5	software	software	PROPN
dad-10670	307	6	ll	ll	AUX
dad-10670	307	7	svml	svml	PROPN
dad-10670	307	8	svml	svml	PROPN
dad-10670	307	9	svmm	svmm	VERB
dad-10670	307	10	ls	ls	PROPN
dad-10670	307	11	svml	svml	ADJ
dad-10670	307	12	training	training	NOUN
dad-10670	307	13	time	time	NOUN
dad-10670	307	14	21.4s	21.4s	NUM
dad-10670	307	15	5m53s	5m53s	NUM
dad-10670	307	16	12	12	NUM
dad-10670	307	17	m	m	NUM
dad-10670	307	18	15	15	NUM
dad-10670	307	19	m	m	NOUN
dad-10670	307	20	23	23	NUM
dad-10670	307	21	m	m	NUM
dad-10670	307	22	216	216	NUM
dad-10670	307	23	m	m	NOUN
dad-10670	307	24	accuracy	accuracy	NOUN
dad-10670	307	25	82.2	82.2	NUM
dad-10670	307	26	85.0	85.0	NUM
dad-10670	307	27	82.9	82.9	NUM
dad-10670	307	28	65.8	65.8	NUM
dad-10670	307	29	66.8	66.8	NUM
dad-10670	307	30	61.0	61.0	NUM
dad-10670	307	31	a	a	DET
dad-10670	307	32	http://www.csie.ntu.edu.tw/˜cjlin/liblinear/	http://www.csie.ntu.edu.tw/˜cjlin/liblinear/	NOUN
dad-10670	307	33	b	b	PROPN
dad-10670	307	34	http://svmlight.joachims.org/	http://svmlight.joachims.org/	NOUN
dad-10670	307	35	c	c	PROPN
dad-10670	307	36	http://svmlight.joachims.org/svm	http://svmlight.joachims.org/svm	VERB
dad-10670	308	1	multiclass.html	multiclass.html	NOUN
dad-10670	308	2	d	d	PROPN
dad-10670	308	3	http://www.csie.ntu.edu.tw/˜cjlin/libsvm/	http://www.csie.ntu.edu.tw/˜cjlin/libsvm/	NOUN
dad-10670	308	4	despite	despite	SCONJ
dad-10670	308	5	their	their	PRON
dad-10670	308	6	simplicity	simplicity	NOUN
dad-10670	308	7	,	,	PUNCT
dad-10670	308	8	the	the	DET
dad-10670	308	9	noticeably	noticeably	ADV
dad-10670	308	10	good	good	ADJ
dad-10670	308	11	performances	performance	NOUN
dad-10670	308	12	of	of	ADP
dad-10670	308	13	linear	linear	ADJ
dad-10670	308	14	kernel	kernel	NOUN
dad-10670	308	15	methods	method	NOUN
dad-10670	308	16	in	in	ADP
dad-10670	308	17	the	the	DET
dad-10670	308	18	results	result	NOUN
dad-10670	308	19	presented	present	VERB
dad-10670	308	20	in	in	ADP
dad-10670	308	21	table	table	NOUN
dad-10670	308	22	5	5	NUM
dad-10670	308	23	,	,	PUNCT
dad-10670	308	24	as	as	SCONJ
dad-10670	308	25	compared	compare	VERB
dad-10670	308	26	to	to	ADP
dad-10670	308	27	the	the	DET
dad-10670	308	28	more	more	ADV
dad-10670	308	29	complex	complex	ADJ
dad-10670	308	30	polynomial	polynomial	ADJ
dad-10670	308	31	and	and	CCONJ
dad-10670	308	32	rbf	rbf	PROPN
dad-10670	308	33	kernels	kernels	PROPN
dad-10670	308	34	,	,	PUNCT
dad-10670	308	35	indicate	indicate	VERB
dad-10670	308	36	that	that	SCONJ
dad-10670	308	37	our	our	PRON
dad-10670	308	38	data	datum	NOUN
dad-10670	308	39	separates	separate	VERB
dad-10670	308	40	fairly	fairly	ADV
dad-10670	308	41	well	well	ADV
dad-10670	308	42	linearly	linearly	ADV
dad-10670	308	43	.	.	PUNCT
dad-10670	309	1	this	this	PRON
dad-10670	309	2	is	be	AUX
dad-10670	309	3	a	a	DET
dad-10670	309	4	commonly	commonly	ADV
dad-10670	309	5	observed	observe	VERB
dad-10670	309	6	effect	effect	NOUN
dad-10670	309	7	of	of	ADP
dad-10670	309	8	high	high	ADJ
dad-10670	309	9	-	-	PUNCT
dad-10670	309	10	dimensional	dimensional	ADJ
dad-10670	309	11	input	input	NOUN
dad-10670	309	12	such	such	ADJ
dad-10670	309	13	as	as	ADP
dad-10670	309	14	ours	our	NOUN
dad-10670	309	15	(	(	PUNCT
dad-10670	309	16	over	over	ADP
dad-10670	309	17	100,000	100,000	NUM
dad-10670	309	18	features	feature	NOUN
dad-10670	309	19	)	)	PUNCT
dad-10670	309	20	(	(	PUNCT
dad-10670	309	21	chen	chen	PROPN
dad-10670	309	22	et	et	PROPN
dad-10670	309	23	al	al	PROPN
dad-10670	309	24	.	.	PROPN
dad-10670	309	25	,	,	PUNCT
dad-10670	309	26	2007	2007	NUM
dad-10670	309	27	)	)	PUNCT
dad-10670	309	28	.	.	PUNCT
dad-10670	310	1	reitter	reitter	NOUN
dad-10670	310	2	(	(	PUNCT
dad-10670	310	3	2003a	2003a	NUM
dad-10670	310	4	)	)	PUNCT
dad-10670	310	5	provides	provide	VERB
dad-10670	310	6	a	a	DET
dad-10670	310	7	baseline	baseline	NOUN
dad-10670	310	8	for	for	ADP
dad-10670	310	9	absolute	absolute	ADJ
dad-10670	310	10	comparison	comparison	NOUN
dad-10670	310	11	on	on	ADP
dad-10670	310	12	the	the	DET
dad-10670	310	13	multi	multi	ADJ
dad-10670	310	14	-	-	ADJ
dad-10670	310	15	label	label	ADJ
dad-10670	310	16	classification	classification	NOUN
dad-10670	310	17	task	task	NOUN
dad-10670	310	18	for	for	ADP
dad-10670	310	19	a	a	DET
dad-10670	310	20	similar	similar	ADJ
dad-10670	310	21	classifier	classifier	NOUN
dad-10670	310	22	that	that	PRON
dad-10670	310	23	assumes	assume	VERB
dad-10670	310	24	perfect	perfect	ADJ
dad-10670	310	25	segmentation	segmentation	NOUN
dad-10670	310	26	of	of	ADP
dad-10670	310	27	the	the	DET
dad-10670	310	28	input	input	NOUN
dad-10670	310	29	,	,	PUNCT
dad-10670	310	30	as	as	SCONJ
dad-10670	310	31	ours	ours	PRON
dad-10670	310	32	does	do	VERB
dad-10670	310	33	.	.	PUNCT
dad-10670	311	1	reitter	reitter	PROPN
dad-10670	311	2	’s	’s	PART
dad-10670	311	3	accuracy	accuracy	NOUN
dad-10670	311	4	results	result	NOUN
dad-10670	311	5	of	of	ADP
dad-10670	311	6	61	61	NUM
dad-10670	311	7	%	%	NOUN
dad-10670	311	8	for	for	ADP
dad-10670	311	9	a	a	DET
dad-10670	311	10	smaller	small	ADJ
dad-10670	311	11	set	set	NOUN
dad-10670	311	12	of	of	ADP
dad-10670	311	13	training	training	NOUN
dad-10670	311	14	instances	instance	NOUN
dad-10670	311	15	(	(	PUNCT
dad-10670	311	16	7976	7976	NUM
dad-10670	311	17	instances	instance	NOUN
dad-10670	311	18	from	from	ADP
dad-10670	311	19	240	240	NUM
dad-10670	311	20	documents	document	NOUN
dad-10670	311	21	compared	compare	VERB
dad-10670	311	22	to	to	ADP
dad-10670	311	23	17,742	17,742	NUM
dad-10670	311	24	instances	instance	NOUN
dad-10670	311	25	in	in	ADP
dad-10670	311	26	our	our	PRON
dad-10670	311	27	case	case	NOUN
dad-10670	311	28	)	)	PUNCT
dad-10670	311	29	with	with	ADP
dad-10670	311	30	considerably	considerably	ADV
dad-10670	311	31	less	less	ADJ
dad-10670	311	32	classes	class	NOUN
dad-10670	311	33	(	(	PUNCT
dad-10670	311	34	16	16	NUM
dad-10670	311	35	rhetorical	rhetorical	ADJ
dad-10670	311	36	relation	relation	NOUN
dad-10670	311	37	labels	label	NOUN
dad-10670	311	38	with	with	ADP
dad-10670	311	39	no	no	DET
dad-10670	311	40	nuclearity	nuclearity	NOUN
dad-10670	311	41	,	,	PUNCT
dad-10670	311	42	as	as	SCONJ
dad-10670	311	43	opposed	oppose	VERB
dad-10670	311	44	to	to	ADP
dad-10670	311	45	our	our	PRON
dad-10670	311	46	41	41	NUM
dad-10670	311	47	nuclearized	nuclearize	VERB
dad-10670	311	48	relation	relation	NOUN
dad-10670	311	49	classes	class	NOUN
dad-10670	311	50	)	)	PUNCT
dad-10670	311	51	seems	seem	VERB
dad-10670	311	52	to	to	PART
dad-10670	311	53	indicate	indicate	VERB
dad-10670	311	54	that	that	SCONJ
dad-10670	311	55	this	this	DET
dad-10670	311	56	sub	sub	NOUN
dad-10670	311	57	-	-	NOUN
dad-10670	311	58	component	component	NOUN
dad-10670	311	59	of	of	ADP
dad-10670	311	60	our	our	PRON
dad-10670	311	61	system	system	NOUN
dad-10670	311	62	with	with	ADP
dad-10670	311	63	an	an	DET
dad-10670	311	64	accuracy	accuracy	NOUN
dad-10670	311	65	of	of	ADP
dad-10670	311	66	66.8	66.8	NUM
dad-10670	311	67	%	%	NOUN
dad-10670	311	68	performs	perform	VERB
dad-10670	311	69	well	well	ADV
dad-10670	311	70	.	.	PUNCT
dad-10670	312	1	in	in	ADP
dad-10670	312	2	hilda	hilda	PROPN
dad-10670	312	3	,	,	PUNCT
dad-10670	312	4	we	we	PRON
dad-10670	312	5	select	select	VERB
dad-10670	312	6	a	a	DET
dad-10670	312	7	linear	linear	ADJ
dad-10670	312	8	kernel	kernel	NOUN
dad-10670	312	9	for	for	ADP
dad-10670	312	10	struct	struct	NOUN
dad-10670	312	11	and	and	CCONJ
dad-10670	312	12	an	an	DET
dad-10670	312	13	optimally	optimally	ADV
dad-10670	312	14	parameterized	parameterized	ADJ
dad-10670	312	15	rbf	rbf	PROPN
dad-10670	312	16	kernel	kernel	PROPN
dad-10670	312	17	for	for	ADP
dad-10670	312	18	lab	lab	NOUN
dad-10670	312	19	,	,	PUNCT
dad-10670	312	20	considering	consider	VERB
dad-10670	312	21	performance	performance	NOUN
dad-10670	312	22	and	and	CCONJ
dad-10670	312	23	runtime	runtime	NOUN
dad-10670	312	24	complexity	complexity	NOUN
dad-10670	312	25	.	.	PUNCT
dad-10670	313	1	we	we	PRON
dad-10670	313	2	use	use	VERB
dad-10670	313	3	modified	modified	ADJ
dad-10670	313	4	versions	version	NOUN
dad-10670	313	5	of	of	ADP
dad-10670	313	6	the	the	DET
dad-10670	313	7	liblinear	liblinear	NOUN
dad-10670	313	8	(	(	PUNCT
dad-10670	313	9	optimized	optimize	VERB
dad-10670	313	10	for	for	ADP
dad-10670	313	11	large	large	ADJ
dad-10670	313	12	-	-	PUNCT
dad-10670	313	13	instances	instance	NOUN
dad-10670	313	14	)	)	PUNCT
dad-10670	313	15	and	and	CCONJ
dad-10670	313	16	libsvm	libsvm	VERB
dad-10670	313	17	packages	package	NOUN
dad-10670	313	18	.	.	PUNCT
dad-10670	314	1	all	all	DET
dad-10670	314	2	further	further	ADJ
dad-10670	314	3	evaluation	evaluation	NOUN
dad-10670	314	4	results	result	NOUN
dad-10670	314	5	reported	report	VERB
dad-10670	314	6	in	in	ADP
dad-10670	314	7	this	this	DET
dad-10670	314	8	article	article	NOUN
dad-10670	314	9	were	be	AUX
dad-10670	314	10	conducted	conduct	VERB
dad-10670	314	11	using	use	VERB
dad-10670	314	12	these	these	DET
dad-10670	314	13	kernels	kernel	NOUN
dad-10670	314	14	.	.	PUNCT
dad-10670	315	1	to	to	PART
dad-10670	315	2	give	give	VERB
dad-10670	315	3	a	a	DET
dad-10670	315	4	more	more	ADV
dad-10670	315	5	precise	precise	ADJ
dad-10670	315	6	idea	idea	NOUN
dad-10670	315	7	of	of	ADP
dad-10670	315	8	the	the	DET
dad-10670	315	9	classifiers	classifier	NOUN
dad-10670	315	10	’	’	PART
dad-10670	315	11	performance	performance	NOUN
dad-10670	315	12	,	,	PUNCT
dad-10670	315	13	we	we	PRON
dad-10670	315	14	look	look	VERB
dad-10670	315	15	into	into	ADP
dad-10670	315	16	the	the	DET
dad-10670	315	17	results	result	NOUN
dad-10670	315	18	for	for	ADP
dad-10670	315	19	each	each	DET
dad-10670	315	20	class	class	NOUN
dad-10670	315	21	,	,	PUNCT
dad-10670	315	22	when	when	SCONJ
dad-10670	315	23	trained	train	VERB
dad-10670	315	24	on	on	ADP
dad-10670	315	25	rst	rst	PROPN
dad-10670	315	26	-	-	PUNCT
dad-10670	315	27	dt	dt	PROPN
dad-10670	315	28	’s	’s	PART
dad-10670	315	29	standard	standard	ADJ
dad-10670	315	30	training	training	NOUN
dad-10670	315	31	set	set	NOUN
dad-10670	315	32	and	and	CCONJ
dad-10670	315	33	evaluated	evaluate	VERB
dad-10670	315	34	on	on	ADP
dad-10670	315	35	the	the	DET
dad-10670	315	36	standard	standard	ADJ
dad-10670	315	37	test	test	NOUN
dad-10670	315	38	set	set	VERB
dad-10670	315	39	.	.	PUNCT
dad-10670	316	1	table	table	NOUN
dad-10670	316	2	6	6	NUM
dad-10670	316	3	indicates	indicate	VERB
dad-10670	316	4	the	the	DET
dad-10670	316	5	results	result	NOUN
dad-10670	316	6	for	for	ADP
dad-10670	316	7	struct	struct	NOUN
dad-10670	316	8	,	,	PUNCT
dad-10670	316	9	while	while	SCONJ
dad-10670	316	10	table	table	NOUN
dad-10670	316	11	7	7	NUM
dad-10670	316	12	indicates	indicate	VERB
dad-10670	316	13	the	the	DET
dad-10670	316	14	results	result	NOUN
dad-10670	316	15	for	for	ADP
dad-10670	316	16	label	label	NOUN
dad-10670	316	17	.	.	PUNCT
dad-10670	317	1	in	in	ADP
dad-10670	317	2	table	table	NOUN
dad-10670	317	3	6	6	NUM
dad-10670	317	4	,	,	PUNCT
dad-10670	317	5	we	we	PRON
dad-10670	317	6	notice	notice	VERB
dad-10670	317	7	in	in	ADP
dad-10670	317	8	particular	particular	ADJ
dad-10670	317	9	that	that	SCONJ
dad-10670	317	10	the	the	DET
dad-10670	317	11	performance	performance	NOUN
dad-10670	317	12	for	for	ADP
dad-10670	317	13	the	the	DET
dad-10670	317	14	class	class	NOUN
dad-10670	317	15	‘	'	PUNCT
dad-10670	317	16	+1	+1	NOUN
dad-10670	317	17	’	'	PUNCT
dad-10670	317	18	,	,	PUNCT
dad-10670	317	19	corresponding	correspond	VERB
dad-10670	317	20	to	to	ADP
dad-10670	317	21	the	the	DET
dad-10670	317	22	presence	presence	NOUN
dad-10670	317	23	of	of	ADP
dad-10670	317	24	a	a	DET
dad-10670	317	25	structural	structural	ADJ
dad-10670	317	26	relation	relation	NOUN
dad-10670	317	27	between	between	ADP
dad-10670	317	28	to	to	ADP
dad-10670	317	29	rs	r	NOUN
dad-10670	317	30	-	-	PUNCT
dad-10670	317	31	trees	tree	NOUN
dad-10670	317	32	is	be	AUX
dad-10670	317	33	lower	low	ADJ
dad-10670	317	34	(	(	PUNCT
dad-10670	317	35	f	f	X
dad-10670	317	36	-	-	PUNCT
dad-10670	317	37	score	score	NOUN
dad-10670	317	38	of	of	ADP
dad-10670	317	39	73.3	73.3	NUM
dad-10670	317	40	%	%	NOUN
dad-10670	317	41	)	)	PUNCT
dad-10670	317	42	than	than	ADP
dad-10670	317	43	for	for	ADP
dad-10670	317	44	the	the	DET
dad-10670	317	45	absence	absence	NOUN
dad-10670	317	46	of	of	ADP
dad-10670	317	47	a	a	DET
dad-10670	317	48	relation	relation	NOUN
dad-10670	317	49	(	(	PUNCT
dad-10670	317	50	f	f	X
dad-10670	317	51	-	-	PUNCT
dad-10670	317	52	score	score	NOUN
dad-10670	317	53	of	of	ADP
dad-10670	317	54	86.7	86.7	NUM
dad-10670	317	55	%	%	NOUN
dad-10670	317	56	)	)	PUNCT
dad-10670	317	57	.	.	PUNCT
dad-10670	318	1	because	because	SCONJ
dad-10670	318	2	rst	rst	VERB
dad-10670	318	3	-	-	PUNCT
dad-10670	318	4	dt	dt	PROPN
dad-10670	318	5	’s	’s	PART
dad-10670	318	6	test	test	NOUN
dad-10670	318	7	set	set	NOUN
dad-10670	318	8	is	be	AUX
dad-10670	318	9	relatively	relatively	ADV
dad-10670	318	10	small	small	ADJ
dad-10670	318	11	(	(	PUNCT
dad-10670	318	12	38	38	NUM
dad-10670	318	13	documents	document	NOUN
dad-10670	318	14	)	)	PUNCT
dad-10670	318	15	,	,	PUNCT
dad-10670	318	16	and	and	CCONJ
dad-10670	318	17	as	as	SCONJ
dad-10670	318	18	certain	certain	ADJ
dad-10670	318	19	discourse	discourse	NOUN
dad-10670	318	20	relations	relation	NOUN
dad-10670	318	21	occur	occur	VERB
dad-10670	318	22	rarely	rarely	ADV
dad-10670	318	23	in	in	ADP
dad-10670	318	24	natural	natural	ADJ
dad-10670	318	25	language	language	NOUN
dad-10670	318	26	texts	text	NOUN
dad-10670	318	27	,	,	PUNCT
dad-10670	318	28	several	several	ADJ
dad-10670	318	29	of	of	ADP
dad-10670	318	30	label	label	NOUN
dad-10670	318	31	’s	’s	PART
dad-10670	318	32	classes	class	NOUN
dad-10670	318	33	are	be	AUX
dad-10670	318	34	not	not	PART
dad-10670	318	35	present	present	ADJ
dad-10670	318	36	in	in	ADP
dad-10670	318	37	the	the	DET
dad-10670	318	38	test	test	NOUN
dad-10670	318	39	set	set	VERB
dad-10670	318	40	and	and	CCONJ
dad-10670	318	41	thus	thus	ADV
dad-10670	318	42	not	not	PART
dad-10670	318	43	presented	present	VERB
dad-10670	318	44	in	in	ADP
dad-10670	318	45	table	table	NOUN
dad-10670	318	46	7	7	NUM
dad-10670	318	47	.	.	PUNCT
dad-10670	319	1	we	we	PRON
dad-10670	319	2	observe	observe	VERB
dad-10670	319	3	that	that	SCONJ
dad-10670	319	4	performance	performance	NOUN
dad-10670	319	5	varies	vary	VERB
dad-10670	319	6	widely	widely	ADV
dad-10670	319	7	across	across	ADP
dad-10670	319	8	classes	class	NOUN
dad-10670	319	9	.	.	PUNCT
dad-10670	320	1	for	for	ADP
dad-10670	320	2	instance	instance	NOUN
dad-10670	320	3	,	,	PUNCT
dad-10670	320	4	attribution	attribution	NOUN
dad-10670	320	5	is	be	AUX
dad-10670	320	6	very	very	ADV
dad-10670	320	7	well	well	ADV
dad-10670	320	8	detected	detect	VERB
dad-10670	320	9	,	,	PUNCT
dad-10670	320	10	with	with	ADP
dad-10670	320	11	f	f	PROPN
dad-10670	320	12	-	-	PUNCT
dad-10670	320	13	scores	score	NOUN
dad-10670	320	14	close	close	ADV
dad-10670	320	15	to	to	ADP
dad-10670	320	16	95	95	NUM
dad-10670	320	17	%	%	NOUN
dad-10670	320	18	,	,	PUNCT
dad-10670	320	19	whereas	whereas	SCONJ
dad-10670	320	20	cause[n][s	cause[n][s	NOUN
dad-10670	320	21	]	]	PUNCT
dad-10670	320	22	has	have	VERB
dad-10670	320	23	a	a	DET
dad-10670	320	24	very	very	ADV
dad-10670	320	25	low	low	ADJ
dad-10670	320	26	f	f	NOUN
dad-10670	320	27	-	-	PUNCT
dad-10670	320	28	score	score	NOUN
dad-10670	320	29	of	of	ADP
dad-10670	320	30	3.9	3.9	NUM
dad-10670	320	31	%	%	NOUN
dad-10670	320	32	.	.	PUNCT
dad-10670	321	1	here	here	ADV
dad-10670	321	2	,	,	PUNCT
dad-10670	321	3	the	the	DET
dad-10670	321	4	average	average	ADJ
dad-10670	321	5	f	f	NOUN
dad-10670	321	6	-	-	PUNCT
dad-10670	321	7	score	score	NOUN
dad-10670	321	8	is	be	AUX
dad-10670	321	9	47.7	47.7	NUM
dad-10670	321	10	%	%	NOUN
dad-10670	321	11	.	.	PUNCT
dad-10670	322	1	as	as	SCONJ
dad-10670	322	2	we	we	PRON
dad-10670	322	3	employed	employ	VERB
dad-10670	322	4	a	a	DET
dad-10670	322	5	linear	linear	ADJ
dad-10670	322	6	kernel	kernel	NOUN
dad-10670	322	7	in	in	ADP
dad-10670	322	8	struct	struct	NOUN
dad-10670	322	9	,	,	PUNCT
dad-10670	322	10	it	it	PRON
dad-10670	322	11	is	be	AUX
dad-10670	322	12	possible	possible	ADJ
dad-10670	322	13	to	to	PART
dad-10670	322	14	identify	identify	VERB
dad-10670	322	15	which	which	DET
dad-10670	322	16	features	feature	NOUN
dad-10670	322	17	were	be	AUX
dad-10670	322	18	useful	useful	ADJ
dad-10670	322	19	,	,	PUNCT
dad-10670	322	20	by	by	ADP
dad-10670	322	21	looking	look	VERB
dad-10670	322	22	at	at	ADP
dad-10670	322	23	the	the	DET
dad-10670	322	24	weight	weight	NOUN
dad-10670	322	25	assigned	assign	VERB
dad-10670	322	26	by	by	ADP
dad-10670	322	27	the	the	DET
dad-10670	322	28	classifier	classifier	NOUN
dad-10670	322	29	to	to	ADP
dad-10670	322	30	each	each	DET
dad-10670	322	31	feature	feature	NOUN
dad-10670	322	32	in	in	ADP
dad-10670	322	33	the	the	DET
dad-10670	322	34	svm	svm	ADJ
dad-10670	322	35	model	model	NOUN
dad-10670	322	36	file	file	NOUN
dad-10670	322	37	.	.	PUNCT
dad-10670	323	1	table	table	NOUN
dad-10670	323	2	8	8	NUM
dad-10670	323	3	contains	contain	VERB
dad-10670	323	4	the	the	DET
dad-10670	323	5	25	25	NUM
dad-10670	323	6	features	feature	NOUN
dad-10670	323	7	with	with	ADP
dad-10670	323	8	the	the	DET
dad-10670	323	9	highest	high	ADJ
dad-10670	323	10	weights	weight	NOUN
dad-10670	323	11	in	in	ADP
dad-10670	323	12	absolute	absolute	ADJ
dad-10670	323	13	value	value	NOUN
dad-10670	323	14	.	.	PUNCT
dad-10670	324	1	these	these	DET
dad-10670	324	2	features	feature	NOUN
dad-10670	324	3	are	be	AUX
dad-10670	324	4	thus	thus	ADV
dad-10670	324	5	useful	useful	ADJ
dad-10670	324	6	for	for	ADP
dad-10670	324	7	detecting	detect	VERB
dad-10670	324	8	the	the	DET
dad-10670	324	9	presence	presence	NOUN
dad-10670	324	10	of	of	ADP
dad-10670	324	11	a	a	DET
dad-10670	324	12	discourse	discourse	NOUN
dad-10670	324	13	relation	relation	NOUN
dad-10670	324	14	between	between	ADP
dad-10670	324	15	two	two	NUM
dad-10670	324	16	rs	rs	ADJ
dad-10670	324	17	trees	tree	NOUN
dad-10670	324	18	.	.	PUNCT
dad-10670	325	1	in	in	ADP
dad-10670	325	2	particular	particular	ADJ
dad-10670	325	3	,	,	PUNCT
dad-10670	325	4	we	we	PRON
dad-10670	325	5	observe	observe	VERB
dad-10670	325	6	15	15	NUM
dad-10670	325	7	hernault	hernault	ADJ
dad-10670	325	8	,	,	PUNCT
dad-10670	325	9	prendinger	prendinger	NOUN
dad-10670	325	10	,	,	PUNCT
dad-10670	325	11	duverle	duverle	NOUN
dad-10670	325	12	,	,	PUNCT
dad-10670	325	13	and	and	CCONJ
dad-10670	325	14	ishizuka	ishizuka	NOUN
dad-10670	325	15	table	table	NOUN
dad-10670	325	16	6	6	NUM
dad-10670	325	17	:	:	PUNCT
dad-10670	325	18	class	class	NOUN
dad-10670	325	19	-	-	PUNCT
dad-10670	325	20	specific	specific	ADJ
dad-10670	325	21	performance	performance	NOUN
dad-10670	325	22	for	for	ADP
dad-10670	325	23	struct	struct	NOUN
dad-10670	325	24	,	,	PUNCT
dad-10670	325	25	evaluated	evaluate	VERB
dad-10670	325	26	on	on	ADP
dad-10670	325	27	rst	rst	PROPN
dad-10670	325	28	-	-	PUNCT
dad-10670	325	29	dt	dt	PROPN
dad-10670	325	30	’s	’s	PART
dad-10670	325	31	test	test	NOUN
dad-10670	325	32	set	set	VERB
dad-10670	325	33	svm	svm	PROPN
dad-10670	325	34	class	class	NOUN
dad-10670	325	35	precision	precision	NOUN
dad-10670	325	36	recall	recall	NOUN
dad-10670	325	37	f	f	NOUN
dad-10670	325	38	-	-	PUNCT
dad-10670	325	39	score	score	NOUN
dad-10670	325	40	+1	+1	PROPN
dad-10670	325	41	74.0	74.0	NUM
dad-10670	325	42	72.7	72.7	NUM
dad-10670	325	43	73.3	73.3	NUM
dad-10670	325	44	−1	−1	NOUN
dad-10670	325	45	86.3	86.3	NUM
dad-10670	325	46	87.1	87.1	NUM
dad-10670	325	47	86.7	86.7	NUM
dad-10670	325	48	table	table	NOUN
dad-10670	325	49	7	7	NUM
dad-10670	325	50	:	:	PUNCT
dad-10670	325	51	class	class	NOUN
dad-10670	325	52	-	-	PUNCT
dad-10670	325	53	specific	specific	ADJ
dad-10670	325	54	performance	performance	NOUN
dad-10670	325	55	for	for	ADP
dad-10670	325	56	label	label	NOUN
dad-10670	325	57	,	,	PUNCT
dad-10670	325	58	evaluated	evaluate	VERB
dad-10670	325	59	on	on	ADP
dad-10670	325	60	rst	rst	PROPN
dad-10670	325	61	-	-	PUNCT
dad-10670	325	62	dt	dt	PROPN
dad-10670	325	63	’s	’s	PART
dad-10670	325	64	test	test	NOUN
dad-10670	325	65	set	set	VERB
dad-10670	325	66	svm	svm	PROPN
dad-10670	325	67	class	class	NOUN
dad-10670	325	68	precision	precision	NOUN
dad-10670	325	69	recall	recall	VERB
dad-10670	325	70	f	f	NOUN
dad-10670	325	71	-	-	PUNCT
dad-10670	325	72	score	score	NOUN
dad-10670	325	73	attribution[n][s	attribution[n][s	NOUN
dad-10670	325	74	]	]	X
dad-10670	325	75	93.6	93.6	NUM
dad-10670	325	76	96.2	96.2	NUM
dad-10670	325	77	94.9	94.9	NUM
dad-10670	325	78	attribution[s][n	attribution[s][n	NOUN
dad-10670	325	79	]	]	PUNCT
dad-10670	325	80	95.7	95.7	NUM
dad-10670	325	81	93.7	93.7	NUM
dad-10670	325	82	94.7	94.7	NUM
dad-10670	325	83	background[n][s	background[n][s	NOUN
dad-10670	325	84	]	]	PUNCT
dad-10670	325	85	47.8	47.8	NUM
dad-10670	325	86	41.5	41.5	NUM
dad-10670	325	87	44.4	44.4	NUM
dad-10670	325	88	background[s][n	background[s][n	VERB
dad-10670	325	89	]	]	PUNCT
dad-10670	325	90	38.7	38.7	NUM
dad-10670	325	91	20.7	20.7	NUM
dad-10670	325	92	27.0	27.0	NUM
dad-10670	325	93	cause[n][s	cause[n][s	NOUN
dad-10670	325	94	]	]	PUNCT
dad-10670	325	95	33.3	33.3	NUM
dad-10670	325	96	2.1	2.1	NUM
dad-10670	325	97	3.9	3.9	NUM
dad-10670	325	98	comparison[n][s	comparison[n][s	NOUN
dad-10670	325	99	]	]	X
dad-10670	325	100	50.0	50.0	NUM
dad-10670	325	101	5.9	5.9	NUM
dad-10670	325	102	10.5	10.5	NUM
dad-10670	325	103	condition[n][s	condition[n][s	PROPN
dad-10670	325	104	]	]	PUNCT
dad-10670	325	105	100.0	100.0	NUM
dad-10670	325	106	47.8	47.8	NUM
dad-10670	325	107	64.7	64.7	NUM
dad-10670	325	108	condition[s][n	condition[s][n	ADJ
dad-10670	325	109	]	]	PUNCT
dad-10670	325	110	85.7	85.7	NUM
dad-10670	325	111	72.0	72.0	NUM
dad-10670	325	112	78.3	78.3	NUM
dad-10670	325	113	contrast[n][n	contrast[n][n	VERB
dad-10670	325	114	]	]	X
dad-10670	325	115	31.1	31.1	NUM
dad-10670	325	116	21.9	21.9	NUM
dad-10670	325	117	25.7	25.7	NUM
dad-10670	325	118	contrast[n][s	contrast[n][s	NOUN
dad-10670	325	119	]	]	X
dad-10670	325	120	50.0	50.0	NUM
dad-10670	325	121	20.8	20.8	NUM
dad-10670	325	122	29.4	29.4	NUM
dad-10670	325	123	contrast[s][n	contrast[s][n	VERB
dad-10670	325	124	]	]	PUNCT
dad-10670	325	125	51.1	51.1	NUM
dad-10670	325	126	39.7	39.7	NUM
dad-10670	325	127	44.7	44.7	NUM
dad-10670	325	128	elaboration[n][s	elaboration[n][s	ADJ
dad-10670	325	129	]	]	PUNCT
dad-10670	325	130	58.1	58.1	NUM
dad-10670	325	131	94.5	94.5	NUM
dad-10670	325	132	72.0	72.0	NUM
dad-10670	325	133	enablement[n][s	enablement[n][s	NOUN
dad-10670	325	134	]	]	PUNCT
dad-10670	325	135	61.9	61.9	NUM
dad-10670	325	136	59.1	59.1	NUM
dad-10670	325	137	60.5	60.5	NUM
dad-10670	325	138	enablement[s][n	enablement[s][n	NOUN
dad-10670	325	139	]	]	PUNCT
dad-10670	325	140	50.0	50.0	NUM
dad-10670	325	141	50.0	50.0	NUM
dad-10670	325	142	50.0	50.0	NUM
dad-10670	325	143	explanation[n][s	explanation[n][s	NOUN
dad-10670	325	144	]	]	PUNCT
dad-10670	325	145	66.7	66.7	NUM
dad-10670	325	146	16.8	16.8	NUM
dad-10670	325	147	26.9	26.9	NUM
dad-10670	325	148	joint[n][n	joint[n][n	NOUN
dad-10670	325	149	]	]	PUNCT
dad-10670	325	150	62.2	62.2	NUM
dad-10670	325	151	55.2	55.2	NUM
dad-10670	325	152	58.5	58.5	NUM
dad-10670	325	153	manner	manner	NOUN
dad-10670	325	154	-	-	PUNCT
dad-10670	325	155	means[n][s	means[n][s	NOUN
dad-10670	325	156	]	]	X
dad-10670	325	157	53.8	53.8	NUM
dad-10670	325	158	29.2	29.2	NUM
dad-10670	325	159	37.8	37.8	NUM
dad-10670	325	160	same	same	ADJ
dad-10670	325	161	-	-	PUNCT
dad-10670	325	162	unit[n][n	unit[n][n	NOUN
dad-10670	325	163	]	]	X
dad-10670	325	164	78.8	78.8	NUM
dad-10670	325	165	96.9	96.9	NUM
dad-10670	325	166	86.9	86.9	NUM
dad-10670	325	167	summary[n][s	summary[n][s	NOUN
dad-10670	325	168	]	]	PUNCT
dad-10670	325	169	100.0	100.0	NUM
dad-10670	325	170	31.2	31.2	NUM
dad-10670	325	171	47.6	47.6	NUM
dad-10670	325	172	temporal[n][n	temporal[n][n	NOUN
dad-10670	325	173	]	]	X
dad-10670	325	174	83.3	83.3	NUM
dad-10670	325	175	11.9	11.9	NUM
dad-10670	325	176	20.8	20.8	NUM
dad-10670	325	177	temporal[n][s	temporal[n][s	NOUN
dad-10670	325	178	]	]	PUNCT
dad-10670	325	179	85.7	85.7	NUM
dad-10670	325	180	24.0	24.0	NUM
dad-10670	325	181	37.5	37.5	NUM
dad-10670	325	182	textual	textual	ADV
dad-10670	325	183	-	-	PUNCT
dad-10670	325	184	organization[n][n	organization[n][n	VERB
dad-10670	325	185	]	]	PUNCT
dad-10670	325	186	33.3	33.3	NUM
dad-10670	325	187	22.2	22.2	NUM
dad-10670	325	188	26.7	26.7	NUM
dad-10670	325	189	topic	topic	NOUN
dad-10670	325	190	-	-	PUNCT
dad-10670	325	191	change[n][n	change[n][n	VERB
dad-10670	325	192	]	]	X
dad-10670	325	193	83.3	83.3	NUM
dad-10670	325	194	38.5	38.5	NUM
dad-10670	325	195	52.6	52.6	NUM
dad-10670	325	196	that	that	SCONJ
dad-10670	325	197	the	the	DET
dad-10670	325	198	most	most	ADV
dad-10670	325	199	heavily	heavily	ADV
dad-10670	325	200	-	-	PUNCT
dad-10670	325	201	weighted	weight	VERB
dad-10670	325	202	feature	feature	NOUN
dad-10670	325	203	is	be	AUX
dad-10670	325	204	the	the	DET
dad-10670	325	205	binary	binary	ADJ
dad-10670	325	206	feature	feature	NOUN
dad-10670	325	207	indicating	indicate	VERB
dad-10670	325	208	if	if	SCONJ
dad-10670	325	209	the	the	DET
dad-10670	325	210	two	two	NUM
dad-10670	325	211	input	input	NOUN
dad-10670	325	212	spans	span	NOUN
dad-10670	325	213	belong	belong	VERB
dad-10670	325	214	to	to	ADP
dad-10670	325	215	the	the	DET
dad-10670	325	216	same	same	ADJ
dad-10670	325	217	sentence	sentence	NOUN
dad-10670	325	218	.	.	PUNCT
dad-10670	326	1	among	among	ADP
dad-10670	326	2	these	these	DET
dad-10670	326	3	features	feature	NOUN
dad-10670	326	4	,	,	PUNCT
dad-10670	326	5	we	we	PRON
dad-10670	326	6	notice	notice	VERB
dad-10670	326	7	several	several	ADJ
dad-10670	326	8	measures	measure	NOUN
dad-10670	326	9	of	of	ADP
dad-10670	326	10	distance	distance	NOUN
dad-10670	326	11	,	,	PUNCT
dad-10670	326	12	such	such	ADJ
dad-10670	326	13	as	as	ADP
dad-10670	326	14	the	the	DET
dad-10670	326	15	distance	distance	NOUN
dad-10670	326	16	of	of	ADP
dad-10670	326	17	the	the	DET
dad-10670	326	18	left	left	ADJ
dad-10670	326	19	span	span	NOUN
dad-10670	326	20	to	to	ADP
dad-10670	326	21	the	the	DET
dad-10670	326	22	sentence	sentence	NOUN
dad-10670	326	23	beginning	beginning	NOUN
dad-10670	326	24	,	,	PUNCT
dad-10670	326	25	the	the	DET
dad-10670	326	26	relative	relative	ADJ
dad-10670	326	27	size	size	NOUN
dad-10670	326	28	of	of	ADP
dad-10670	326	29	the	the	DET
dad-10670	326	30	left	left	ADJ
dad-10670	326	31	and	and	CCONJ
dad-10670	326	32	right	right	ADJ
dad-10670	326	33	span	span	NOUN
dad-10670	326	34	’s	’s	PART
dad-10670	326	35	edus	edus	NOUN
dad-10670	326	36	over	over	ADP
dad-10670	326	37	the	the	DET
dad-10670	326	38	sentence	sentence	NOUN
dad-10670	326	39	length	length	NOUN
dad-10670	326	40	,	,	PUNCT
dad-10670	326	41	or	or	CCONJ
dad-10670	326	42	the	the	DET
dad-10670	326	43	number	number	NOUN
dad-10670	326	44	of	of	ADP
dad-10670	326	45	paragraphs	paragraph	NOUN
dad-10670	326	46	in	in	ADP
dad-10670	326	47	the	the	DET
dad-10670	326	48	right	right	ADJ
dad-10670	326	49	span	span	NOUN
dad-10670	326	50	.	.	PUNCT
dad-10670	327	1	we	we	PRON
dad-10670	327	2	also	also	ADV
dad-10670	327	3	find	find	VERB
dad-10670	327	4	several	several	ADJ
dad-10670	327	5	features	feature	NOUN
dad-10670	327	6	related	relate	VERB
dad-10670	327	7	to	to	ADP
dad-10670	327	8	dominance	dominance	NOUN
dad-10670	327	9	sets	set	NOUN
dad-10670	327	10	,	,	PUNCT
dad-10670	327	11	such	such	ADJ
dad-10670	327	12	as	as	ADP
dad-10670	327	13	features	feature	NOUN
dad-10670	327	14	encoding	encode	VERB
dad-10670	327	15	the	the	DET
dad-10670	327	16	dominated	dominate	VERB
dad-10670	327	17	and	and	CCONJ
dad-10670	327	18	dominating	dominate	VERB
dad-10670	327	19	node	node	NOUN
dad-10670	327	20	’s	’s	PART
dad-10670	327	21	pos	pos	NOUN
dad-10670	327	22	tags	tag	NOUN
dad-10670	327	23	and	and	CCONJ
dad-10670	327	24	lexical	lexical	ADJ
dad-10670	327	25	heads	head	NOUN
dad-10670	327	26	.	.	PUNCT
dad-10670	328	1	16	16	NUM
dad-10670	328	2	hilda	hilda	VERB
dad-10670	328	3	:	:	PUNCT
dad-10670	328	4	a	a	DET
dad-10670	328	5	discourse	discourse	NOUN
dad-10670	328	6	parser	parser	NOUN
dad-10670	328	7	using	use	VERB
dad-10670	328	8	support	support	NOUN
dad-10670	328	9	vector	vector	NOUN
dad-10670	328	10	machine	machine	NOUN
dad-10670	328	11	classification	classification	NOUN
dad-10670	328	12	table	table	NOUN
dad-10670	328	13	8	8	NUM
dad-10670	328	14	:	:	PUNCT
dad-10670	328	15	top	top	ADJ
dad-10670	328	16	25	25	NUM
dad-10670	328	17	svm	svm	ADJ
dad-10670	328	18	weights	weight	NOUN
dad-10670	328	19	of	of	ADP
dad-10670	328	20	the	the	DET
dad-10670	328	21	linear	linear	ADJ
dad-10670	328	22	kernel	kernel	NOUN
dad-10670	328	23	for	for	ADP
dad-10670	328	24	struct	struct	NOUN
dad-10670	328	25	.	.	PUNCT
dad-10670	329	1	feature	feature	NOUN
dad-10670	329	2	weight	weight	NOUN
dad-10670	329	3	both	both	DET
dad-10670	329	4	spans	span	NOUN
dad-10670	329	5	belong	belong	VERB
dad-10670	329	6	to	to	ADP
dad-10670	329	7	the	the	DET
dad-10670	329	8	same	same	ADJ
dad-10670	329	9	sentence	sentence	NOUN
dad-10670	329	10	4.118836	4.118836	NUM
dad-10670	329	11	size	size	NOUN
dad-10670	329	12	of	of	ADP
dad-10670	329	13	span	span	NOUN
dad-10670	329	14	over	over	ADP
dad-10670	329	15	sentence	sentence	NOUN
dad-10670	329	16	in	in	ADP
dad-10670	329	17	edus	edus	NOUN
dad-10670	329	18	3.582545	3.582545	NUM
dad-10670	329	19	distance	distance	NOUN
dad-10670	329	20	of	of	ADP
dad-10670	329	21	the	the	DET
dad-10670	329	22	left	left	ADJ
dad-10670	329	23	span	span	NOUN
dad-10670	329	24	to	to	ADP
dad-10670	329	25	beginning	beginning	NOUN
dad-10670	329	26	of	of	ADP
dad-10670	329	27	sentence	sentence	NOUN
dad-10670	329	28	in	in	ADP
dad-10670	329	29	edus	edus	PROPN
dad-10670	329	30	-3.437157	-3.437157	PROPN
dad-10670	329	31	common	common	PROPN
dad-10670	329	32	ancestor	ancestor	NOUN
dad-10670	329	33	’s	’s	PART
dad-10670	329	34	pos	pos	NOUN
dad-10670	329	35	tag	tag	NOUN
dad-10670	329	36	is	be	AUX
dad-10670	329	37	‘	'	PUNCT
dad-10670	329	38	prn	prn	PRON
dad-10670	329	39	’	'	PUNCT
dad-10670	329	40	-2.911269	-2.911269	NOUN
dad-10670	329	41	dominating	dominate	VERB
dad-10670	329	42	node	node	PROPN
dad-10670	329	43	’s	’s	PART
dad-10670	329	44	lexical	lexical	ADJ
dad-10670	329	45	head	head	NOUN
dad-10670	329	46	is	be	AUX
dad-10670	329	47	‘	'	PUNCT
dad-10670	329	48	which	which	PRON
dad-10670	329	49	’	'	PUNCT
dad-10670	329	50	-2.668148	-2.668148	NOUN
dad-10670	329	51	pos	pos	NOUN
dad-10670	329	52	tag	tag	NOUN
dad-10670	329	53	of	of	ADP
dad-10670	329	54	the	the	DET
dad-10670	329	55	right	right	ADJ
dad-10670	329	56	span	span	NOUN
dad-10670	329	57	’s	’s	PART
dad-10670	329	58	last	last	ADJ
dad-10670	329	59	token	token	NOUN
dad-10670	329	60	is	be	AUX
dad-10670	329	61	‘	'	PUNCT
dad-10670	329	62	.	.	PUNCT
dad-10670	329	63	’	'	PUNCT
dad-10670	330	1	2.636921	2.636921	NUM
dad-10670	330	2	size	size	NOUN
dad-10670	330	3	of	of	ADP
dad-10670	330	4	left	left	ADJ
dad-10670	330	5	span	span	NOUN
dad-10670	330	6	over	over	ADP
dad-10670	330	7	sentence	sentence	NOUN
dad-10670	330	8	in	in	ADP
dad-10670	330	9	tokens	token	NOUN
dad-10670	330	10	-2.341654	-2.341654	ADJ
dad-10670	330	11	size	size	NOUN
dad-10670	330	12	of	of	ADP
dad-10670	330	13	both	both	DET
dad-10670	330	14	spans	span	NOUN
dad-10670	330	15	over	over	ADP
dad-10670	330	16	sentence	sentence	NOUN
dad-10670	330	17	in	in	ADP
dad-10670	330	18	tokens	token	NOUN
dad-10670	330	19	-2.222655	-2.222655	NOUN
dad-10670	330	20	left	leave	VERB
dad-10670	330	21	and	and	CCONJ
dad-10670	330	22	right	right	ADJ
dad-10670	330	23	span	span	NOUN
dad-10670	330	24	belong	belong	VERB
dad-10670	330	25	to	to	ADP
dad-10670	330	26	the	the	DET
dad-10670	330	27	same	same	ADJ
dad-10670	330	28	sentence	sentence	NOUN
dad-10670	330	29	-2.217709	-2.217709	NOUN
dad-10670	330	30	pos	pos	NOUN
dad-10670	330	31	tag	tag	NOUN
dad-10670	330	32	of	of	ADP
dad-10670	330	33	the	the	DET
dad-10670	330	34	left	left	ADJ
dad-10670	330	35	span	span	NOUN
dad-10670	330	36	’s	’s	PART
dad-10670	330	37	last	last	ADJ
dad-10670	330	38	token	token	NOUN
dad-10670	330	39	is	be	AUX
dad-10670	330	40	‘	'	PUNCT
dad-10670	330	41	.	.	PUNCT
dad-10670	330	42	’	'	PUNCT
dad-10670	331	1	2.170483	2.170483	NUM
dad-10670	331	2	number	number	NOUN
dad-10670	331	3	of	of	ADP
dad-10670	331	4	paragraphs	paragraph	NOUN
dad-10670	331	5	in	in	ADP
dad-10670	331	6	the	the	DET
dad-10670	331	7	right	right	ADJ
dad-10670	331	8	span	span	NOUN
dad-10670	331	9	2.123776	2.123776	NUM
dad-10670	331	10	relative	relative	ADJ
dad-10670	331	11	position	position	NOUN
dad-10670	331	12	of	of	ADP
dad-10670	331	13	lexical	lexical	ADJ
dad-10670	331	14	head	head	NOUN
dad-10670	331	15	in	in	ADP
dad-10670	331	16	sentence	sentence	NOUN
dad-10670	331	17	2.084602	2.084602	NUM
dad-10670	331	18	pos	pos	NOUN
dad-10670	331	19	tag	tag	NOUN
dad-10670	331	20	of	of	ADP
dad-10670	331	21	the	the	DET
dad-10670	331	22	left	left	ADJ
dad-10670	331	23	span	span	NOUN
dad-10670	331	24	’s	’s	PART
dad-10670	331	25	last	last	ADJ
dad-10670	331	26	token	token	NOUN
dad-10670	331	27	is	be	AUX
dad-10670	331	28	‘	'	PUNCT
dad-10670	331	29	”	"	PUNCT
dad-10670	331	30	’	'	PUNCT
dad-10670	331	31	2.051767	2.051767	NUM
dad-10670	331	32	dominated	dominate	VERB
dad-10670	331	33	node	node	NOUN
dad-10670	331	34	’s	’s	PART
dad-10670	331	35	sibling	sible	VERB
dad-10670	331	36	’s	’s	PART
dad-10670	331	37	pos	pos	NOUN
dad-10670	331	38	tag	tag	NOUN
dad-10670	331	39	is	be	AUX
dad-10670	331	40	‘	'	PUNCT
dad-10670	331	41	vbg	vbg	NOUN
dad-10670	331	42	’	'	PUNCT
dad-10670	331	43	-1.875610	-1.875610	NOUN
dad-10670	331	44	pos	pos	NOUN
dad-10670	331	45	tag	tag	NOUN
dad-10670	331	46	of	of	ADP
dad-10670	331	47	the	the	DET
dad-10670	331	48	right	right	PROPN
dad-10670	331	49	span	span	NOUN
dad-10670	331	50	’s	’s	PART
dad-10670	331	51	first	first	ADJ
dad-10670	331	52	token	token	VERB
dad-10670	331	53	is	be	AUX
dad-10670	331	54	‘	'	PUNCT
dad-10670	331	55	to	to	ADP
dad-10670	331	56	’	'	PUNCT
dad-10670	331	57	-1.869705	-1.869705	NOUN
dad-10670	331	58	distance	distance	NOUN
dad-10670	331	59	of	of	ADP
dad-10670	331	60	the	the	DET
dad-10670	331	61	right	right	ADJ
dad-10670	331	62	span	span	NOUN
dad-10670	331	63	from	from	ADP
dad-10670	331	64	the	the	DET
dad-10670	331	65	sentence	sentence	NOUN
dad-10670	331	66	beginning	begin	VERB
dad-10670	331	67	-1.839178	-1.839178	ADJ
dad-10670	331	68	pos	pos	NOUN
dad-10670	331	69	tag	tag	NOUN
dad-10670	331	70	of	of	ADP
dad-10670	331	71	the	the	DET
dad-10670	331	72	right	right	ADJ
dad-10670	331	73	span	span	NOUN
dad-10670	331	74	’s	’s	PART
dad-10670	331	75	last	last	ADJ
dad-10670	331	76	token	token	NOUN
dad-10670	331	77	is	be	AUX
dad-10670	331	78	‘	'	PUNCT
dad-10670	331	79	,	,	PUNCT
dad-10670	331	80	’	'	PUNCT
dad-10670	331	81	-1.814842	-1.814842	VERB
dad-10670	331	82	top	top	ADJ
dad-10670	331	83	lexical	lexical	ADJ
dad-10670	331	84	head	head	NOUN
dad-10670	331	85	of	of	ADP
dad-10670	331	86	the	the	DET
dad-10670	331	87	left	left	ADJ
dad-10670	331	88	span	span	NOUN
dad-10670	331	89	is	be	AUX
dad-10670	331	90	‘	'	PUNCT
dad-10670	331	91	which	which	PRON
dad-10670	331	92	’	'	PUNCT
dad-10670	331	93	1.777975	1.777975	NUM
dad-10670	331	94	top	top	ADJ
dad-10670	331	95	lexical	lexical	ADJ
dad-10670	331	96	head	head	NOUN
dad-10670	331	97	of	of	ADP
dad-10670	331	98	the	the	DET
dad-10670	331	99	sentence	sentence	NOUN
dad-10670	331	100	belongs	belong	VERB
dad-10670	331	101	to	to	ADP
dad-10670	331	102	the	the	DET
dad-10670	331	103	right	right	ADJ
dad-10670	331	104	span	span	NOUN
dad-10670	331	105	-1.774898	-1.774898	PROPN
dad-10670	331	106	top	top	ADJ
dad-10670	331	107	lexical	lexical	ADJ
dad-10670	331	108	head	head	NOUN
dad-10670	331	109	of	of	ADP
dad-10670	331	110	the	the	DET
dad-10670	331	111	left	left	ADJ
dad-10670	331	112	span	span	NOUN
dad-10670	331	113	is	be	AUX
dad-10670	331	114	‘	'	PUNCT
dad-10670	331	115	of	of	ADP
dad-10670	331	116	’	'	PUNCT
dad-10670	331	117	-1.706626	-1.706626	NOUN
dad-10670	331	118	dominating	dominate	VERB
dad-10670	331	119	node	node	NOUN
dad-10670	331	120	’s	’s	PART
dad-10670	331	121	lexical	lexical	ADJ
dad-10670	331	122	head	head	NOUN
dad-10670	331	123	is	be	AUX
dad-10670	331	124	‘	'	PUNCT
dad-10670	331	125	set	set	ADJ
dad-10670	331	126	’	'	PUNCT
dad-10670	331	127	-1.670822	-1.670822	NOUN
dad-10670	331	128	dominated	dominate	VERB
dad-10670	331	129	node	node	PROPN
dad-10670	331	130	’s	’s	PART
dad-10670	331	131	lexical	lexical	ADJ
dad-10670	331	132	head	head	NOUN
dad-10670	331	133	is	be	AUX
dad-10670	331	134	‘	'	PUNCT
dad-10670	331	135	ideas	idea	NOUN
dad-10670	331	136	’	'	PUNCT
dad-10670	331	137	-1.669270	-1.669270	NOUN
dad-10670	331	138	dominating	dominate	VERB
dad-10670	331	139	node	node	NOUN
dad-10670	331	140	’s	’s	PART
dad-10670	331	141	pos	pos	NOUN
dad-10670	331	142	tag	tag	NOUN
dad-10670	331	143	is	be	AUX
dad-10670	331	144	‘	'	PUNCT
dad-10670	331	145	nn	nn	NUM
dad-10670	331	146	’	'	PUNCT
dad-10670	331	147	-1.650378	-1.650378	ADV
dad-10670	331	148	dominated	dominate	VERB
dad-10670	331	149	node	node	PROPN
dad-10670	331	150	’s	’s	PART
dad-10670	331	151	sibling	sible	VERB
dad-10670	331	152	’s	’s	PART
dad-10670	331	153	pos	pos	NOUN
dad-10670	331	154	tag	tag	NOUN
dad-10670	331	155	is	be	AUX
dad-10670	331	156	‘	'	PUNCT
dad-10670	331	157	vbn	vbn	NOUN
dad-10670	331	158	’	'	PUNCT
dad-10670	331	159	1.628170	1.628170	NUM
dad-10670	331	160	dominating	dominate	VERB
dad-10670	331	161	node	node	NOUN
dad-10670	331	162	’s	’s	PART
dad-10670	331	163	pos	pos	NOUN
dad-10670	331	164	tag	tag	NOUN
dad-10670	331	165	is	be	AUX
dad-10670	331	166	‘	'	PUNCT
dad-10670	331	167	vbn	vbn	NOUN
dad-10670	331	168	’	'	PUNCT
dad-10670	331	169	-1.617607	-1.617607	NOUN
dad-10670	331	170	4.3	4.3	NUM
dad-10670	331	171	full	full	ADJ
dad-10670	331	172	system	system	NOUN
dad-10670	331	173	performance	performance	NOUN
dad-10670	331	174	a	a	DET
dad-10670	331	175	measure	measure	NOUN
dad-10670	331	176	of	of	ADP
dad-10670	331	177	the	the	DET
dad-10670	331	178	performance	performance	NOUN
dad-10670	331	179	of	of	ADP
dad-10670	331	180	our	our	PRON
dad-10670	331	181	full	full	ADJ
dad-10670	331	182	system	system	NOUN
dad-10670	331	183	is	be	AUX
dad-10670	331	184	realized	realize	VERB
dad-10670	331	185	by	by	ADP
dad-10670	331	186	comparing	compare	VERB
dad-10670	331	187	the	the	DET
dad-10670	331	188	structure	structure	NOUN
dad-10670	331	189	and	and	CCONJ
dad-10670	331	190	labeling	labeling	NOUN
dad-10670	331	191	of	of	ADP
dad-10670	331	192	the	the	DET
dad-10670	331	193	rs	rs	NOUN
dad-10670	331	194	-	-	PUNCT
dad-10670	331	195	tree	tree	NOUN
dad-10670	331	196	produced	produce	VERB
dad-10670	331	197	by	by	ADP
dad-10670	331	198	our	our	PRON
dad-10670	331	199	algorithm	algorithm	NOUN
dad-10670	331	200	to	to	ADP
dad-10670	331	201	that	that	PRON
dad-10670	331	202	obtained	obtain	VERB
dad-10670	331	203	through	through	ADP
dad-10670	331	204	manual	manual	ADJ
dad-10670	331	205	annotation	annotation	NOUN
dad-10670	331	206	(	(	PUNCT
dad-10670	331	207	our	our	PRON
dad-10670	331	208	gold	gold	ADJ
dad-10670	331	209	standard	standard	NOUN
dad-10670	331	210	)	)	PUNCT
dad-10670	331	211	.	.	PUNCT
dad-10670	332	1	here	here	ADV
dad-10670	332	2	,	,	PUNCT
dad-10670	332	3	we	we	PRON
dad-10670	332	4	use	use	VERB
dad-10670	332	5	the	the	DET
dad-10670	332	6	parseval	parseval	NOUN
dad-10670	332	7	metrics	metric	NOUN
dad-10670	332	8	(	(	PUNCT
dad-10670	332	9	black	black	NOUN
dad-10670	332	10	et	et	PROPN
dad-10670	332	11	al	al	PROPN
dad-10670	332	12	.	.	PROPN
dad-10670	332	13	,	,	PUNCT
dad-10670	332	14	1991	1991	NUM
dad-10670	332	15	)	)	PUNCT
dad-10670	332	16	that	that	PRON
dad-10670	332	17	defines	define	VERB
dad-10670	332	18	the	the	DET
dad-10670	332	19	performance	performance	NOUN
dad-10670	332	20	metrics	metric	NOUN
dad-10670	332	21	(	(	PUNCT
dad-10670	332	22	precision	precision	NOUN
dad-10670	332	23	,	,	PUNCT
dad-10670	332	24	recall	recall	NOUN
dad-10670	332	25	,	,	PUNCT
dad-10670	332	26	f	f	X
dad-10670	332	27	-	-	PUNCT
dad-10670	332	28	score	score	NOUN
dad-10670	332	29	)	)	PUNCT
dad-10670	332	30	of	of	ADP
dad-10670	332	31	a	a	DET
dad-10670	332	32	parsing	parsing	NOUN
dad-10670	332	33	system	system	NOUN
dad-10670	332	34	.	.	PUNCT
dad-10670	333	1	for	for	ADP
dad-10670	333	2	comparability	comparability	NOUN
dad-10670	333	3	,	,	PUNCT
dad-10670	333	4	the	the	DET
dad-10670	333	5	tree	tree	NOUN
dad-10670	333	6	constituents	constituent	NOUN
dad-10670	333	7	used	use	VERB
dad-10670	333	8	in	in	ADP
dad-10670	333	9	the	the	DET
dad-10670	333	10	method	method	NOUN
dad-10670	333	11	are	be	AUX
dad-10670	333	12	enumerated	enumerate	VERB
dad-10670	333	13	using	use	VERB
dad-10670	333	14	marcu	marcu	NOUN
dad-10670	333	15	’s	’s	PART
dad-10670	333	16	formalism	formalism	NOUN
dad-10670	333	17	(	(	PUNCT
dad-10670	333	18	see	see	VERB
dad-10670	333	19	marcu	marcu	NOUN
dad-10670	333	20	,	,	PUNCT
dad-10670	333	21	2000	2000	NUM
dad-10670	333	22	,	,	PUNCT
dad-10670	333	23	143–144	143–144	NUM
dad-10670	333	24	)	)	PUNCT
dad-10670	333	25	,	,	PUNCT
dad-10670	333	26	which	which	PRON
dad-10670	333	27	assigns	assign	VERB
dad-10670	333	28	relation	relation	NOUN
dad-10670	333	29	labels	label	NOUN
dad-10670	333	30	to	to	ADP
dad-10670	333	31	the	the	DET
dad-10670	333	32	child	child	NOUN
dad-10670	333	33	nodes	node	NOUN
dad-10670	333	34	of	of	ADP
dad-10670	333	35	the	the	DET
dad-10670	333	36	relation	relation	NOUN
dad-10670	333	37	instead	instead	ADV
dad-10670	333	38	of	of	ADP
dad-10670	333	39	the	the	DET
dad-10670	333	40	parent	parent	NOUN
dad-10670	333	41	node	node	NOUN
dad-10670	333	42	,	,	PUNCT
dad-10670	333	43	thus	thus	ADV
dad-10670	333	44	making	make	VERB
dad-10670	333	45	it	it	PRON
dad-10670	333	46	easier	easy	ADJ
dad-10670	333	47	to	to	PART
dad-10670	333	48	represent	represent	VERB
dad-10670	333	49	n	n	NUM
dad-10670	333	50	-	-	PUNCT
dad-10670	333	51	ary	ary	PROPN
dad-10670	333	52	relations	relation	NOUN
dad-10670	333	53	.	.	PUNCT
dad-10670	334	1	in	in	ADP
dad-10670	334	2	practice	practice	NOUN
dad-10670	334	3	,	,	PUNCT
dad-10670	334	4	this	this	DET
dad-10670	334	5	formalism	formalism	NOUN
dad-10670	334	6	yields	yield	VERB
dad-10670	334	7	lower	low	ADJ
dad-10670	334	8	scores	score	NOUN
dad-10670	334	9	than	than	ADP
dad-10670	334	10	our	our	PRON
dad-10670	334	11	own	own	ADJ
dad-10670	334	12	,	,	PUNCT
dad-10670	334	13	likely	likely	ADJ
dad-10670	334	14	because	because	SCONJ
dad-10670	334	15	of	of	ADP
dad-10670	334	16	the	the	DET
dad-10670	334	17	greater	great	ADJ
dad-10670	334	18	emphasis	emphasis	NOUN
dad-10670	334	19	placed	place	VERB
dad-10670	334	20	on	on	ADP
dad-10670	334	21	tree	tree	NOUN
dad-10670	334	22	structure	structure	NOUN
dad-10670	334	23	accuracy	accuracy	NOUN
dad-10670	334	24	over	over	ADP
dad-10670	334	25	relation	relation	NOUN
dad-10670	334	26	labeling	labeling	NOUN
dad-10670	334	27	.	.	PUNCT
dad-10670	335	1	hilda	hilda	PROPN
dad-10670	335	2	is	be	AUX
dad-10670	335	3	evaluated	evaluate	VERB
dad-10670	335	4	based	base	VERB
dad-10670	335	5	on	on	ADP
dad-10670	335	6	three	three	NUM
dad-10670	335	7	experiments	experiment	NOUN
dad-10670	335	8	.	.	PUNCT
dad-10670	336	1	1	1	X
dad-10670	336	2	.	.	X
dad-10670	336	3	perfectly	perfectly	ADV
dad-10670	336	4	segmented	segment	VERB
dad-10670	336	5	input	input	NOUN
dad-10670	336	6	taken	take	VERB
dad-10670	336	7	from	from	ADP
dad-10670	336	8	the	the	DET
dad-10670	336	9	rst	rst	NOUN
dad-10670	336	10	-	-	PUNCT
dad-10670	336	11	dt	dt	PROPN
dad-10670	336	12	(	(	PUNCT
dad-10670	336	13	indicated	indicate	VERB
dad-10670	336	14	as	as	ADP
dad-10670	336	15	‘	'	PUNCT
dad-10670	336	16	manual	manual	NOUN
dad-10670	336	17	’	'	PUNCT
dad-10670	336	18	in	in	ADP
dad-10670	336	19	the	the	DET
dad-10670	336	20	following	following	ADJ
dad-10670	336	21	tables	table	NOUN
dad-10670	336	22	)	)	PUNCT
dad-10670	336	23	.	.	PUNCT
dad-10670	337	1	in	in	ADP
dad-10670	337	2	this	this	DET
dad-10670	337	3	case	case	NOUN
dad-10670	337	4	,	,	PUNCT
dad-10670	337	5	we	we	PRON
dad-10670	337	6	use	use	VERB
dad-10670	337	7	parse	parse	NOUN
dad-10670	337	8	trees	tree	NOUN
dad-10670	337	9	from	from	ADP
dad-10670	337	10	the	the	DET
dad-10670	337	11	penn	penn	PROPN
dad-10670	337	12	treebank	treebank	NOUN
dad-10670	337	13	in	in	ADP
dad-10670	337	14	all	all	DET
dad-10670	337	15	steps	step	NOUN
dad-10670	337	16	of	of	ADP
dad-10670	337	17	the	the	DET
dad-10670	337	18	parsing	parsing	NOUN
dad-10670	337	19	.	.	PUNCT
dad-10670	338	1	17	17	NUM
dad-10670	338	2	hernault	hernault	ADJ
dad-10670	338	3	,	,	PUNCT
dad-10670	338	4	prendinger	prendinger	NOUN
dad-10670	338	5	,	,	PUNCT
dad-10670	338	6	duverle	duverle	NOUN
dad-10670	338	7	,	,	PUNCT
dad-10670	338	8	and	and	CCONJ
dad-10670	338	9	ishizuka	ishizuka	VERB
dad-10670	338	10	2	2	NUM
dad-10670	338	11	.	.	PUNCT
dad-10670	339	1	the	the	DET
dad-10670	339	2	system	system	NOUN
dad-10670	339	3	is	be	AUX
dad-10670	339	4	run	run	VERB
dad-10670	339	5	using	use	VERB
dad-10670	339	6	the	the	DET
dad-10670	339	7	output	output	NOUN
dad-10670	339	8	of	of	ADP
dad-10670	339	9	spade	spade	NOUN
dad-10670	339	10	’s	’s	PART
dad-10670	339	11	segmenter	segmenter	NOUN
dad-10670	339	12	(	(	PUNCT
dad-10670	339	13	spade	spade	NOUN
dad-10670	339	14	-	-	PUNCT
dad-10670	339	15	seg	seg	NOUN
dad-10670	339	16	)	)	PUNCT
dad-10670	339	17	,	,	PUNCT
dad-10670	339	18	in	in	ADP
dad-10670	339	19	which	which	DET
dad-10670	339	20	case	case	NOUN
dad-10670	339	21	charniak	charniak	NOUN
dad-10670	339	22	’s	’s	PART
dad-10670	339	23	parser	parser	NOUN
dad-10670	339	24	is	be	AUX
dad-10670	339	25	used	use	VERB
dad-10670	339	26	for	for	ADP
dad-10670	339	27	generating	generate	VERB
dad-10670	339	28	parse	parse	NOUN
dad-10670	339	29	trees	tree	NOUN
dad-10670	339	30	.	.	PUNCT
dad-10670	340	1	3	3	X
dad-10670	340	2	.	.	X
dad-10670	340	3	the	the	DET
dad-10670	340	4	system	system	NOUN
dad-10670	340	5	is	be	AUX
dad-10670	340	6	run	run	VERB
dad-10670	340	7	with	with	ADP
dad-10670	340	8	our	our	PRON
dad-10670	340	9	own	own	ADJ
dad-10670	340	10	segmenter	segmenter	NOUN
dad-10670	340	11	(	(	PUNCT
dad-10670	340	12	hilda	hilda	PROPN
dad-10670	340	13	-	-	PUNCT
dad-10670	340	14	seg	seg	PROPN
dad-10670	340	15	)	)	PUNCT
dad-10670	340	16	,	,	PUNCT
dad-10670	340	17	presented	present	VERB
dad-10670	340	18	in	in	ADP
dad-10670	340	19	section	section	NOUN
dad-10670	340	20	4.1	4.1	NUM
dad-10670	340	21	.	.	PUNCT
dad-10670	341	1	in	in	ADP
dad-10670	341	2	this	this	DET
dad-10670	341	3	case	case	NOUN
dad-10670	341	4	,	,	PUNCT
dad-10670	341	5	we	we	PRON
dad-10670	341	6	use	use	VERB
dad-10670	341	7	parse	parse	NOUN
dad-10670	341	8	trees	tree	NOUN
dad-10670	341	9	from	from	ADP
dad-10670	341	10	the	the	DET
dad-10670	341	11	stanford	stanford	PROPN
dad-10670	341	12	parser	parser	NOUN
dad-10670	341	13	.	.	PUNCT
dad-10670	342	1	the	the	DET
dad-10670	342	2	first	first	ADJ
dad-10670	342	3	measure	measure	NOUN
dad-10670	342	4	gives	give	VERB
dad-10670	342	5	us	we	PRON
dad-10670	342	6	a	a	DET
dad-10670	342	7	good	good	ADJ
dad-10670	342	8	estimate	estimate	NOUN
dad-10670	342	9	of	of	ADP
dad-10670	342	10	our	our	PRON
dad-10670	342	11	system	system	NOUN
dad-10670	342	12	’s	’s	PART
dad-10670	342	13	optimal	optimal	ADJ
dad-10670	342	14	performance	performance	NOUN
dad-10670	342	15	(	(	PUNCT
dad-10670	342	16	given	give	VERB
dad-10670	342	17	optimal	optimal	ADJ
dad-10670	342	18	input	input	NOUN
dad-10670	342	19	)	)	PUNCT
dad-10670	342	20	,	,	PUNCT
dad-10670	342	21	while	while	SCONJ
dad-10670	342	22	the	the	DET
dad-10670	342	23	others	other	NOUN
dad-10670	342	24	provide	provide	VERB
dad-10670	342	25	a	a	DET
dad-10670	342	26	concrete	concrete	ADJ
dad-10670	342	27	performance	performance	NOUN
dad-10670	342	28	evaluation	evaluation	NOUN
dad-10670	342	29	under	under	ADP
dad-10670	342	30	practical	practical	ADJ
dad-10670	342	31	conditions	condition	NOUN
dad-10670	342	32	.	.	PUNCT
dad-10670	343	1	our	our	PRON
dad-10670	343	2	measurements	measurement	NOUN
dad-10670	343	3	are	be	AUX
dad-10670	343	4	taken	take	VERB
dad-10670	343	5	based	base	VERB
dad-10670	343	6	on	on	ADP
dad-10670	343	7	the	the	DET
dad-10670	343	8	standard	standard	ADJ
dad-10670	343	9	test	test	NOUN
dad-10670	343	10	subset	subset	NOUN
dad-10670	343	11	of	of	ADP
dad-10670	343	12	41	41	NUM
dad-10670	343	13	files	file	NOUN
dad-10670	343	14	provided	provide	VERB
dad-10670	343	15	by	by	ADP
dad-10670	343	16	the	the	DET
dad-10670	343	17	rst	rst	PROPN
dad-10670	343	18	-	-	PUNCT
dad-10670	343	19	dt	dt	NOUN
dad-10670	343	20	corpus	corpus	NOUN
dad-10670	343	21	.	.	PUNCT
dad-10670	344	1	for	for	ADP
dad-10670	344	2	each	each	DET
dad-10670	344	3	experiment	experiment	NOUN
dad-10670	344	4	,	,	PUNCT
dad-10670	344	5	parse	parse	NOUN
dad-10670	344	6	trees	tree	NOUN
dad-10670	344	7	are	be	AUX
dad-10670	344	8	evaluated	evaluate	VERB
dad-10670	344	9	on	on	ADP
dad-10670	344	10	four	four	NUM
dad-10670	344	11	successive	successive	ADJ
dad-10670	344	12	,	,	PUNCT
dad-10670	344	13	increasingly	increasingly	ADV
dad-10670	344	14	complex	complex	ADJ
dad-10670	344	15	criteria	criterion	NOUN
dad-10670	344	16	.	.	PUNCT
dad-10670	345	1	first	first	ADV
dad-10670	345	2	,	,	PUNCT
dad-10670	345	3	on	on	ADP
dad-10670	345	4	the	the	DET
dad-10670	345	5	blank	blank	ADJ
dad-10670	345	6	tree	tree	NOUN
dad-10670	345	7	structure	structure	NOUN
dad-10670	345	8	(	(	PUNCT
dad-10670	345	9	‘s	‘s	NOUN
dad-10670	345	10	’	'	PUNCT
dad-10670	345	11	)	)	PUNCT
dad-10670	345	12	,	,	PUNCT
dad-10670	345	13	then	then	ADV
dad-10670	345	14	on	on	ADP
dad-10670	345	15	the	the	DET
dad-10670	345	16	tree	tree	NOUN
dad-10670	345	17	structure	structure	NOUN
dad-10670	345	18	with	with	ADP
dad-10670	345	19	nuclearity	nuclearity	NOUN
dad-10670	345	20	indication	indication	NOUN
dad-10670	345	21	(	(	PUNCT
dad-10670	345	22	‘	'	PUNCT
dad-10670	345	23	n	n	CCONJ
dad-10670	345	24	’	'	PUNCT
dad-10670	345	25	)	)	PUNCT
dad-10670	345	26	,	,	PUNCT
dad-10670	345	27	then	then	ADV
dad-10670	345	28	the	the	DET
dad-10670	345	29	tree	tree	NOUN
dad-10670	345	30	structure	structure	NOUN
dad-10670	345	31	with	with	ADP
dad-10670	345	32	rhetorical	rhetorical	ADJ
dad-10670	345	33	relation	relation	NOUN
dad-10670	345	34	indication	indication	NOUN
dad-10670	345	35	but	but	CCONJ
dad-10670	345	36	no	no	DET
dad-10670	345	37	nuclearity	nuclearity	NOUN
dad-10670	345	38	indication	indication	NOUN
dad-10670	345	39	(	(	PUNCT
dad-10670	345	40	‘	'	PUNCT
dad-10670	345	41	r	r	NOUN
dad-10670	345	42	’	'	PUNCT
dad-10670	345	43	)	)	PUNCT
dad-10670	345	44	,	,	PUNCT
dad-10670	345	45	and	and	CCONJ
dad-10670	345	46	finally	finally	ADV
dad-10670	345	47	on	on	ADP
dad-10670	345	48	the	the	DET
dad-10670	345	49	fully	fully	ADV
dad-10670	345	50	-	-	PUNCT
dad-10670	345	51	labeled	label	VERB
dad-10670	345	52	tree	tree	NOUN
dad-10670	345	53	structure	structure	NOUN
dad-10670	345	54	with	with	ADP
dad-10670	345	55	both	both	DET
dad-10670	345	56	nuclearity	nuclearity	NOUN
dad-10670	345	57	and	and	CCONJ
dad-10670	345	58	rhetorical	rhetorical	ADJ
dad-10670	345	59	relation	relation	NOUN
dad-10670	345	60	labels	label	NOUN
dad-10670	345	61	(	(	PUNCT
dad-10670	345	62	‘	'	PUNCT
dad-10670	345	63	f	f	X
dad-10670	345	64	’	'	PUNCT
dad-10670	345	65	)	)	PUNCT
dad-10670	345	66	.	.	PUNCT
dad-10670	346	1	for	for	ADP
dad-10670	346	2	each	each	DET
dad-10670	346	3	criteria	criterion	NOUN
dad-10670	346	4	,	,	PUNCT
dad-10670	346	5	precision	precision	NOUN
dad-10670	346	6	is	be	AUX
dad-10670	346	7	calculated	calculate	VERB
dad-10670	346	8	by	by	ADP
dad-10670	346	9	taking	take	VERB
dad-10670	346	10	the	the	DET
dad-10670	346	11	ratio	ratio	NOUN
dad-10670	346	12	of	of	ADP
dad-10670	346	13	the	the	DET
dad-10670	346	14	number	number	NOUN
dad-10670	346	15	of	of	ADP
dad-10670	346	16	identical	identical	ADJ
dad-10670	346	17	tree	tree	NOUN
dad-10670	346	18	constituents	constituent	NOUN
dad-10670	346	19	found	find	VERB
dad-10670	346	20	in	in	ADP
dad-10670	346	21	both	both	CCONJ
dad-10670	346	22	the	the	DET
dad-10670	346	23	generated	generate	VERB
dad-10670	346	24	rs	rs	NOUN
dad-10670	346	25	-	-	PUNCT
dad-10670	346	26	tree	tree	NOUN
dad-10670	346	27	and	and	CCONJ
dad-10670	346	28	the	the	DET
dad-10670	346	29	gold	gold	NOUN
dad-10670	346	30	-	-	PUNCT
dad-10670	346	31	standard	standard	NOUN
dad-10670	346	32	tree	tree	NOUN
dad-10670	346	33	,	,	PUNCT
dad-10670	346	34	against	against	ADP
dad-10670	346	35	the	the	DET
dad-10670	346	36	total	total	ADJ
dad-10670	346	37	number	number	NOUN
dad-10670	346	38	of	of	ADP
dad-10670	346	39	constituents	constituent	NOUN
dad-10670	346	40	in	in	ADP
dad-10670	346	41	the	the	DET
dad-10670	346	42	generated	generate	VERB
dad-10670	346	43	discourse	discourse	NOUN
dad-10670	346	44	tree	tree	NOUN
dad-10670	346	45	.	.	PUNCT
dad-10670	347	1	recall	recall	PROPN
dad-10670	347	2	is	be	AUX
dad-10670	347	3	calculated	calculate	VERB
dad-10670	347	4	by	by	ADP
dad-10670	347	5	taking	take	VERB
dad-10670	347	6	the	the	DET
dad-10670	347	7	ratio	ratio	NOUN
dad-10670	347	8	of	of	ADP
dad-10670	347	9	the	the	DET
dad-10670	347	10	number	number	NOUN
dad-10670	347	11	of	of	ADP
dad-10670	347	12	identical	identical	ADJ
dad-10670	347	13	tree	tree	NOUN
dad-10670	347	14	constituents	constituent	NOUN
dad-10670	347	15	found	find	VERB
dad-10670	347	16	in	in	ADP
dad-10670	347	17	both	both	CCONJ
dad-10670	347	18	the	the	DET
dad-10670	347	19	generated	generate	VERB
dad-10670	347	20	rs	rs	NOUN
dad-10670	347	21	-	-	PUNCT
dad-10670	347	22	tree	tree	NOUN
dad-10670	347	23	and	and	CCONJ
dad-10670	347	24	the	the	DET
dad-10670	347	25	gold	gold	NOUN
dad-10670	347	26	-	-	PUNCT
dad-10670	347	27	standard	standard	NOUN
dad-10670	347	28	tree	tree	NOUN
dad-10670	347	29	,	,	PUNCT
dad-10670	347	30	against	against	ADP
dad-10670	347	31	the	the	DET
dad-10670	347	32	total	total	ADJ
dad-10670	347	33	number	number	NOUN
dad-10670	347	34	of	of	ADP
dad-10670	347	35	constituents	constituent	NOUN
dad-10670	347	36	in	in	ADP
dad-10670	347	37	the	the	DET
dad-10670	347	38	gold	gold	NOUN
dad-10670	347	39	-	-	PUNCT
dad-10670	347	40	standard	standard	ADJ
dad-10670	347	41	discourse	discourse	NOUN
dad-10670	347	42	tree	tree	NOUN
dad-10670	347	43	.	.	PUNCT
dad-10670	348	1	the	the	DET
dad-10670	348	2	results	result	NOUN
dad-10670	348	3	are	be	AUX
dad-10670	348	4	summarized	summarize	VERB
dad-10670	348	5	in	in	ADP
dad-10670	348	6	figure	figure	NOUN
dad-10670	348	7	9	9	NUM
dad-10670	348	8	.	.	PUNCT
dad-10670	349	1	note	note	VERB
dad-10670	349	2	that	that	SCONJ
dad-10670	349	3	when	when	SCONJ
dad-10670	349	4	using	use	VERB
dad-10670	349	5	perfect	perfect	ADJ
dad-10670	349	6	segmentation	segmentation	NOUN
dad-10670	349	7	,	,	PUNCT
dad-10670	349	8	precision	precision	NOUN
dad-10670	349	9	and	and	CCONJ
dad-10670	349	10	recall	recall	NOUN
dad-10670	349	11	are	be	AUX
dad-10670	349	12	identical	identical	ADJ
dad-10670	349	13	since	since	SCONJ
dad-10670	349	14	both	both	DET
dad-10670	349	15	trees	tree	NOUN
dad-10670	349	16	have	have	VERB
dad-10670	349	17	same	same	ADJ
dad-10670	349	18	number	number	NOUN
dad-10670	349	19	of	of	ADP
dad-10670	349	20	constituents	constituent	NOUN
dad-10670	349	21	.	.	PUNCT
dad-10670	350	1	table	table	NOUN
dad-10670	350	2	9	9	NUM
dad-10670	350	3	:	:	PUNCT
dad-10670	350	4	full	full	ADJ
dad-10670	350	5	parser	parser	NOUN
dad-10670	350	6	evaluation	evaluation	NOUN
dad-10670	350	7	,	,	PUNCT
dad-10670	350	8	using	use	VERB
dad-10670	350	9	the	the	DET
dad-10670	350	10	standard	standard	ADJ
dad-10670	350	11	test	test	NOUN
dad-10670	350	12	subset	subset	PROPN
dad-10670	350	13	seg	seg	PROPN
dad-10670	350	14	.	.	PROPN
dad-10670	350	15	manual	manual	ADJ
dad-10670	350	16	spade	spade	NOUN
dad-10670	350	17	-	-	PUNCT
dad-10670	350	18	seg	seg	NOUN
dad-10670	350	19	hilda	hilda	PROPN
dad-10670	350	20	-	-	PUNCT
dad-10670	350	21	seg	seg	PROPN
dad-10670	350	22	s	s	PROPN
dad-10670	350	23	n	n	NOUN
dad-10670	350	24	r	r	NOUN
dad-10670	350	25	f	f	PROPN
dad-10670	350	26	s	s	PART
dad-10670	350	27	n	n	ADP
dad-10670	350	28	r	r	NOUN
dad-10670	350	29	f	f	PROPN
dad-10670	350	30	s	s	PART
dad-10670	350	31	n	n	ADP
dad-10670	350	32	r	r	NOUN
dad-10670	350	33	f	f	PROPN
dad-10670	350	34	precision	precision	NOUN
dad-10670	350	35	83.0	83.0	NUM
dad-10670	350	36	68.4	68.4	NUM
dad-10670	350	37	55.3	55.3	NUM
dad-10670	350	38	54.8	54.8	NUM
dad-10670	350	39	69.5	69.5	NUM
dad-10670	350	40	56.1	56.1	NUM
dad-10670	350	41	44.9	44.9	NUM
dad-10670	350	42	44.4	44.4	NUM
dad-10670	350	43	73.0	73.0	NUM
dad-10670	350	44	59.7	59.7	NUM
dad-10670	350	45	48.2	48.2	NUM
dad-10670	350	46	47.7	47.7	NUM
dad-10670	350	47	recall	recall	NOUN
dad-10670	350	48	83.0	83.0	NUM
dad-10670	350	49	68.4	68.4	NUM
dad-10670	350	50	55.3	55.3	NUM
dad-10670	350	51	54.8	54.8	NUM
dad-10670	350	52	69.2	69.2	NUM
dad-10670	350	53	55.8	55.8	NUM
dad-10670	350	54	44.7	44.7	NUM
dad-10670	350	55	44.2	44.2	NUM
dad-10670	350	56	71.7	71.7	NUM
dad-10670	351	1	58.6	58.6	NUM
dad-10670	351	2	47.4	47.4	NUM
dad-10670	351	3	46.9	46.9	NUM
dad-10670	351	4	f	f	NOUN
dad-10670	351	5	-	-	PUNCT
dad-10670	351	6	score	score	VERB
dad-10670	351	7	83.0	83.0	NUM
dad-10670	351	8	68.4	68.4	NUM
dad-10670	351	9	55.3	55.3	NUM
dad-10670	351	10	54.8	54.8	NUM
dad-10670	351	11	69.3	69.3	NUM
dad-10670	351	12	56.0	56.0	NUM
dad-10670	351	13	44.8	44.8	NUM
dad-10670	351	14	44.3	44.3	NUM
dad-10670	351	15	72.3	72.3	NUM
dad-10670	351	16	59.1	59.1	NUM
dad-10670	351	17	47.8	47.8	NUM
dad-10670	351	18	47.3	47.3	NUM
dad-10670	351	19	as	as	SCONJ
dad-10670	351	20	expected	expect	VERB
dad-10670	351	21	,	,	PUNCT
dad-10670	351	22	we	we	PRON
dad-10670	351	23	observe	observe	VERB
dad-10670	351	24	the	the	DET
dad-10670	351	25	highest	high	ADJ
dad-10670	351	26	performance	performance	NOUN
dad-10670	351	27	when	when	SCONJ
dad-10670	351	28	using	use	VERB
dad-10670	351	29	manual	manual	ADJ
dad-10670	351	30	segmentation	segmentation	NOUN
dad-10670	351	31	,	,	PUNCT
dad-10670	351	32	with	with	ADP
dad-10670	351	33	an	an	DET
dad-10670	351	34	f	f	NOUN
dad-10670	351	35	-	-	PUNCT
dad-10670	351	36	score	score	NOUN
dad-10670	351	37	of	of	ADP
dad-10670	351	38	83	83	NUM
dad-10670	351	39	%	%	NOUN
dad-10670	351	40	for	for	ADP
dad-10670	351	41	structure	structure	NOUN
dad-10670	351	42	annotation	annotation	NOUN
dad-10670	351	43	,	,	PUNCT
dad-10670	351	44	and	and	CCONJ
dad-10670	351	45	54.8	54.8	NUM
dad-10670	351	46	%	%	NOUN
dad-10670	351	47	for	for	ADP
dad-10670	351	48	the	the	DET
dad-10670	351	49	fully	fully	ADV
dad-10670	351	50	-	-	PUNCT
dad-10670	351	51	annotated	annotate	VERB
dad-10670	351	52	tree	tree	NOUN
dad-10670	351	53	.	.	PUNCT
dad-10670	352	1	using	use	VERB
dad-10670	352	2	spade	spade	NOUN
dad-10670	352	3	yields	yield	NOUN
dad-10670	352	4	lower	low	ADJ
dad-10670	352	5	results	result	NOUN
dad-10670	352	6	,	,	PUNCT
dad-10670	352	7	with	with	ADP
dad-10670	352	8	an	an	DET
dad-10670	352	9	f	f	NOUN
dad-10670	352	10	-	-	PUNCT
dad-10670	352	11	score	score	NOUN
dad-10670	352	12	of	of	ADP
dad-10670	352	13	69.3	69.3	NUM
dad-10670	352	14	%	%	NOUN
dad-10670	352	15	for	for	ADP
dad-10670	352	16	structure	structure	NOUN
dad-10670	352	17	annotation	annotation	NOUN
dad-10670	352	18	,	,	PUNCT
dad-10670	352	19	and	and	CCONJ
dad-10670	352	20	44.3	44.3	NUM
dad-10670	352	21	%	%	NOUN
dad-10670	352	22	for	for	ADP
dad-10670	352	23	the	the	DET
dad-10670	352	24	full	full	ADJ
dad-10670	352	25	discourse	discourse	NOUN
dad-10670	352	26	tree	tree	NOUN
dad-10670	352	27	.	.	PUNCT
dad-10670	353	1	on	on	ADP
dad-10670	353	2	the	the	DET
dad-10670	353	3	other	other	ADJ
dad-10670	353	4	hand	hand	NOUN
dad-10670	353	5	,	,	PUNCT
dad-10670	353	6	when	when	SCONJ
dad-10670	353	7	using	use	VERB
dad-10670	353	8	the	the	DET
dad-10670	353	9	discourse	discourse	NOUN
dad-10670	353	10	segmenter	segmenter	NOUN
dad-10670	353	11	presented	present	VERB
dad-10670	353	12	in	in	ADP
dad-10670	353	13	section	section	NOUN
dad-10670	353	14	4.1	4.1	NUM
dad-10670	353	15	,	,	PUNCT
dad-10670	353	16	structure	structure	NOUN
dad-10670	353	17	annotation	annotation	NOUN
dad-10670	353	18	produces	produce	VERB
dad-10670	353	19	f	f	NOUN
dad-10670	353	20	-	-	PUNCT
dad-10670	353	21	scores	score	NOUN
dad-10670	353	22	of	of	ADP
dad-10670	353	23	72.3	72.3	NUM
dad-10670	353	24	%	%	NOUN
dad-10670	353	25	for	for	ADP
dad-10670	353	26	structure	structure	NOUN
dad-10670	353	27	annotation	annotation	NOUN
dad-10670	353	28	,	,	PUNCT
dad-10670	353	29	and	and	CCONJ
dad-10670	353	30	47.3	47.3	NUM
dad-10670	353	31	%	%	NOUN
dad-10670	353	32	for	for	ADP
dad-10670	353	33	the	the	DET
dad-10670	353	34	full	full	ADJ
dad-10670	353	35	tree	tree	NOUN
dad-10670	353	36	.	.	PUNCT
dad-10670	354	1	these	these	DET
dad-10670	354	2	differences	difference	NOUN
dad-10670	354	3	indicate	indicate	VERB
dad-10670	354	4	that	that	SCONJ
dad-10670	354	5	,	,	PUNCT
dad-10670	354	6	although	although	SCONJ
dad-10670	354	7	these	these	DET
dad-10670	354	8	segmenters	segmenter	NOUN
dad-10670	354	9	score	score	VERB
dad-10670	354	10	high	high	ADV
dad-10670	354	11	on	on	ADP
dad-10670	354	12	the	the	DET
dad-10670	354	13	segmentation	segmentation	NOUN
dad-10670	354	14	task	task	NOUN
dad-10670	354	15	(	(	PUNCT
dad-10670	354	16	above	above	ADP
dad-10670	354	17	80	80	NUM
dad-10670	354	18	%	%	NOUN
dad-10670	354	19	)	)	PUNCT
dad-10670	354	20	,	,	PUNCT
dad-10670	354	21	errors	error	NOUN
dad-10670	354	22	occurring	occur	VERB
dad-10670	354	23	at	at	ADP
dad-10670	354	24	the	the	DET
dad-10670	354	25	beginning	beginning	NOUN
dad-10670	354	26	of	of	ADP
dad-10670	354	27	the	the	DET
dad-10670	354	28	pipeline	pipeline	NOUN
dad-10670	354	29	are	be	AUX
dad-10670	354	30	amplified	amplify	VERB
dad-10670	354	31	and	and	CCONJ
dad-10670	354	32	weight	weight	NOUN
dad-10670	354	33	heavily	heavily	ADV
dad-10670	354	34	on	on	ADP
dad-10670	354	35	the	the	DET
dad-10670	354	36	performance	performance	NOUN
dad-10670	354	37	of	of	ADP
dad-10670	354	38	the	the	DET
dad-10670	354	39	output	output	NOUN
dad-10670	354	40	.	.	PUNCT
dad-10670	355	1	furthermore	furthermore	ADV
dad-10670	355	2	,	,	PUNCT
dad-10670	355	3	the	the	DET
dad-10670	355	4	parse	parse	NOUN
dad-10670	355	5	trees	tree	NOUN
dad-10670	355	6	of	of	ADP
dad-10670	355	7	spade	spade	NOUN
dad-10670	355	8	-	-	PUNCT
dad-10670	355	9	seg	seg	NOUN
dad-10670	355	10	and	and	CCONJ
dad-10670	355	11	hilda	hilda	PROPN
dad-10670	355	12	-	-	PUNCT
dad-10670	355	13	seg	seg	PROPN
dad-10670	355	14	are	be	AUX
dad-10670	355	15	generated	generate	VERB
dad-10670	355	16	automatically	automatically	ADV
dad-10670	355	17	,	,	PUNCT
dad-10670	355	18	which	which	PRON
dad-10670	355	19	also	also	ADV
dad-10670	355	20	explains	explain	VERB
dad-10670	355	21	the	the	DET
dad-10670	355	22	lower	low	ADJ
dad-10670	355	23	results	result	NOUN
dad-10670	355	24	.	.	PUNCT
dad-10670	356	1	in	in	ADP
dad-10670	356	2	order	order	NOUN
dad-10670	356	3	to	to	PART
dad-10670	356	4	compare	compare	VERB
dad-10670	356	5	our	our	PRON
dad-10670	356	6	results	result	NOUN
dad-10670	356	7	to	to	ADP
dad-10670	356	8	the	the	DET
dad-10670	356	9	gold	gold	ADJ
dad-10670	356	10	standard	standard	NOUN
dad-10670	356	11	(	(	PUNCT
dad-10670	356	12	defined	define	VERB
dad-10670	356	13	as	as	ADP
dad-10670	356	14	manual	manual	ADJ
dad-10670	356	15	agreement	agreement	NOUN
dad-10670	356	16	between	between	ADP
dad-10670	356	17	human	human	ADJ
dad-10670	356	18	annotators	annotator	NOUN
dad-10670	356	19	)	)	PUNCT
dad-10670	356	20	more	more	ADV
dad-10670	356	21	accurately	accurately	ADV
dad-10670	356	22	,	,	PUNCT
dad-10670	356	23	we	we	PRON
dad-10670	356	24	also	also	ADV
dad-10670	356	25	evaluated	evaluate	VERB
dad-10670	356	26	performances	performance	NOUN
dad-10670	356	27	using	use	VERB
dad-10670	356	28	another	another	DET
dad-10670	356	29	test	test	NOUN
dad-10670	356	30	set	set	NOUN
dad-10670	356	31	(	(	PUNCT
dad-10670	356	32	see	see	VERB
dad-10670	356	33	table	table	NOUN
dad-10670	356	34	10	10	NUM
dad-10670	356	35	)	)	PUNCT
dad-10670	356	36	of	of	ADP
dad-10670	356	37	52	52	NUM
dad-10670	356	38	files	file	NOUN
dad-10670	356	39	annotated	annotate	VERB
dad-10670	356	40	by	by	ADP
dad-10670	356	41	two	two	NUM
dad-10670	356	42	labelers	labeler	NOUN
dad-10670	356	43	of	of	ADP
dad-10670	356	44	the	the	DET
dad-10670	356	45	rst	rst	NOUN
dad-10670	356	46	-	-	PUNCT
dad-10670	356	47	dt	dt	NOUN
dad-10670	356	48	.	.	PUNCT
dad-10670	357	1	in	in	ADP
dad-10670	357	2	each	each	DET
dad-10670	357	3	case	case	NOUN
dad-10670	357	4	,	,	PUNCT
dad-10670	357	5	the	the	DET
dad-10670	357	6	remaining	remain	VERB
dad-10670	357	7	340	340	NUM
dad-10670	357	8	-	-	SYM
dad-10670	357	9	350	350	NUM
dad-10670	357	10	files	file	NOUN
dad-10670	357	11	are	be	AUX
dad-10670	357	12	used	use	VERB
dad-10670	357	13	for	for	ADP
dad-10670	357	14	training	training	NOUN
dad-10670	357	15	.	.	PUNCT
dad-10670	358	1	the	the	DET
dad-10670	358	2	f	f	NOUN
dad-10670	358	3	-	-	PUNCT
dad-10670	358	4	score	score	NOUN
dad-10670	358	5	for	for	ADP
dad-10670	358	6	human	human	ADJ
dad-10670	358	7	agreement	agreement	NOUN
dad-10670	358	8	is	be	AUX
dad-10670	358	9	65.3	65.3	NUM
dad-10670	358	10	%	%	NOUN
dad-10670	358	11	.	.	PUNCT
dad-10670	359	1	when	when	SCONJ
dad-10670	359	2	using	use	VERB
dad-10670	359	3	manual	manual	ADJ
dad-10670	359	4	segmentation	segmentation	NOUN
dad-10670	359	5	,	,	PUNCT
dad-10670	359	6	we	we	PRON
dad-10670	359	7	obtain	obtain	VERB
dad-10670	359	8	an	an	DET
dad-10670	359	9	f	f	NOUN
dad-10670	359	10	-	-	PUNCT
dad-10670	359	11	score	score	NOUN
dad-10670	359	12	of	of	ADP
dad-10670	359	13	55.1	55.1	NUM
dad-10670	359	14	%	%	NOUN
dad-10670	359	15	on	on	ADP
dad-10670	359	16	this	this	DET
dad-10670	359	17	test	test	NOUN
dad-10670	359	18	set	set	NOUN
dad-10670	359	19	,	,	PUNCT
dad-10670	359	20	which	which	PRON
dad-10670	359	21	is	be	AUX
dad-10670	359	22	84	84	NUM
dad-10670	359	23	%	%	NOUN
dad-10670	359	24	of	of	ADP
dad-10670	359	25	human	human	ADJ
dad-10670	359	26	performance	performance	NOUN
dad-10670	359	27	level	level	NOUN
dad-10670	359	28	.	.	PUNCT
dad-10670	360	1	when	when	SCONJ
dad-10670	360	2	using	use	VERB
dad-10670	360	3	our	our	PRON
dad-10670	360	4	own	own	ADJ
dad-10670	360	5	segmenter	segmenter	NOUN
dad-10670	360	6	,	,	PUNCT
dad-10670	360	7	we	we	PRON
dad-10670	360	8	obtain	obtain	VERB
dad-10670	360	9	an	an	DET
dad-10670	360	10	f	f	NOUN
dad-10670	360	11	-	-	PUNCT
dad-10670	360	12	score	score	NOUN
dad-10670	360	13	of	of	ADP
dad-10670	360	14	48.8	48.8	NUM
dad-10670	360	15	%	%	NOUN
dad-10670	360	16	,	,	PUNCT
dad-10670	360	17	which	which	PRON
dad-10670	360	18	is	be	AUX
dad-10670	360	19	74.7	74.7	NUM
dad-10670	360	20	%	%	NOUN
dad-10670	360	21	of	of	ADP
dad-10670	360	22	human	human	ADJ
dad-10670	360	23	performance	performance	NOUN
dad-10670	360	24	level	level	NOUN
dad-10670	360	25	.	.	PUNCT
dad-10670	361	1	18	18	NUM
dad-10670	361	2	hilda	hilda	VERB
dad-10670	361	3	:	:	PUNCT
dad-10670	361	4	a	a	DET
dad-10670	361	5	discourse	discourse	NOUN
dad-10670	361	6	parser	parser	NOUN
dad-10670	361	7	using	use	VERB
dad-10670	361	8	support	support	NOUN
dad-10670	361	9	vector	vector	NOUN
dad-10670	361	10	machine	machine	NOUN
dad-10670	361	11	classification	classification	NOUN
dad-10670	361	12	table	table	NOUN
dad-10670	361	13	10	10	NUM
dad-10670	361	14	:	:	PUNCT
dad-10670	361	15	evaluation	evaluation	NOUN
dad-10670	361	16	on	on	ADP
dad-10670	361	17	a	a	DET
dad-10670	361	18	doubly	doubly	ADV
dad-10670	361	19	-	-	PUNCT
dad-10670	361	20	annotated	annotate	VERB
dad-10670	361	21	subset	subset	NOUN
dad-10670	361	22	and	and	CCONJ
dad-10670	361	23	comparison	comparison	NOUN
dad-10670	361	24	to	to	ADP
dad-10670	361	25	human	human	ADJ
dad-10670	361	26	agreement	agreement	NOUN
dad-10670	361	27	system	system	NOUN
dad-10670	361	28	performance	performance	NOUN
dad-10670	361	29	human	human	PROPN
dad-10670	361	30	agreement	agreement	PROPN
dad-10670	361	31	seg	seg	PROPN
dad-10670	361	32	.	.	PUNCT
dad-10670	362	1	manual	manual	PROPN
dad-10670	362	2	hilda	hilda	PROPN
dad-10670	362	3	-	-	PUNCT
dad-10670	362	4	seg	seg	PROPN
dad-10670	362	5	–	–	PUNCT
dad-10670	362	6	s	s	NOUN
dad-10670	362	7	n	n	PRON
dad-10670	362	8	r	r	NOUN
dad-10670	362	9	f	f	PROPN
dad-10670	362	10	s	s	PART
dad-10670	363	1	n	n	ADP
dad-10670	363	2	r	r	NOUN
dad-10670	363	3	f	f	PROPN
dad-10670	363	4	s	s	PART
dad-10670	363	5	n	n	PRON
dad-10670	363	6	r	r	NOUN
dad-10670	363	7	f	f	PROPN
dad-10670	363	8	precision	precision	NOUN
dad-10670	363	9	84.1	84.1	NUM
dad-10670	363	10	70.6	70.6	NUM
dad-10670	363	11	55.6	55.6	NUM
dad-10670	363	12	55.1	55.1	NUM
dad-10670	363	13	74.0	74.0	NUM
dad-10670	363	14	61.7	61.7	NUM
dad-10670	363	15	49.4	49.4	NUM
dad-10670	363	16	48.9	48.9	NUM
dad-10670	363	17	88.0	88.0	NUM
dad-10670	363	18	77.5	77.5	NUM
dad-10670	363	19	66.0	66.0	NUM
dad-10670	363	20	65.2	65.2	NUM
dad-10670	363	21	recall	recall	NOUN
dad-10670	363	22	84.1	84.1	NUM
dad-10670	363	23	70.6	70.6	NUM
dad-10670	363	24	55.6	55.6	NUM
dad-10670	363	25	55.1	55.1	NUM
dad-10670	363	26	73.7	73.7	NUM
dad-10670	363	27	61.5	61.5	NUM
dad-10670	363	28	49.1	49.1	NUM
dad-10670	363	29	48.7	48.7	NUM
dad-10670	363	30	88.1	88.1	NUM
dad-10670	363	31	77.6	77.6	NUM
dad-10670	363	32	66.1	66.1	NUM
dad-10670	363	33	65.3	65.3	NUM
dad-10670	363	34	f	f	NOUN
dad-10670	363	35	-	-	PUNCT
dad-10670	363	36	score	score	NOUN
dad-10670	363	37	84.1	84.1	NUM
dad-10670	363	38	70.6	70.6	NUM
dad-10670	363	39	55.6	55.6	NUM
dad-10670	363	40	55.1	55.1	NUM
dad-10670	363	41	73.8	73.8	NUM
dad-10670	363	42	61.6	61.6	NUM
dad-10670	363	43	49.2	49.2	NUM
dad-10670	363	44	48.8	48.8	NUM
dad-10670	363	45	88.1	88.1	NUM
dad-10670	363	46	77.5	77.5	NUM
dad-10670	363	47	66.0	66.0	NUM
dad-10670	363	48	65.3	65.3	NUM
dad-10670	363	49	4.4	4.4	NUM
dad-10670	363	50	comparison	comparison	NOUN
dad-10670	363	51	to	to	ADP
dad-10670	363	52	other	other	ADJ
dad-10670	363	53	systems	system	NOUN
dad-10670	363	54	in	in	ADP
dad-10670	363	55	this	this	DET
dad-10670	363	56	section	section	NOUN
dad-10670	363	57	,	,	PUNCT
dad-10670	363	58	we	we	PRON
dad-10670	363	59	attempt	attempt	VERB
dad-10670	363	60	to	to	PART
dad-10670	363	61	compare	compare	VERB
dad-10670	363	62	our	our	PRON
dad-10670	363	63	system	system	NOUN
dad-10670	363	64	to	to	ADP
dad-10670	363	65	other	other	ADJ
dad-10670	363	66	text	text	NOUN
dad-10670	363	67	level	level	NOUN
dad-10670	363	68	discourse	discourse	NOUN
dad-10670	363	69	parsers	parser	NOUN
dad-10670	363	70	,	,	PUNCT
dad-10670	363	71	namely	namely	ADV
dad-10670	363	72	marcu	marcu	NOUN
dad-10670	363	73	’s	’s	PART
dad-10670	363	74	decision	decision	NOUN
dad-10670	363	75	-	-	PUNCT
dad-10670	363	76	tree	tree	NOUN
dad-10670	363	77	based	base	VERB
dad-10670	363	78	parser	parser	NOUN
dad-10670	363	79	(	(	PUNCT
dad-10670	363	80	marcu	marcu	NOUN
dad-10670	363	81	,	,	PUNCT
dad-10670	363	82	2000	2000	NUM
dad-10670	363	83	)	)	PUNCT
dad-10670	363	84	,	,	PUNCT
dad-10670	363	85	the	the	DET
dad-10670	363	86	multi	multi	ADJ
dad-10670	363	87	-	-	ADJ
dad-10670	363	88	level	level	ADJ
dad-10670	363	89	rule	rule	NOUN
dad-10670	363	90	-	-	PUNCT
dad-10670	363	91	based	base	VERB
dad-10670	363	92	system	system	NOUN
dad-10670	363	93	built	build	VERB
dad-10670	363	94	by	by	ADP
dad-10670	363	95	le	le	X
dad-10670	363	96	thanh	thanh	PROPN
dad-10670	363	97	et	et	PROPN
dad-10670	363	98	al	al	PROPN
dad-10670	363	99	.	.	PROPN
dad-10670	363	100	(	(	PUNCT
dad-10670	363	101	2004b	2004b	NUM
dad-10670	363	102	)	)	PUNCT
dad-10670	363	103	,	,	PUNCT
dad-10670	363	104	and	and	CCONJ
dad-10670	363	105	the	the	DET
dad-10670	363	106	shift	shift	NOUN
dad-10670	363	107	-	-	PUNCT
dad-10670	363	108	reduce	reduce	NOUN
dad-10670	363	109	parser	parser	NOUN
dad-10670	363	110	of	of	ADP
dad-10670	363	111	sagae	sagae	NOUN
dad-10670	363	112	(	(	PUNCT
dad-10670	363	113	2009	2009	NUM
dad-10670	363	114	)	)	PUNCT
dad-10670	363	115	.	.	PUNCT
dad-10670	364	1	for	for	ADP
dad-10670	364	2	comparison	comparison	NOUN
dad-10670	364	3	with	with	ADP
dad-10670	364	4	marcu	marcu	NOUN
dad-10670	364	5	and	and	CCONJ
dad-10670	364	6	lethanh	lethanh	ADJ
dad-10670	364	7	’s	’s	PART
dad-10670	364	8	systems	system	NOUN
dad-10670	364	9	,	,	PUNCT
dad-10670	364	10	we	we	PRON
dad-10670	364	11	face	face	VERB
dad-10670	364	12	several	several	ADJ
dad-10670	364	13	difficulties	difficulty	NOUN
dad-10670	364	14	:	:	PUNCT
dad-10670	364	15	marcu	marcu	NOUN
dad-10670	364	16	(	(	PUNCT
dad-10670	364	17	2000	2000	NUM
dad-10670	364	18	)	)	PUNCT
dad-10670	364	19	reports	report	VERB
dad-10670	364	20	his	his	PRON
dad-10670	364	21	results	result	NOUN
dad-10670	364	22	on	on	ADP
dad-10670	364	23	unspecified	unspecified	ADJ
dad-10670	364	24	documents	document	NOUN
dad-10670	364	25	and	and	CCONJ
dad-10670	364	26	we	we	PRON
dad-10670	364	27	do	do	AUX
dad-10670	364	28	not	not	PART
dad-10670	364	29	have	have	VERB
dad-10670	364	30	access	access	NOUN
dad-10670	364	31	to	to	ADP
dad-10670	364	32	the	the	DET
dad-10670	364	33	parser	parser	NOUN
dad-10670	364	34	software	software	NOUN
dad-10670	364	35	.	.	PUNCT
dad-10670	365	1	therefore	therefore	ADV
dad-10670	365	2	we	we	PRON
dad-10670	365	3	could	could	AUX
dad-10670	365	4	not	not	PART
dad-10670	365	5	evaluate	evaluate	VERB
dad-10670	365	6	his	his	PRON
dad-10670	365	7	parser	parser	NOUN
dad-10670	365	8	on	on	ADP
dad-10670	365	9	our	our	PRON
dad-10670	365	10	test	test	NOUN
dad-10670	365	11	set	set	NOUN
dad-10670	365	12	.	.	PUNCT
dad-10670	366	1	then	then	ADV
dad-10670	366	2	,	,	PUNCT
dad-10670	366	3	in	in	ADP
dad-10670	366	4	the	the	DET
dad-10670	366	5	case	case	NOUN
dad-10670	366	6	of	of	ADP
dad-10670	366	7	lethanh	lethanh	ADJ
dad-10670	366	8	’s	’s	PART
dad-10670	366	9	system	system	NOUN
dad-10670	366	10	,	,	PUNCT
dad-10670	366	11	we	we	PRON
dad-10670	366	12	do	do	AUX
dad-10670	366	13	have	have	VERB
dad-10670	366	14	access	access	NOUN
dad-10670	366	15	to	to	ADP
dad-10670	366	16	the	the	DET
dad-10670	366	17	documents	document	NOUN
dad-10670	366	18	on	on	ADP
dad-10670	366	19	which	which	PRON
dad-10670	366	20	the	the	DET
dad-10670	366	21	system	system	NOUN
dad-10670	366	22	was	be	AUX
dad-10670	366	23	evaluated	evaluate	VERB
dad-10670	366	24	,	,	PUNCT
dad-10670	366	25	but	but	CCONJ
dad-10670	366	26	we	we	PRON
dad-10670	366	27	do	do	AUX
dad-10670	366	28	not	not	PART
dad-10670	366	29	have	have	VERB
dad-10670	366	30	access	access	NOUN
dad-10670	366	31	to	to	ADP
dad-10670	366	32	the	the	DET
dad-10670	366	33	parser	parser	NOUN
dad-10670	366	34	software	software	NOUN
dad-10670	366	35	either	either	ADV
dad-10670	366	36	.	.	PUNCT
dad-10670	367	1	another	another	DET
dad-10670	367	2	issue	issue	NOUN
dad-10670	367	3	comes	come	VERB
dad-10670	367	4	from	from	ADP
dad-10670	367	5	the	the	DET
dad-10670	367	6	number	number	NOUN
dad-10670	367	7	of	of	ADP
dad-10670	367	8	discourse	discourse	NOUN
dad-10670	367	9	relations	relation	NOUN
dad-10670	367	10	employed	employ	VERB
dad-10670	367	11	:	:	PUNCT
dad-10670	367	12	in	in	ADP
dad-10670	367	13	marcu	marcu	NOUN
dad-10670	367	14	’s	’s	PART
dad-10670	367	15	case	case	NOUN
dad-10670	367	16	,	,	PUNCT
dad-10670	367	17	15	15	NUM
dad-10670	367	18	relations	relation	NOUN
dad-10670	367	19	were	be	AUX
dad-10670	367	20	employed	employ	VERB
dad-10670	367	21	,	,	PUNCT
dad-10670	367	22	while	while	SCONJ
dad-10670	367	23	lethanh	lethanh	ADJ
dad-10670	367	24	used	use	VERB
dad-10670	367	25	14	14	NUM
dad-10670	367	26	,	,	PUNCT
dad-10670	367	27	and	and	CCONJ
dad-10670	367	28	we	we	PRON
dad-10670	367	29	used	use	VERB
dad-10670	367	30	18	18	NUM
dad-10670	367	31	.	.	PUNCT
dad-10670	368	1	for	for	ADP
dad-10670	368	2	these	these	DET
dad-10670	368	3	two	two	NUM
dad-10670	368	4	systems	system	NOUN
dad-10670	368	5	,	,	PUNCT
dad-10670	368	6	we	we	PRON
dad-10670	368	7	have	have	VERB
dad-10670	368	8	no	no	DET
dad-10670	368	9	choice	choice	NOUN
dad-10670	368	10	but	but	SCONJ
dad-10670	368	11	to	to	PART
dad-10670	368	12	reproduce	reproduce	VERB
dad-10670	368	13	the	the	DET
dad-10670	368	14	results	result	NOUN
dad-10670	368	15	provided	provide	VERB
dad-10670	368	16	by	by	ADP
dad-10670	368	17	each	each	DET
dad-10670	368	18	author	author	NOUN
dad-10670	368	19	,	,	PUNCT
dad-10670	368	20	in	in	ADP
dad-10670	368	21	separate	separate	ADJ
dad-10670	368	22	tables	table	NOUN
dad-10670	368	23	for	for	ADP
dad-10670	368	24	fairness	fairness	NOUN
dad-10670	368	25	.	.	PUNCT
dad-10670	369	1	we	we	PRON
dad-10670	369	2	define	define	VERB
dad-10670	369	3	a	a	DET
dad-10670	369	4	measure	measure	NOUN
dad-10670	369	5	,	,	PUNCT
dad-10670	369	6	falgo	falgo	NOUN
dad-10670	369	7	/	/	SYM
dad-10670	369	8	fhuman	fhuman	NOUN
dad-10670	369	9	,	,	PUNCT
dad-10670	369	10	computed	compute	VERB
dad-10670	369	11	for	for	ADP
dad-10670	369	12	each	each	DET
dad-10670	369	13	system	system	NOUN
dad-10670	369	14	,	,	PUNCT
dad-10670	369	15	as	as	ADP
dad-10670	369	16	an	an	DET
dad-10670	369	17	informal	informal	ADJ
dad-10670	369	18	indicator	indicator	NOUN
dad-10670	369	19	of	of	ADP
dad-10670	369	20	how	how	SCONJ
dad-10670	369	21	each	each	DET
dad-10670	369	22	parser	parser	NOUN
dad-10670	369	23	is	be	AUX
dad-10670	369	24	estimated	estimate	VERB
dad-10670	369	25	to	to	PART
dad-10670	369	26	perform	perform	VERB
dad-10670	369	27	compared	compare	VERB
dad-10670	369	28	to	to	ADP
dad-10670	369	29	the	the	DET
dad-10670	369	30	human	human	ADJ
dad-10670	369	31	agreement	agreement	NOUN
dad-10670	369	32	.	.	PUNCT
dad-10670	370	1	as	as	SCONJ
dad-10670	370	2	each	each	DET
dad-10670	370	3	author	author	NOUN
dad-10670	370	4	found	find	VERB
dad-10670	370	5	a	a	DET
dad-10670	370	6	different	different	ADJ
dad-10670	370	7	value	value	NOUN
dad-10670	370	8	for	for	ADP
dad-10670	370	9	human	human	ADJ
dad-10670	370	10	agreement	agreement	NOUN
dad-10670	370	11	f	f	NOUN
dad-10670	370	12	-	-	PUNCT
dad-10670	370	13	score	score	NOUN
dad-10670	370	14	,	,	PUNCT
dad-10670	370	15	we	we	PRON
dad-10670	370	16	scale	scale	VERB
dad-10670	370	17	the	the	DET
dad-10670	370	18	score	score	NOUN
dad-10670	370	19	of	of	ADP
dad-10670	370	20	each	each	DET
dad-10670	370	21	system	system	NOUN
dad-10670	370	22	by	by	ADP
dad-10670	370	23	the	the	DET
dad-10670	370	24	human	human	ADJ
dad-10670	370	25	agreement	agreement	NOUN
dad-10670	370	26	f	f	NOUN
dad-10670	370	27	-	-	PUNCT
dad-10670	370	28	score	score	NOUN
dad-10670	370	29	found	find	VERB
dad-10670	370	30	by	by	ADP
dad-10670	370	31	its	its	PRON
dad-10670	370	32	respective	respective	ADJ
dad-10670	370	33	author	author	NOUN
dad-10670	370	34	.	.	PUNCT
dad-10670	371	1	for	for	ADP
dad-10670	371	2	instance	instance	NOUN
dad-10670	371	3	,	,	PUNCT
dad-10670	371	4	le	le	PROPN
dad-10670	371	5	thanh	thanh	VERB
dad-10670	371	6	et	et	PROPN
dad-10670	371	7	al	al	PROPN
dad-10670	371	8	.	.	PROPN
dad-10670	372	1	(	(	PUNCT
dad-10670	372	2	2004b	2004b	NUM
dad-10670	372	3	)	)	PUNCT
dad-10670	372	4	report	report	VERB
dad-10670	372	5	a	a	DET
dad-10670	372	6	surprisingly	surprisingly	ADV
dad-10670	372	7	low	low	ADJ
dad-10670	372	8	72.7	72.7	NUM
dad-10670	372	9	%	%	NOUN
dad-10670	372	10	f	f	NOUN
dad-10670	372	11	-	-	PUNCT
dad-10670	372	12	score	score	NOUN
dad-10670	372	13	for	for	ADP
dad-10670	372	14	structure	structure	NOUN
dad-10670	372	15	,	,	PUNCT
dad-10670	372	16	when	when	SCONJ
dad-10670	372	17	marcu	marcu	NOUN
dad-10670	372	18	(	(	PUNCT
dad-10670	372	19	2000	2000	NUM
dad-10670	372	20	)	)	PUNCT
dad-10670	372	21	notes	note	VERB
dad-10670	372	22	81	81	NUM
dad-10670	372	23	%	%	NOUN
dad-10670	372	24	,	,	PUNCT
dad-10670	372	25	and	and	CCONJ
dad-10670	372	26	we	we	PRON
dad-10670	372	27	found	find	VERB
dad-10670	372	28	87	87	NUM
dad-10670	372	29	%	%	NOUN
dad-10670	372	30	.	.	PUNCT
dad-10670	373	1	this	this	PRON
dad-10670	373	2	suggests	suggest	VERB
dad-10670	373	3	that	that	SCONJ
dad-10670	373	4	,	,	PUNCT
dad-10670	373	5	despite	despite	SCONJ
dad-10670	373	6	our	our	PRON
dad-10670	373	7	best	good	ADJ
dad-10670	373	8	efforts	effort	NOUN
dad-10670	373	9	,	,	PUNCT
dad-10670	373	10	evaluation	evaluation	NOUN
dad-10670	373	11	metrics	metric	NOUN
dad-10670	373	12	used	use	VERB
dad-10670	373	13	by	by	ADP
dad-10670	373	14	each	each	DET
dad-10670	373	15	author	author	NOUN
dad-10670	373	16	might	might	AUX
dad-10670	373	17	differ	differ	VERB
dad-10670	373	18	.	.	PUNCT
dad-10670	374	1	again	again	ADV
dad-10670	374	2	,	,	PUNCT
dad-10670	374	3	this	this	DET
dad-10670	374	4	measure	measure	NOUN
dad-10670	374	5	is	be	AUX
dad-10670	374	6	an	an	DET
dad-10670	374	7	informal	informal	ADJ
dad-10670	374	8	indicator	indicator	NOUN
dad-10670	374	9	of	of	ADP
dad-10670	374	10	performance	performance	NOUN
dad-10670	374	11	.	.	PUNCT
dad-10670	375	1	table	table	NOUN
dad-10670	375	2	11	11	NUM
dad-10670	375	3	indicates	indicate	VERB
dad-10670	375	4	the	the	DET
dad-10670	375	5	results	result	NOUN
dad-10670	375	6	given	give	VERB
dad-10670	375	7	by	by	ADP
dad-10670	375	8	marcu	marcu	NOUN
dad-10670	375	9	(	(	PUNCT
dad-10670	375	10	2000	2000	NUM
dad-10670	375	11	)	)	PUNCT
dad-10670	375	12	,	,	PUNCT
dad-10670	375	13	table	table	NOUN
dad-10670	375	14	12	12	NUM
dad-10670	375	15	the	the	DET
dad-10670	375	16	results	result	NOUN
dad-10670	375	17	given	give	VERB
dad-10670	375	18	by	by	ADP
dad-10670	375	19	le	le	X
dad-10670	375	20	thanh	thanh	PROPN
dad-10670	375	21	et	et	PROPN
dad-10670	375	22	al	al	PROPN
dad-10670	375	23	.	.	PROPN
dad-10670	375	24	(	(	PUNCT
dad-10670	375	25	2004b	2004b	NUM
dad-10670	375	26	)	)	PUNCT
dad-10670	375	27	,	,	PUNCT
dad-10670	375	28	and	and	CCONJ
dad-10670	375	29	table	table	NOUN
dad-10670	375	30	13	13	NUM
dad-10670	375	31	the	the	DET
dad-10670	375	32	results	result	NOUN
dad-10670	375	33	for	for	ADP
dad-10670	375	34	our	our	PRON
dad-10670	375	35	system	system	NOUN
dad-10670	375	36	.	.	PUNCT
dad-10670	376	1	note	note	VERB
dad-10670	376	2	that	that	SCONJ
dad-10670	376	3	in	in	ADP
dad-10670	376	4	table	table	NOUN
dad-10670	376	5	13	13	NUM
dad-10670	376	6	,	,	PUNCT
dad-10670	376	7	in	in	ADP
dad-10670	376	8	order	order	NOUN
dad-10670	376	9	to	to	PART
dad-10670	376	10	make	make	VERB
dad-10670	376	11	our	our	PRON
dad-10670	376	12	experimental	experimental	ADJ
dad-10670	376	13	setting	setting	NOUN
dad-10670	376	14	as	as	ADV
dad-10670	376	15	close	close	ADJ
dad-10670	376	16	as	as	ADP
dad-10670	376	17	possible	possible	ADJ
dad-10670	376	18	to	to	ADP
dad-10670	376	19	lethanh	lethanh	VERB
dad-10670	376	20	’s	’s	ADV
dad-10670	376	21	,	,	PUNCT
dad-10670	376	22	we	we	PRON
dad-10670	376	23	used	use	VERB
dad-10670	376	24	lethanh	lethanh	PROPN
dad-10670	376	25	’s	’s	PART
dad-10670	376	26	set	set	VERB
dad-10670	376	27	of	of	ADP
dad-10670	376	28	21	21	NUM
dad-10670	376	29	documents	document	NOUN
dad-10670	376	30	as	as	ADP
dad-10670	376	31	testing	testing	NOUN
dad-10670	376	32	subset	subset	NOUN
dad-10670	376	33	,	,	PUNCT
dad-10670	376	34	and	and	CCONJ
dad-10670	376	35	the	the	DET
dad-10670	376	36	rest	rest	NOUN
dad-10670	376	37	of	of	ADP
dad-10670	376	38	the	the	DET
dad-10670	376	39	corpus	corpus	NOUN
dad-10670	376	40	for	for	ADP
dad-10670	376	41	training	training	NOUN
dad-10670	376	42	.	.	PUNCT
dad-10670	377	1	using	use	VERB
dad-10670	377	2	our	our	PRON
dad-10670	377	3	calculated	calculate	VERB
dad-10670	377	4	human	human	ADJ
dad-10670	377	5	-	-	PUNCT
dad-10670	377	6	agreement	agreement	NOUN
dad-10670	377	7	f	f	NOUN
dad-10670	377	8	-	-	PUNCT
dad-10670	377	9	scores	score	NOUN
dad-10670	377	10	,	,	PUNCT
dad-10670	377	11	we	we	PRON
dad-10670	377	12	estimate	estimate	VERB
dad-10670	377	13	that	that	SCONJ
dad-10670	377	14	our	our	PRON
dad-10670	377	15	system	system	NOUN
dad-10670	377	16	reaches	reach	VERB
dad-10670	377	17	86.4	86.4	NUM
dad-10670	377	18	%	%	NOUN
dad-10670	377	19	of	of	ADP
dad-10670	377	20	the	the	DET
dad-10670	377	21	human	human	ADJ
dad-10670	377	22	performance	performance	NOUN
dad-10670	377	23	level	level	NOUN
dad-10670	377	24	for	for	ADP
dad-10670	377	25	structure	structure	NOUN
dad-10670	377	26	labeling	labeling	NOUN
dad-10670	377	27	,	,	PUNCT
dad-10670	377	28	while	while	SCONJ
dad-10670	377	29	for	for	ADP
dad-10670	377	30	nuclearity	nuclearity	NOUN
dad-10670	377	31	and	and	CCONJ
dad-10670	377	32	relation	relation	NOUN
dad-10670	377	33	labeling	labeling	NOUN
dad-10670	377	34	,	,	PUNCT
dad-10670	377	35	we	we	PRON
dad-10670	377	36	obtain	obtain	VERB
dad-10670	377	37	79.8	79.8	NUM
dad-10670	377	38	%	%	NOUN
dad-10670	377	39	and	and	CCONJ
dad-10670	377	40	78.3	78.3	NUM
dad-10670	377	41	%	%	NOUN
dad-10670	377	42	,	,	PUNCT
dad-10670	377	43	respectively	respectively	ADV
dad-10670	377	44	,	,	PUNCT
dad-10670	377	45	of	of	ADP
dad-10670	377	46	the	the	DET
dad-10670	377	47	human	human	ADJ
dad-10670	377	48	f	f	PROPN
dad-10670	377	49	-	-	PUNCT
dad-10670	377	50	score	score	NOUN
dad-10670	377	51	.	.	PUNCT
dad-10670	378	1	marcu	marcu	NOUN
dad-10670	378	2	(	(	PUNCT
dad-10670	378	3	2000	2000	NUM
dad-10670	378	4	)	)	PUNCT
dad-10670	378	5	calculated	calculate	VERB
dad-10670	378	6	reaching	reach	VERB
dad-10670	378	7	25.7	25.7	NUM
dad-10670	378	8	%	%	NOUN
dad-10670	378	9	of	of	ADP
dad-10670	378	10	the	the	DET
dad-10670	378	11	human	human	ADJ
dad-10670	378	12	performance	performance	NOUN
dad-10670	378	13	level	level	NOUN
dad-10670	378	14	for	for	ADP
dad-10670	378	15	relation	relation	NOUN
dad-10670	378	16	labeling	labeling	NOUN
dad-10670	378	17	,	,	PUNCT
dad-10670	378	18	while	while	SCONJ
dad-10670	378	19	le	le	PROPN
dad-10670	378	20	thanh	thanh	VERB
dad-10670	378	21	et	et	PROPN
dad-10670	378	22	al	al	PROPN
dad-10670	378	23	.	.	PROPN
dad-10670	379	1	(	(	PUNCT
dad-10670	379	2	2004b	2004b	NUM
dad-10670	379	3	)	)	PUNCT
dad-10670	379	4	calculated	calculate	VERB
dad-10670	379	5	reaching	reach	VERB
dad-10670	379	6	39.9	39.9	NUM
dad-10670	379	7	%	%	NOUN
dad-10670	379	8	of	of	ADP
dad-10670	379	9	this	this	DET
dad-10670	379	10	level	level	NOUN
dad-10670	379	11	.	.	PUNCT
dad-10670	380	1	direct	direct	ADJ
dad-10670	380	2	comparison	comparison	NOUN
dad-10670	380	3	to	to	ADP
dad-10670	380	4	our	our	PRON
dad-10670	380	5	system	system	NOUN
dad-10670	380	6	is	be	AUX
dad-10670	380	7	only	only	ADV
dad-10670	380	8	possible	possible	ADJ
dad-10670	380	9	in	in	ADP
dad-10670	380	10	the	the	DET
dad-10670	380	11	case	case	NOUN
dad-10670	380	12	of	of	ADP
dad-10670	380	13	sagae	sagae	NOUN
dad-10670	380	14	(	(	PUNCT
dad-10670	380	15	2009	2009	NUM
dad-10670	380	16	)	)	PUNCT
dad-10670	380	17	,	,	PUNCT
dad-10670	380	18	as	as	SCONJ
dad-10670	380	19	sagae	sagae	NOUN
dad-10670	380	20	employed	employ	VERB
dad-10670	380	21	the	the	DET
dad-10670	380	22	same	same	ADJ
dad-10670	380	23	18	18	NUM
dad-10670	380	24	standard	standard	ADJ
dad-10670	380	25	rst	rst	ADJ
dad-10670	380	26	-	-	PUNCT
dad-10670	380	27	dt	dt	NOUN
dad-10670	380	28	relations	relation	NOUN
dad-10670	380	29	,	,	PUNCT
dad-10670	380	30	and	and	CCONJ
dad-10670	380	31	evaluation	evaluation	NOUN
dad-10670	380	32	was	be	AUX
dad-10670	380	33	done	do	VERB
dad-10670	380	34	on	on	ADP
dad-10670	380	35	the	the	DET
dad-10670	380	36	rst	rst	PROPN
dad-10670	380	37	-	-	PUNCT
dad-10670	380	38	dt	dt	PROPN
dad-10670	380	39	’s	’s	PART
dad-10670	380	40	test	test	NOUN
dad-10670	380	41	subset	subset	VERB
dad-10670	380	42	.	.	PUNCT
dad-10670	381	1	a	a	DET
dad-10670	381	2	performance	performance	NOUN
dad-10670	381	3	comparison	comparison	NOUN
dad-10670	381	4	of	of	ADP
dad-10670	381	5	both	both	DET
dad-10670	381	6	systems	system	NOUN
dad-10670	381	7	is	be	AUX
dad-10670	381	8	presented	present	VERB
dad-10670	381	9	in	in	ADP
dad-10670	381	10	table	table	NOUN
dad-10670	381	11	14	14	NUM
dad-10670	381	12	.	.	PUNCT
dad-10670	382	1	the	the	DET
dad-10670	382	2	scores	score	NOUN
dad-10670	382	3	are	be	AUX
dad-10670	382	4	for	for	ADP
dad-10670	382	5	full	full	ADJ
dad-10670	382	6	discourse	discourse	NOUN
dad-10670	382	7	tree	tree	NOUN
dad-10670	382	8	creation	creation	NOUN
dad-10670	382	9	.	.	PUNCT
dad-10670	383	1	we	we	PRON
dad-10670	383	2	observe	observe	VERB
dad-10670	383	3	that	that	SCONJ
dad-10670	383	4	the	the	DET
dad-10670	383	5	performance	performance	NOUN
dad-10670	383	6	of	of	ADP
dad-10670	383	7	the	the	DET
dad-10670	383	8	two	two	NUM
dad-10670	383	9	parsers	parser	NOUN
dad-10670	383	10	is	be	AUX
dad-10670	383	11	relatively	relatively	ADV
dad-10670	383	12	similar	similar	ADJ
dad-10670	383	13	.	.	PUNCT
dad-10670	384	1	in	in	ADP
dad-10670	384	2	particular	particular	ADJ
dad-10670	384	3	,	,	PUNCT
dad-10670	384	4	recall	recall	NOUN
dad-10670	384	5	is	be	AUX
dad-10670	384	6	46.2	46.2	NUM
dad-10670	384	7	%	%	NOUN
dad-10670	384	8	in	in	ADP
dad-10670	384	9	sag	sag	NOUN
dad-10670	384	10	,	,	PUNCT
dad-10670	384	11	while	while	SCONJ
dad-10670	384	12	we	we	PRON
dad-10670	384	13	obtained	obtain	VERB
dad-10670	384	14	46.9	46.9	NUM
dad-10670	384	15	%	%	NOUN
dad-10670	384	16	in	in	ADP
dad-10670	384	17	hilda	hilda	PROPN
dad-10670	384	18	.	.	PUNCT
dad-10670	385	1	precision	precision	NOUN
dad-10670	385	2	for	for	ADP
dad-10670	385	3	hilda	hilda	PROPN
dad-10670	385	4	is	be	AUX
dad-10670	385	5	however	however	ADV
dad-10670	385	6	11	11	NUM
dad-10670	385	7	%	%	NOUN
dad-10670	385	8	higher	high	ADJ
dad-10670	385	9	than	than	ADP
dad-10670	385	10	sag	sag	NOUN
dad-10670	385	11	,	,	PUNCT
dad-10670	385	12	at	at	ADP
dad-10670	385	13	47.7	47.7	NUM
dad-10670	385	14	%	%	NOUN
dad-10670	385	15	,	,	PUNCT
dad-10670	385	16	against	against	ADP
dad-10670	385	17	42.9	42.9	NUM
dad-10670	385	18	%	%	NOUN
dad-10670	385	19	for	for	ADP
dad-10670	385	20	sagae	sagae	NOUN
dad-10670	385	21	’s	’s	PART
dad-10670	385	22	system	system	NOUN
dad-10670	385	23	.	.	PUNCT
dad-10670	386	1	19	19	NUM
dad-10670	386	2	hernault	hernault	ADJ
dad-10670	386	3	,	,	PUNCT
dad-10670	386	4	prendinger	prendinger	NOUN
dad-10670	386	5	,	,	PUNCT
dad-10670	386	6	duverle	duverle	NOUN
dad-10670	386	7	,	,	PUNCT
dad-10670	386	8	and	and	CCONJ
dad-10670	386	9	ishizuka	ishizuka	NOUN
dad-10670	386	10	table	table	NOUN
dad-10670	386	11	11	11	NUM
dad-10670	386	12	:	:	PUNCT
dad-10670	386	13	performance	performance	NOUN
dad-10670	386	14	for	for	ADP
dad-10670	386	15	the	the	DET
dad-10670	386	16	parser	parser	NOUN
dad-10670	386	17	of	of	ADP
dad-10670	386	18	marcu	marcu	NOUN
dad-10670	386	19	(	(	PUNCT
dad-10670	386	20	2000	2000	NUM
dad-10670	386	21	)	)	PUNCT
dad-10670	386	22	,	,	PUNCT
dad-10670	386	23	using	use	VERB
dad-10670	386	24	15	15	NUM
dad-10670	386	25	discourse	discourse	NOUN
dad-10670	386	26	relations	relation	NOUN
dad-10670	386	27	,	,	PUNCT
dad-10670	386	28	evaluated	evaluate	VERB
dad-10670	386	29	on	on	ADP
dad-10670	386	30	unspecified	unspecified	ADJ
dad-10670	386	31	documents	document	NOUN
dad-10670	386	32	structure	structure	NOUN
dad-10670	386	33	nuclearity	nuclearity	NOUN
dad-10670	386	34	relations	relation	NOUN
dad-10670	386	35	precision	precision	VERB
dad-10670	386	36	65.8	65.8	NUM
dad-10670	386	37	54.0	54.0	NUM
dad-10670	386	38	34.3	34.3	NUM
dad-10670	386	39	recall	recall	NOUN
dad-10670	386	40	34.0	34.0	NUM
dad-10670	386	41	21.6	21.6	NUM
dad-10670	386	42	13.0	13.0	NUM
dad-10670	386	43	f	f	NOUN
dad-10670	386	44	-	-	PUNCT
dad-10670	386	45	score	score	VERB
dad-10670	386	46	44.8	44.8	NUM
dad-10670	386	47	30.9	30.9	NUM
dad-10670	386	48	18.8	18.8	NUM
dad-10670	386	49	falgo	falgo	NOUN
dad-10670	386	50	/	/	SYM
dad-10670	386	51	fhuman	fhuman	NOUN
dad-10670	386	52	56.0	56.0	NUM
dad-10670	386	53	42.9	42.9	NUM
dad-10670	386	54	25.7	25.7	NUM
dad-10670	386	55	table	table	NOUN
dad-10670	386	56	12	12	NUM
dad-10670	386	57	:	:	PUNCT
dad-10670	386	58	performance	performance	NOUN
dad-10670	386	59	for	for	ADP
dad-10670	386	60	the	the	DET
dad-10670	386	61	parser	parser	NOUN
dad-10670	386	62	of	of	ADP
dad-10670	386	63	le	le	X
dad-10670	386	64	thanh	thanh	PROPN
dad-10670	386	65	et	et	PROPN
dad-10670	386	66	al	al	PROPN
dad-10670	386	67	.	.	PROPN
dad-10670	387	1	(	(	PUNCT
dad-10670	387	2	2004b	2004b	NUM
dad-10670	387	3	)	)	PUNCT
dad-10670	388	1	,	,	PUNCT
dad-10670	388	2	using	use	VERB
dad-10670	388	3	14	14	NUM
dad-10670	388	4	discourse	discourse	NOUN
dad-10670	388	5	relations	relation	NOUN
dad-10670	388	6	,	,	PUNCT
dad-10670	388	7	evaluated	evaluate	VERB
dad-10670	388	8	on	on	ADP
dad-10670	388	9	21	21	NUM
dad-10670	388	10	documents	document	NOUN
dad-10670	388	11	from	from	ADP
dad-10670	388	12	the	the	DET
dad-10670	388	13	rst	rst	PROPN
dad-10670	388	14	-	-	PUNCT
dad-10670	388	15	dt	dt	PROPN
dad-10670	388	16	structure	structure	NOUN
dad-10670	388	17	nuclearity	nuclearity	NOUN
dad-10670	388	18	relations	relation	NOUN
dad-10670	388	19	precision	precision	VERB
dad-10670	388	20	54.5	54.5	NUM
dad-10670	388	21	47.8	47.8	NUM
dad-10670	388	22	40.5	40.5	NUM
dad-10670	388	23	recall	recall	NOUN
dad-10670	388	24	52.9	52.9	NUM
dad-10670	388	25	46.4	46.4	NUM
dad-10670	388	26	39.3	39.3	NUM
dad-10670	388	27	f	f	NOUN
dad-10670	388	28	-	-	PUNCT
dad-10670	388	29	score	score	VERB
dad-10670	388	30	53.7	53.7	NUM
dad-10670	388	31	47.1	47.1	NUM
dad-10670	388	32	39.9	39.9	NUM
dad-10670	388	33	falgo	falgo	NOUN
dad-10670	388	34	/	/	SYM
dad-10670	388	35	fhuman	fhuman	NOUN
dad-10670	388	36	73.9	73.9	NUM
dad-10670	388	37	71.8	71.8	NUM
dad-10670	388	38	70.1	70.1	NUM
dad-10670	388	39	4.5	4.5	NUM
dad-10670	388	40	time	time	NOUN
dad-10670	388	41	performance	performance	NOUN
dad-10670	388	42	to	to	PART
dad-10670	388	43	complete	complete	VERB
dad-10670	388	44	the	the	DET
dad-10670	388	45	discussion	discussion	NOUN
dad-10670	388	46	of	of	ADP
dad-10670	388	47	the	the	DET
dad-10670	388	48	parser	parser	NOUN
dad-10670	388	49	’s	’s	PART
dad-10670	388	50	actual	actual	ADJ
dad-10670	388	51	performance	performance	NOUN
dad-10670	388	52	,	,	PUNCT
dad-10670	388	53	we	we	PRON
dad-10670	388	54	measure	measure	VERB
dad-10670	388	55	the	the	DET
dad-10670	388	56	time	time	NOUN
dad-10670	388	57	taken	take	VERB
dad-10670	388	58	to	to	PART
dad-10670	388	59	parse	parse	VERB
dad-10670	388	60	standard	standard	ADJ
dad-10670	388	61	texts	text	NOUN
dad-10670	388	62	from	from	ADP
dad-10670	388	63	the	the	DET
dad-10670	388	64	test	test	NOUN
dad-10670	388	65	section	section	NOUN
dad-10670	388	66	of	of	ADP
dad-10670	388	67	the	the	DET
dad-10670	388	68	rst	rst	NOUN
dad-10670	388	69	-	-	PUNCT
dad-10670	388	70	dt	dt	NOUN
dad-10670	388	71	.	.	PUNCT
dad-10670	389	1	the	the	DET
dad-10670	389	2	computer	computer	NOUN
dad-10670	389	3	used	use	VERB
dad-10670	389	4	in	in	ADP
dad-10670	389	5	the	the	DET
dad-10670	389	6	tests	test	NOUN
dad-10670	389	7	has	have	VERB
dad-10670	389	8	a	a	DET
dad-10670	389	9	2.53	2.53	NUM
dad-10670	389	10	ghz	ghz	NOUN
dad-10670	389	11	processor	processor	NOUN
dad-10670	389	12	and	and	CCONJ
dad-10670	389	13	4	4	NUM
dad-10670	389	14	gb	gb	NOUN
dad-10670	389	15	of	of	ADP
dad-10670	389	16	ram	ram	NOUN
dad-10670	389	17	.	.	PUNCT
dad-10670	390	1	on	on	ADP
dad-10670	390	2	average	average	ADJ
dad-10670	390	3	,	,	PUNCT
dad-10670	390	4	loading	load	VERB
dad-10670	390	5	the	the	DET
dad-10670	390	6	stanford	stanford	PROPN
dad-10670	390	7	syntax	syntax	PROPN
dad-10670	390	8	parser	parser	NOUN
dad-10670	390	9	and	and	CCONJ
dad-10670	390	10	svm	svm	ADJ
dad-10670	390	11	models	model	NOUN
dad-10670	390	12	altogether	altogether	ADV
dad-10670	390	13	takes	take	VERB
dad-10670	390	14	four	four	NUM
dad-10670	390	15	seconds	second	NOUN
dad-10670	390	16	.	.	PUNCT
dad-10670	391	1	this	this	PRON
dad-10670	391	2	is	be	AUX
dad-10670	391	3	done	do	VERB
dad-10670	391	4	only	only	ADV
dad-10670	391	5	once	once	ADV
dad-10670	391	6	.	.	PUNCT
dad-10670	392	1	we	we	PRON
dad-10670	392	2	use	use	VERB
dad-10670	392	3	modified	modified	ADJ
dad-10670	392	4	command	command	NOUN
dad-10670	392	5	-	-	PUNCT
dad-10670	392	6	line	line	NOUN
dad-10670	392	7	versions	version	NOUN
dad-10670	392	8	of	of	ADP
dad-10670	392	9	libsvm	libsvm	NOUN
dad-10670	392	10	and	and	CCONJ
dad-10670	392	11	liblinear	liblinear	NOUN
dad-10670	392	12	to	to	PART
dad-10670	392	13	accept	accept	VERB
dad-10670	392	14	data	datum	NOUN
dad-10670	392	15	from	from	ADP
dad-10670	392	16	stdin	stdin	PROPN
dad-10670	392	17	,	,	PUNCT
dad-10670	392	18	so	so	SCONJ
dad-10670	392	19	that	that	SCONJ
dad-10670	392	20	loading	loading	NOUN
dad-10670	392	21	of	of	ADP
dad-10670	392	22	models	model	NOUN
dad-10670	392	23	is	be	AUX
dad-10670	392	24	performed	perform	VERB
dad-10670	392	25	only	only	ADV
dad-10670	392	26	once	once	ADV
dad-10670	392	27	.	.	PUNCT
dad-10670	393	1	after	after	ADP
dad-10670	393	2	this	this	PRON
dad-10670	393	3	,	,	PUNCT
dad-10670	393	4	we	we	PRON
dad-10670	393	5	can	can	AUX
dad-10670	393	6	perform	perform	VERB
dad-10670	393	7	fast	fast	ADJ
dad-10670	393	8	classification	classification	NOUN
dad-10670	393	9	decisions	decision	NOUN
dad-10670	393	10	,	,	PUNCT
dad-10670	393	11	taking	take	VERB
dad-10670	393	12	a	a	DET
dad-10670	393	13	few	few	ADJ
dad-10670	393	14	tenths	tenth	NOUN
dad-10670	393	15	of	of	ADP
dad-10670	393	16	a	a	DET
dad-10670	393	17	second	second	ADJ
dad-10670	393	18	each	each	PRON
dad-10670	393	19	.	.	PUNCT
dad-10670	394	1	table	table	NOUN
dad-10670	394	2	15	15	NUM
dad-10670	394	3	shows	show	VERB
dad-10670	394	4	the	the	DET
dad-10670	394	5	performance	performance	NOUN
dad-10670	394	6	of	of	ADP
dad-10670	394	7	the	the	DET
dad-10670	394	8	parser	parser	NOUN
dad-10670	394	9	on	on	ADP
dad-10670	394	10	various	various	ADJ
dad-10670	394	11	texts	text	NOUN
dad-10670	394	12	.	.	PUNCT
dad-10670	395	1	tseg	tseg	PRON
dad-10670	395	2	is	be	AUX
dad-10670	395	3	the	the	DET
dad-10670	395	4	time	time	NOUN
dad-10670	395	5	taken	take	VERB
dad-10670	395	6	to	to	PART
dad-10670	395	7	segment	segment	VERB
dad-10670	395	8	the	the	DET
dad-10670	395	9	text	text	NOUN
dad-10670	395	10	,	,	PUNCT
dad-10670	395	11	and	and	CCONJ
dad-10670	395	12	tbuild	tbuild	VERB
dad-10670	395	13	the	the	DET
dad-10670	395	14	time	time	NOUN
dad-10670	395	15	taken	take	VERB
dad-10670	395	16	to	to	PART
dad-10670	395	17	build	build	VERB
dad-10670	395	18	the	the	DET
dad-10670	395	19	tree	tree	NOUN
dad-10670	395	20	.	.	PUNCT
dad-10670	396	1	as	as	SCONJ
dad-10670	396	2	expected	expect	VERB
dad-10670	396	3	,	,	PUNCT
dad-10670	396	4	we	we	PRON
dad-10670	396	5	empirically	empirically	ADV
dad-10670	396	6	observe	observe	VERB
dad-10670	396	7	that	that	SCONJ
dad-10670	396	8	tseg	tseg	PROPN
dad-10670	396	9	and	and	CCONJ
dad-10670	396	10	tbuild	tbuild	PROPN
dad-10670	396	11	follow	follow	VERB
dad-10670	396	12	a	a	DET
dad-10670	396	13	o(n	o(n	NOUN
dad-10670	396	14	)	)	PUNCT
dad-10670	396	15	time	time	NOUN
dad-10670	396	16	complexity	complexity	NOUN
dad-10670	396	17	.	.	PUNCT
dad-10670	397	1	however	however	ADV
dad-10670	397	2	,	,	PUNCT
dad-10670	397	3	in	in	ADP
dad-10670	397	4	the	the	DET
dad-10670	397	5	system	system	NOUN
dad-10670	397	6	,	,	PUNCT
dad-10670	397	7	the	the	DET
dad-10670	397	8	only	only	ADJ
dad-10670	397	9	process	process	NOUN
dad-10670	397	10	not	not	PART
dad-10670	397	11	under	under	ADP
dad-10670	397	12	our	our	PRON
dad-10670	397	13	control	control	NOUN
dad-10670	397	14	is	be	AUX
dad-10670	397	15	the	the	DET
dad-10670	397	16	syntax	syntax	NOUN
dad-10670	397	17	parser	parser	NOUN
dad-10670	397	18	.	.	PUNCT
dad-10670	398	1	it	it	PRON
dad-10670	398	2	is	be	AUX
dad-10670	398	3	important	important	ADJ
dad-10670	398	4	to	to	PART
dad-10670	398	5	ensure	ensure	VERB
dad-10670	398	6	that	that	SCONJ
dad-10670	398	7	the	the	DET
dad-10670	398	8	time	time	NOUN
dad-10670	398	9	taken	take	VERB
dad-10670	398	10	to	to	PART
dad-10670	398	11	generate	generate	VERB
dad-10670	398	12	parse	parse	NOUN
dad-10670	398	13	trees	tree	NOUN
dad-10670	398	14	is	be	AUX
dad-10670	398	15	reasonable	reasonable	ADJ
dad-10670	398	16	.	.	PUNCT
dad-10670	399	1	on	on	ADP
dad-10670	399	2	average	average	ADJ
dad-10670	399	3	,	,	PUNCT
dad-10670	399	4	on	on	ADP
dad-10670	399	5	our	our	PRON
dad-10670	399	6	test	test	NOUN
dad-10670	399	7	computer	computer	NOUN
dad-10670	399	8	,	,	PUNCT
dad-10670	399	9	the	the	DET
dad-10670	399	10	stanford	stanford	PROPN
dad-10670	399	11	parser	parser	NOUN
dad-10670	399	12	takes	take	VERB
dad-10670	399	13	10	10	NUM
dad-10670	399	14	seconds	second	NOUN
dad-10670	399	15	to	to	PART
dad-10670	399	16	parse	parse	VERB
dad-10670	399	17	300	300	NUM
dad-10670	399	18	words	word	NOUN
dad-10670	399	19	.	.	PUNCT
dad-10670	400	1	of	of	ADP
dad-10670	400	2	the	the	DET
dad-10670	400	3	three	three	NUM
dad-10670	400	4	processes	process	NOUN
dad-10670	400	5	(	(	PUNCT
dad-10670	400	6	syntax	syntax	NOUN
dad-10670	400	7	parsing	parsing	NOUN
dad-10670	400	8	,	,	PUNCT
dad-10670	400	9	discourse	discourse	NOUN
dad-10670	400	10	segmentation	segmentation	NOUN
dad-10670	400	11	,	,	PUNCT
dad-10670	400	12	tree	tree	NOUN
dad-10670	400	13	building	building	NOUN
dad-10670	400	14	)	)	PUNCT
dad-10670	400	15	,	,	PUNCT
dad-10670	400	16	syntax	syntax	NOUN
dad-10670	400	17	parsing	parsing	NOUN
dad-10670	400	18	is	be	AUX
dad-10670	400	19	always	always	ADV
dad-10670	400	20	the	the	DET
dad-10670	400	21	slowest	slow	ADJ
dad-10670	400	22	task	task	NOUN
dad-10670	400	23	,	,	PUNCT
dad-10670	400	24	and	and	CCONJ
dad-10670	400	25	thus	thus	ADV
dad-10670	400	26	the	the	DET
dad-10670	400	27	bottleneck	bottleneck	NOUN
dad-10670	400	28	of	of	ADP
dad-10670	400	29	the	the	DET
dad-10670	400	30	system	system	NOUN
dad-10670	400	31	.	.	PUNCT
dad-10670	401	1	still	still	ADV
dad-10670	401	2	,	,	PUNCT
dad-10670	401	3	this	this	DET
dad-10670	401	4	level	level	NOUN
dad-10670	401	5	of	of	ADP
dad-10670	401	6	performance	performance	NOUN
dad-10670	401	7	makes	make	VERB
dad-10670	401	8	the	the	DET
dad-10670	401	9	parser	parser	NOUN
dad-10670	401	10	usable	usable	ADJ
dad-10670	401	11	as	as	ADP
dad-10670	401	12	part	part	NOUN
dad-10670	401	13	of	of	ADP
dad-10670	401	14	an	an	DET
dad-10670	401	15	interactive	interactive	ADJ
dad-10670	401	16	system	system	NOUN
dad-10670	401	17	.	.	PUNCT
dad-10670	402	1	20	20	NUM
dad-10670	402	2	hilda	hilda	VERB
dad-10670	402	3	:	:	PUNCT
dad-10670	402	4	a	a	DET
dad-10670	402	5	discourse	discourse	NOUN
dad-10670	402	6	parser	parser	NOUN
dad-10670	402	7	using	use	VERB
dad-10670	402	8	support	support	NOUN
dad-10670	402	9	vector	vector	NOUN
dad-10670	402	10	machine	machine	NOUN
dad-10670	402	11	classification	classification	NOUN
dad-10670	402	12	table	table	NOUN
dad-10670	402	13	13	13	NUM
dad-10670	402	14	:	:	PUNCT
dad-10670	402	15	performance	performance	NOUN
dad-10670	402	16	for	for	ADP
dad-10670	402	17	hilda	hilda	NOUN
dad-10670	402	18	,	,	PUNCT
dad-10670	402	19	using	use	VERB
dad-10670	402	20	18	18	NUM
dad-10670	402	21	discourse	discourse	NOUN
dad-10670	402	22	relations	relation	NOUN
dad-10670	402	23	,	,	PUNCT
dad-10670	402	24	evaluated	evaluate	VERB
dad-10670	402	25	on	on	ADP
dad-10670	402	26	21	21	NUM
dad-10670	402	27	documents	document	NOUN
dad-10670	402	28	from	from	ADP
dad-10670	402	29	the	the	DET
dad-10670	402	30	rst	rst	NOUN
dad-10670	402	31	-	-	PUNCT
dad-10670	402	32	dt	dt	PROPN
dad-10670	402	33	(	(	PUNCT
dad-10670	402	34	lethanh	lethanh	VERB
dad-10670	402	35	’s	’s	PART
dad-10670	402	36	)	)	PUNCT
dad-10670	402	37	structure	structure	NOUN
dad-10670	402	38	nuclearity	nuclearity	NOUN
dad-10670	402	39	relations	relation	NOUN
dad-10670	402	40	precision	precision	VERB
dad-10670	402	41	76.0	76.0	NUM
dad-10670	402	42	61.4	61.4	NUM
dad-10670	402	43	51.2	51.2	NUM
dad-10670	402	44	recall	recall	NOUN
dad-10670	402	45	75.6	75.6	NUM
dad-10670	402	46	61.2	61.2	NUM
dad-10670	402	47	50.6	50.6	NUM
dad-10670	402	48	f	f	NOUN
dad-10670	402	49	-	-	PUNCT
dad-10670	402	50	score	score	NOUN
dad-10670	402	51	75.8	75.8	NUM
dad-10670	402	52	61.3	61.3	NUM
dad-10670	402	53	50.9	50.9	NUM
dad-10670	402	54	falgo	falgo	NOUN
dad-10670	402	55	/	/	SYM
dad-10670	402	56	fhuman	fhuman	NOUN
dad-10670	402	57	86.4	86.4	NUM
dad-10670	402	58	79.8	79.8	NUM
dad-10670	402	59	78.3	78.3	NUM
dad-10670	402	60	table	table	NOUN
dad-10670	402	61	14	14	NUM
dad-10670	402	62	:	:	PUNCT
dad-10670	402	63	performance	performance	NOUN
dad-10670	402	64	comparison	comparison	NOUN
dad-10670	402	65	between	between	ADP
dad-10670	402	66	the	the	DET
dad-10670	402	67	proposed	propose	VERB
dad-10670	402	68	parser	parser	NOUN
dad-10670	402	69	(	(	PUNCT
dad-10670	402	70	hilda	hilda	PROPN
dad-10670	402	71	)	)	PUNCT
dad-10670	402	72	and	and	CCONJ
dad-10670	402	73	the	the	DET
dad-10670	402	74	parser	parser	NOUN
dad-10670	402	75	of	of	ADP
dad-10670	402	76	sagae	sagae	NOUN
dad-10670	402	77	(	(	PUNCT
dad-10670	402	78	2009	2009	NUM
dad-10670	402	79	)	)	PUNCT
dad-10670	402	80	(	(	PUNCT
dad-10670	402	81	sag	sag	NOUN
dad-10670	402	82	)	)	PUNCT
dad-10670	402	83	,	,	PUNCT
dad-10670	402	84	for	for	ADP
dad-10670	402	85	full	full	ADJ
dad-10670	402	86	discourse	discourse	NOUN
dad-10670	402	87	tree	tree	NOUN
dad-10670	402	88	creation	creation	NOUN
dad-10670	402	89	,	,	PUNCT
dad-10670	402	90	using	use	VERB
dad-10670	402	91	18	18	NUM
dad-10670	402	92	discourse	discourse	NOUN
dad-10670	402	93	relations	relation	NOUN
dad-10670	402	94	,	,	PUNCT
dad-10670	402	95	evaluated	evaluate	VERB
dad-10670	402	96	on	on	ADP
dad-10670	402	97	the	the	DET
dad-10670	402	98	rst	rst	PROPN
dad-10670	402	99	-	-	PUNCT
dad-10670	402	100	dt	dt	PROPN
dad-10670	402	101	’s	’s	PART
dad-10670	402	102	test	test	NOUN
dad-10670	402	103	set	set	VERB
dad-10670	402	104	sag	sag	PROPN
dad-10670	402	105	hilda	hilda	PROPN
dad-10670	402	106	precision	precision	NOUN
dad-10670	402	107	42.9	42.9	NUM
dad-10670	402	108	47.7	47.7	NUM
dad-10670	402	109	recall	recall	VERB
dad-10670	402	110	46.2	46.2	NUM
dad-10670	402	111	46.9	46.9	NUM
dad-10670	402	112	f	f	NOUN
dad-10670	402	113	-	-	PUNCT
dad-10670	402	114	score	score	VERB
dad-10670	402	115	44.5	44.5	NUM
dad-10670	402	116	47.3	47.3	NUM
dad-10670	402	117	4.6	4.6	NUM
dad-10670	402	118	limitations	limitation	NOUN
dad-10670	402	119	in	in	ADP
dad-10670	402	120	this	this	DET
dad-10670	402	121	section	section	NOUN
dad-10670	402	122	,	,	PUNCT
dad-10670	402	123	we	we	PRON
dad-10670	402	124	present	present	VERB
dad-10670	402	125	some	some	DET
dad-10670	402	126	limitations	limitation	NOUN
dad-10670	402	127	of	of	ADP
dad-10670	402	128	the	the	DET
dad-10670	402	129	parser	parser	NOUN
dad-10670	402	130	and	and	CCONJ
dad-10670	402	131	discuss	discuss	VERB
dad-10670	402	132	possible	possible	ADJ
dad-10670	402	133	improvements	improvement	NOUN
dad-10670	402	134	and	and	CCONJ
dad-10670	402	135	directions	direction	NOUN
dad-10670	402	136	for	for	ADP
dad-10670	402	137	future	future	ADJ
dad-10670	402	138	work	work	NOUN
dad-10670	402	139	.	.	PUNCT
dad-10670	403	1	when	when	SCONJ
dad-10670	403	2	evaluating	evaluate	VERB
dad-10670	403	3	the	the	DET
dad-10670	403	4	proposed	propose	VERB
dad-10670	403	5	discourse	discourse	NOUN
dad-10670	403	6	segmenter	segmenter	NOUN
dad-10670	403	7	on	on	ADP
dad-10670	403	8	the	the	DET
dad-10670	403	9	rst	rst	PROPN
dad-10670	403	10	-	-	PUNCT
dad-10670	403	11	dt	dt	PROPN
dad-10670	403	12	’s	’s	PART
dad-10670	403	13	test	test	NOUN
dad-10670	403	14	set	set	NOUN
dad-10670	403	15	,	,	PUNCT
dad-10670	403	16	we	we	PRON
dad-10670	403	17	observe	observe	VERB
dad-10670	403	18	that	that	SCONJ
dad-10670	403	19	all	all	DET
dad-10670	403	20	segmentation	segmentation	NOUN
dad-10670	403	21	errors	error	NOUN
dad-10670	403	22	are	be	AUX
dad-10670	403	23	due	due	ADJ
dad-10670	403	24	to	to	ADP
dad-10670	403	25	over	over	ADP
dad-10670	403	26	-	-	PUNCT
dad-10670	403	27	segmentation	segmentation	NOUN
dad-10670	403	28	,	,	PUNCT
dad-10670	403	29	i.e.	i.e.	X
dad-10670	403	30	words	word	NOUN
dad-10670	403	31	which	which	PRON
dad-10670	403	32	are	be	AUX
dad-10670	403	33	not	not	PART
dad-10670	403	34	edu	edu	PROPN
dad-10670	403	35	boundaries	boundary	NOUN
dad-10670	403	36	are	be	AUX
dad-10670	403	37	mistaken	mistaken	ADJ
dad-10670	403	38	for	for	ADP
dad-10670	403	39	boundaries	boundary	NOUN
dad-10670	403	40	.	.	PUNCT
dad-10670	404	1	table	table	NOUN
dad-10670	404	2	16	16	NUM
dad-10670	404	3	shows	show	VERB
dad-10670	404	4	the	the	DET
dad-10670	404	5	ten	ten	NUM
dad-10670	404	6	most	most	ADV
dad-10670	404	7	frequent	frequent	ADJ
dad-10670	404	8	words	word	NOUN
dad-10670	404	9	on	on	ADP
dad-10670	404	10	which	which	PRON
dad-10670	404	11	segmentation	segmentation	NOUN
dad-10670	404	12	errors	error	NOUN
dad-10670	404	13	occur	occur	VERB
dad-10670	404	14	,	,	PUNCT
dad-10670	404	15	as	as	ADV
dad-10670	404	16	well	well	ADV
dad-10670	404	17	as	as	ADP
dad-10670	404	18	how	how	SCONJ
dad-10670	404	19	prevalent	prevalent	ADJ
dad-10670	404	20	these	these	DET
dad-10670	404	21	errors	error	NOUN
dad-10670	404	22	are	be	AUX
dad-10670	404	23	among	among	ADP
dad-10670	404	24	all	all	DET
dad-10670	404	25	segmentation	segmentation	NOUN
dad-10670	404	26	errors	error	NOUN
dad-10670	404	27	.	.	PUNCT
dad-10670	405	1	several	several	ADJ
dad-10670	405	2	of	of	ADP
dad-10670	405	3	these	these	DET
dad-10670	405	4	mistakes	mistake	NOUN
dad-10670	405	5	are	be	AUX
dad-10670	405	6	linked	link	VERB
dad-10670	405	7	to	to	ADP
dad-10670	405	8	punctuation	punctuation	NOUN
dad-10670	405	9	,	,	PUNCT
dad-10670	405	10	particularly	particularly	ADV
dad-10670	405	11	to	to	ADP
dad-10670	405	12	quotes	quote	NOUN
dad-10670	405	13	and	and	CCONJ
dad-10670	405	14	dash	dash	NOUN
dad-10670	405	15	marks	mark	NOUN
dad-10670	405	16	.	.	PUNCT
dad-10670	406	1	also	also	ADV
dad-10670	406	2	,	,	PUNCT
dad-10670	406	3	several	several	ADJ
dad-10670	406	4	mistakes	mistake	NOUN
dad-10670	406	5	seem	seem	VERB
dad-10670	406	6	related	related	ADJ
dad-10670	406	7	to	to	ADP
dad-10670	406	8	the	the	DET
dad-10670	406	9	segmenter	segmenter	NOUN
dad-10670	406	10	creating	create	VERB
dad-10670	406	11	excessively	excessively	ADV
dad-10670	406	12	small	small	ADJ
dad-10670	406	13	clause	clause	NOUN
dad-10670	406	14	-	-	PUNCT
dad-10670	406	15	like	like	ADJ
dad-10670	406	16	units	unit	NOUN
dad-10670	406	17	.	.	PUNCT
dad-10670	407	1	typical	typical	ADJ
dad-10670	407	2	improper	improper	ADJ
dad-10670	407	3	seg21	seg21	PROPN
dad-10670	407	4	hernault	hernault	PROPN
dad-10670	407	5	,	,	PUNCT
dad-10670	407	6	prendinger	prendinger	NOUN
dad-10670	407	7	,	,	PUNCT
dad-10670	407	8	duverle	duverle	NOUN
dad-10670	407	9	,	,	PUNCT
dad-10670	407	10	and	and	CCONJ
dad-10670	407	11	ishizuka	ishizuka	NOUN
dad-10670	407	12	table	table	NOUN
dad-10670	407	13	15	15	NUM
dad-10670	407	14	:	:	PUNCT
dad-10670	407	15	time	time	NOUN
dad-10670	407	16	taken	take	VERB
dad-10670	407	17	to	to	PART
dad-10670	407	18	parse	parse	VERB
dad-10670	407	19	various	various	ADJ
dad-10670	407	20	texts	text	NOUN
dad-10670	407	21	(	(	PUNCT
dad-10670	407	22	in	in	ADP
dad-10670	407	23	seconds	second	NOUN
dad-10670	407	24	)	)	PUNCT
dad-10670	407	25	text	text	NOUN
dad-10670	407	26	sentences	sentence	VERB
dad-10670	407	27	words	word	NOUN
dad-10670	407	28	tseg	tseg	PROPN
dad-10670	407	29	(	(	PUNCT
dad-10670	407	30	s	s	NOUN
dad-10670	407	31	)	)	PUNCT
dad-10670	407	32	tbuild	tbuild	NOUN
dad-10670	407	33	(	(	PUNCT
dad-10670	407	34	s	s	NOUN
dad-10670	407	35	)	)	PUNCT
dad-10670	407	36	wsj0616	wsj0616	VERB
dad-10670	407	37	36	36	NUM
dad-10670	407	38	838	838	NUM
dad-10670	407	39	5.02	5.02	NUM
dad-10670	407	40	11.30	11.30	NUM
dad-10670	407	41	wsj1113	wsj1113	NOUN
dad-10670	407	42	5	5	NUM
dad-10670	407	43	102	102	NUM
dad-10670	407	44	0.40	0.40	NUM
dad-10670	407	45	0.51	0.51	NUM
dad-10670	407	46	wsj1129	wsj1129	NOUN
dad-10670	407	47	2	2	NUM
dad-10670	407	48	45	45	NUM
dad-10670	407	49	0.26	0.26	NUM
dad-10670	407	50	0.40	0.40	NUM
dad-10670	407	51	wsj2336	wsj2336	NOUN
dad-10670	407	52	10	10	NUM
dad-10670	407	53	262	262	NUM
dad-10670	407	54	1.71	1.71	NUM
dad-10670	407	55	3.72	3.72	NUM
dad-10670	407	56	wsj2385	wsj2385	NOUN
dad-10670	407	57	50	50	NUM
dad-10670	407	58	516	516	NUM
dad-10670	407	59	3.05	3.05	NUM
dad-10670	407	60	7.17	7.17	NUM
dad-10670	407	61	mentation	mentation	NOUN
dad-10670	407	62	decisions	decision	NOUN
dad-10670	407	63	(	(	PUNCT
dad-10670	407	64	marked	mark	VERB
dad-10670	407	65	↑	↑	NOUN
dad-10670	407	66	)	)	PUNCT
dad-10670	407	67	related	relate	VERB
dad-10670	407	68	to	to	ADP
dad-10670	407	69	these	these	DET
dad-10670	407	70	words	word	NOUN
dad-10670	407	71	are	be	AUX
dad-10670	407	72	seen	see	VERB
dad-10670	407	73	in	in	ADP
dad-10670	407	74	the	the	DET
dad-10670	407	75	following	follow	VERB
dad-10670	407	76	sentences	sentence	NOUN
dad-10670	407	77	,	,	PUNCT
dad-10670	407	78	where	where	SCONJ
dad-10670	407	79	the	the	DET
dad-10670	407	80	true	true	ADJ
dad-10670	407	81	segmentation	segmentation	NOUN
dad-10670	407	82	decisions	decision	NOUN
dad-10670	407	83	(	(	PUNCT
dad-10670	407	84	when	when	SCONJ
dad-10670	407	85	present	present	ADJ
dad-10670	407	86	)	)	PUNCT
dad-10670	407	87	are	be	AUX
dad-10670	407	88	marked	mark	VERB
dad-10670	407	89	⇑	⇑	PROPN
dad-10670	407	90	:	:	PUNCT
dad-10670	407	91	‘	'	PUNCT
dad-10670	407	92	‘	'	PUNCT
dad-10670	407	93	[	[	X
dad-10670	407	94	...	...	X
dad-10670	407	95	]	]	X
dad-10670	407	96	the	the	DET
dad-10670	407	97	banco	banco	PROPN
dad-10670	407	98	exterior	exterior	PROPN
dad-10670	407	99	group	group	NOUN
dad-10670	407	100	has	have	VERB
dad-10670	407	101	a	a	DET
dad-10670	407	102	lot	lot	NOUN
dad-10670	407	103	↑	↑	NOUN
dad-10670	407	104	to	to	PART
dad-10670	407	105	offer	offer	VERB
dad-10670	407	106	a	a	DET
dad-10670	407	107	potential	potential	ADJ
dad-10670	407	108	suitor	suitor	NOUN
dad-10670	407	109	.	.	PUNCT
dad-10670	407	110	’’	’'	PUNCT
dad-10670	408	1	‘	'	PUNCT
dad-10670	408	2	‘	'	PUNCT
dad-10670	408	3	[	[	X
dad-10670	408	4	...	...	X
dad-10670	408	5	]	]	X
dad-10670	408	6	when	when	SCONJ
dad-10670	408	7	the	the	DET
dad-10670	408	8	market	market	NOUN
dad-10670	408	9	does	do	AUX
dad-10670	408	10	recover	recover	VERB
dad-10670	408	11	,	,	PUNCT
dad-10670	408	12	↑	↑	PROPN
dad-10670	408	13	the	the	DET
dad-10670	408	14	damage	damage	NOUN
dad-10670	408	15	is	be	AUX
dad-10670	408	16	done	do	VERB
dad-10670	408	17	[	[	X
dad-10670	408	18	...	...	PUNCT
dad-10670	408	19	]	]	X
dad-10670	408	20	’’	’'	PUNCT
dad-10670	408	21	‘	'	PUNCT
dad-10670	408	22	‘	'	PUNCT
dad-10670	408	23	[	[	X
dad-10670	408	24	...	...	X
dad-10670	408	25	]	]	X
dad-10670	408	26	cut	cut	NOUN
dad-10670	408	27	costs	cost	NOUN
dad-10670	408	28	,	,	PUNCT
dad-10670	408	29	increase	increase	VERB
dad-10670	408	30	capital	capital	NOUN
dad-10670	408	31	↑	↑	NOUN
dad-10670	408	32	and	and	CCONJ
dad-10670	408	33	build	build	VERB
dad-10670	408	34	new	new	ADJ
dad-10670	408	35	areas	area	NOUN
dad-10670	408	36	of	of	ADP
dad-10670	408	37	business	business	NOUN
dad-10670	408	38	[	[	X
dad-10670	408	39	...	...	PUNCT
dad-10670	408	40	]	]	X
dad-10670	408	41	’’	’'	PUNCT
dad-10670	408	42	‘	'	PUNCT
dad-10670	408	43	‘	'	PUNCT
dad-10670	408	44	[	[	X
dad-10670	408	45	...	...	X
dad-10670	408	46	]	]	PUNCT
dad-10670	408	47	and	and	CCONJ
dad-10670	408	48	perhaps	perhaps	ADV
dad-10670	408	49	even	even	ADV
dad-10670	408	50	destroy	destroy	VERB
dad-10670	408	51	↑	↑	NOUN
dad-10670	408	52	-	-	PUNCT
dad-10670	408	53	⇑	⇑	PROPN
dad-10670	408	54	a	a	DET
dad-10670	408	55	$	$	SYM
dad-10670	408	56	2.38	2.38	NUM
dad-10670	408	57	settlement	settlement	NOUN
dad-10670	408	58	fund	fund	NOUN
dad-10670	409	1	[	[	X
dad-10670	409	2	...	...	PUNCT
dad-10670	409	3	]	]	X
dad-10670	409	4	’’	’'	PUNCT
dad-10670	409	5	‘	'	PUNCT
dad-10670	409	6	‘	'	PUNCT
dad-10670	409	7	in	in	ADP
dad-10670	409	8	august	august	PROPN
dad-10670	409	9	,	,	PUNCT
dad-10670	409	10	↑	↑	PROPN
dad-10670	409	11	mr	mr	PROPN
dad-10670	409	12	.	.	PROPN
dad-10670	409	13	lewis	lewis	PROPN
dad-10670	409	14	pleaded	plead	VERB
dad-10670	409	15	guilty	guilty	ADJ
dad-10670	409	16	to	to	ADP
dad-10670	409	17	three	three	NUM
dad-10670	409	18	felony	felony	NOUN
dad-10670	409	19	counts	count	NOUN
dad-10670	409	20	.	.	PUNCT
dad-10670	409	21	’’	’'	PUNCT
dad-10670	410	1	‘	'	PUNCT
dad-10670	410	2	‘	'	PUNCT
dad-10670	410	3	when	when	SCONJ
dad-10670	410	4	it	it	PRON
dad-10670	410	5	misses	miss	VERB
dad-10670	410	6	one	one	NUM
dad-10670	410	7	month	month	NOUN
dad-10670	410	8	↑	↑	NOUN
dad-10670	410	9	it	it	PRON
dad-10670	410	10	tends	tend	VERB
dad-10670	410	11	to	to	PART
dad-10670	410	12	miss	miss	VERB
dad-10670	410	13	the	the	DET
dad-10670	410	14	next	next	ADJ
dad-10670	410	15	month	month	NOUN
dad-10670	410	16	.	.	PUNCT
dad-10670	411	1	’’	’'	PUNCT
dad-10670	411	2	table	table	NOUN
dad-10670	411	3	16	16	NUM
dad-10670	411	4	:	:	PUNCT
dad-10670	411	5	words	word	NOUN
dad-10670	411	6	on	on	ADP
dad-10670	411	7	which	which	PRON
dad-10670	411	8	erroneous	erroneous	ADJ
dad-10670	411	9	segmentation	segmentation	NOUN
dad-10670	411	10	decisions	decision	NOUN
dad-10670	411	11	are	be	AUX
dad-10670	411	12	most	most	ADV
dad-10670	411	13	frequent	frequent	ADJ
dad-10670	411	14	,	,	PUNCT
dad-10670	411	15	and	and	CCONJ
dad-10670	411	16	their	their	PRON
dad-10670	411	17	proportion	proportion	NOUN
dad-10670	411	18	among	among	ADP
dad-10670	411	19	all	all	DET
dad-10670	411	20	segmentation	segmentation	NOUN
dad-10670	411	21	errors	error	NOUN
dad-10670	411	22	on	on	ADP
dad-10670	411	23	the	the	DET
dad-10670	411	24	rst	rst	PROPN
dad-10670	411	25	-	-	PUNCT
dad-10670	411	26	dt	dt	PROPN
dad-10670	411	27	’s	’s	PART
dad-10670	411	28	test	test	NOUN
dad-10670	411	29	set	set	VERB
dad-10670	411	30	word	word	NOUN
dad-10670	411	31	pos	pos	NOUN
dad-10670	411	32	prevalence	prevalence	NOUN
dad-10670	411	33	among	among	ADP
dad-10670	411	34	errors	error	NOUN
dad-10670	411	35	(	(	PUNCT
dad-10670	411	36	%	%	INTJ
dad-10670	411	37	)	)	PUNCT
dad-10670	411	38	to	to	ADP
dad-10670	411	39	to	to	ADP
dad-10670	411	40	7.7	7.7	NUM
dad-10670	411	41	the	the	DET
dad-10670	411	42	dt	dt	PROPN
dad-10670	411	43	6.6	6.6	NUM
dad-10670	411	44	and	and	CCONJ
dad-10670	411	45	cc	cc	ADP
dad-10670	411	46	4.6	4.6	NUM
dad-10670	411	47	“	"	PUNCT
dad-10670	411	48	“	"	PUNCT
dad-10670	411	49	3.4	3.4	NUM
dad-10670	411	50	–	–	PUNCT
dad-10670	411	51	:	:	PUNCT
dad-10670	411	52	2.3	2.3	NUM
dad-10670	411	53	mr	mr	PROPN
dad-10670	411	54	.	.	PROPN
dad-10670	411	55	nnp	nnp	PROPN
dad-10670	411	56	2.3	2.3	NUM
dad-10670	412	1	it	it	PRON
dad-10670	412	2	prp	prp	VERB
dad-10670	412	3	2.1	2.1	NUM
dad-10670	412	4	for	for	ADP
dad-10670	412	5	in	in	ADP
dad-10670	412	6	2.0	2.0	NUM
dad-10670	412	7	if	if	SCONJ
dad-10670	412	8	in	in	ADP
dad-10670	412	9	1.8	1.8	NUM
dad-10670	412	10	said	say	VERB
dad-10670	412	11	vbd	vbd	NOUN
dad-10670	412	12	1.6	1.6	NUM
dad-10670	412	13	22	22	NUM
dad-10670	412	14	hilda	hilda	VERB
dad-10670	412	15	:	:	PUNCT
dad-10670	412	16	a	a	DET
dad-10670	412	17	discourse	discourse	NOUN
dad-10670	412	18	parser	parser	NOUN
dad-10670	412	19	using	use	VERB
dad-10670	412	20	support	support	NOUN
dad-10670	412	21	vector	vector	NOUN
dad-10670	412	22	machine	machine	NOUN
dad-10670	412	23	classification	classification	NOUN
dad-10670	412	24	although	although	SCONJ
dad-10670	412	25	a	a	DET
dad-10670	412	26	quantitative	quantitative	ADJ
dad-10670	412	27	evaluation	evaluation	NOUN
dad-10670	412	28	of	of	ADP
dad-10670	412	29	the	the	DET
dad-10670	412	30	discourse	discourse	NOUN
dad-10670	412	31	trees	tree	NOUN
dad-10670	412	32	produced	produce	VERB
dad-10670	412	33	by	by	ADP
dad-10670	412	34	our	our	PRON
dad-10670	412	35	parser	parser	NOUN
dad-10670	412	36	is	be	AUX
dad-10670	412	37	made	make	VERB
dad-10670	412	38	possible	possible	ADJ
dad-10670	412	39	by	by	ADP
dad-10670	412	40	the	the	DET
dad-10670	412	41	usage	usage	NOUN
dad-10670	412	42	of	of	ADP
dad-10670	412	43	metrics	metric	NOUN
dad-10670	412	44	traditionally	traditionally	ADV
dad-10670	412	45	employed	employ	VERB
dad-10670	412	46	for	for	ADP
dad-10670	412	47	evaluating	evaluate	VERB
dad-10670	412	48	the	the	DET
dad-10670	412	49	performance	performance	NOUN
dad-10670	412	50	of	of	ADP
dad-10670	412	51	syntax	syntax	NOUN
dad-10670	412	52	parsers	parser	NOUN
dad-10670	412	53	,	,	PUNCT
dad-10670	412	54	we	we	PRON
dad-10670	412	55	found	find	VERB
dad-10670	412	56	that	that	SCONJ
dad-10670	412	57	a	a	DET
dad-10670	412	58	qualitative	qualitative	ADJ
dad-10670	412	59	evaluation	evaluation	NOUN
dad-10670	412	60	of	of	ADP
dad-10670	412	61	the	the	DET
dad-10670	412	62	generated	generate	VERB
dad-10670	412	63	discourse	discourse	NOUN
dad-10670	412	64	trees	tree	NOUN
dad-10670	412	65	was	be	AUX
dad-10670	412	66	difficult	difficult	ADJ
dad-10670	412	67	.	.	PUNCT
dad-10670	413	1	in	in	ADP
dad-10670	413	2	particular	particular	ADJ
dad-10670	413	3	,	,	PUNCT
dad-10670	413	4	we	we	PRON
dad-10670	413	5	hypothesize	hypothesize	VERB
dad-10670	413	6	that	that	SCONJ
dad-10670	413	7	finding	find	VERB
dad-10670	413	8	patterns	pattern	NOUN
dad-10670	413	9	of	of	ADP
dad-10670	413	10	errors	error	NOUN
dad-10670	413	11	commonly	commonly	ADV
dad-10670	413	12	present	present	ADJ
dad-10670	413	13	in	in	ADP
dad-10670	413	14	the	the	DET
dad-10670	413	15	generated	generate	VERB
dad-10670	413	16	discourse	discourse	NOUN
dad-10670	413	17	trees	tree	NOUN
dad-10670	413	18	is	be	AUX
dad-10670	413	19	made	make	VERB
dad-10670	413	20	difficult	difficult	ADJ
dad-10670	413	21	by	by	ADP
dad-10670	413	22	the	the	DET
dad-10670	413	23	greedy	greedy	ADJ
dad-10670	413	24	approach	approach	NOUN
dad-10670	413	25	adopted	adopt	VERB
dad-10670	413	26	.	.	PUNCT
dad-10670	414	1	because	because	SCONJ
dad-10670	414	2	of	of	ADP
dad-10670	414	3	the	the	DET
dad-10670	414	4	proposed	propose	VERB
dad-10670	414	5	algorithm	algorithm	NOUN
dad-10670	414	6	and	and	CCONJ
dad-10670	414	7	the	the	DET
dad-10670	414	8	way	way	NOUN
dad-10670	414	9	the	the	DET
dad-10670	414	10	classifiers	classifier	NOUN
dad-10670	414	11	are	be	AUX
dad-10670	414	12	cascaded	cascade	VERB
dad-10670	414	13	,	,	PUNCT
dad-10670	414	14	classification	classification	NOUN
dad-10670	414	15	errors	error	NOUN
dad-10670	414	16	occurring	occur	VERB
dad-10670	414	17	anywhere	anywhere	ADV
dad-10670	414	18	when	when	SCONJ
dad-10670	414	19	building	build	VERB
dad-10670	414	20	the	the	DET
dad-10670	414	21	discourse	discourse	NOUN
dad-10670	414	22	tree	tree	NOUN
dad-10670	414	23	will	will	AUX
dad-10670	414	24	affect	affect	VERB
dad-10670	414	25	both	both	DET
dad-10670	414	26	sibling	sible	VERB
dad-10670	414	27	nodes	node	NOUN
dad-10670	414	28	and	and	CCONJ
dad-10670	414	29	higher	high	ADJ
dad-10670	414	30	level	level	NOUN
dad-10670	414	31	relations	relation	NOUN
dad-10670	414	32	of	of	ADP
dad-10670	414	33	the	the	DET
dad-10670	414	34	tree	tree	NOUN
dad-10670	414	35	,	,	PUNCT
dad-10670	414	36	making	make	VERB
dad-10670	414	37	the	the	DET
dad-10670	414	38	real	real	ADJ
dad-10670	414	39	cause	cause	NOUN
dad-10670	414	40	of	of	ADP
dad-10670	414	41	errors	error	NOUN
dad-10670	414	42	difficult	difficult	ADJ
dad-10670	414	43	to	to	PART
dad-10670	414	44	identify	identify	VERB
dad-10670	414	45	.	.	PUNCT
dad-10670	415	1	indeed	indeed	ADV
dad-10670	415	2	,	,	PUNCT
dad-10670	415	3	the	the	DET
dad-10670	415	4	presented	present	VERB
dad-10670	415	5	greedy	greedy	ADJ
dad-10670	415	6	algorithm	algorithm	NOUN
dad-10670	415	7	is	be	AUX
dad-10670	415	8	time	time	NOUN
dad-10670	415	9	-	-	PUNCT
dad-10670	415	10	efficient	efficient	ADJ
dad-10670	415	11	,	,	PUNCT
dad-10670	415	12	but	but	CCONJ
dad-10670	415	13	provides	provide	VERB
dad-10670	415	14	only	only	ADV
dad-10670	415	15	locally	locally	ADV
dad-10670	415	16	optimal	optimal	ADJ
dad-10670	415	17	solutions	solution	NOUN
dad-10670	415	18	.	.	PUNCT
dad-10670	416	1	using	use	VERB
dad-10670	416	2	generic	generic	ADJ
dad-10670	416	3	global	global	ADJ
dad-10670	416	4	probabilistic	probabilistic	ADJ
dad-10670	416	5	optimization	optimization	NOUN
dad-10670	416	6	meta	meta	ADJ
dad-10670	416	7	-	-	PUNCT
dad-10670	416	8	algorithms	algorithm	NOUN
dad-10670	416	9	such	such	ADJ
dad-10670	416	10	as	as	ADP
dad-10670	416	11	simulated	simulated	ADJ
dad-10670	416	12	annealing	annealing	NOUN
dad-10670	416	13	(	(	PUNCT
dad-10670	416	14	kirkpatrick	kirkpatrick	PROPN
dad-10670	416	15	et	et	PROPN
dad-10670	416	16	al	al	PROPN
dad-10670	416	17	.	.	PROPN
dad-10670	416	18	,	,	PUNCT
dad-10670	416	19	1983	1983	NUM
dad-10670	416	20	)	)	PUNCT
dad-10670	416	21	,	,	PUNCT
dad-10670	416	22	we	we	PRON
dad-10670	416	23	hope	hope	VERB
dad-10670	416	24	to	to	PART
dad-10670	416	25	address	address	VERB
dad-10670	416	26	the	the	DET
dad-10670	416	27	problem	problem	NOUN
dad-10670	416	28	of	of	ADP
dad-10670	416	29	local	local	ADJ
dad-10670	416	30	optimality	optimality	NOUN
dad-10670	416	31	while	while	SCONJ
dad-10670	416	32	maintaining	maintain	VERB
dad-10670	416	33	reasonable	reasonable	ADJ
dad-10670	416	34	time	time	NOUN
dad-10670	416	35	complexity	complexity	NOUN
dad-10670	416	36	.	.	PUNCT
dad-10670	417	1	secondly	secondly	ADV
dad-10670	417	2	,	,	PUNCT
dad-10670	417	3	we	we	PRON
dad-10670	417	4	plan	plan	VERB
dad-10670	417	5	to	to	PART
dad-10670	417	6	investigate	investigate	VERB
dad-10670	417	7	the	the	DET
dad-10670	417	8	use	use	NOUN
dad-10670	417	9	of	of	ADP
dad-10670	417	10	sequential	sequential	ADJ
dad-10670	417	11	labeling	labeling	NOUN
dad-10670	417	12	methods	method	NOUN
dad-10670	417	13	(	(	PUNCT
dad-10670	417	14	lafferty	lafferty	NOUN
dad-10670	417	15	et	et	PROPN
dad-10670	417	16	al	al	PROPN
dad-10670	417	17	.	.	PROPN
dad-10670	417	18	,	,	PUNCT
dad-10670	417	19	2001	2001	NUM
dad-10670	417	20	)	)	PUNCT
dad-10670	417	21	for	for	ADP
dad-10670	417	22	finding	find	VERB
dad-10670	417	23	optimal	optimal	ADJ
dad-10670	417	24	sequences	sequence	NOUN
dad-10670	417	25	of	of	ADP
dad-10670	417	26	discourse	discourse	NOUN
dad-10670	417	27	relations	relation	NOUN
dad-10670	417	28	.	.	PUNCT
dad-10670	418	1	another	another	DET
dad-10670	418	2	related	related	ADJ
dad-10670	418	3	shortcoming	shortcoming	NOUN
dad-10670	418	4	of	of	ADP
dad-10670	418	5	the	the	DET
dad-10670	418	6	current	current	ADJ
dad-10670	418	7	algorithm	algorithm	NOUN
dad-10670	418	8	,	,	PUNCT
dad-10670	418	9	as	as	SCONJ
dad-10670	418	10	presented	present	VERB
dad-10670	418	11	in	in	ADP
dad-10670	418	12	figure	figure	NOUN
dad-10670	418	13	6	6	NUM
dad-10670	418	14	,	,	PUNCT
dad-10670	418	15	occurs	occur	VERB
dad-10670	418	16	when	when	SCONJ
dad-10670	418	17	two	two	NUM
dad-10670	418	18	pairs	pair	NOUN
dad-10670	418	19	of	of	ADP
dad-10670	418	20	edus	edus	NOUN
dad-10670	418	21	have	have	VERB
dad-10670	418	22	an	an	DET
dad-10670	418	23	equal	equal	ADJ
dad-10670	418	24	probability	probability	NOUN
dad-10670	418	25	of	of	ADP
dad-10670	418	26	being	be	AUX
dad-10670	418	27	connected	connect	VERB
dad-10670	418	28	by	by	ADP
dad-10670	418	29	a	a	DET
dad-10670	418	30	discourse	discourse	NOUN
dad-10670	418	31	relation	relation	NOUN
dad-10670	418	32	.	.	PUNCT
dad-10670	419	1	in	in	ADP
dad-10670	419	2	the	the	DET
dad-10670	419	3	present	present	ADJ
dad-10670	419	4	system	system	NOUN
dad-10670	419	5	,	,	PUNCT
dad-10670	419	6	the	the	DET
dad-10670	419	7	default	default	NOUN
dad-10670	419	8	behavior	behavior	NOUN
dad-10670	419	9	in	in	ADP
dad-10670	419	10	case	case	NOUN
dad-10670	419	11	of	of	ADP
dad-10670	419	12	conflict	conflict	NOUN
dad-10670	419	13	is	be	AUX
dad-10670	419	14	to	to	PART
dad-10670	419	15	select	select	VERB
dad-10670	419	16	the	the	DET
dad-10670	419	17	first	first	ADJ
dad-10670	419	18	pair	pair	NOUN
dad-10670	419	19	of	of	ADP
dad-10670	419	20	units	unit	NOUN
dad-10670	419	21	in	in	ADP
dad-10670	419	22	text	text	NOUN
dad-10670	419	23	order	order	NOUN
dad-10670	419	24	.	.	PUNCT
dad-10670	420	1	however	however	ADV
dad-10670	420	2	,	,	PUNCT
dad-10670	420	3	ideally	ideally	ADV
dad-10670	420	4	,	,	PUNCT
dad-10670	420	5	all	all	DET
dad-10670	420	6	candidates	candidate	NOUN
dad-10670	420	7	should	should	AUX
dad-10670	420	8	be	be	AUX
dad-10670	420	9	considered	consider	VERB
dad-10670	420	10	,	,	PUNCT
dad-10670	420	11	with	with	SCONJ
dad-10670	420	12	their	their	PRON
dad-10670	420	13	respective	respective	ADJ
dad-10670	420	14	trees	tree	NOUN
dad-10670	420	15	built	build	VERB
dad-10670	420	16	independently	independently	ADV
dad-10670	420	17	,	,	PUNCT
dad-10670	420	18	somehow	somehow	ADV
dad-10670	420	19	scored	score	VERB
dad-10670	420	20	,	,	PUNCT
dad-10670	420	21	and	and	CCONJ
dad-10670	420	22	the	the	DET
dad-10670	420	23	best	good	ADJ
dad-10670	420	24	candidate	candidate	NOUN
dad-10670	420	25	being	be	AUX
dad-10670	420	26	finally	finally	ADV
dad-10670	420	27	returned	return	VERB
dad-10670	420	28	.	.	PUNCT
dad-10670	421	1	we	we	PRON
dad-10670	421	2	hope	hope	VERB
dad-10670	421	3	to	to	PART
dad-10670	421	4	address	address	VERB
dad-10670	421	5	this	this	DET
dad-10670	421	6	point	point	NOUN
dad-10670	421	7	by	by	ADP
dad-10670	421	8	employing	employ	VERB
dad-10670	421	9	the	the	DET
dad-10670	421	10	aforementioned	aforementioned	ADJ
dad-10670	421	11	optimization	optimization	NOUN
dad-10670	421	12	methods	method	NOUN
dad-10670	421	13	.	.	PUNCT
dad-10670	422	1	in	in	ADP
dad-10670	422	2	section	section	NOUN
dad-10670	422	3	4.2	4.2	NUM
dad-10670	422	4	,	,	PUNCT
dad-10670	422	5	we	we	PRON
dad-10670	422	6	noticed	notice	VERB
dad-10670	422	7	important	important	ADJ
dad-10670	422	8	discrepancies	discrepancy	NOUN
dad-10670	422	9	in	in	ADP
dad-10670	422	10	f	f	NOUN
dad-10670	422	11	-	-	PUNCT
dad-10670	422	12	scores	score	NOUN
dad-10670	422	13	(	(	PUNCT
dad-10670	422	14	see	see	VERB
dad-10670	422	15	table	table	NOUN
dad-10670	422	16	7	7	NUM
dad-10670	422	17	)	)	PUNCT
dad-10670	422	18	for	for	ADP
dad-10670	422	19	the	the	DET
dad-10670	422	20	different	different	ADJ
dad-10670	422	21	classes	class	NOUN
dad-10670	422	22	of	of	ADP
dad-10670	422	23	label	label	NOUN
dad-10670	422	24	.	.	PUNCT
dad-10670	423	1	we	we	PRON
dad-10670	423	2	hypothesize	hypothesize	VERB
dad-10670	423	3	that	that	SCONJ
dad-10670	423	4	this	this	DET
dad-10670	423	5	issue	issue	NOUN
dad-10670	423	6	is	be	AUX
dad-10670	423	7	mainly	mainly	ADV
dad-10670	423	8	related	relate	VERB
dad-10670	423	9	to	to	ADP
dad-10670	423	10	the	the	DET
dad-10670	423	11	nature	nature	NOUN
dad-10670	423	12	of	of	ADP
dad-10670	423	13	discourse	discourse	NOUN
dad-10670	423	14	relations	relation	NOUN
dad-10670	423	15	.	.	PUNCT
dad-10670	424	1	indeed	indeed	ADV
dad-10670	424	2	,	,	PUNCT
dad-10670	424	3	a	a	DET
dad-10670	424	4	notoriously	notoriously	ADV
dad-10670	424	5	challenging	challenging	ADJ
dad-10670	424	6	task	task	NOUN
dad-10670	424	7	for	for	ADP
dad-10670	424	8	discourse	discourse	NOUN
dad-10670	424	9	parsers	parser	NOUN
dad-10670	424	10	is	be	AUX
dad-10670	424	11	to	to	PART
dad-10670	424	12	detect	detect	VERB
dad-10670	424	13	‘	'	PUNCT
dad-10670	424	14	implicit	implicit	ADJ
dad-10670	424	15	relations	relation	NOUN
dad-10670	424	16	’	'	PUNCT
dad-10670	424	17	,	,	PUNCT
dad-10670	424	18	i.e.	i.e.	X
dad-10670	424	19	relations	relation	NOUN
dad-10670	424	20	not	not	PART
dad-10670	424	21	signaled	signal	VERB
dad-10670	424	22	by	by	ADP
dad-10670	424	23	discourse	discourse	NOUN
dad-10670	424	24	cues	cue	NOUN
dad-10670	424	25	,	,	PUNCT
dad-10670	424	26	such	such	ADJ
dad-10670	424	27	as	as	ADP
dad-10670	424	28	‘	'	PUNCT
dad-10670	424	29	thus	thus	ADV
dad-10670	424	30	’	'	PUNCT
dad-10670	424	31	,	,	PUNCT
dad-10670	424	32	‘	'	PUNCT
dad-10670	424	33	however	however	ADV
dad-10670	424	34	’	'	PUNCT
dad-10670	424	35	,	,	PUNCT
dad-10670	424	36	‘	'	PUNCT
dad-10670	424	37	but	but	CCONJ
dad-10670	424	38	’	'	PUNCT
dad-10670	424	39	,	,	PUNCT
dad-10670	424	40	etc	etc	X
dad-10670	424	41	.	.	X
dad-10670	424	42	certain	certain	ADJ
dad-10670	424	43	relations	relation	NOUN
dad-10670	424	44	,	,	PUNCT
dad-10670	424	45	such	such	ADJ
dad-10670	424	46	as	as	ADP
dad-10670	424	47	attribution	attribution	NOUN
dad-10670	424	48	,	,	PUNCT
dad-10670	424	49	are	be	AUX
dad-10670	424	50	invariably	invariably	ADV
dad-10670	424	51	realized	realize	VERB
dad-10670	424	52	with	with	ADP
dad-10670	424	53	a	a	DET
dad-10670	424	54	discourse	discourse	NOUN
dad-10670	424	55	cue	cue	NOUN
dad-10670	424	56	,	,	PUNCT
dad-10670	424	57	indicated	indicate	VERB
dad-10670	424	58	by	by	ADP
dad-10670	424	59	the	the	DET
dad-10670	424	60	verb	verb	NOUN
dad-10670	424	61	‘	'	PUNCT
dad-10670	424	62	say	say	INTJ
dad-10670	424	63	’	'	PUNCT
dad-10670	424	64	,	,	PUNCT
dad-10670	424	65	such	such	ADJ
dad-10670	424	66	as	as	ADP
dad-10670	424	67	‘	'	PUNCT
dad-10670	424	68	x	x	PROPN
dad-10670	424	69	said	say	VERB
dad-10670	424	70	that	that	SCONJ
dad-10670	424	71	y	y	NOUN
dad-10670	424	72	’	'	PUNCT
dad-10670	424	73	.	.	PUNCT
dad-10670	425	1	this	this	DET
dad-10670	425	2	relation	relation	NOUN
dad-10670	425	3	is	be	AUX
dad-10670	425	4	easy	easy	ADJ
dad-10670	425	5	to	to	PART
dad-10670	425	6	detect	detect	VERB
dad-10670	425	7	,	,	PUNCT
dad-10670	425	8	and	and	CCONJ
dad-10670	425	9	we	we	PRON
dad-10670	425	10	empirically	empirically	ADV
dad-10670	425	11	observe	observe	VERB
dad-10670	425	12	high	high	ADJ
dad-10670	425	13	performance	performance	NOUN
dad-10670	425	14	on	on	ADP
dad-10670	425	15	attribution	attribution	NOUN
dad-10670	425	16	’s	’s	PART
dad-10670	425	17	classes	class	NOUN
dad-10670	425	18	(	(	PUNCT
dad-10670	425	19	see	see	VERB
dad-10670	425	20	table	table	NOUN
dad-10670	425	21	7	7	NUM
dad-10670	425	22	)	)	PUNCT
dad-10670	425	23	.	.	PUNCT
dad-10670	426	1	on	on	ADP
dad-10670	426	2	the	the	DET
dad-10670	426	3	other	other	ADJ
dad-10670	426	4	hand	hand	NOUN
dad-10670	426	5	,	,	PUNCT
dad-10670	426	6	an	an	DET
dad-10670	426	7	implicit	implicit	ADJ
dad-10670	426	8	contrast	contrast	NOUN
dad-10670	426	9	relation	relation	NOUN
dad-10670	426	10	holds	hold	VERB
dad-10670	426	11	between	between	ADP
dad-10670	426	12	the	the	DET
dad-10670	426	13	two	two	NUM
dad-10670	426	14	following	follow	VERB
dad-10670	426	15	discourse	discourse	NOUN
dad-10670	426	16	units	unit	NOUN
dad-10670	426	17	:	:	PUNCT
dad-10670	427	1	[	[	X
dad-10670	427	2	mr	mr	PROPN
dad-10670	427	3	.	.	PROPN
dad-10670	427	4	roman	roman	PROPN
dad-10670	427	5	comes	come	VERB
dad-10670	427	6	across	across	ADP
dad-10670	427	7	as	as	ADP
dad-10670	427	8	a	a	DET
dad-10670	427	9	low	low	ADJ
dad-10670	427	10	-	-	PUNCT
dad-10670	427	11	key	key	NOUN
dad-10670	427	12	executive;]2a	executive;]2a	PROPN
dad-10670	427	13	[	[	X
dad-10670	427	14	mr	mr	PROPN
dad-10670	427	15	.	.	PROPN
dad-10670	427	16	phillips	phillips	PROPN
dad-10670	427	17	has	have	VERB
dad-10670	427	18	a	a	DET
dad-10670	427	19	flashier	flashy	ADJ
dad-10670	427	20	personality.]2b	personality.]2b	NOUN
dad-10670	427	21	in	in	ADP
dad-10670	427	22	this	this	DET
dad-10670	427	23	example	example	NOUN
dad-10670	427	24	,	,	PUNCT
dad-10670	427	25	the	the	DET
dad-10670	427	26	contrast	contrast	NOUN
dad-10670	427	27	is	be	AUX
dad-10670	427	28	captured	capture	VERB
dad-10670	427	29	by	by	ADP
dad-10670	427	30	the	the	DET
dad-10670	427	31	antonymy	antonymy	NOUN
dad-10670	427	32	between	between	ADP
dad-10670	427	33	the	the	DET
dad-10670	427	34	adjectives	adjective	NOUN
dad-10670	427	35	low	low	ADJ
dad-10670	427	36	-	-	PUNCT
dad-10670	427	37	key	key	NOUN
dad-10670	427	38	and	and	CCONJ
dad-10670	427	39	flashy	flashy	ADJ
dad-10670	427	40	.	.	PUNCT
dad-10670	428	1	as	as	SCONJ
dad-10670	428	2	there	there	PRON
dad-10670	428	3	is	be	VERB
dad-10670	428	4	no	no	DET
dad-10670	428	5	cue	cue	NOUN
dad-10670	428	6	,	,	PUNCT
dad-10670	428	7	it	it	PRON
dad-10670	428	8	is	be	AUX
dad-10670	428	9	harder	hard	ADJ
dad-10670	428	10	to	to	PART
dad-10670	428	11	detect	detect	VERB
dad-10670	428	12	,	,	PUNCT
dad-10670	428	13	and	and	CCONJ
dad-10670	428	14	we	we	PRON
dad-10670	428	15	observe	observe	VERB
dad-10670	428	16	in	in	ADP
dad-10670	428	17	practice	practice	NOUN
dad-10670	428	18	a	a	DET
dad-10670	428	19	lower	low	ADJ
dad-10670	428	20	performance	performance	NOUN
dad-10670	428	21	for	for	ADP
dad-10670	428	22	contrast	contrast	NOUN
dad-10670	428	23	relations	relation	NOUN
dad-10670	428	24	.	.	PUNCT
dad-10670	429	1	taboada	taboada	NOUN
dad-10670	429	2	(	(	PUNCT
dad-10670	429	3	2006	2006	NUM
dad-10670	429	4	)	)	PUNCT
dad-10670	429	5	estimates	estimate	VERB
dad-10670	429	6	that	that	SCONJ
dad-10670	429	7	60	60	NUM
dad-10670	429	8	%	%	NOUN
dad-10670	429	9	to	to	PART
dad-10670	429	10	70	70	NUM
dad-10670	429	11	%	%	NOUN
dad-10670	429	12	of	of	ADP
dad-10670	429	13	naturally	naturally	ADV
dad-10670	429	14	-	-	PUNCT
dad-10670	429	15	occurring	occur	VERB
dad-10670	429	16	discourse	discourse	NOUN
dad-10670	429	17	relations	relation	NOUN
dad-10670	429	18	are	be	AUX
dad-10670	429	19	implicit	implicit	ADJ
dad-10670	429	20	.	.	PUNCT
dad-10670	430	1	in	in	ADP
dad-10670	430	2	general	general	ADJ
dad-10670	430	3	,	,	PUNCT
dad-10670	430	4	while	while	SCONJ
dad-10670	430	5	explicit	explicit	ADJ
dad-10670	430	6	discourse	discourse	NOUN
dad-10670	430	7	relations	relation	NOUN
dad-10670	430	8	are	be	AUX
dad-10670	430	9	well	well	ADV
dad-10670	430	10	detected	detect	VERB
dad-10670	430	11	(	(	PUNCT
dad-10670	430	12	pitler	pitler	NOUN
dad-10670	430	13	et	et	PROPN
dad-10670	430	14	al	al	PROPN
dad-10670	430	15	.	.	PROPN
dad-10670	430	16	,	,	PUNCT
dad-10670	430	17	2008	2008	NUM
dad-10670	430	18	)	)	PUNCT
dad-10670	430	19	,	,	PUNCT
dad-10670	430	20	implicit	implicit	ADJ
dad-10670	430	21	relations	relation	NOUN
dad-10670	430	22	are	be	AUX
dad-10670	430	23	challenging	challenge	VERB
dad-10670	430	24	to	to	PART
dad-10670	430	25	detect	detect	VERB
dad-10670	430	26	.	.	PUNCT
dad-10670	431	1	for	for	ADP
dad-10670	431	2	instance	instance	NOUN
dad-10670	431	3	,	,	PUNCT
dad-10670	431	4	we	we	PRON
dad-10670	431	5	fed	feed	VERB
dad-10670	431	6	the	the	DET
dad-10670	431	7	proposed	propose	VERB
dad-10670	431	8	parser	parser	NOUN
dad-10670	431	9	ten	ten	NUM
dad-10670	431	10	short	short	ADJ
dad-10670	431	11	sentences	sentence	NOUN
dad-10670	431	12	containing	contain	VERB
dad-10670	431	13	an	an	DET
dad-10670	431	14	implicit	implicit	ADJ
dad-10670	431	15	contrast	contrast	NOUN
dad-10670	431	16	relation	relation	NOUN
dad-10670	431	17	.	.	PUNCT
dad-10670	432	1	all	all	DET
dad-10670	432	2	relations	relation	NOUN
dad-10670	432	3	were	be	AUX
dad-10670	432	4	wrongly	wrongly	ADV
dad-10670	432	5	classified	classify	VERB
dad-10670	432	6	,	,	PUNCT
dad-10670	432	7	either	either	CCONJ
dad-10670	432	8	as	as	ADP
dad-10670	432	9	elaboration	elaboration	NOUN
dad-10670	432	10	in	in	ADP
dad-10670	432	11	six	six	NUM
dad-10670	432	12	cases	case	NOUN
dad-10670	432	13	or	or	CCONJ
dad-10670	432	14	as	as	ADP
dad-10670	432	15	joint	joint	ADJ
dad-10670	432	16	in	in	ADP
dad-10670	432	17	four	four	NUM
dad-10670	432	18	cases	case	NOUN
dad-10670	432	19	.	.	PUNCT
dad-10670	433	1	similar	similar	ADJ
dad-10670	433	2	errors	error	NOUN
dad-10670	433	3	were	be	AUX
dad-10670	433	4	seen	see	VERB
dad-10670	433	5	when	when	SCONJ
dad-10670	433	6	trying	try	VERB
dad-10670	433	7	to	to	PART
dad-10670	433	8	classify	classify	VERB
dad-10670	433	9	implicit	implicit	ADJ
dad-10670	433	10	comparison	comparison	NOUN
dad-10670	433	11	and	and	CCONJ
dad-10670	433	12	temporal	temporal	ADJ
dad-10670	433	13	relations	relation	NOUN
dad-10670	433	14	.	.	PUNCT
dad-10670	434	1	this	this	PRON
dad-10670	434	2	indicates	indicate	VERB
dad-10670	434	3	that	that	SCONJ
dad-10670	434	4	our	our	PRON
dad-10670	434	5	set	set	NOUN
dad-10670	434	6	of	of	ADP
dad-10670	434	7	features	feature	NOUN
dad-10670	434	8	is	be	AUX
dad-10670	434	9	not	not	PART
dad-10670	434	10	fully	fully	ADV
dad-10670	434	11	-	-	PUNCT
dad-10670	434	12	adapted	adapt	VERB
dad-10670	434	13	for	for	ADP
dad-10670	434	14	implicit	implicit	ADJ
dad-10670	434	15	relations	relation	NOUN
dad-10670	434	16	.	.	PUNCT
dad-10670	435	1	recent	recent	ADJ
dad-10670	435	2	evidence	evidence	NOUN
dad-10670	435	3	(	(	PUNCT
dad-10670	435	4	webber	webber	PROPN
dad-10670	435	5	,	,	PUNCT
dad-10670	435	6	2009	2009	NUM
dad-10670	435	7	)	)	PUNCT
dad-10670	435	8	even	even	ADV
dad-10670	435	9	suggests	suggest	VERB
dad-10670	435	10	that	that	SCONJ
dad-10670	435	11	the	the	DET
dad-10670	435	12	features	feature	NOUN
dad-10670	435	13	present	present	ADJ
dad-10670	435	14	when	when	SCONJ
dad-10670	435	15	an	an	DET
dad-10670	435	16	explicit	explicit	ADJ
dad-10670	435	17	discourse	discourse	NOUN
dad-10670	435	18	relation	relation	NOUN
dad-10670	435	19	is	be	AUX
dad-10670	435	20	realized	realize	VERB
dad-10670	435	21	differ	differ	VERB
dad-10670	435	22	significantly	significantly	ADV
dad-10670	435	23	from	from	ADP
dad-10670	435	24	those	those	PRON
dad-10670	435	25	present	present	ADJ
dad-10670	435	26	in	in	ADP
dad-10670	435	27	the	the	DET
dad-10670	435	28	case	case	NOUN
dad-10670	435	29	of	of	ADP
dad-10670	435	30	an	an	DET
dad-10670	435	31	implicit	implicit	ADJ
dad-10670	435	32	relation	relation	NOUN
dad-10670	435	33	.	.	PUNCT
dad-10670	436	1	a	a	DET
dad-10670	436	2	possible	possible	ADJ
dad-10670	436	3	solution	solution	NOUN
dad-10670	436	4	is	be	AUX
dad-10670	436	5	to	to	PART
dad-10670	436	6	train	train	VERB
dad-10670	436	7	a	a	DET
dad-10670	436	8	separate	separate	ADJ
dad-10670	436	9	discourse	discourse	NOUN
dad-10670	436	10	relation	relation	NOUN
dad-10670	436	11	classifier	classifier	NOUN
dad-10670	436	12	for	for	ADP
dad-10670	436	13	implicit	implicit	ADJ
dad-10670	436	14	discourse	discourse	NOUN
dad-10670	436	15	relations	relation	NOUN
dad-10670	436	16	.	.	PUNCT
dad-10670	437	1	a	a	DET
dad-10670	437	2	different	different	ADJ
dad-10670	437	3	set	set	NOUN
dad-10670	437	4	of	of	ADP
dad-10670	437	5	features	feature	NOUN
dad-10670	437	6	will	will	AUX
dad-10670	437	7	be	be	AUX
dad-10670	437	8	required	require	VERB
dad-10670	437	9	.	.	PUNCT
dad-10670	438	1	for	for	ADP
dad-10670	438	2	this	this	DET
dad-10670	438	3	task	task	NOUN
dad-10670	438	4	,	,	PUNCT
dad-10670	438	5	several	several	ADJ
dad-10670	438	6	researchers	researcher	NOUN
dad-10670	438	7	(	(	PUNCT
dad-10670	438	8	marcu	marcu	NOUN
dad-10670	438	9	and	and	CCONJ
dad-10670	438	10	echihabi	echihabi	NOUN
dad-10670	438	11	,	,	PUNCT
dad-10670	438	12	2002	2002	NUM
dad-10670	438	13	;	;	PUNCT
dad-10670	438	14	pitler	pitler	NOUN
dad-10670	438	15	et	et	PROPN
dad-10670	438	16	al	al	PROPN
dad-10670	438	17	.	.	PROPN
dad-10670	438	18	,	,	PUNCT
dad-10670	438	19	2009	2009	NUM
dad-10670	438	20	;	;	PUNCT
dad-10670	438	21	lin	lin	PROPN
dad-10670	438	22	et	et	PROPN
dad-10670	438	23	al	al	PROPN
dad-10670	438	24	.	.	PROPN
dad-10670	438	25	,	,	PUNCT
dad-10670	438	26	2009	2009	NUM
dad-10670	438	27	)	)	PUNCT
dad-10670	438	28	have	have	AUX
dad-10670	438	29	noted	note	VERB
dad-10670	438	30	relatively	relatively	ADV
dad-10670	438	31	good	good	ADJ
dad-10670	438	32	performance	performance	NOUN
dad-10670	438	33	of	of	ADP
dad-10670	438	34	word	word	NOUN
dad-10670	438	35	pair	pair	NOUN
dad-10670	438	36	features	feature	NOUN
dad-10670	438	37	taken	take	VERB
dad-10670	438	38	between	between	ADP
dad-10670	438	39	the	the	DET
dad-10670	438	40	relation	relation	NOUN
dad-10670	438	41	’s	’s	PART
dad-10670	438	42	two	two	NUM
dad-10670	438	43	argument	argument	NOUN
dad-10670	438	44	23	23	NUM
dad-10670	438	45	hernault	hernault	ADJ
dad-10670	438	46	,	,	PUNCT
dad-10670	438	47	prendinger	prendinger	NOUN
dad-10670	438	48	,	,	PUNCT
dad-10670	438	49	duverle	duverle	NOUN
dad-10670	438	50	,	,	PUNCT
dad-10670	438	51	and	and	CCONJ
dad-10670	438	52	ishizuka	ishizuka	PROPN
dad-10670	438	53	spans	span	NOUN
dad-10670	438	54	.	.	PUNCT
dad-10670	439	1	for	for	ADP
dad-10670	439	2	instance	instance	NOUN
dad-10670	439	3	,	,	PUNCT
dad-10670	439	4	using	use	VERB
dad-10670	439	5	the	the	DET
dad-10670	439	6	same	same	ADJ
dad-10670	439	7	example	example	NOUN
dad-10670	439	8	,	,	PUNCT
dad-10670	439	9	we	we	PRON
dad-10670	439	10	would	would	AUX
dad-10670	439	11	encode	encode	VERB
dad-10670	439	12	lemmatized	lemmatize	VERB
dad-10670	439	13	word	word	NOUN
dad-10670	439	14	pairs	pair	NOUN
dad-10670	439	15	such	such	ADJ
dad-10670	439	16	as	as	ADP
dad-10670	439	17	(	(	PUNCT
dad-10670	439	18	mr	mr	PROPN
dad-10670	439	19	.	.	PROPN
dad-10670	439	20	,	,	PUNCT
dad-10670	439	21	mr	mr	PROPN
dad-10670	439	22	.	.	PROPN
dad-10670	439	23	)	)	PUNCT
dad-10670	439	24	,	,	PUNCT
dad-10670	439	25	(	(	PUNCT
dad-10670	439	26	mr	mr	PROPN
dad-10670	439	27	.	.	PROPN
dad-10670	439	28	,	,	PUNCT
dad-10670	439	29	phillips	phillips	PROPN
dad-10670	439	30	)	)	PUNCT
dad-10670	439	31	,	,	PUNCT
dad-10670	439	32	.	.	PUNCT
dad-10670	439	33	.	.	PUNCT
dad-10670	440	1	.	.	PUNCT
dad-10670	441	1	,	,	PUNCT
dad-10670	441	2	(	(	PUNCT
dad-10670	441	3	low	low	ADJ
dad-10670	441	4	-	-	PUNCT
dad-10670	441	5	key	key	ADJ
dad-10670	441	6	,	,	PUNCT
dad-10670	441	7	flashy	flashy	ADJ
dad-10670	441	8	)	)	PUNCT
dad-10670	441	9	,	,	PUNCT
dad-10670	441	10	.	.	PUNCT
dad-10670	441	11	.	.	PUNCT
dad-10670	441	12	.	.	PUNCT
dad-10670	442	1	,	,	PUNCT
dad-10670	442	2	(	(	PUNCT
dad-10670	442	3	executive	executive	NOUN
dad-10670	442	4	,	,	PUNCT
dad-10670	442	5	personality	personality	NOUN
dad-10670	442	6	)	)	PUNCT
dad-10670	442	7	,	,	PUNCT
dad-10670	442	8	etc	etc	X
dad-10670	442	9	.	.	X
dad-10670	443	1	however	however	ADV
dad-10670	443	2	,	,	PUNCT
dad-10670	443	3	word	word	NOUN
dad-10670	443	4	pair	pair	NOUN
dad-10670	443	5	features	feature	VERB
dad-10670	443	6	alone	alone	ADV
dad-10670	443	7	are	be	AUX
dad-10670	443	8	not	not	PART
dad-10670	443	9	sufficient	sufficient	ADJ
dad-10670	443	10	to	to	PART
dad-10670	443	11	obtain	obtain	VERB
dad-10670	443	12	high	high	ADJ
dad-10670	443	13	performance	performance	NOUN
dad-10670	443	14	rates	rate	NOUN
dad-10670	443	15	.	.	PUNCT
dad-10670	444	1	we	we	PRON
dad-10670	444	2	plan	plan	VERB
dad-10670	444	3	to	to	PART
dad-10670	444	4	investigate	investigate	VERB
dad-10670	444	5	implicit	implicit	ADJ
dad-10670	444	6	discourse	discourse	NOUN
dad-10670	444	7	relation	relation	NOUN
dad-10670	444	8	classification	classification	NOUN
dad-10670	444	9	by	by	ADP
dad-10670	444	10	employing	employ	VERB
dad-10670	444	11	features	feature	NOUN
dad-10670	444	12	based	base	VERB
dad-10670	444	13	on	on	ADP
dad-10670	444	14	measures	measure	NOUN
dad-10670	444	15	of	of	ADP
dad-10670	444	16	semantic	semantic	ADJ
dad-10670	444	17	similarity	similarity	NOUN
dad-10670	444	18	between	between	ADP
dad-10670	444	19	edus	edus	NOUN
dad-10670	444	20	,	,	PUNCT
dad-10670	444	21	using	use	VERB
dad-10670	444	22	for	for	ADP
dad-10670	444	23	instance	instance	NOUN
dad-10670	444	24	wordnet	wordnet	NOUN
dad-10670	444	25	(	(	PUNCT
dad-10670	444	26	fellbaum	fellbaum	NOUN
dad-10670	444	27	,	,	PUNCT
dad-10670	444	28	1998	1998	NUM
dad-10670	444	29	)	)	PUNCT
dad-10670	444	30	.	.	PUNCT
dad-10670	445	1	finally	finally	ADV
dad-10670	445	2	,	,	PUNCT
dad-10670	445	3	we	we	PRON
dad-10670	445	4	plan	plan	VERB
dad-10670	445	5	to	to	PART
dad-10670	445	6	apply	apply	VERB
dad-10670	445	7	feature	feature	NOUN
dad-10670	445	8	selection	selection	NOUN
dad-10670	445	9	in	in	ADP
dad-10670	445	10	order	order	NOUN
dad-10670	445	11	to	to	PART
dad-10670	445	12	effectively	effectively	ADV
dad-10670	445	13	reduce	reduce	VERB
dad-10670	445	14	the	the	DET
dad-10670	445	15	size	size	NOUN
dad-10670	445	16	of	of	ADP
dad-10670	445	17	the	the	DET
dad-10670	445	18	feature	feature	NOUN
dad-10670	445	19	set	set	NOUN
dad-10670	445	20	.	.	PUNCT
dad-10670	446	1	this	this	DET
dad-10670	446	2	method	method	NOUN
dad-10670	446	3	has	have	VERB
dad-10670	446	4	the	the	DET
dad-10670	446	5	promise	promise	NOUN
dad-10670	446	6	of	of	ADP
dad-10670	446	7	shorter	short	ADJ
dad-10670	446	8	training	training	NOUN
dad-10670	446	9	and	and	CCONJ
dad-10670	446	10	testing	testing	NOUN
dad-10670	446	11	times	time	NOUN
dad-10670	446	12	.	.	PUNCT
dad-10670	447	1	another	another	DET
dad-10670	447	2	interesting	interesting	ADJ
dad-10670	447	3	aspect	aspect	NOUN
dad-10670	447	4	of	of	ADP
dad-10670	447	5	feature	feature	NOUN
dad-10670	447	6	selection	selection	NOUN
dad-10670	447	7	is	be	AUX
dad-10670	447	8	the	the	DET
dad-10670	447	9	potential	potential	ADJ
dad-10670	447	10	reduction	reduction	NOUN
dad-10670	447	11	of	of	ADP
dad-10670	447	12	noise	noise	NOUN
dad-10670	447	13	,	,	PUNCT
dad-10670	447	14	which	which	PRON
dad-10670	447	15	may	may	AUX
dad-10670	447	16	lead	lead	VERB
dad-10670	447	17	to	to	ADP
dad-10670	447	18	an	an	DET
dad-10670	447	19	improvement	improvement	NOUN
dad-10670	447	20	in	in	ADP
dad-10670	447	21	accuracy	accuracy	NOUN
dad-10670	447	22	.	.	PUNCT
dad-10670	448	1	5	5	X
dad-10670	448	2	.	.	X
dad-10670	448	3	conclusions	conclusion	NOUN
dad-10670	448	4	we	we	PRON
dad-10670	448	5	presented	present	VERB
dad-10670	448	6	hilda	hilda	PROPN
dad-10670	448	7	,	,	PUNCT
dad-10670	448	8	an	an	DET
dad-10670	448	9	automated	automate	VERB
dad-10670	448	10	discourse	discourse	NOUN
dad-10670	448	11	parser	parser	NOUN
dad-10670	448	12	that	that	PRON
dad-10670	448	13	analyzes	analyze	VERB
dad-10670	448	14	text	text	NOUN
dad-10670	448	15	in	in	ADP
dad-10670	448	16	the	the	DET
dad-10670	448	17	framework	framework	NOUN
dad-10670	448	18	of	of	ADP
dad-10670	448	19	the	the	DET
dad-10670	448	20	rhetorical	rhetorical	ADJ
dad-10670	448	21	structure	structure	NOUN
dad-10670	448	22	theory	theory	NOUN
dad-10670	448	23	.	.	PUNCT
dad-10670	449	1	the	the	DET
dad-10670	449	2	system	system	NOUN
dad-10670	449	3	is	be	AUX
dad-10670	449	4	composed	compose	VERB
dad-10670	449	5	of	of	ADP
dad-10670	449	6	two	two	NUM
dad-10670	449	7	modules	module	NOUN
dad-10670	449	8	,	,	PUNCT
dad-10670	449	9	(	(	PUNCT
dad-10670	449	10	1	1	X
dad-10670	449	11	)	)	PUNCT
dad-10670	449	12	a	a	DET
dad-10670	449	13	discourse	discourse	NOUN
dad-10670	449	14	segmenter	segmenter	NOUN
dad-10670	449	15	,	,	PUNCT
dad-10670	449	16	and	and	CCONJ
dad-10670	449	17	(	(	PUNCT
dad-10670	449	18	2	2	X
dad-10670	449	19	)	)	PUNCT
dad-10670	449	20	a	a	DET
dad-10670	449	21	relation	relation	NOUN
dad-10670	449	22	classifier	classifier	NOUN
dad-10670	449	23	and	and	CCONJ
dad-10670	449	24	tree	tree	NOUN
dad-10670	449	25	builder	builder	NOUN
dad-10670	449	26	.	.	PUNCT
dad-10670	450	1	both	both	DET
dad-10670	450	2	tasks	task	NOUN
dad-10670	450	3	are	be	AUX
dad-10670	450	4	based	base	VERB
dad-10670	450	5	on	on	ADP
dad-10670	450	6	support	support	NOUN
dad-10670	450	7	vector	vector	NOUN
dad-10670	450	8	machine	machine	NOUN
dad-10670	450	9	classification	classification	NOUN
dad-10670	450	10	.	.	PUNCT
dad-10670	451	1	the	the	DET
dad-10670	451	2	discourse	discourse	NOUN
dad-10670	451	3	segmenter	segmenter	NOUN
dad-10670	451	4	performs	perform	VERB
dad-10670	451	5	at	at	ADP
dad-10670	451	6	around	around	ADP
dad-10670	451	7	96	96	NUM
dad-10670	451	8	%	%	NOUN
dad-10670	451	9	of	of	ADP
dad-10670	451	10	the	the	DET
dad-10670	451	11	human	human	ADJ
dad-10670	451	12	level	level	NOUN
dad-10670	451	13	,	,	PUNCT
dad-10670	451	14	while	while	SCONJ
dad-10670	451	15	the	the	DET
dad-10670	451	16	relation	relation	NOUN
dad-10670	451	17	classification	classification	NOUN
dad-10670	451	18	and	and	CCONJ
dad-10670	451	19	tree	tree	NOUN
dad-10670	451	20	-	-	PUNCT
dad-10670	451	21	building	building	NOUN
dad-10670	451	22	module	module	NOUN
dad-10670	451	23	reaches	reach	VERB
dad-10670	451	24	84	84	NUM
dad-10670	451	25	%	%	NOUN
dad-10670	451	26	of	of	ADP
dad-10670	451	27	human	human	ADJ
dad-10670	451	28	level	level	NOUN
dad-10670	451	29	.	.	PUNCT
dad-10670	452	1	our	our	PRON
dad-10670	452	2	system	system	NOUN
dad-10670	452	3	combines	combine	VERB
dad-10670	452	4	both	both	DET
dad-10670	452	5	modules	module	NOUN
dad-10670	452	6	for	for	ADP
dad-10670	452	7	creating	create	VERB
dad-10670	452	8	a	a	DET
dad-10670	452	9	fully	fully	ADV
dad-10670	452	10	automatic	automatic	ADJ
dad-10670	452	11	text	text	NOUN
dad-10670	452	12	level	level	NOUN
dad-10670	452	13	parser	parser	NOUN
dad-10670	452	14	,	,	PUNCT
dad-10670	452	15	reaching	reach	VERB
dad-10670	452	16	78.3	78.3	NUM
dad-10670	452	17	%	%	NOUN
dad-10670	452	18	of	of	ADP
dad-10670	452	19	the	the	DET
dad-10670	452	20	human	human	ADJ
dad-10670	452	21	performance	performance	NOUN
dad-10670	452	22	level	level	NOUN
dad-10670	452	23	.	.	PUNCT
dad-10670	453	1	importantly	importantly	ADV
dad-10670	453	2	,	,	PUNCT
dad-10670	453	3	the	the	DET
dad-10670	453	4	tree	tree	NOUN
dad-10670	453	5	-	-	PUNCT
dad-10670	453	6	building	building	NOUN
dad-10670	453	7	process	process	NOUN
dad-10670	453	8	performs	perform	VERB
dad-10670	453	9	in	in	ADP
dad-10670	453	10	linear	linear	ADJ
dad-10670	453	11	time	time	NOUN
dad-10670	453	12	,	,	PUNCT
dad-10670	453	13	which	which	PRON
dad-10670	453	14	allows	allow	VERB
dad-10670	453	15	us	we	PRON
dad-10670	453	16	to	to	PART
dad-10670	453	17	use	use	VERB
dad-10670	453	18	the	the	DET
dad-10670	453	19	system	system	NOUN
dad-10670	453	20	in	in	ADP
dad-10670	453	21	the	the	DET
dad-10670	453	22	context	context	NOUN
dad-10670	453	23	of	of	ADP
dad-10670	453	24	a	a	DET
dad-10670	453	25	(	(	PUNCT
dad-10670	453	26	near	near	ADJ
dad-10670	453	27	)	)	PUNCT
dad-10670	453	28	real	real	ADJ
dad-10670	453	29	-	-	PUNCT
dad-10670	453	30	time	time	NOUN
dad-10670	453	31	or	or	CCONJ
dad-10670	453	32	interactive	interactive	ADJ
dad-10670	453	33	system	system	NOUN
dad-10670	453	34	,	,	PUNCT
dad-10670	453	35	such	such	ADJ
dad-10670	453	36	as	as	ADP
dad-10670	453	37	dialogue	dialogue	NOUN
dad-10670	453	38	generation	generation	NOUN
dad-10670	453	39	or	or	CCONJ
dad-10670	453	40	questionanswering	questionanswering	NOUN
dad-10670	453	41	.	.	PUNCT
dad-10670	454	1	for	for	ADP
dad-10670	454	2	instance	instance	NOUN
dad-10670	454	3	,	,	PUNCT
dad-10670	454	4	the	the	DET
dad-10670	454	5	proposed	propose	VERB
dad-10670	454	6	parser	parser	NOUN
dad-10670	454	7	can	can	AUX
dad-10670	454	8	be	be	AUX
dad-10670	454	9	integrated	integrate	VERB
dad-10670	454	10	in	in	ADP
dad-10670	454	11	a	a	DET
dad-10670	454	12	dialogue	dialogue	NOUN
dad-10670	454	13	generation	generation	NOUN
dad-10670	454	14	system	system	NOUN
dad-10670	454	15	such	such	ADJ
dad-10670	454	16	as	as	ADP
dad-10670	454	17	text	text	NOUN
dad-10670	454	18	-	-	PUNCT
dad-10670	454	19	to	to	ADP
dad-10670	454	20	-	-	PUNCT
dad-10670	454	21	dialogue	dialogue	NOUN
dad-10670	454	22	(	(	PUNCT
dad-10670	454	23	prendinger	prendinger	NOUN
dad-10670	454	24	et	et	PROPN
dad-10670	454	25	al	al	PROPN
dad-10670	454	26	.	.	PROPN
dad-10670	454	27	,	,	PUNCT
dad-10670	454	28	2007	2007	NUM
dad-10670	454	29	;	;	PUNCT
dad-10670	454	30	hernault	hernault	PROPN
dad-10670	454	31	et	et	PROPN
dad-10670	454	32	al	al	PROPN
dad-10670	454	33	.	.	PROPN
dad-10670	454	34	,	,	PUNCT
dad-10670	454	35	2008	2008	NUM
dad-10670	454	36	)	)	PUNCT
dad-10670	454	37	.	.	PUNCT
dad-10670	455	1	in	in	ADP
dad-10670	455	2	this	this	DET
dad-10670	455	3	system	system	NOUN
dad-10670	455	4	,	,	PUNCT
dad-10670	455	5	rst	rst	PROPN
dad-10670	455	6	structures	structure	NOUN
dad-10670	455	7	are	be	AUX
dad-10670	455	8	used	use	VERB
dad-10670	455	9	as	as	ADP
dad-10670	455	10	the	the	DET
dad-10670	455	11	backbone	backbone	NOUN
dad-10670	455	12	for	for	ADP
dad-10670	455	13	generating	generate	VERB
dad-10670	455	14	dialogues	dialogue	NOUN
dad-10670	455	15	between	between	ADP
dad-10670	455	16	two	two	NUM
dad-10670	455	17	agents	agent	NOUN
dad-10670	455	18	representing	represent	VERB
dad-10670	455	19	a	a	DET
dad-10670	455	20	layperson	layperson	NOUN
dad-10670	455	21	and	and	CCONJ
dad-10670	455	22	an	an	DET
dad-10670	455	23	expert	expert	NOUN
dad-10670	455	24	.	.	PUNCT
dad-10670	456	1	question	question	NOUN
dad-10670	456	2	-	-	PUNCT
dad-10670	456	3	answer	answer	NOUN
dad-10670	456	4	pairs	pair	NOUN
dad-10670	456	5	between	between	ADP
dad-10670	456	6	the	the	DET
dad-10670	456	7	two	two	NUM
dad-10670	456	8	agents	agent	NOUN
dad-10670	456	9	are	be	AUX
dad-10670	456	10	created	create	VERB
dad-10670	456	11	by	by	ADP
dad-10670	456	12	applying	apply	VERB
dad-10670	456	13	relation	relation	NOUN
dad-10670	456	14	-	-	PUNCT
dad-10670	456	15	specific	specific	ADJ
dad-10670	456	16	mapping	mapping	NOUN
dad-10670	456	17	rules	rule	NOUN
dad-10670	456	18	manipulating	manipulate	VERB
dad-10670	456	19	the	the	DET
dad-10670	456	20	leaves	leave	NOUN
dad-10670	456	21	of	of	ADP
dad-10670	456	22	the	the	DET
dad-10670	456	23	rst	rst	PROPN
dad-10670	456	24	tree	tree	NOUN
dad-10670	456	25	.	.	PUNCT
dad-10670	457	1	if	if	SCONJ
dad-10670	457	2	the	the	DET
dad-10670	457	3	rst	rst	PROPN
dad-10670	457	4	tree	tree	NOUN
dad-10670	457	5	utilized	utilize	VERB
dad-10670	457	6	contains	contain	VERB
dad-10670	457	7	errors	error	NOUN
dad-10670	457	8	,	,	PUNCT
dad-10670	457	9	such	such	ADJ
dad-10670	457	10	as	as	ADP
dad-10670	457	11	improperly	improperly	ADV
dad-10670	457	12	-	-	PUNCT
dad-10670	457	13	segmented	segment	VERB
dad-10670	457	14	edus	edus	NOUN
dad-10670	457	15	or	or	CCONJ
dad-10670	457	16	erroneous	erroneous	ADJ
dad-10670	457	17	discourse	discourse	NOUN
dad-10670	457	18	relations	relation	NOUN
dad-10670	457	19	,	,	PUNCT
dad-10670	457	20	the	the	DET
dad-10670	457	21	generated	generate	VERB
dad-10670	457	22	dialogue	dialogue	NOUN
dad-10670	457	23	will	will	AUX
dad-10670	457	24	be	be	AUX
dad-10670	457	25	semantically	semantically	ADV
dad-10670	457	26	incorrect	incorrect	ADJ
dad-10670	457	27	.	.	PUNCT
dad-10670	458	1	hence	hence	ADV
dad-10670	458	2	,	,	PUNCT
dad-10670	458	3	a	a	DET
dad-10670	458	4	sound	sound	ADJ
dad-10670	458	5	discourse	discourse	NOUN
dad-10670	458	6	parser	parser	NOUN
dad-10670	458	7	is	be	AUX
dad-10670	458	8	important	important	ADJ
dad-10670	458	9	.	.	PUNCT
dad-10670	459	1	more	more	ADV
dad-10670	459	2	recently	recently	ADV
dad-10670	459	3	,	,	PUNCT
dad-10670	459	4	the	the	DET
dad-10670	459	5	coda	coda	ADJ
dad-10670	459	6	corpus	corpus	NOUN
dad-10670	459	7	(	(	PUNCT
dad-10670	459	8	stoyanchev	stoyanchev	NOUN
dad-10670	459	9	and	and	CCONJ
dad-10670	459	10	piwek	piwek	PROPN
dad-10670	459	11	,	,	PUNCT
dad-10670	459	12	2010	2010	NUM
dad-10670	459	13	)	)	PUNCT
dad-10670	459	14	was	be	AUX
dad-10670	459	15	released	release	VERB
dad-10670	459	16	.	.	PUNCT
dad-10670	460	1	this	this	DET
dad-10670	460	2	corpus	corpus	NOUN
dad-10670	460	3	contains	contain	VERB
dad-10670	460	4	700	700	NUM
dad-10670	460	5	expository	expository	ADJ
dad-10670	460	6	dialogues	dialogue	NOUN
dad-10670	460	7	labeled	label	VERB
dad-10670	460	8	with	with	ADP
dad-10670	460	9	dialogue	dialogue	NOUN
dad-10670	460	10	acts	act	NOUN
dad-10670	460	11	,	,	PUNCT
dad-10670	460	12	paired	pair	VERB
dad-10670	460	13	with	with	ADP
dad-10670	460	14	human	human	NOUN
dad-10670	460	15	-	-	PUNCT
dad-10670	460	16	authored	author	VERB
dad-10670	460	17	,	,	PUNCT
dad-10670	460	18	equivalent	equivalent	ADJ
dad-10670	460	19	monologues	monologue	NOUN
dad-10670	460	20	annotated	annotate	VERB
dad-10670	460	21	with	with	ADP
dad-10670	460	22	rst	rst	PROPN
dad-10670	460	23	discourse	discourse	NOUN
dad-10670	460	24	relations	relation	NOUN
dad-10670	460	25	.	.	PUNCT
dad-10670	461	1	using	use	VERB
dad-10670	461	2	this	this	DET
dad-10670	461	3	corpus	corpus	NOUN
dad-10670	461	4	,	,	PUNCT
dad-10670	461	5	one	one	PRON
dad-10670	461	6	can	can	AUX
dad-10670	461	7	learn	learn	VERB
dad-10670	461	8	how	how	SCONJ
dad-10670	461	9	to	to	PART
dad-10670	461	10	generate	generate	VERB
dad-10670	461	11	an	an	DET
dad-10670	461	12	expository	expository	ADJ
dad-10670	461	13	dialogue	dialogue	NOUN
dad-10670	461	14	from	from	ADP
dad-10670	461	15	an	an	DET
dad-10670	461	16	rst	rst	ADJ
dad-10670	461	17	-	-	PUNCT
dad-10670	461	18	annotated	annotated	ADJ
dad-10670	461	19	monologue	monologue	NOUN
dad-10670	461	20	.	.	PUNCT
dad-10670	462	1	an	an	DET
dad-10670	462	2	ambitious	ambitious	ADJ
dad-10670	462	3	next	next	ADJ
dad-10670	462	4	step	step	NOUN
dad-10670	462	5	is	be	AUX
dad-10670	462	6	to	to	PART
dad-10670	462	7	automatize	automatize	VERB
dad-10670	462	8	the	the	DET
dad-10670	462	9	generation	generation	NOUN
dad-10670	462	10	process	process	NOUN
dad-10670	462	11	,	,	PUNCT
dad-10670	462	12	i.e.	i.e.	X
dad-10670	462	13	creating	create	VERB
dad-10670	462	14	a	a	DET
dad-10670	462	15	meaningful	meaningful	ADJ
dad-10670	462	16	expository	expository	ADJ
dad-10670	462	17	dialogue	dialogue	NOUN
dad-10670	462	18	given	give	VERB
dad-10670	462	19	any	any	DET
dad-10670	462	20	plain	plain	ADJ
dad-10670	462	21	,	,	PUNCT
dad-10670	462	22	non	non	ADJ
dad-10670	462	23	-	-	ADJ
dad-10670	462	24	annotated	annotated	ADJ
dad-10670	462	25	monologue	monologue	NOUN
dad-10670	462	26	.	.	PUNCT
dad-10670	463	1	this	this	PRON
dad-10670	463	2	is	be	AUX
dad-10670	463	3	a	a	DET
dad-10670	463	4	promising	promising	ADJ
dad-10670	463	5	perspective	perspective	NOUN
dad-10670	463	6	for	for	ADP
dad-10670	463	7	creating	create	VERB
dad-10670	463	8	systems	system	NOUN
dad-10670	463	9	that	that	PRON
dad-10670	463	10	are	be	AUX
dad-10670	463	11	able	able	ADJ
dad-10670	463	12	to	to	PART
dad-10670	463	13	present	present	VERB
dad-10670	463	14	information	information	NOUN
dad-10670	463	15	automatically	automatically	ADV
dad-10670	463	16	,	,	PUNCT
dad-10670	463	17	such	such	ADJ
dad-10670	463	18	as	as	ADP
dad-10670	463	19	tutoring	tutoring	NOUN
dad-10670	463	20	applications	application	NOUN
dad-10670	463	21	.	.	PUNCT
dad-10670	464	1	a	a	DET
dad-10670	464	2	prerequisite	prerequisite	NOUN
dad-10670	464	3	is	be	AUX
dad-10670	464	4	the	the	DET
dad-10670	464	5	accurate	accurate	ADJ
dad-10670	464	6	detection	detection	NOUN
dad-10670	464	7	of	of	ADP
dad-10670	464	8	the	the	DET
dad-10670	464	9	discourse	discourse	NOUN
dad-10670	464	10	relations	relation	NOUN
dad-10670	464	11	in	in	ADP
dad-10670	464	12	the	the	DET
dad-10670	464	13	input	input	NOUN
dad-10670	464	14	monologues	monologue	NOUN
dad-10670	464	15	,	,	PUNCT
dad-10670	464	16	and	and	CCONJ
dad-10670	464	17	hence	hence	ADV
dad-10670	464	18	,	,	PUNCT
dad-10670	464	19	the	the	DET
dad-10670	464	20	role	role	NOUN
dad-10670	464	21	of	of	ADP
dad-10670	464	22	the	the	DET
dad-10670	464	23	discourse	discourse	NOUN
dad-10670	464	24	parser	parser	NOUN
dad-10670	464	25	is	be	AUX
dad-10670	464	26	once	once	ADV
dad-10670	464	27	again	again	ADV
dad-10670	464	28	central	central	ADJ
dad-10670	464	29	.	.	PUNCT
dad-10670	465	1	with	with	ADP
dad-10670	465	2	these	these	DET
dad-10670	465	3	applications	application	NOUN
dad-10670	465	4	in	in	ADP
dad-10670	465	5	mind	mind	NOUN
dad-10670	465	6	,	,	PUNCT
dad-10670	465	7	we	we	PRON
dad-10670	465	8	plan	plan	VERB
dad-10670	465	9	to	to	PART
dad-10670	465	10	make	make	VERB
dad-10670	465	11	the	the	DET
dad-10670	465	12	proposed	propose	VERB
dad-10670	465	13	parser	parser	NOUN
dad-10670	465	14	more	more	ADV
dad-10670	465	15	efficient	efficient	ADJ
dad-10670	465	16	,	,	PUNCT
dad-10670	465	17	by	by	ADP
dad-10670	465	18	addressing	address	VERB
dad-10670	465	19	the	the	DET
dad-10670	465	20	issues	issue	NOUN
dad-10670	465	21	presented	present	VERB
dad-10670	465	22	in	in	ADP
dad-10670	465	23	section	section	NOUN
dad-10670	465	24	4.6	4.6	NUM
dad-10670	465	25	.	.	PUNCT
dad-10670	466	1	in	in	ADP
dad-10670	466	2	particular	particular	ADJ
dad-10670	466	3	,	,	PUNCT
dad-10670	466	4	we	we	PRON
dad-10670	466	5	intend	intend	VERB
dad-10670	466	6	to	to	PART
dad-10670	466	7	improve	improve	VERB
dad-10670	466	8	the	the	DET
dad-10670	466	9	tree	tree	NOUN
dad-10670	466	10	-	-	PUNCT
dad-10670	466	11	building	building	NOUN
dad-10670	466	12	method	method	NOUN
dad-10670	466	13	,	,	PUNCT
dad-10670	466	14	in	in	ADP
dad-10670	466	15	order	order	NOUN
dad-10670	466	16	to	to	PART
dad-10670	466	17	create	create	VERB
dad-10670	466	18	globally	globally	ADV
dad-10670	466	19	optimal	optimal	ADJ
dad-10670	466	20	discourse	discourse	NOUN
dad-10670	466	21	structures	structure	NOUN
dad-10670	466	22	.	.	PUNCT
dad-10670	467	1	finally	finally	ADV
dad-10670	467	2	,	,	PUNCT
dad-10670	467	3	we	we	PRON
dad-10670	467	4	plan	plan	VERB
dad-10670	467	5	to	to	PART
dad-10670	467	6	train	train	VERB
dad-10670	467	7	a	a	DET
dad-10670	467	8	separate	separate	ADJ
dad-10670	467	9	classifier	classifier	NOUN
dad-10670	467	10	for	for	ADP
dad-10670	467	11	detecting	detect	VERB
dad-10670	467	12	implicit	implicit	ADJ
dad-10670	467	13	discourse	discourse	NOUN
dad-10670	467	14	relations	relation	NOUN
dad-10670	467	15	as	as	ADV
dad-10670	467	16	well	well	ADV
dad-10670	467	17	.	.	PUNCT
dad-10670	468	1	24	24	NUM
dad-10670	468	2	hilda	hilda	VERB
dad-10670	468	3	:	:	PUNCT
dad-10670	468	4	a	a	DET
dad-10670	468	5	discourse	discourse	NOUN
dad-10670	468	6	parser	parser	NOUN
dad-10670	468	7	using	use	VERB
dad-10670	468	8	support	support	NOUN
dad-10670	468	9	vector	vector	NOUN
dad-10670	468	10	machine	machine	NOUN
dad-10670	468	11	classification	classification	NOUN
dad-10670	468	12	appendix	appendix	NOUN
dad-10670	468	13	a.	a.	NOUN
dad-10670	468	14	system	system	NOUN
dad-10670	468	15	workflow	workflow	NOUN
dad-10670	468	16	test	test	NOUN
dad-10670	468	17	document	document	NOUN
dad-10670	468	18	syntax	syntax	NOUN
dad-10670	468	19	parsing	parsing	NOUN
dad-10670	468	20	(	(	PUNCT
dad-10670	468	21	stanford	stanford	PROPN
dad-10670	468	22	parser	parser	PROPN
dad-10670	468	23	)	)	PUNCT
dad-10670	468	24	syntax	syntax	NOUN
dad-10670	468	25	trees	tree	NOUN
dad-10670	468	26	lexicalization	lexicalization	NOUN
dad-10670	468	27	lexicalized	lexicalize	VERB
dad-10670	468	28	syntax	syntax	NOUN
dad-10670	468	29	trees	tree	NOUN
dad-10670	468	30	segmentation	segmentation	NOUN
dad-10670	468	31	feature	feature	NOUN
dad-10670	468	32	extraction	extraction	NOUN
dad-10670	468	33	svm	svm	ADJ
dad-10670	468	34	classification	classification	NOUN
dad-10670	468	35	tokenized	tokenize	VERB
dad-10670	468	36	edus	edus	NOUN
dad-10670	468	37	svm	svm	PROPN
dad-10670	468	38	classification	classification	NOUN
dad-10670	468	39	relation	relation	NOUN
dad-10670	468	40	labeling	labeling	NOUN
dad-10670	468	41	feature	feature	NOUN
dad-10670	468	42	extraction	extraction	NOUN
dad-10670	468	43	scored	score	VERB
dad-10670	468	44	rs	rs	ADJ
dad-10670	468	45	subtrees	subtree	NOUN
dad-10670	468	46	rhetorical	rhetorical	ADJ
dad-10670	468	47	structure	structure	NOUN
dad-10670	468	48	tree	tree	NOUN
dad-10670	468	49	bottom	bottom	ADJ
dad-10670	468	50	-	-	PUNCT
dad-10670	468	51	up	up	ADP
dad-10670	468	52	tree	tree	NOUN
dad-10670	468	53	construction	construction	NOUN
dad-10670	468	54	rst	rst	PROPN
dad-10670	468	55	discourse	discourse	NOUN
dad-10670	468	56	treebank	treebank	PROPN
dad-10670	468	57	penn	penn	PROPN
dad-10670	468	58	treebank	treebank	PROPN
dad-10670	468	59	edus	edus	NOUN
dad-10670	468	60	lexicalization	lexicalization	NOUN
dad-10670	468	61	lexicalized	lexicalize	VERB
dad-10670	468	62	syntax	syntax	NOUN
dad-10670	468	63	trees	tree	NOUN
dad-10670	468	64	syntax	syntax	NOUN
dad-10670	468	65	trees	tree	NOUN
dad-10670	468	66	alignment	alignment	NOUN
dad-10670	468	67	segmentation	segmentation	NOUN
dad-10670	468	68	feature	feature	NOUN
dad-10670	468	69	extraction	extraction	NOUN
dad-10670	468	70	svm	svm	ADJ
dad-10670	468	71	training	training	NOUN
dad-10670	468	72	tokenization	tokenization	NOUN
dad-10670	468	73	tokenized	tokenize	VERB
dad-10670	468	74	edus	edus	NOUN
dad-10670	468	75	relation	relation	NOUN
dad-10670	468	76	labeling	labeling	NOUN
dad-10670	468	77	feature	feature	NOUN
dad-10670	468	78	extraction	extraction	NOUN
dad-10670	468	79	svm	svm	ADJ
dad-10670	468	80	training	training	NOUN
dad-10670	468	81	segmentation	segmentation	NOUN
dad-10670	468	82	modelrelation	modelrelation	NOUN
dad-10670	468	83	labeling	labeling	NOUN
dad-10670	468	84	model	model	NOUN
dad-10670	468	85	alignment	alignment	NOUN
dad-10670	468	86	25	25	NUM
dad-10670	468	87	hernault	hernault	ADJ
dad-10670	468	88	,	,	PUNCT
dad-10670	468	89	prendinger	prendinger	NOUN
dad-10670	468	90	,	,	PUNCT
dad-10670	468	91	duverle	duverle	NOUN
dad-10670	468	92	,	,	PUNCT
dad-10670	468	93	and	and	CCONJ
dad-10670	468	94	ishizuka	ishizuka	PROPN
dad-10670	468	95	appendix	appendix	PROPN
dad-10670	468	96	b.	b.	PROPN
dad-10670	468	97	relation	relation	PROPN
dad-10670	468	98	list	list	NOUN
dad-10670	468	99	relation	relation	NOUN
dad-10670	468	100	nuclearity	nuclearity	NOUN
dad-10670	468	101	options	option	NOUN
dad-10670	468	102	attribution	attribution	NOUN
dad-10670	468	103	nucleus	nucleus	PROPN
dad-10670	468	104	satellite	satellite	PROPN
dad-10670	468	105	satellite	satellite	PROPN
dad-10670	468	106	nucleus	nucleus	PROPN
dad-10670	468	107	background	background	NOUN
dad-10670	468	108	nucleus	nucleus	PROPN
dad-10670	468	109	satellite	satellite	PROPN
dad-10670	468	110	satellite	satellite	PROPN
dad-10670	468	111	nucleus	nucleus	PROPN
dad-10670	468	112	cause	cause	SCONJ
dad-10670	468	113	nucleus	nucleus	PROPN
dad-10670	468	114	nucleus	nucleus	PROPN
dad-10670	468	115	nucleus	nucleus	PROPN
dad-10670	468	116	satellite	satellite	PROPN
dad-10670	468	117	satellite	satellite	PROPN
dad-10670	468	118	nucleus	nucleus	PROPN
dad-10670	468	119	comparison	comparison	NOUN
dad-10670	468	120	nucleus	nucleus	PROPN
dad-10670	468	121	nucleus	nucleus	PROPN
dad-10670	468	122	nucleus	nucleus	PROPN
dad-10670	468	123	satellite	satellite	PROPN
dad-10670	468	124	satellite	satellite	PROPN
dad-10670	468	125	nucleus	nucleus	PROPN
dad-10670	468	126	condition	condition	NOUN
dad-10670	468	127	nucleus	nucleus	PROPN
dad-10670	468	128	nucleus	nucleus	PROPN
dad-10670	468	129	nucleus	nucleus	PROPN
dad-10670	468	130	satellite	satellite	PROPN
dad-10670	468	131	satellite	satellite	PROPN
dad-10670	468	132	nucleus	nucleus	PROPN
dad-10670	468	133	contrast	contrast	NOUN
dad-10670	468	134	nucleus	nucleus	PROPN
dad-10670	468	135	nucleus	nucleus	PROPN
dad-10670	468	136	nucleus	nucleus	PROPN
dad-10670	468	137	satellite	satellite	PROPN
dad-10670	468	138	satellite	satellite	PROPN
dad-10670	468	139	nucleus	nucleus	PROPN
dad-10670	468	140	elaboration	elaboration	NOUN
dad-10670	468	141	nucleus	nucleus	PROPN
dad-10670	468	142	satellite	satellite	PROPN
dad-10670	468	143	satellite	satellite	PROPN
dad-10670	468	144	nucleus	nucleus	PROPN
dad-10670	468	145	enablement	enablement	PROPN
dad-10670	468	146	nucleus	nucleus	PROPN
dad-10670	468	147	satellite	satellite	PROPN
dad-10670	468	148	satellite	satellite	PROPN
dad-10670	468	149	nucleus	nucleus	PROPN
dad-10670	468	150	evaluation	evaluation	NOUN
dad-10670	468	151	nucleus	nucleus	PROPN
dad-10670	468	152	nucleus	nucleus	PROPN
dad-10670	468	153	nucleus	nucleus	PROPN
dad-10670	468	154	satellite	satellite	PROPN
dad-10670	468	155	satellite	satellite	PROPN
dad-10670	468	156	nucleus	nucleus	PROPN
dad-10670	468	157	explanation	explanation	PROPN
dad-10670	468	158	nucleus	nucleus	PROPN
dad-10670	468	159	nucleus	nucleus	PROPN
dad-10670	468	160	nucleus	nucleus	PROPN
dad-10670	468	161	satellite	satellite	PROPN
dad-10670	468	162	satellite	satellite	PROPN
dad-10670	468	163	nucleus	nucleus	PROPN
dad-10670	468	164	joint	joint	PROPN
dad-10670	468	165	nucleus	nucleus	PROPN
dad-10670	468	166	nucleus	nucleus	PROPN
dad-10670	468	167	manner	manner	NOUN
dad-10670	468	168	-	-	PUNCT
dad-10670	468	169	means	mean	VERB
dad-10670	468	170	nucleus	nucleus	PROPN
dad-10670	468	171	satellite	satellite	PROPN
dad-10670	468	172	satellite	satellite	PROPN
dad-10670	468	173	nucleus	nucleus	PROPN
dad-10670	468	174	summary	summary	PROPN
dad-10670	468	175	nucleus	nucleus	PROPN
dad-10670	468	176	satellite	satellite	PROPN
dad-10670	468	177	satellite	satellite	PROPN
dad-10670	468	178	nucleus	nucleus	PROPN
dad-10670	468	179	temporal	temporal	ADJ
dad-10670	468	180	nucleus	nucleus	PROPN
dad-10670	468	181	nucleus	nucleus	PROPN
dad-10670	468	182	nucleus	nucleus	PROPN
dad-10670	468	183	satellite	satellite	PROPN
dad-10670	468	184	satellite	satellite	PROPN
dad-10670	468	185	nucleus	nucleus	PROPN
dad-10670	468	186	topic	topic	NOUN
dad-10670	468	187	-	-	PUNCT
dad-10670	468	188	change	change	NOUN
dad-10670	468	189	nucleus	nucleus	ADJ
dad-10670	468	190	nucleus	nucleus	PROPN
dad-10670	468	191	nucleus	nucleus	PROPN
dad-10670	468	192	satellite	satellite	PROPN
dad-10670	468	193	topic	topic	NOUN
dad-10670	468	194	-	-	PUNCT
dad-10670	468	195	comment	comment	NOUN
dad-10670	468	196	nucleus	nucleus	PROPN
dad-10670	468	197	nucleus	nucleus	PROPN
dad-10670	468	198	nucleus	nucleus	PROPN
dad-10670	468	199	satellite	satellite	PROPN
dad-10670	468	200	satellite	satellite	PROPN
dad-10670	468	201	nucleus	nucleus	PROPN
dad-10670	468	202	same	same	ADJ
dad-10670	468	203	-	-	PUNCT
dad-10670	468	204	unit	unit	NOUN
dad-10670	468	205	nucleus	nucleus	PROPN
dad-10670	468	206	nucleus	nucleus	PROPN
dad-10670	468	207	textual	textual	ADJ
dad-10670	468	208	-	-	PUNCT
dad-10670	468	209	organization	organization	NOUN
dad-10670	468	210	nucleus	nucleus	ADJ
dad-10670	468	211	nucleus	nucleus	PROPN
dad-10670	468	212	26	26	NUM
dad-10670	468	213	hilda	hilda	VERB
dad-10670	468	214	:	:	PUNCT
dad-10670	468	215	a	a	DET
dad-10670	468	216	discourse	discourse	NOUN
dad-10670	468	217	parser	parser	NOUN
dad-10670	468	218	using	use	VERB
dad-10670	468	219	support	support	NOUN
dad-10670	468	220	vector	vector	NOUN
dad-10670	468	221	machine	machine	NOUN
dad-10670	468	222	classification	classification	NOUN
dad-10670	468	223	references	reference	NOUN
dad-10670	468	224	nicholas	nicholas	PROPN
dad-10670	468	225	asher	asher	PROPN
dad-10670	468	226	and	and	CCONJ
dad-10670	468	227	alex	alex	PROPN
dad-10670	468	228	lascarides	lascarides	PROPN
dad-10670	468	229	.	.	PUNCT
dad-10670	469	1	logics	logic	NOUN
dad-10670	469	2	of	of	ADP
dad-10670	469	3	conversation	conversation	NOUN
dad-10670	469	4	.	.	PUNCT
dad-10670	470	1	cambridge	cambridge	PROPN
dad-10670	470	2	university	university	PROPN
dad-10670	470	3	press	press	NOUN
dad-10670	470	4	,	,	PUNCT
dad-10670	470	5	2003	2003	NUM
dad-10670	470	6	.	.	PUNCT
dad-10670	471	1	jason	jason	PROPN
dad-10670	471	2	baldridge	baldridge	PROPN
dad-10670	471	3	and	and	CCONJ
dad-10670	471	4	alex	alex	PROPN
dad-10670	471	5	lascarides	lascarides	PROPN
dad-10670	471	6	.	.	PUNCT
dad-10670	472	1	probabilistic	probabilistic	ADJ
dad-10670	472	2	head	head	NOUN
dad-10670	472	3	-	-	PUNCT
dad-10670	472	4	driven	drive	VERB
dad-10670	472	5	parsing	parsing	NOUN
dad-10670	472	6	for	for	ADP
dad-10670	472	7	discourse	discourse	NOUN
dad-10670	472	8	structure	structure	NOUN
dad-10670	472	9	.	.	PUNCT
dad-10670	473	1	in	in	ADP
dad-10670	473	2	proceedings	proceeding	NOUN
dad-10670	473	3	of	of	ADP
dad-10670	473	4	the	the	DET
dad-10670	473	5	ninth	ninth	ADJ
dad-10670	473	6	conference	conference	NOUN
dad-10670	473	7	on	on	ADP
dad-10670	473	8	computational	computational	ADJ
dad-10670	473	9	natural	natural	ADJ
dad-10670	473	10	language	language	NOUN
dad-10670	473	11	learning	learning	NOUN
dad-10670	473	12	(	(	PUNCT
dad-10670	473	13	conll2005	conll2005	PROPN
dad-10670	473	14	)	)	PUNCT
dad-10670	473	15	,	,	PUNCT
dad-10670	473	16	pages	page	NOUN
dad-10670	473	17	96–103	96–103	NUM
dad-10670	473	18	,	,	PUNCT
dad-10670	473	19	ann	ann	PROPN
dad-10670	473	20	arbor	arbor	PROPN
dad-10670	473	21	,	,	PUNCT
dad-10670	473	22	michigan	michigan	PROPN
dad-10670	473	23	,	,	PUNCT
dad-10670	473	24	june	june	PROPN
dad-10670	473	25	2005	2005	NUM
dad-10670	473	26	.	.	PUNCT
dad-10670	474	1	association	association	NOUN
dad-10670	474	2	for	for	ADP
dad-10670	474	3	computational	computational	ADJ
dad-10670	474	4	linguistics	linguistic	NOUN
dad-10670	474	5	.	.	PUNCT
dad-10670	475	1	ezra	ezra	PROPN
dad-10670	475	2	black	black	PROPN
dad-10670	475	3	,	,	PUNCT
dad-10670	475	4	steven	steven	PROPN
dad-10670	475	5	p.	p.	PROPN
dad-10670	475	6	abney	abney	PROPN
dad-10670	475	7	,	,	PUNCT
dad-10670	475	8	d.	d.	PROPN
dad-10670	475	9	flickenger	flickenger	PROPN
dad-10670	475	10	,	,	PUNCT
dad-10670	475	11	claudia	claudia	PROPN
dad-10670	475	12	gdaniec	gdaniec	PROPN
dad-10670	475	13	,	,	PUNCT
dad-10670	475	14	ralph	ralph	PROPN
dad-10670	475	15	grishman	grishman	PROPN
dad-10670	475	16	,	,	PUNCT
dad-10670	475	17	p.	p.	PROPN
dad-10670	475	18	harrison	harrison	PROPN
dad-10670	475	19	,	,	PUNCT
dad-10670	475	20	donald	donald	PROPN
dad-10670	475	21	hindle	hindle	PROPN
dad-10670	475	22	,	,	PUNCT
dad-10670	475	23	robert	robert	PROPN
dad-10670	475	24	ingria	ingria	PROPN
dad-10670	475	25	,	,	PUNCT
dad-10670	475	26	frederick	frederick	PROPN
dad-10670	475	27	jelinek	jelinek	PROPN
dad-10670	475	28	,	,	PUNCT
dad-10670	475	29	judith	judith	PROPN
dad-10670	475	30	l.	l.	PROPN
dad-10670	475	31	klavans	klavans	PROPN
dad-10670	475	32	,	,	PUNCT
dad-10670	475	33	mark	mark	PROPN
dad-10670	475	34	liberman	liberman	PROPN
dad-10670	475	35	,	,	PUNCT
dad-10670	475	36	mitchell	mitchell	PROPN
dad-10670	475	37	p.	p.	PROPN
dad-10670	475	38	marcus	marcus	PROPN
dad-10670	475	39	,	,	PUNCT
dad-10670	475	40	salim	salim	PROPN
dad-10670	475	41	roukos	roukos	PROPN
dad-10670	475	42	,	,	PUNCT
dad-10670	475	43	beatrice	beatrice	PROPN
dad-10670	475	44	santorini	santorini	PROPN
dad-10670	475	45	,	,	PUNCT
dad-10670	475	46	and	and	CCONJ
dad-10670	475	47	tomek	tomek	PROPN
dad-10670	475	48	strzalkowski	strzalkowski	PROPN
dad-10670	475	49	.	.	PUNCT
dad-10670	476	1	a	a	DET
dad-10670	476	2	procedure	procedure	NOUN
dad-10670	476	3	for	for	ADP
dad-10670	476	4	quantitatively	quantitatively	ADV
dad-10670	476	5	comparing	compare	VERB
dad-10670	476	6	the	the	DET
dad-10670	476	7	syntactic	syntactic	ADJ
dad-10670	476	8	coverage	coverage	NOUN
dad-10670	476	9	of	of	ADP
dad-10670	476	10	english	english	ADJ
dad-10670	476	11	grammars	grammar	NOUN
dad-10670	476	12	.	.	PUNCT
dad-10670	477	1	in	in	ADP
dad-10670	477	2	proceedings	proceeding	NOUN
dad-10670	477	3	of	of	ADP
dad-10670	477	4	workshop	workshop	NOUN
dad-10670	477	5	on	on	ADP
dad-10670	477	6	speech	speech	NOUN
dad-10670	477	7	and	and	CCONJ
dad-10670	477	8	natural	natural	ADJ
dad-10670	477	9	language	language	NOUN
dad-10670	477	10	,	,	PUNCT
dad-10670	477	11	pages	page	VERB
dad-10670	477	12	306–311	306–311	NUM
dad-10670	477	13	.	.	PUNCT
dad-10670	478	1	association	association	NOUN
dad-10670	478	2	for	for	ADP
dad-10670	478	3	computational	computational	ADJ
dad-10670	478	4	linguistics	linguistic	NOUN
dad-10670	478	5	morristown	morristown	PROPN
dad-10670	478	6	,	,	PUNCT
dad-10670	478	7	nj	nj	PROPN
dad-10670	478	8	,	,	PUNCT
dad-10670	478	9	usa	usa	PROPN
dad-10670	478	10	,	,	PUNCT
dad-10670	478	11	1991	1991	NUM
dad-10670	478	12	.	.	PUNCT
dad-10670	479	1	leo	leo	PROPN
dad-10670	479	2	breiman	breiman	PROPN
dad-10670	479	3	.	.	PUNCT
dad-10670	480	1	bagging	bag	VERB
dad-10670	480	2	predictors	predictor	NOUN
dad-10670	480	3	.	.	PUNCT
dad-10670	481	1	machine	machine	NOUN
dad-10670	481	2	learning	learning	PROPN
dad-10670	481	3	,	,	PUNCT
dad-10670	481	4	24(2):123–140	24(2):123–140	NUM
dad-10670	481	5	,	,	PUNCT
dad-10670	481	6	1996	1996	NUM
dad-10670	481	7	.	.	PUNCT
dad-10670	482	1	lynn	lynn	PROPN
dad-10670	482	2	carlson	carlson	PROPN
dad-10670	482	3	,	,	PUNCT
dad-10670	482	4	daniel	daniel	PROPN
dad-10670	482	5	marcu	marcu	PROPN
dad-10670	482	6	,	,	PUNCT
dad-10670	482	7	and	and	CCONJ
dad-10670	482	8	mary	mary	PROPN
dad-10670	482	9	ellen	ellen	PROPN
dad-10670	482	10	okurowski	okurowski	PROPN
dad-10670	482	11	.	.	PUNCT
dad-10670	483	1	building	build	VERB
dad-10670	483	2	a	a	DET
dad-10670	483	3	discourse	discourse	NOUN
dad-10670	483	4	-	-	PUNCT
dad-10670	483	5	tagged	tag	VERB
dad-10670	483	6	corpus	corpus	NOUN
dad-10670	483	7	in	in	ADP
dad-10670	483	8	the	the	DET
dad-10670	483	9	framework	framework	NOUN
dad-10670	483	10	of	of	ADP
dad-10670	483	11	rhetorical	rhetorical	ADJ
dad-10670	483	12	structure	structure	NOUN
dad-10670	483	13	theory	theory	NOUN
dad-10670	483	14	.	.	PUNCT
dad-10670	484	1	proceedings	proceeding	NOUN
dad-10670	484	2	of	of	ADP
dad-10670	484	3	second	second	ADJ
dad-10670	484	4	sigdial	sigdial	ADJ
dad-10670	484	5	workshop	workshop	NOUN
dad-10670	484	6	on	on	ADP
dad-10670	484	7	discourse	discourse	NOUN
dad-10670	484	8	and	and	CCONJ
dad-10670	484	9	dialogue	dialogue	NOUN
dad-10670	484	10	-	-	PUNCT
dad-10670	484	11	volume	volume	NOUN
dad-10670	484	12	16	16	NUM
dad-10670	484	13	,	,	PUNCT
dad-10670	484	14	pages	page	NOUN
dad-10670	484	15	1–10	1–10	NOUN
dad-10670	484	16	,	,	PUNCT
dad-10670	484	17	2001	2001	NUM
dad-10670	484	18	.	.	PUNCT
dad-10670	485	1	joyce	joyce	PROPN
dad-10670	485	2	y.	y.	PROPN
dad-10670	485	3	chai	chai	PROPN
dad-10670	485	4	and	and	CCONJ
dad-10670	485	5	rong	rong	PROPN
dad-10670	485	6	jin	jin	PROPN
dad-10670	485	7	.	.	PUNCT
dad-10670	486	1	discourse	discourse	NOUN
dad-10670	486	2	structure	structure	NOUN
dad-10670	486	3	for	for	ADP
dad-10670	486	4	context	context	NOUN
dad-10670	486	5	question	question	NOUN
dad-10670	486	6	answering	answering	NOUN
dad-10670	486	7	.	.	PUNCT
dad-10670	487	1	in	in	ADP
dad-10670	487	2	sanda	sanda	NOUN
dad-10670	487	3	harabagiu	harabagiu	NOUN
dad-10670	487	4	and	and	CCONJ
dad-10670	487	5	finley	finley	PROPN
dad-10670	487	6	lacatusu	lacatusu	PROPN
dad-10670	487	7	,	,	PUNCT
dad-10670	487	8	editors	editor	NOUN
dad-10670	487	9	,	,	PUNCT
dad-10670	487	10	hlt	hlt	PROPN
dad-10670	487	11	-	-	PUNCT
dad-10670	487	12	naacl	naacl	PROPN
dad-10670	487	13	2004	2004	NUM
dad-10670	487	14	:	:	PUNCT
dad-10670	487	15	workshop	workshop	NOUN
dad-10670	487	16	on	on	ADP
dad-10670	487	17	pragmatics	pragmatic	NOUN
dad-10670	487	18	of	of	ADP
dad-10670	487	19	question	question	NOUN
dad-10670	487	20	answering	answering	NOUN
dad-10670	487	21	,	,	PUNCT
dad-10670	487	22	pages	page	NOUN
dad-10670	487	23	23–30	23–30	NUM
dad-10670	487	24	,	,	PUNCT
dad-10670	487	25	boston	boston	PROPN
dad-10670	487	26	,	,	PUNCT
dad-10670	487	27	massachusetts	massachusetts	PROPN
dad-10670	487	28	,	,	PUNCT
dad-10670	487	29	usa	usa	PROPN
dad-10670	487	30	,	,	PUNCT
dad-10670	487	31	may	may	AUX
dad-10670	487	32	2	2	NUM
dad-10670	487	33	may	may	PROPN
dad-10670	487	34	7	7	NUM
dad-10670	487	35	2004	2004	NUM
dad-10670	487	36	.	.	PUNCT
dad-10670	488	1	association	association	NOUN
dad-10670	488	2	for	for	ADP
dad-10670	488	3	computational	computational	ADJ
dad-10670	488	4	linguistics	linguistic	NOUN
dad-10670	488	5	.	.	PUNCT
dad-10670	489	1	eugene	eugene	PROPN
dad-10670	489	2	charniak	charniak	PROPN
dad-10670	489	3	.	.	PUNCT
dad-10670	490	1	a	a	DET
dad-10670	490	2	maximum	maximum	ADJ
dad-10670	490	3	-	-	PUNCT
dad-10670	490	4	entropy	entropy	NOUN
dad-10670	490	5	-	-	PUNCT
dad-10670	490	6	inspired	inspire	VERB
dad-10670	490	7	parser	parser	NOUN
dad-10670	490	8	.	.	PUNCT
dad-10670	491	1	in	in	ADP
dad-10670	491	2	proceedings	proceeding	NOUN
dad-10670	491	3	of	of	ADP
dad-10670	491	4	the	the	DET
dad-10670	491	5	1st	1st	ADJ
dad-10670	491	6	north	north	PROPN
dad-10670	491	7	american	american	ADJ
dad-10670	491	8	chapter	chapter	NOUN
dad-10670	491	9	of	of	ADP
dad-10670	491	10	the	the	DET
dad-10670	491	11	association	association	NOUN
dad-10670	491	12	for	for	ADP
dad-10670	491	13	computational	computational	ADJ
dad-10670	491	14	linguistics	linguistic	NOUN
dad-10670	491	15	conference	conference	NOUN
dad-10670	491	16	,	,	PUNCT
dad-10670	491	17	pages	page	NOUN
dad-10670	491	18	132–139	132–139	NUM
dad-10670	491	19	,	,	PUNCT
dad-10670	491	20	san	san	PROPN
dad-10670	491	21	francisco	francisco	PROPN
dad-10670	491	22	,	,	PUNCT
dad-10670	491	23	ca	ca	PROPN
dad-10670	491	24	,	,	PUNCT
dad-10670	491	25	usa	usa	PROPN
dad-10670	491	26	,	,	PUNCT
dad-10670	491	27	2000	2000	NUM
dad-10670	491	28	.	.	PUNCT
dad-10670	492	1	morgan	morgan	PROPN
dad-10670	492	2	kaufmann	kaufmann	PROPN
dad-10670	492	3	publishers	publishers	PROPN
dad-10670	492	4	inc	inc	PROPN
dad-10670	492	5	.	.	PROPN
dad-10670	492	6	degang	degang	PROPN
dad-10670	492	7	chen	chen	PROPN
dad-10670	492	8	,	,	PUNCT
dad-10670	492	9	qiang	qiang	PROPN
dad-10670	492	10	he	he	PRON
dad-10670	492	11	,	,	PUNCT
dad-10670	492	12	and	and	CCONJ
dad-10670	492	13	xizhao	xizhao	PROPN
dad-10670	492	14	wang	wang	PROPN
dad-10670	492	15	.	.	PUNCT
dad-10670	493	1	on	on	ADP
dad-10670	493	2	linear	linear	ADJ
dad-10670	493	3	separability	separability	NOUN
dad-10670	493	4	of	of	ADP
dad-10670	493	5	data	datum	NOUN
dad-10670	493	6	sets	set	NOUN
dad-10670	493	7	in	in	ADP
dad-10670	493	8	feature	feature	NOUN
dad-10670	493	9	space	space	NOUN
dad-10670	493	10	.	.	PUNCT
dad-10670	494	1	neurocomputing	neurocomputing	NOUN
dad-10670	494	2	,	,	PUNCT
dad-10670	494	3	70(13	70(13	PROPN
dad-10670	494	4	-	-	SYM
dad-10670	494	5	15):2441–2448	15):2441–2448	NUM
dad-10670	494	6	,	,	PUNCT
dad-10670	494	7	2007	2007	NUM
dad-10670	494	8	.	.	PUNCT
dad-10670	495	1	david	david	PROPN
dad-10670	495	2	duverle	duverle	PROPN
dad-10670	495	3	and	and	CCONJ
dad-10670	495	4	helmut	helmut	PROPN
dad-10670	495	5	prendinger	prendinger	NOUN
dad-10670	495	6	.	.	PUNCT
dad-10670	496	1	a	a	DET
dad-10670	496	2	novel	novel	ADJ
dad-10670	496	3	discourse	discourse	NOUN
dad-10670	496	4	parser	parser	NOUN
dad-10670	496	5	based	base	VERB
dad-10670	496	6	on	on	ADP
dad-10670	496	7	support	support	NOUN
dad-10670	496	8	vector	vector	NOUN
dad-10670	496	9	machine	machine	NOUN
dad-10670	496	10	classification	classification	NOUN
dad-10670	496	11	.	.	PUNCT
dad-10670	497	1	in	in	ADP
dad-10670	497	2	proceedings	proceeding	NOUN
dad-10670	497	3	of	of	ADP
dad-10670	497	4	the	the	DET
dad-10670	497	5	joint	joint	ADJ
dad-10670	497	6	conference	conference	NOUN
dad-10670	497	7	of	of	ADP
dad-10670	497	8	the	the	DET
dad-10670	497	9	47th	47th	ADJ
dad-10670	497	10	annual	annual	ADJ
dad-10670	497	11	meeting	meeting	NOUN
dad-10670	497	12	of	of	ADP
dad-10670	497	13	the	the	DET
dad-10670	497	14	acl	acl	PROPN
dad-10670	497	15	and	and	CCONJ
dad-10670	497	16	the	the	DET
dad-10670	497	17	4th	4th	ADJ
dad-10670	497	18	international	international	ADJ
dad-10670	497	19	joint	joint	ADJ
dad-10670	497	20	conference	conference	NOUN
dad-10670	497	21	on	on	ADP
dad-10670	497	22	natural	natural	ADJ
dad-10670	497	23	language	language	NOUN
dad-10670	497	24	processing	processing	NOUN
dad-10670	497	25	of	of	ADP
dad-10670	497	26	the	the	DET
dad-10670	497	27	afnlp	afnlp	NOUN
dad-10670	497	28	,	,	PUNCT
dad-10670	497	29	pages	page	NOUN
dad-10670	497	30	665–673	665–673	NUM
dad-10670	497	31	,	,	PUNCT
dad-10670	497	32	suntec	suntec	ADJ
dad-10670	497	33	,	,	PUNCT
dad-10670	497	34	singapore	singapore	PROPN
dad-10670	497	35	,	,	PUNCT
dad-10670	497	36	august	august	PROPN
dad-10670	497	37	2009	2009	NUM
dad-10670	497	38	.	.	PUNCT
dad-10670	498	1	association	association	NOUN
dad-10670	498	2	for	for	ADP
dad-10670	498	3	computational	computational	ADJ
dad-10670	498	4	linguistics	linguistic	NOUN
dad-10670	498	5	.	.	PUNCT
dad-10670	499	1	christiane	christiane	PROPN
dad-10670	499	2	fellbaum	fellbaum	PROPN
dad-10670	499	3	,	,	PUNCT
dad-10670	499	4	editor	editor	NOUN
dad-10670	499	5	.	.	PUNCT
dad-10670	500	1	wordnet	wordnet	NOUN
dad-10670	500	2	:	:	PUNCT
dad-10670	500	3	an	an	DET
dad-10670	500	4	electronic	electronic	ADJ
dad-10670	500	5	lexical	lexical	ADJ
dad-10670	500	6	database	database	NOUN
dad-10670	500	7	.	.	PUNCT
dad-10670	501	1	mit	mit	PROPN
dad-10670	501	2	press	press	PROPN
dad-10670	501	3	,	,	PUNCT
dad-10670	501	4	1998	1998	NUM
dad-10670	501	5	.	.	PUNCT
dad-10670	502	1	hugo	hugo	PROPN
dad-10670	502	2	hernault	hernault	PROPN
dad-10670	502	3	,	,	PUNCT
dad-10670	502	4	paul	paul	PROPN
dad-10670	502	5	piwek	piwek	PROPN
dad-10670	502	6	,	,	PUNCT
dad-10670	502	7	helmut	helmut	PROPN
dad-10670	502	8	prendinger	prendinger	NOUN
dad-10670	502	9	,	,	PUNCT
dad-10670	502	10	and	and	CCONJ
dad-10670	502	11	mitsuru	mitsuru	PROPN
dad-10670	502	12	ishizuka	ishizuka	PROPN
dad-10670	502	13	.	.	PUNCT
dad-10670	503	1	generating	generate	VERB
dad-10670	503	2	dialogues	dialogue	NOUN
dad-10670	503	3	for	for	ADP
dad-10670	503	4	virtual	virtual	ADJ
dad-10670	503	5	agents	agent	NOUN
dad-10670	503	6	using	use	VERB
dad-10670	503	7	nested	nested	ADJ
dad-10670	503	8	textual	textual	ADJ
dad-10670	503	9	coherence	coherence	NOUN
dad-10670	503	10	relations	relation	NOUN
dad-10670	503	11	.	.	PUNCT
dad-10670	504	1	in	in	ADP
dad-10670	504	2	iva	iva	PROPN
dad-10670	504	3	’	'	PUNCT
dad-10670	504	4	08	08	NUM
dad-10670	504	5	:	:	PUNCT
dad-10670	504	6	proceedings	proceeding	NOUN
dad-10670	504	7	of	of	ADP
dad-10670	504	8	the	the	DET
dad-10670	504	9	8th	8th	ADJ
dad-10670	504	10	international	international	ADJ
dad-10670	504	11	conference	conference	NOUN
dad-10670	504	12	on	on	ADP
dad-10670	504	13	intelligent	intelligent	ADJ
dad-10670	504	14	virtual	virtual	ADJ
dad-10670	504	15	agents	agent	NOUN
dad-10670	504	16	,	,	PUNCT
dad-10670	504	17	pages	page	NOUN
dad-10670	504	18	139–145	139–145	NUM
dad-10670	504	19	,	,	PUNCT
dad-10670	504	20	berlin	berlin	PROPN
dad-10670	504	21	,	,	PUNCT
dad-10670	504	22	heidelberg	heidelberg	PROPN
dad-10670	504	23	,	,	PUNCT
dad-10670	504	24	2008	2008	NUM
dad-10670	504	25	.	.	PUNCT
dad-10670	505	1	springer	springer	NOUN
dad-10670	505	2	-	-	PUNCT
dad-10670	505	3	verlag	verlag	PROPN
dad-10670	505	4	.	.	PUNCT
dad-10670	506	1	scott	scott	PROPN
dad-10670	506	2	kirkpatrick	kirkpatrick	PROPN
dad-10670	506	3	,	,	PUNCT
dad-10670	506	4	c.	c.	PROPN
dad-10670	506	5	daniel	daniel	PROPN
dad-10670	506	6	gelatt	gelatt	PROPN
dad-10670	506	7	,	,	PUNCT
dad-10670	506	8	and	and	CCONJ
dad-10670	506	9	mario	mario	PROPN
dad-10670	506	10	p.	p.	PROPN
dad-10670	506	11	vecchi	vecchi	PROPN
dad-10670	506	12	.	.	PUNCT
dad-10670	507	1	optimization	optimization	NOUN
dad-10670	507	2	by	by	ADP
dad-10670	507	3	simulated	simulated	ADJ
dad-10670	507	4	annealing	annealing	NOUN
dad-10670	507	5	.	.	PUNCT
dad-10670	508	1	science	science	NOUN
dad-10670	508	2	,	,	PUNCT
dad-10670	508	3	220(4598):671–680	220(4598):671–680	NUM
dad-10670	508	4	,	,	PUNCT
dad-10670	508	5	1983	1983	NUM
dad-10670	508	6	.	.	PUNCT
dad-10670	509	1	27	27	NUM
dad-10670	509	2	hernault	hernault	ADJ
dad-10670	509	3	,	,	PUNCT
dad-10670	509	4	prendinger	prendinger	NOUN
dad-10670	509	5	,	,	PUNCT
dad-10670	509	6	duverle	duverle	NOUN
dad-10670	509	7	,	,	PUNCT
dad-10670	509	8	and	and	CCONJ
dad-10670	509	9	ishizuka	ishizuka	PROPN
dad-10670	509	10	dan	dan	PROPN
dad-10670	509	11	klein	klein	PROPN
dad-10670	509	12	and	and	CCONJ
dad-10670	509	13	christopher	christopher	PROPN
dad-10670	509	14	d.	d.	PROPN
dad-10670	509	15	manning	manning	PROPN
dad-10670	509	16	.	.	PUNCT
dad-10670	510	1	fast	fast	ADJ
dad-10670	510	2	exact	exact	ADJ
dad-10670	510	3	inference	inference	NOUN
dad-10670	510	4	with	with	ADP
dad-10670	510	5	a	a	DET
dad-10670	510	6	factored	factored	ADJ
dad-10670	510	7	model	model	NOUN
dad-10670	510	8	for	for	ADP
dad-10670	510	9	natural	natural	ADJ
dad-10670	510	10	language	language	NOUN
dad-10670	510	11	parsing	parsing	NOUN
dad-10670	510	12	.	.	PUNCT
dad-10670	511	1	in	in	ADP
dad-10670	511	2	advances	advance	NOUN
dad-10670	511	3	in	in	ADP
dad-10670	511	4	neural	neural	ADJ
dad-10670	511	5	information	information	NOUN
dad-10670	511	6	processing	processing	NOUN
dad-10670	511	7	systems	system	NOUN
dad-10670	511	8	,	,	PUNCT
dad-10670	511	9	volume	volume	NOUN
dad-10670	511	10	15	15	NUM
dad-10670	511	11	,	,	PUNCT
dad-10670	511	12	pages	page	NOUN
dad-10670	511	13	3–10	3–10	PROPN
dad-10670	511	14	.	.	PUNCT
dad-10670	512	1	mit	mit	PROPN
dad-10670	512	2	press	press	PROPN
dad-10670	512	3	,	,	PUNCT
dad-10670	512	4	2003	2003	NUM
dad-10670	512	5	.	.	PUNCT
dad-10670	513	1	john	john	PROPN
dad-10670	513	2	d.	d.	PROPN
dad-10670	513	3	lafferty	lafferty	PROPN
dad-10670	513	4	,	,	PUNCT
dad-10670	513	5	andrew	andrew	PROPN
dad-10670	513	6	mccallum	mccallum	PROPN
dad-10670	513	7	,	,	PUNCT
dad-10670	513	8	and	and	CCONJ
dad-10670	513	9	fernando	fernando	PROPN
dad-10670	513	10	c.	c.	PROPN
dad-10670	513	11	n.	n.	PROPN
dad-10670	513	12	pereira	pereira	PROPN
dad-10670	513	13	.	.	PUNCT
dad-10670	514	1	conditional	conditional	ADJ
dad-10670	514	2	random	random	ADJ
dad-10670	514	3	fields	field	NOUN
dad-10670	514	4	:	:	PUNCT
dad-10670	514	5	probabilistic	probabilistic	ADJ
dad-10670	514	6	models	model	NOUN
dad-10670	514	7	for	for	ADP
dad-10670	514	8	segmenting	segment	VERB
dad-10670	514	9	and	and	CCONJ
dad-10670	514	10	labeling	labeling	NOUN
dad-10670	514	11	sequence	sequence	NOUN
dad-10670	514	12	data	datum	NOUN
dad-10670	514	13	.	.	PUNCT
dad-10670	515	1	in	in	ADP
dad-10670	515	2	icml’01	icml’01	PROPN
dad-10670	515	3	:	:	PUNCT
dad-10670	515	4	proceedings	proceeding	NOUN
dad-10670	515	5	of	of	ADP
dad-10670	515	6	the	the	DET
dad-10670	515	7	eighteenth	eighteenth	ADJ
dad-10670	515	8	international	international	ADJ
dad-10670	515	9	conference	conference	NOUN
dad-10670	515	10	on	on	ADP
dad-10670	515	11	machine	machine	NOUN
dad-10670	515	12	learning	learning	NOUN
dad-10670	515	13	,	,	PUNCT
dad-10670	515	14	pages	page	NOUN
dad-10670	515	15	282–289	282–289	NUM
dad-10670	515	16	,	,	PUNCT
dad-10670	515	17	san	san	PROPN
dad-10670	515	18	francisco	francisco	PROPN
dad-10670	515	19	,	,	PUNCT
dad-10670	515	20	ca	ca	PROPN
dad-10670	515	21	,	,	PUNCT
dad-10670	515	22	usa	usa	PROPN
dad-10670	515	23	,	,	PUNCT
dad-10670	515	24	2001	2001	NUM
dad-10670	515	25	.	.	PUNCT
dad-10670	516	1	morgan	morgan	PROPN
dad-10670	516	2	kaufmann	kaufmann	PROPN
dad-10670	516	3	publishers	publishers	PROPN
dad-10670	516	4	inc	inc	PROPN
dad-10670	516	5	.	.	PROPN
dad-10670	516	6	huong	huong	PROPN
dad-10670	516	7	le	le	PROPN
dad-10670	516	8	thanh	thanh	PROPN
dad-10670	516	9	,	,	PUNCT
dad-10670	516	10	geetha	geetha	PROPN
dad-10670	516	11	abeysinghe	abeysinghe	NOUN
dad-10670	516	12	,	,	PUNCT
dad-10670	516	13	and	and	CCONJ
dad-10670	516	14	christian	christian	PROPN
dad-10670	516	15	huyck	huyck	PROPN
dad-10670	516	16	.	.	PUNCT
dad-10670	517	1	automated	automate	VERB
dad-10670	517	2	discourse	discourse	NOUN
dad-10670	517	3	segmentation	segmentation	NOUN
dad-10670	517	4	by	by	ADP
dad-10670	517	5	syntactic	syntactic	ADJ
dad-10670	517	6	information	information	NOUN
dad-10670	517	7	and	and	CCONJ
dad-10670	517	8	cue	cue	NOUN
dad-10670	517	9	phrases	phrase	NOUN
dad-10670	517	10	.	.	PUNCT
dad-10670	518	1	in	in	ADP
dad-10670	518	2	proceedings	proceeding	NOUN
dad-10670	518	3	of	of	ADP
dad-10670	518	4	aia’04	aia’04	PROPN
dad-10670	518	5	,	,	PUNCT
dad-10670	518	6	innsbruck	innsbruck	PROPN
dad-10670	518	7	,	,	PUNCT
dad-10670	518	8	austria	austria	PROPN
dad-10670	518	9	,	,	PUNCT
dad-10670	518	10	february	february	PROPN
dad-10670	518	11	16–18	16–18	NUM
dad-10670	518	12	2004a	2004a	NUM
dad-10670	518	13	.	.	PUNCT
dad-10670	519	1	huong	huong	PROPN
dad-10670	519	2	le	le	PROPN
dad-10670	519	3	thanh	thanh	PROPN
dad-10670	519	4	,	,	PUNCT
dad-10670	519	5	geetha	geetha	PROPN
dad-10670	519	6	abeysinghe	abeysinghe	NOUN
dad-10670	519	7	,	,	PUNCT
dad-10670	519	8	and	and	CCONJ
dad-10670	519	9	christian	christian	PROPN
dad-10670	519	10	huyck	huyck	PROPN
dad-10670	519	11	.	.	PUNCT
dad-10670	520	1	generating	generate	VERB
dad-10670	520	2	discourse	discourse	NOUN
dad-10670	520	3	structures	structure	NOUN
dad-10670	520	4	for	for	ADP
dad-10670	520	5	written	write	VERB
dad-10670	520	6	texts	text	NOUN
dad-10670	520	7	.	.	PUNCT
dad-10670	521	1	in	in	ADP
dad-10670	521	2	proceedings	proceeding	NOUN
dad-10670	521	3	of	of	ADP
dad-10670	521	4	the	the	DET
dad-10670	521	5	20th	20th	ADJ
dad-10670	521	6	international	international	ADJ
dad-10670	521	7	conference	conference	NOUN
dad-10670	521	8	on	on	ADP
dad-10670	521	9	computational	computational	ADJ
dad-10670	521	10	linguistics	linguistic	NOUN
dad-10670	521	11	,	,	PUNCT
dad-10670	521	12	pages	page	NOUN
dad-10670	521	13	329–335	329–335	NUM
dad-10670	521	14	,	,	PUNCT
dad-10670	521	15	geneva	geneva	PROPN
dad-10670	521	16	,	,	PUNCT
dad-10670	521	17	switzerland	switzerland	PROPN
dad-10670	521	18	,	,	PUNCT
dad-10670	521	19	aug	aug	PROPN
dad-10670	521	20	23	23	NUM
dad-10670	521	21	–	–	PUNCT
dad-10670	521	22	aug	aug	PROPN
dad-10670	521	23	27	27	NUM
dad-10670	521	24	2004b	2004b	NOUN
dad-10670	521	25	.	.	PUNCT
dad-10670	522	1	coling	cole	VERB
dad-10670	522	2	.	.	PUNCT
dad-10670	523	1	ziheng	ziheng	PROPN
dad-10670	523	2	lin	lin	PROPN
dad-10670	523	3	,	,	PUNCT
dad-10670	523	4	min	min	PROPN
dad-10670	523	5	-	-	ADJ
dad-10670	523	6	yen	yen	NOUN
dad-10670	523	7	kan	kan	PROPN
dad-10670	523	8	,	,	PUNCT
dad-10670	523	9	and	and	CCONJ
dad-10670	523	10	hwee	hwee	VERB
dad-10670	523	11	tou	tou	PROPN
dad-10670	523	12	ng	ng	PROPN
dad-10670	523	13	.	.	PUNCT
dad-10670	524	1	recognizing	recognize	VERB
dad-10670	524	2	implicit	implicit	ADJ
dad-10670	524	3	discourse	discourse	NOUN
dad-10670	524	4	relations	relation	NOUN
dad-10670	524	5	in	in	ADP
dad-10670	524	6	the	the	DET
dad-10670	524	7	penn	penn	PROPN
dad-10670	524	8	discourse	discourse	PROPN
dad-10670	524	9	treebank	treebank	PROPN
dad-10670	524	10	.	.	PUNCT
dad-10670	525	1	in	in	ADP
dad-10670	525	2	proceedings	proceeding	NOUN
dad-10670	525	3	of	of	ADP
dad-10670	525	4	the	the	DET
dad-10670	525	5	2009	2009	NUM
dad-10670	525	6	conference	conference	NOUN
dad-10670	525	7	on	on	ADP
dad-10670	525	8	empirical	empirical	ADJ
dad-10670	525	9	methods	method	NOUN
dad-10670	525	10	in	in	ADP
dad-10670	525	11	natural	natural	ADJ
dad-10670	525	12	language	language	NOUN
dad-10670	525	13	processing	processing	NOUN
dad-10670	525	14	,	,	PUNCT
dad-10670	525	15	pages	page	NOUN
dad-10670	525	16	343–351	343–351	NUM
dad-10670	525	17	,	,	PUNCT
dad-10670	525	18	singapore	singapore	PROPN
dad-10670	525	19	,	,	PUNCT
dad-10670	525	20	august	august	PROPN
dad-10670	525	21	2009	2009	NUM
dad-10670	525	22	.	.	PUNCT
dad-10670	526	1	association	association	NOUN
dad-10670	526	2	for	for	ADP
dad-10670	526	3	computational	computational	ADJ
dad-10670	526	4	linguistics	linguistic	NOUN
dad-10670	526	5	.	.	PUNCT
dad-10670	527	1	david	david	PROPN
dad-10670	527	2	m.	m.	PROPN
dad-10670	527	3	magerman	magerman	PROPN
dad-10670	527	4	.	.	PUNCT
dad-10670	528	1	statistical	statistical	ADJ
dad-10670	528	2	decision	decision	NOUN
dad-10670	528	3	-	-	PUNCT
dad-10670	528	4	tree	tree	NOUN
dad-10670	528	5	models	model	NOUN
dad-10670	528	6	for	for	ADP
dad-10670	528	7	parsing	parsing	NOUN
dad-10670	528	8	.	.	PUNCT
dad-10670	529	1	proceedings	proceeding	NOUN
dad-10670	529	2	of	of	ADP
dad-10670	529	3	the	the	DET
dad-10670	529	4	33rd	33rd	ADJ
dad-10670	529	5	annual	annual	ADJ
dad-10670	529	6	meeting	meeting	NOUN
dad-10670	529	7	on	on	ADP
dad-10670	529	8	association	association	NOUN
dad-10670	529	9	for	for	ADP
dad-10670	529	10	computational	computational	ADJ
dad-10670	529	11	linguistics	linguistic	NOUN
dad-10670	529	12	,	,	PUNCT
dad-10670	529	13	pages	page	NOUN
dad-10670	529	14	276–283	276–283	NUM
dad-10670	529	15	,	,	PUNCT
dad-10670	529	16	1995	1995	NUM
dad-10670	529	17	.	.	PUNCT
dad-10670	530	1	william	william	PROPN
dad-10670	530	2	c.	c.	PROPN
dad-10670	530	3	mann	mann	PROPN
dad-10670	530	4	and	and	CCONJ
dad-10670	530	5	sandra	sandra	PROPN
dad-10670	530	6	a.	a.	PROPN
dad-10670	530	7	thompson	thompson	PROPN
dad-10670	530	8	.	.	PUNCT
dad-10670	531	1	rhetorical	rhetorical	ADJ
dad-10670	531	2	structure	structure	NOUN
dad-10670	531	3	theory	theory	NOUN
dad-10670	531	4	:	:	PUNCT
dad-10670	531	5	toward	toward	ADP
dad-10670	531	6	a	a	DET
dad-10670	531	7	functional	functional	ADJ
dad-10670	531	8	theory	theory	NOUN
dad-10670	531	9	of	of	ADP
dad-10670	531	10	text	text	NOUN
dad-10670	531	11	organization	organization	NOUN
dad-10670	531	12	.	.	PUNCT
dad-10670	532	1	text	text	NOUN
dad-10670	532	2	,	,	PUNCT
dad-10670	532	3	8(3):243–281	8(3):243–281	NUM
dad-10670	532	4	,	,	PUNCT
dad-10670	532	5	1988	1988	NUM
dad-10670	532	6	.	.	PUNCT
dad-10670	533	1	christopher	christopher	PROPN
dad-10670	533	2	d.	d.	PROPN
dad-10670	533	3	manning	manning	PROPN
dad-10670	533	4	and	and	CCONJ
dad-10670	533	5	hinrich	hinrich	PROPN
dad-10670	533	6	schütze	schütze	PROPN
dad-10670	533	7	.	.	PUNCT
dad-10670	534	1	foundations	foundation	NOUN
dad-10670	534	2	of	of	ADP
dad-10670	534	3	statistical	statistical	ADJ
dad-10670	534	4	natural	natural	ADJ
dad-10670	534	5	language	language	NOUN
dad-10670	534	6	processing	processing	NOUN
dad-10670	534	7	.	.	PUNCT
dad-10670	535	1	mit	mit	PROPN
dad-10670	535	2	press	press	PROPN
dad-10670	535	3	,	,	PUNCT
dad-10670	535	4	1999	1999	NUM
dad-10670	535	5	.	.	PUNCT
dad-10670	536	1	daniel	daniel	PROPN
dad-10670	536	2	marcu	marcu	PROPN
dad-10670	536	3	.	.	PUNCT
dad-10670	537	1	building	build	VERB
dad-10670	537	2	up	up	ADP
dad-10670	537	3	rhetorical	rhetorical	ADJ
dad-10670	537	4	structure	structure	NOUN
dad-10670	537	5	trees	tree	NOUN
dad-10670	537	6	.	.	PUNCT
dad-10670	538	1	proceedings	proceeding	NOUN
dad-10670	538	2	of	of	ADP
dad-10670	538	3	the	the	DET
dad-10670	538	4	national	national	ADJ
dad-10670	538	5	conference	conference	NOUN
dad-10670	538	6	on	on	ADP
dad-10670	538	7	artificial	artificial	ADJ
dad-10670	538	8	intelligence	intelligence	NOUN
dad-10670	538	9	,	,	PUNCT
dad-10670	538	10	pages	page	NOUN
dad-10670	538	11	1069–1074	1069–1074	NUM
dad-10670	538	12	,	,	PUNCT
dad-10670	538	13	1996	1996	NUM
dad-10670	538	14	.	.	PUNCT
dad-10670	539	1	daniel	daniel	PROPN
dad-10670	539	2	marcu	marcu	PROPN
dad-10670	539	3	.	.	PUNCT
dad-10670	540	1	the	the	DET
dad-10670	540	2	theory	theory	NOUN
dad-10670	540	3	and	and	CCONJ
dad-10670	540	4	practice	practice	NOUN
dad-10670	540	5	of	of	ADP
dad-10670	540	6	discourse	discourse	NOUN
dad-10670	540	7	parsing	parsing	NOUN
dad-10670	540	8	and	and	CCONJ
dad-10670	540	9	summarization	summarization	NOUN
dad-10670	540	10	.	.	PUNCT
dad-10670	541	1	mit	mit	PROPN
dad-10670	541	2	press	press	PROPN
dad-10670	541	3	,	,	PUNCT
dad-10670	541	4	2000	2000	NUM
dad-10670	541	5	.	.	PUNCT
dad-10670	542	1	daniel	daniel	PROPN
dad-10670	542	2	marcu	marcu	PROPN
dad-10670	542	3	and	and	CCONJ
dad-10670	542	4	abdessamad	abdessamad	NOUN
dad-10670	542	5	echihabi	echihabi	NOUN
dad-10670	542	6	.	.	PUNCT
dad-10670	543	1	an	an	DET
dad-10670	543	2	unsupervised	unsupervised	ADJ
dad-10670	543	3	approach	approach	NOUN
dad-10670	543	4	to	to	ADP
dad-10670	543	5	recognizing	recognize	VERB
dad-10670	543	6	discourse	discourse	NOUN
dad-10670	543	7	relations	relation	NOUN
dad-10670	543	8	.	.	PUNCT
dad-10670	544	1	in	in	ADP
dad-10670	544	2	proceedings	proceeding	NOUN
dad-10670	544	3	of	of	ADP
dad-10670	544	4	40th	40th	ADJ
dad-10670	544	5	annual	annual	ADJ
dad-10670	544	6	meeting	meeting	NOUN
dad-10670	544	7	of	of	ADP
dad-10670	544	8	the	the	DET
dad-10670	544	9	association	association	NOUN
dad-10670	544	10	for	for	ADP
dad-10670	544	11	computational	computational	ADJ
dad-10670	544	12	linguistics	linguistic	NOUN
dad-10670	544	13	,	,	PUNCT
dad-10670	544	14	pages	page	NOUN
dad-10670	544	15	368–375	368–375	NUM
dad-10670	544	16	,	,	PUNCT
dad-10670	544	17	philadelphia	philadelphia	PROPN
dad-10670	544	18	,	,	PUNCT
dad-10670	544	19	pennsylvania	pennsylvania	PROPN
dad-10670	544	20	,	,	PUNCT
dad-10670	544	21	usa	usa	PROPN
dad-10670	544	22	,	,	PUNCT
dad-10670	544	23	july	july	PROPN
dad-10670	544	24	2002	2002	NUM
dad-10670	544	25	.	.	PUNCT
dad-10670	545	1	association	association	NOUN
dad-10670	545	2	for	for	ADP
dad-10670	545	3	computational	computational	ADJ
dad-10670	545	4	linguistics	linguistic	NOUN
dad-10670	545	5	.	.	PUNCT
dad-10670	546	1	doi	doi	NOUN
dad-10670	546	2	:	:	PUNCT
dad-10670	546	3	10.3115/1073083.1073145	10.3115/1073083.1073145	NUM
dad-10670	546	4	.	.	PUNCT
dad-10670	547	1	mitchell	mitchell	PROPN
dad-10670	547	2	p.	p.	PROPN
dad-10670	547	3	marcus	marcus	PROPN
dad-10670	547	4	,	,	PUNCT
dad-10670	547	5	mary	mary	PROPN
dad-10670	547	6	ann	ann	PROPN
dad-10670	547	7	marcinkiewicz	marcinkiewicz	PROPN
dad-10670	547	8	,	,	PUNCT
dad-10670	547	9	and	and	CCONJ
dad-10670	547	10	beatrice	beatrice	PROPN
dad-10670	547	11	santorini	santorini	PROPN
dad-10670	547	12	.	.	PUNCT
dad-10670	548	1	building	build	VERB
dad-10670	548	2	a	a	DET
dad-10670	548	3	large	large	ADJ
dad-10670	548	4	annotated	annotated	ADJ
dad-10670	548	5	corpus	corpus	NOUN
dad-10670	548	6	of	of	ADP
dad-10670	548	7	english	english	PROPN
dad-10670	548	8	:	:	PUNCT
dad-10670	548	9	the	the	DET
dad-10670	548	10	penn	penn	PROPN
dad-10670	548	11	treebank	treebank	PROPN
dad-10670	548	12	.	.	PUNCT
dad-10670	549	1	computational	computational	ADJ
dad-10670	549	2	linguistics	linguistic	NOUN
dad-10670	549	3	,	,	PUNCT
dad-10670	549	4	19(2):313–330	19(2):313–330	NUM
dad-10670	549	5	,	,	PUNCT
dad-10670	549	6	1993	1993	NUM
dad-10670	549	7	.	.	PUNCT
dad-10670	550	1	johanna	johanna	PROPN
dad-10670	550	2	d.	d.	PROPN
dad-10670	550	3	moore	moore	PROPN
dad-10670	550	4	and	and	CCONJ
dad-10670	550	5	peter	peter	PROPN
dad-10670	550	6	wiemer	wiemer	PROPN
dad-10670	550	7	-	-	PUNCT
dad-10670	550	8	hastings	hastings	PROPN
dad-10670	550	9	.	.	PUNCT
dad-10670	551	1	discourse	discourse	PROPN
dad-10670	551	2	in	in	ADP
dad-10670	551	3	computational	computational	ADJ
dad-10670	551	4	linguistics	linguistic	NOUN
dad-10670	551	5	and	and	CCONJ
dad-10670	551	6	artificial	artificial	ADJ
dad-10670	551	7	intelligence	intelligence	NOUN
dad-10670	551	8	.	.	PUNCT
dad-10670	552	1	in	in	ADP
dad-10670	552	2	a.	a.	NOUN
dad-10670	552	3	graesser	graesser	NOUN
dad-10670	552	4	,	,	PUNCT
dad-10670	552	5	m.	m.	NOUN
dad-10670	552	6	gernsbacher	gernsbacher	NOUN
dad-10670	552	7	,	,	PUNCT
dad-10670	552	8	and	and	CCONJ
dad-10670	552	9	s.	s.	PROPN
dad-10670	552	10	goldman	goldman	PROPN
dad-10670	552	11	,	,	PUNCT
dad-10670	552	12	editors	editor	NOUN
dad-10670	552	13	,	,	PUNCT
dad-10670	552	14	handbook	handbook	NOUN
dad-10670	552	15	of	of	ADP
dad-10670	552	16	discourse	discourse	NOUN
dad-10670	552	17	processes	process	NOUN
dad-10670	552	18	,	,	PUNCT
dad-10670	552	19	pages	page	NOUN
dad-10670	552	20	439–486	439–486	NUM
dad-10670	552	21	.	.	PUNCT
dad-10670	552	22	erlbaum	erlbaum	PROPN
dad-10670	552	23	,	,	PUNCT
dad-10670	552	24	mahwah	mahwah	PROPN
dad-10670	552	25	,	,	PUNCT
dad-10670	552	26	nj	nj	PROPN
dad-10670	552	27	,	,	PUNCT
dad-10670	552	28	2003	2003	NUM
dad-10670	552	29	.	.	PUNCT
dad-10670	553	1	28	28	NUM
dad-10670	553	2	hilda	hilda	VERB
dad-10670	553	3	:	:	PUNCT
dad-10670	553	4	a	a	DET
dad-10670	553	5	discourse	discourse	NOUN
dad-10670	553	6	parser	parser	NOUN
dad-10670	553	7	using	use	VERB
dad-10670	553	8	support	support	NOUN
dad-10670	553	9	vector	vector	NOUN
dad-10670	553	10	machine	machine	NOUN
dad-10670	553	11	classification	classification	NOUN
dad-10670	553	12	jon	jon	PROPN
dad-10670	553	13	oberlander	oberlander	PROPN
dad-10670	553	14	and	and	CCONJ
dad-10670	553	15	johanna	johanna	PROPN
dad-10670	553	16	d.	d.	PROPN
dad-10670	553	17	moore	moore	PROPN
dad-10670	553	18	.	.	PUNCT
dad-10670	554	1	cue	cue	PROPN
dad-10670	554	2	phrases	phrase	NOUN
dad-10670	554	3	in	in	ADP
dad-10670	554	4	discourse	discourse	NOUN
dad-10670	554	5	:	:	PUNCT
dad-10670	554	6	further	further	ADJ
dad-10670	554	7	evidence	evidence	NOUN
dad-10670	554	8	for	for	ADP
dad-10670	554	9	the	the	DET
dad-10670	554	10	core	core	NOUN
dad-10670	554	11	:	:	PUNCT
dad-10670	554	12	contributor	contributor	NOUN
dad-10670	554	13	distinction	distinction	NOUN
dad-10670	554	14	.	.	PUNCT
dad-10670	555	1	in	in	ADP
dad-10670	555	2	workshop	workshop	NOUN
dad-10670	555	3	on	on	ADP
dad-10670	555	4	levels	level	NOUN
dad-10670	555	5	of	of	ADP
dad-10670	555	6	representation	representation	NOUN
dad-10670	555	7	in	in	ADP
dad-10670	555	8	discourse	discourse	NOUN
dad-10670	555	9	,	,	PUNCT
dad-10670	555	10	pages	page	NOUN
dad-10670	555	11	87–93	87–93	NUM
dad-10670	555	12	,	,	PUNCT
dad-10670	555	13	edinburgh	edinburgh	PROPN
dad-10670	555	14	,	,	PUNCT
dad-10670	555	15	uk	uk	PROPN
dad-10670	555	16	,	,	PUNCT
dad-10670	555	17	1999	1999	NUM
dad-10670	555	18	.	.	PUNCT
dad-10670	556	1	emily	emily	PROPN
dad-10670	556	2	pitler	pitler	PROPN
dad-10670	556	3	,	,	PUNCT
dad-10670	556	4	mridhula	mridhula	NOUN
dad-10670	556	5	raghupathy	raghupathy	ADJ
dad-10670	556	6	,	,	PUNCT
dad-10670	556	7	hena	hena	PROPN
dad-10670	556	8	mehta	mehta	PROPN
dad-10670	556	9	,	,	PUNCT
dad-10670	556	10	ani	ani	X
dad-10670	556	11	nenkova	nenkova	PROPN
dad-10670	556	12	,	,	PUNCT
dad-10670	556	13	alan	alan	PROPN
dad-10670	556	14	lee	lee	PROPN
dad-10670	556	15	,	,	PUNCT
dad-10670	556	16	and	and	CCONJ
dad-10670	556	17	aravind	aravind	PROPN
dad-10670	556	18	joshi	joshi	PROPN
dad-10670	556	19	.	.	PUNCT
dad-10670	557	1	easily	easily	ADV
dad-10670	557	2	identifiable	identifiable	ADJ
dad-10670	557	3	discourse	discourse	NOUN
dad-10670	557	4	relations	relation	NOUN
dad-10670	557	5	.	.	PUNCT
dad-10670	558	1	in	in	ADP
dad-10670	558	2	coling	cole	VERB
dad-10670	558	3	2008	2008	NUM
dad-10670	558	4	:	:	PUNCT
dad-10670	558	5	companion	companion	NOUN
dad-10670	558	6	volume	volume	NOUN
dad-10670	558	7	:	:	PUNCT
dad-10670	558	8	posters	poster	NOUN
dad-10670	558	9	,	,	PUNCT
dad-10670	558	10	pages	page	NOUN
dad-10670	558	11	87–90	87–90	NUM
dad-10670	558	12	,	,	PUNCT
dad-10670	558	13	manchester	manchester	PROPN
dad-10670	558	14	,	,	PUNCT
dad-10670	558	15	uk	uk	PROPN
dad-10670	558	16	,	,	PUNCT
dad-10670	558	17	august	august	PROPN
dad-10670	558	18	2008	2008	NUM
dad-10670	558	19	.	.	PUNCT
dad-10670	559	1	coling	cole	VERB
dad-10670	559	2	2008	2008	NUM
dad-10670	559	3	organizing	organizing	NOUN
dad-10670	559	4	committee	committee	NOUN
dad-10670	559	5	.	.	PUNCT
dad-10670	560	1	emily	emily	PROPN
dad-10670	560	2	pitler	pitler	PROPN
dad-10670	560	3	,	,	PUNCT
dad-10670	560	4	annie	annie	PROPN
dad-10670	560	5	louis	louis	PROPN
dad-10670	560	6	,	,	PUNCT
dad-10670	560	7	and	and	CCONJ
dad-10670	560	8	ani	ani	X
dad-10670	560	9	nenkova	nenkova	PROPN
dad-10670	560	10	.	.	PUNCT
dad-10670	561	1	automatic	automatic	ADJ
dad-10670	561	2	sense	sense	NOUN
dad-10670	561	3	prediction	prediction	NOUN
dad-10670	561	4	for	for	ADP
dad-10670	561	5	implicit	implicit	ADJ
dad-10670	561	6	discourse	discourse	NOUN
dad-10670	561	7	relations	relation	NOUN
dad-10670	561	8	in	in	ADP
dad-10670	561	9	text	text	NOUN
dad-10670	561	10	.	.	PUNCT
dad-10670	562	1	in	in	ADP
dad-10670	562	2	proceedings	proceeding	NOUN
dad-10670	562	3	of	of	ADP
dad-10670	562	4	the	the	DET
dad-10670	562	5	joint	joint	ADJ
dad-10670	562	6	conference	conference	NOUN
dad-10670	562	7	of	of	ADP
dad-10670	562	8	the	the	DET
dad-10670	562	9	47th	47th	ADJ
dad-10670	562	10	annual	annual	ADJ
dad-10670	562	11	meeting	meeting	NOUN
dad-10670	562	12	of	of	ADP
dad-10670	562	13	the	the	DET
dad-10670	562	14	acl	acl	PROPN
dad-10670	562	15	and	and	CCONJ
dad-10670	562	16	the	the	DET
dad-10670	562	17	4th	4th	ADJ
dad-10670	562	18	international	international	ADJ
dad-10670	562	19	joint	joint	ADJ
dad-10670	562	20	conference	conference	NOUN
dad-10670	562	21	on	on	ADP
dad-10670	562	22	natural	natural	ADJ
dad-10670	562	23	language	language	NOUN
dad-10670	562	24	processing	processing	NOUN
dad-10670	562	25	of	of	ADP
dad-10670	562	26	the	the	DET
dad-10670	562	27	afnlp	afnlp	NOUN
dad-10670	562	28	,	,	PUNCT
dad-10670	562	29	pages	page	NOUN
dad-10670	562	30	683–691	683–691	NUM
dad-10670	562	31	,	,	PUNCT
dad-10670	562	32	suntec	suntec	ADJ
dad-10670	562	33	,	,	PUNCT
dad-10670	562	34	singapore	singapore	PROPN
dad-10670	562	35	,	,	PUNCT
dad-10670	562	36	august	august	PROPN
dad-10670	562	37	2009	2009	NUM
dad-10670	562	38	.	.	PUNCT
dad-10670	563	1	association	association	NOUN
dad-10670	563	2	for	for	ADP
dad-10670	563	3	computational	computational	ADJ
dad-10670	563	4	linguistics	linguistic	NOUN
dad-10670	563	5	.	.	PUNCT
dad-10670	564	1	helmut	helmut	PROPN
dad-10670	564	2	prendinger	prendinger	PROPN
dad-10670	564	3	,	,	PUNCT
dad-10670	564	4	paul	paul	PROPN
dad-10670	564	5	piwek	piwek	PROPN
dad-10670	564	6	,	,	PUNCT
dad-10670	564	7	and	and	CCONJ
dad-10670	564	8	mitsuru	mitsuru	PROPN
dad-10670	564	9	ishizuka	ishizuka	PROPN
dad-10670	564	10	.	.	PUNCT
dad-10670	565	1	a	a	DET
dad-10670	565	2	novel	novel	ADJ
dad-10670	565	3	method	method	NOUN
dad-10670	565	4	for	for	ADP
dad-10670	565	5	automatically	automatically	ADV
dad-10670	565	6	generating	generate	VERB
dad-10670	565	7	multi	multi	ADJ
dad-10670	565	8	-	-	ADJ
dad-10670	565	9	modal	modal	ADJ
dad-10670	565	10	dialogue	dialogue	NOUN
dad-10670	565	11	from	from	ADP
dad-10670	565	12	text	text	NOUN
dad-10670	565	13	.	.	PUNCT
dad-10670	566	1	international	international	ADJ
dad-10670	566	2	journal	journal	NOUN
dad-10670	566	3	of	of	ADP
dad-10670	566	4	semantic	semantic	ADJ
dad-10670	566	5	computing	computing	NOUN
dad-10670	566	6	,	,	PUNCT
dad-10670	566	7	1(3):319–334	1(3):319–334	NUM
dad-10670	566	8	,	,	PUNCT
dad-10670	566	9	2007	2007	NUM
dad-10670	566	10	.	.	PUNCT
dad-10670	567	1	david	david	PROPN
dad-10670	567	2	reitter	reitter	PROPN
dad-10670	567	3	.	.	PUNCT
dad-10670	568	1	rhetorical	rhetorical	ADJ
dad-10670	568	2	analysis	analysis	NOUN
dad-10670	568	3	with	with	ADP
dad-10670	568	4	rich	rich	ADJ
dad-10670	568	5	-	-	PUNCT
dad-10670	568	6	feature	feature	NOUN
dad-10670	568	7	support	support	NOUN
dad-10670	568	8	vector	vector	NOUN
dad-10670	568	9	models	model	NOUN
dad-10670	568	10	.	.	PUNCT
dad-10670	569	1	unpublished	unpublished	ADJ
dad-10670	569	2	master	master	NOUN
dad-10670	569	3	’s	’s	PART
dad-10670	569	4	thesis	thesis	NOUN
dad-10670	569	5	,	,	PUNCT
dad-10670	569	6	university	university	NOUN
dad-10670	569	7	of	of	ADP
dad-10670	569	8	potsdam	potsdam	PROPN
dad-10670	569	9	,	,	PUNCT
dad-10670	569	10	potsdam	potsdam	PROPN
dad-10670	569	11	,	,	PUNCT
dad-10670	569	12	germany	germany	PROPN
dad-10670	569	13	,	,	PUNCT
dad-10670	569	14	2003a	2003a	NUM
dad-10670	569	15	.	.	PUNCT
dad-10670	570	1	david	david	PROPN
dad-10670	570	2	reitter	reitter	PROPN
dad-10670	570	3	.	.	PUNCT
dad-10670	571	1	simple	simple	ADJ
dad-10670	571	2	signals	signal	NOUN
dad-10670	571	3	for	for	ADP
dad-10670	571	4	complex	complex	ADJ
dad-10670	571	5	rhetorics	rhetoric	NOUN
dad-10670	571	6	:	:	PUNCT
dad-10670	571	7	on	on	ADP
dad-10670	571	8	rhetorical	rhetorical	ADJ
dad-10670	571	9	analysis	analysis	NOUN
dad-10670	571	10	with	with	ADP
dad-10670	571	11	rich	rich	ADJ
dad-10670	571	12	-	-	PUNCT
dad-10670	571	13	feature	feature	NOUN
dad-10670	571	14	support	support	NOUN
dad-10670	571	15	vector	vector	NOUN
dad-10670	571	16	models	model	NOUN
dad-10670	571	17	.	.	PUNCT
dad-10670	572	1	ldv	ldv	PROPN
dad-10670	572	2	forum	forum	PROPN
dad-10670	572	3	,	,	PUNCT
dad-10670	572	4	18(1/2):38–52	18(1/2):38–52	NUM
dad-10670	572	5	,	,	PUNCT
dad-10670	572	6	2003b	2003b	NUM
dad-10670	572	7	.	.	PUNCT
dad-10670	573	1	kenji	kenji	PROPN
dad-10670	573	2	sagae	sagae	PROPN
dad-10670	573	3	.	.	PUNCT
dad-10670	574	1	analysis	analysis	NOUN
dad-10670	574	2	of	of	ADP
dad-10670	574	3	discourse	discourse	NOUN
dad-10670	574	4	structure	structure	NOUN
dad-10670	574	5	with	with	ADP
dad-10670	574	6	syntactic	syntactic	ADJ
dad-10670	574	7	dependencies	dependency	NOUN
dad-10670	574	8	and	and	CCONJ
dad-10670	574	9	data	datum	NOUN
dad-10670	574	10	-	-	PUNCT
dad-10670	574	11	driven	drive	VERB
dad-10670	574	12	shiftreduce	shiftreduce	NOUN
dad-10670	574	13	parsing	parsing	NOUN
dad-10670	574	14	.	.	PUNCT
dad-10670	575	1	in	in	ADP
dad-10670	575	2	proceedings	proceeding	NOUN
dad-10670	575	3	of	of	ADP
dad-10670	575	4	the	the	DET
dad-10670	575	5	11th	11th	ADJ
dad-10670	575	6	international	international	ADJ
dad-10670	575	7	conference	conference	NOUN
dad-10670	575	8	on	on	ADP
dad-10670	575	9	parsing	parse	VERB
dad-10670	575	10	technologies	technology	NOUN
dad-10670	575	11	(	(	PUNCT
dad-10670	575	12	iwpt’09	iwpt’09	PROPN
dad-10670	575	13	)	)	PUNCT
dad-10670	575	14	,	,	PUNCT
dad-10670	575	15	pages	page	VERB
dad-10670	575	16	81–84	81–84	NOUN
dad-10670	575	17	,	,	PUNCT
dad-10670	575	18	paris	paris	PROPN
dad-10670	575	19	,	,	PUNCT
dad-10670	575	20	france	france	PROPN
dad-10670	575	21	,	,	PUNCT
dad-10670	575	22	october	october	PROPN
dad-10670	575	23	2009	2009	NUM
dad-10670	575	24	.	.	PUNCT
dad-10670	576	1	association	association	NOUN
dad-10670	576	2	for	for	ADP
dad-10670	576	3	computational	computational	ADJ
dad-10670	576	4	linguistics	linguistic	NOUN
dad-10670	576	5	.	.	PUNCT
dad-10670	577	1	frank	frank	PROPN
dad-10670	577	2	schilder	schilder	PROPN
dad-10670	577	3	.	.	PUNCT
dad-10670	578	1	robust	robust	ADJ
dad-10670	578	2	discourse	discourse	NOUN
dad-10670	578	3	parsing	parsing	NOUN
dad-10670	578	4	via	via	ADP
dad-10670	578	5	discourse	discourse	NOUN
dad-10670	578	6	markers	marker	NOUN
dad-10670	578	7	,	,	PUNCT
dad-10670	578	8	topicality	topicality	NOUN
dad-10670	578	9	and	and	CCONJ
dad-10670	578	10	position	position	NOUN
dad-10670	578	11	.	.	PUNCT
dad-10670	579	1	natural	natural	ADJ
dad-10670	579	2	language	language	NOUN
dad-10670	579	3	engineering	engineering	NOUN
dad-10670	579	4	,	,	PUNCT
dad-10670	579	5	8(2	8(2	NUM
dad-10670	579	6	-	-	SYM
dad-10670	579	7	3):235–255	3):235–255	NUM
dad-10670	579	8	,	,	PUNCT
dad-10670	579	9	2002	2002	NUM
dad-10670	579	10	.	.	PUNCT
dad-10670	580	1	radu	radu	PROPN
dad-10670	580	2	soricut	soricut	VERB
dad-10670	580	3	and	and	CCONJ
dad-10670	580	4	daniel	daniel	PROPN
dad-10670	580	5	marcu	marcu	PROPN
dad-10670	580	6	.	.	PUNCT
dad-10670	581	1	sentence	sentence	NOUN
dad-10670	581	2	level	level	NOUN
dad-10670	581	3	discourse	discourse	NOUN
dad-10670	581	4	parsing	parsing	NOUN
dad-10670	581	5	using	use	VERB
dad-10670	581	6	syntactic	syntactic	ADJ
dad-10670	581	7	and	and	CCONJ
dad-10670	581	8	lexical	lexical	ADJ
dad-10670	581	9	information	information	NOUN
dad-10670	581	10	.	.	PUNCT
dad-10670	582	1	proceedings	proceeding	NOUN
dad-10670	582	2	of	of	ADP
dad-10670	582	3	the	the	DET
dad-10670	582	4	2003	2003	NUM
dad-10670	582	5	conference	conference	NOUN
dad-10670	582	6	of	of	ADP
dad-10670	582	7	the	the	DET
dad-10670	582	8	north	north	ADJ
dad-10670	582	9	american	american	ADJ
dad-10670	582	10	chapter	chapter	NOUN
dad-10670	582	11	of	of	ADP
dad-10670	582	12	the	the	DET
dad-10670	582	13	association	association	NOUN
dad-10670	582	14	for	for	ADP
dad-10670	582	15	computational	computational	ADJ
dad-10670	582	16	linguistics	linguistic	NOUN
dad-10670	582	17	on	on	ADP
dad-10670	582	18	human	human	ADJ
dad-10670	582	19	language	language	NOUN
dad-10670	582	20	technology	technology	NOUN
dad-10670	582	21	,	,	PUNCT
dad-10670	582	22	1:149–156	1:149–156	PROPN
dad-10670	582	23	,	,	PUNCT
dad-10670	582	24	2003	2003	NUM
dad-10670	582	25	.	.	PUNCT
dad-10670	583	1	carl	carl	PROPN
dad-10670	583	2	staelin	staelin	PROPN
dad-10670	583	3	.	.	PUNCT
dad-10670	584	1	parameter	parameter	PROPN
dad-10670	584	2	selection	selection	NOUN
dad-10670	584	3	for	for	ADP
dad-10670	584	4	support	support	NOUN
dad-10670	584	5	vector	vector	NOUN
dad-10670	584	6	machines	machine	NOUN
dad-10670	584	7	.	.	PUNCT
dad-10670	585	1	hewlett	hewlett	PROPN
dad-10670	585	2	-	-	PUNCT
dad-10670	585	3	packard	packard	PROPN
dad-10670	585	4	company	company	NOUN
dad-10670	585	5	,	,	PUNCT
dad-10670	585	6	tech	tech	NOUN
dad-10670	585	7	.	.	PUNCT
dad-10670	586	1	rep	rep	PROPN
dad-10670	586	2	.	.	PROPN
dad-10670	586	3	hpl-2002	hpl-2002	NOUN
dad-10670	586	4	-	-	PUNCT
dad-10670	586	5	354r1	354r1	NUM
dad-10670	586	6	,	,	PUNCT
dad-10670	586	7	2003	2003	NUM
dad-10670	586	8	.	.	PUNCT
dad-10670	587	1	svetlana	svetlana	PROPN
dad-10670	587	2	stoyanchev	stoyanchev	PROPN
dad-10670	587	3	and	and	CCONJ
dad-10670	587	4	paul	paul	PROPN
dad-10670	587	5	piwek	piwek	PROPN
dad-10670	587	6	.	.	PUNCT
dad-10670	588	1	constructing	construct	VERB
dad-10670	588	2	the	the	DET
dad-10670	588	3	coda	coda	ADJ
dad-10670	588	4	corpus	corpus	NOUN
dad-10670	588	5	:	:	PUNCT
dad-10670	588	6	a	a	DET
dad-10670	588	7	parallel	parallel	ADJ
dad-10670	588	8	corpus	corpus	NOUN
dad-10670	588	9	of	of	ADP
dad-10670	588	10	monologues	monologue	NOUN
dad-10670	588	11	and	and	CCONJ
dad-10670	588	12	expository	expository	NOUN
dad-10670	588	13	dialogues	dialogue	NOUN
dad-10670	588	14	.	.	PUNCT
dad-10670	589	1	in	in	ADP
dad-10670	589	2	proceedings	proceeding	NOUN
dad-10670	589	3	of	of	ADP
dad-10670	589	4	the	the	DET
dad-10670	589	5	seventh	seventh	ADJ
dad-10670	589	6	conference	conference	NOUN
dad-10670	589	7	on	on	ADP
dad-10670	589	8	international	international	ADJ
dad-10670	589	9	language	language	NOUN
dad-10670	589	10	resources	resource	NOUN
dad-10670	589	11	and	and	CCONJ
dad-10670	589	12	evaluation	evaluation	NOUN
dad-10670	589	13	(	(	PUNCT
dad-10670	589	14	lrec’10	lrec’10	PROPN
dad-10670	589	15	)	)	PUNCT
dad-10670	589	16	,	,	PUNCT
dad-10670	589	17	valletta	valletta	PROPN
dad-10670	589	18	,	,	PUNCT
dad-10670	589	19	malta	malta	PROPN
dad-10670	589	20	,	,	PUNCT
dad-10670	589	21	may	may	PROPN
dad-10670	589	22	2010	2010	NUM
dad-10670	589	23	.	.	PUNCT
dad-10670	590	1	european	european	ADJ
dad-10670	590	2	language	language	PROPN
dad-10670	590	3	resources	resource	NOUN
dad-10670	590	4	association	association	PROPN
dad-10670	590	5	.	.	PUNCT
dad-10670	591	1	rajen	rajen	PROPN
dad-10670	591	2	subba	subba	PROPN
dad-10670	591	3	and	and	CCONJ
dad-10670	591	4	barbara	barbara	PROPN
dad-10670	591	5	di	di	PROPN
dad-10670	591	6	eugenio	eugenio	PROPN
dad-10670	591	7	.	.	PUNCT
dad-10670	592	1	automatic	automatic	ADJ
dad-10670	592	2	discourse	discourse	NOUN
dad-10670	592	3	segmentation	segmentation	NOUN
dad-10670	592	4	using	use	VERB
dad-10670	592	5	neural	neural	ADJ
dad-10670	592	6	networks	network	NOUN
dad-10670	592	7	.	.	PUNCT
dad-10670	593	1	in	in	ADP
dad-10670	593	2	proceedings	proceeding	NOUN
dad-10670	593	3	of	of	ADP
dad-10670	593	4	11th	11th	ADJ
dad-10670	593	5	workshop	workshop	NOUN
dad-10670	593	6	on	on	ADP
dad-10670	593	7	the	the	DET
dad-10670	593	8	semantics	semantic	NOUN
dad-10670	593	9	and	and	CCONJ
dad-10670	593	10	pragmatics	pragmatic	NOUN
dad-10670	593	11	of	of	ADP
dad-10670	593	12	dialogue	dialogue	NOUN
dad-10670	593	13	,	,	PUNCT
dad-10670	593	14	pages	page	NOUN
dad-10670	593	15	189–190	189–190	NUM
dad-10670	593	16	,	,	PUNCT
dad-10670	593	17	trento	trento	PROPN
dad-10670	593	18	,	,	PUNCT
dad-10670	593	19	italy	italy	PROPN
dad-10670	593	20	,	,	PUNCT
dad-10670	593	21	2007	2007	NUM
dad-10670	593	22	.	.	PUNCT
dad-10670	594	1	rajen	rajen	PROPN
dad-10670	594	2	subba	subba	PROPN
dad-10670	594	3	and	and	CCONJ
dad-10670	594	4	barbara	barbara	PROPN
dad-10670	594	5	di	di	PROPN
dad-10670	594	6	eugenio	eugenio	PROPN
dad-10670	594	7	.	.	PUNCT
dad-10670	595	1	an	an	DET
dad-10670	595	2	effective	effective	ADJ
dad-10670	595	3	discourse	discourse	NOUN
dad-10670	595	4	parser	parser	NOUN
dad-10670	595	5	that	that	PRON
dad-10670	595	6	uses	use	VERB
dad-10670	595	7	rich	rich	ADJ
dad-10670	595	8	linguistic	linguistic	ADJ
dad-10670	595	9	information	information	NOUN
dad-10670	595	10	.	.	PUNCT
dad-10670	596	1	in	in	ADP
dad-10670	596	2	proceedings	proceeding	NOUN
dad-10670	596	3	of	of	ADP
dad-10670	596	4	human	human	ADJ
dad-10670	596	5	language	language	NOUN
dad-10670	596	6	technologies	technology	NOUN
dad-10670	596	7	:	:	PUNCT
dad-10670	596	8	the	the	DET
dad-10670	596	9	2009	2009	NUM
dad-10670	596	10	annual	annual	ADJ
dad-10670	596	11	conference	conference	NOUN
dad-10670	596	12	of	of	ADP
dad-10670	596	13	29	29	NUM
dad-10670	596	14	hernault	hernault	ADJ
dad-10670	596	15	,	,	PUNCT
dad-10670	596	16	prendinger	prendinger	NOUN
dad-10670	596	17	,	,	PUNCT
dad-10670	596	18	duverle	duverle	NOUN
dad-10670	596	19	,	,	PUNCT
dad-10670	596	20	and	and	CCONJ
dad-10670	596	21	ishizuka	ishizuka	VERB
dad-10670	596	22	the	the	DET
dad-10670	596	23	north	north	ADJ
dad-10670	596	24	american	american	ADJ
dad-10670	596	25	chapter	chapter	NOUN
dad-10670	596	26	of	of	ADP
dad-10670	596	27	the	the	DET
dad-10670	596	28	association	association	NOUN
dad-10670	596	29	for	for	ADP
dad-10670	596	30	computational	computational	ADJ
dad-10670	596	31	linguistics	linguistic	NOUN
dad-10670	596	32	,	,	PUNCT
dad-10670	596	33	pages	page	NOUN
dad-10670	596	34	566–574	566–574	NUM
dad-10670	596	35	,	,	PUNCT
dad-10670	596	36	boulder	boulder	NOUN
dad-10670	596	37	,	,	PUNCT
dad-10670	596	38	colorado	colorado	PROPN
dad-10670	596	39	,	,	PUNCT
dad-10670	596	40	june	june	PROPN
dad-10670	596	41	2009	2009	NUM
dad-10670	596	42	.	.	PUNCT
dad-10670	597	1	association	association	NOUN
dad-10670	597	2	for	for	ADP
dad-10670	597	3	computational	computational	ADJ
dad-10670	597	4	linguistics	linguistic	NOUN
dad-10670	597	5	.	.	PUNCT
dad-10670	598	1	maite	maite	PROPN
dad-10670	598	2	taboada	taboada	PROPN
dad-10670	598	3	.	.	PUNCT
dad-10670	599	1	discourse	discourse	PROPN
dad-10670	599	2	markers	marker	NOUN
dad-10670	599	3	as	as	ADP
dad-10670	599	4	signals	signal	NOUN
dad-10670	599	5	(	(	PUNCT
dad-10670	599	6	or	or	CCONJ
dad-10670	599	7	not	not	PART
dad-10670	599	8	)	)	PUNCT
dad-10670	599	9	of	of	ADP
dad-10670	599	10	rhetorical	rhetorical	ADJ
dad-10670	599	11	relations	relation	NOUN
dad-10670	599	12	.	.	PUNCT
dad-10670	600	1	journal	journal	NOUN
dad-10670	600	2	of	of	ADP
dad-10670	600	3	pragmatics	pragmatic	NOUN
dad-10670	600	4	,	,	PUNCT
dad-10670	600	5	38(4):567–592	38(4):567–592	NUM
dad-10670	600	6	,	,	PUNCT
dad-10670	600	7	2006	2006	NUM
dad-10670	600	8	.	.	PUNCT
dad-10670	601	1	milan	milan	PROPN
dad-10670	601	2	tofiloski	tofiloski	PROPN
dad-10670	601	3	,	,	PUNCT
dad-10670	601	4	julian	julian	PROPN
dad-10670	601	5	brooke	brooke	PROPN
dad-10670	601	6	,	,	PUNCT
dad-10670	601	7	and	and	CCONJ
dad-10670	601	8	maite	maite	PROPN
dad-10670	601	9	taboada	taboada	PROPN
dad-10670	601	10	.	.	PUNCT
dad-10670	602	1	a	a	DET
dad-10670	602	2	syntactic	syntactic	ADJ
dad-10670	602	3	and	and	CCONJ
dad-10670	602	4	lexical	lexical	ADJ
dad-10670	602	5	-	-	PUNCT
dad-10670	602	6	based	base	VERB
dad-10670	602	7	discourse	discourse	NOUN
dad-10670	602	8	segmenter	segmenter	NOUN
dad-10670	602	9	.	.	PUNCT
dad-10670	603	1	in	in	ADP
dad-10670	603	2	acl	acl	PROPN
dad-10670	603	3	’	'	PUNCT
dad-10670	603	4	09	09	NUM
dad-10670	603	5	,	,	PUNCT
dad-10670	603	6	pages	page	NOUN
dad-10670	603	7	77–80	77–80	NUM
dad-10670	603	8	,	,	PUNCT
dad-10670	603	9	suntec	suntec	ADJ
dad-10670	603	10	,	,	PUNCT
dad-10670	603	11	singapore	singapore	PROPN
dad-10670	603	12	,	,	PUNCT
dad-10670	603	13	august	august	PROPN
dad-10670	603	14	2009	2009	NUM
dad-10670	603	15	.	.	PUNCT
dad-10670	604	1	association	association	NOUN
dad-10670	604	2	for	for	ADP
dad-10670	604	3	computational	computational	ADJ
dad-10670	604	4	linguistics	linguistic	NOUN
dad-10670	604	5	.	.	PUNCT
dad-10670	605	1	vladimir	vladimir	PROPN
dad-10670	605	2	n.	n.	PROPN
dad-10670	605	3	vapnik	vapnik	PROPN
dad-10670	605	4	.	.	PUNCT
dad-10670	606	1	the	the	DET
dad-10670	606	2	nature	nature	NOUN
dad-10670	606	3	of	of	ADP
dad-10670	606	4	statistical	statistical	ADJ
dad-10670	606	5	learning	learning	NOUN
dad-10670	606	6	theory	theory	NOUN
dad-10670	606	7	.	.	PUNCT
dad-10670	607	1	springer	springer	NOUN
dad-10670	607	2	-	-	PUNCT
dad-10670	607	3	verlag	verlag	PROPN
dad-10670	607	4	new	new	PROPN
dad-10670	607	5	york	york	PROPN
dad-10670	607	6	,	,	PUNCT
dad-10670	607	7	inc	inc	PROPN
dad-10670	607	8	.	.	PROPN
dad-10670	607	9	,	,	PUNCT
dad-10670	607	10	new	new	PROPN
dad-10670	607	11	york	york	PROPN
dad-10670	607	12	,	,	PUNCT
dad-10670	607	13	ny	ny	PROPN
dad-10670	607	14	,	,	PUNCT
dad-10670	607	15	usa	usa	PROPN
dad-10670	607	16	,	,	PUNCT
dad-10670	607	17	1995	1995	NUM
dad-10670	607	18	.	.	PUNCT
dad-10670	608	1	bonnie	bonnie	PROPN
dad-10670	608	2	webber	webber	PROPN
dad-10670	608	3	.	.	PUNCT
dad-10670	609	1	genre	genre	NOUN
dad-10670	609	2	distinctions	distinction	NOUN
dad-10670	609	3	for	for	ADP
dad-10670	609	4	discourse	discourse	NOUN
dad-10670	609	5	in	in	ADP
dad-10670	609	6	the	the	DET
dad-10670	609	7	penn	penn	PROPN
dad-10670	609	8	treebank	treebank	PROPN
dad-10670	609	9	.	.	PUNCT
dad-10670	610	1	in	in	ADP
dad-10670	610	2	proceedings	proceeding	NOUN
dad-10670	610	3	of	of	ADP
dad-10670	610	4	the	the	DET
dad-10670	610	5	joint	joint	ADJ
dad-10670	610	6	conference	conference	NOUN
dad-10670	610	7	of	of	ADP
dad-10670	610	8	the	the	DET
dad-10670	610	9	47th	47th	ADJ
dad-10670	610	10	annual	annual	ADJ
dad-10670	610	11	meeting	meeting	NOUN
dad-10670	610	12	of	of	ADP
dad-10670	610	13	the	the	DET
dad-10670	610	14	acl	acl	PROPN
dad-10670	610	15	and	and	CCONJ
dad-10670	610	16	the	the	DET
dad-10670	610	17	4th	4th	ADJ
dad-10670	610	18	international	international	ADJ
dad-10670	610	19	joint	joint	ADJ
dad-10670	610	20	conference	conference	NOUN
dad-10670	610	21	on	on	ADP
dad-10670	610	22	natural	natural	ADJ
dad-10670	610	23	language	language	NOUN
dad-10670	610	24	processing	processing	NOUN
dad-10670	610	25	of	of	ADP
dad-10670	610	26	the	the	DET
dad-10670	610	27	afnlp	afnlp	NOUN
dad-10670	610	28	,	,	PUNCT
dad-10670	610	29	pages	page	NOUN
dad-10670	610	30	674–682	674–682	NUM
dad-10670	610	31	,	,	PUNCT
dad-10670	610	32	suntec	suntec	ADJ
dad-10670	610	33	,	,	PUNCT
dad-10670	610	34	singapore	singapore	PROPN
dad-10670	610	35	,	,	PUNCT
dad-10670	610	36	august	august	PROPN
dad-10670	610	37	2009	2009	NUM
dad-10670	610	38	.	.	PUNCT
dad-10670	611	1	association	association	NOUN
dad-10670	611	2	for	for	ADP
dad-10670	611	3	computational	computational	ADJ
dad-10670	611	4	linguistics	linguistic	NOUN
dad-10670	611	5	.	.	PUNCT
dad-10670	612	1	30	30	NUM
dad-10670	612	2	hilda	hilda	VERB
dad-10670	612	3	:	:	PUNCT
dad-10670	612	4	a	a	DET
dad-10670	612	5	discourse	discourse	NOUN
dad-10670	612	6	parser	parser	NOUN
dad-10670	612	7	using	use	VERB
dad-10670	612	8	support	support	NOUN
dad-10670	612	9	vector	vector	NOUN
dad-10670	612	10	machine	machine	NOUN
dad-10670	612	11	classification	classification	NOUN
dad-10670	612	12	s	s	PART
dad-10670	612	13	np	np	INTJ
dad-10670	612	14	nn	nn	INTJ
dad-10670	612	15	fa	fa	PROPN
dad-10670	612	16	rm	rm	PROPN
dad-10670	612	17	nn	nn	PROPN
dad-10670	612	18	le	le	PROPN
dad-10670	612	19	nd	nd	PROPN
dad-10670	612	20	in	in	ADP
dad-10670	612	21	g	g	PROPN
dad-10670	612	22	vp	vp	PROPN
dad-10670	612	23	vb	vb	PROPN
dad-10670	612	24	d	d	X
dad-10670	612	25	w	w	PROPN
dad-10670	612	26	as	as	ADP
dad-10670	612	27	vp	vp	PROPN
dad-10670	612	28	vb	vb	PROPN
dad-10670	612	29	n	n	PROPN
dad-10670	612	30	en	en	X
dad-10670	612	31	ac	ac	PROPN
dad-10670	613	1	te	te	PROPN
dad-10670	614	1	d	d	X
dad-10670	614	2	s	s	PROPN
dad-10670	614	3	vp	vp	NOUN
dad-10670	614	4	to	to	PART
dad-10670	614	5	to	to	ADP
dad-10670	614	6	vp	vp	PROPN
dad-10670	614	7	vb	vb	PROPN
dad-10670	614	8	co	co	X
dad-10670	614	9	rre	rre	PROPN
dad-10670	615	1	ct	ct	INTJ
dad-10670	615	2	np	np	INTJ
dad-10670	616	1	dt	dt	X
dad-10670	616	2	th	th	X
dad-10670	616	3	is	be	AUX
dad-10670	616	4	nn	nn	INTJ
dad-10670	616	5	pr	pr	NOUN
dad-10670	616	6	ob	ob	INTJ
dad-10670	616	7	le	le	X
dad-10670	616	8	m	m	VERB
dad-10670	616	9	pp	pp	ADV
dad-10670	616	10	in	in	ADP
dad-10670	616	11	by	by	ADP
dad-10670	616	12	s	s	PROPN
dad-10670	616	13	vp	vp	PROPN
dad-10670	616	14	vb	vb	PROPN
dad-10670	616	15	g	g	PROPN
dad-10670	616	16	pr	pr	INTJ
dad-10670	617	1	ov	ov	INTJ
dad-10670	617	2	i	i	PROPN
dad-10670	617	3	d	d	PROPN
dad-10670	617	4	in	in	ADP
dad-10670	617	5	g	g	PROPN
dad-10670	617	6	np	np	INTJ
dad-10670	617	7	np	np	INTJ
dad-10670	617	8	dt	dt	PROPN
dad-10670	617	9	a	a	DET
dad-10670	617	10	jj	jj	X
dad-10670	617	11	re	re	X
dad-10670	617	12	lia	lia	PROPN
dad-10670	617	13	bl	bl	PROPN
dad-10670	617	14	e	e	PROPN
dad-10670	617	15	nn	nn	PROPN
dad-10670	617	16	flo	flo	PROPN
dad-10670	618	1	w	w	ADV
dad-10670	618	2	pp	pp	ADV
dad-10670	618	3	in	in	ADP
dad-10670	618	4	of	of	ADP
dad-10670	618	5	np	np	INTJ
dad-10670	618	6	jj	jj	PROPN
dad-10670	618	7	le	le	PROPN
dad-10670	618	8	nd	nd	PROPN
dad-10670	618	9	ab	ab	PROPN
dad-10670	618	10	le	le	X
dad-10670	618	11	nn	nn	PROPN
dad-10670	618	12	s	s	PROPN
dad-10670	618	13	fu	fu	PROPN
dad-10670	618	14	nd	nd	NOUN
dad-10670	618	15	s	s	PROPN
dad-10670	618	16	.	.	PUNCT
dad-10670	618	17	.	.	PUNCT
dad-10670	619	1	(	(	PUNCT
dad-10670	619	2	e	e	NOUN
dad-10670	619	3	na	na	ADP
dad-10670	619	4	ct	ct	PROPN
dad-10670	619	5	ed	ed	NOUN
dad-10670	619	6	)	)	PUNCT
dad-10670	619	7	(	(	PUNCT
dad-10670	619	8	e	e	NOUN
dad-10670	619	9	na	na	ADP
dad-10670	619	10	ct	ct	PROPN
dad-10670	619	11	ed	ed	NOUN
dad-10670	619	12	)	)	PUNCT
dad-10670	619	13	(	(	PUNCT
dad-10670	619	14	c	c	NOUN
dad-10670	619	15	or	or	CCONJ
dad-10670	619	16	re	re	NOUN
dad-10670	619	17	ct	ct	NUM
dad-10670	619	18	)	)	PUNCT
dad-10670	619	19	(	(	PUNCT
dad-10670	619	20	w	w	ADP
dad-10670	619	21	as	as	ADP
dad-10670	619	22	)	)	PUNCT
dad-10670	619	23	(	(	PUNCT
dad-10670	619	24	w	w	ADP
dad-10670	619	25	as	as	ADP
dad-10670	619	26	)	)	PUNCT
dad-10670	619	27	(	(	PUNCT
dad-10670	619	28	p	p	X
dad-10670	619	29	ro	ro	INTJ
dad-10670	619	30	bl	bl	INTJ
dad-10670	619	31	em	em	PRON
dad-10670	619	32	)	)	PUNCT
dad-10670	619	33	(	(	PUNCT
dad-10670	619	34	c	c	NOUN
dad-10670	619	35	or	or	CCONJ
dad-10670	619	36	re	re	NOUN
dad-10670	619	37	ct	ct	NUM
dad-10670	619	38	)	)	PUNCT
dad-10670	619	39	(	(	PUNCT
dad-10670	619	40	b	b	X
dad-10670	619	41	y	y	NOUN
dad-10670	619	42	)	)	PUNCT
dad-10670	619	43	fi	fi	NOUN
dad-10670	619	44	gu	gu	NOUN
dad-10670	619	45	re	re	ADP
dad-10670	619	46	7	7	NUM
dad-10670	619	47	:	:	PUNCT
dad-10670	619	48	pa	pa	PROPN
dad-10670	619	49	rt	rt	PROPN
dad-10670	619	50	ia	ia	PROPN
dad-10670	619	51	lly	lly	ADV
dad-10670	619	52	-l	-l	PRON
dad-10670	619	53	ex	ex	INTJ
dad-10670	619	54	ic	ic	PROPN
dad-10670	619	55	al	al	PROPN
dad-10670	619	56	iz	iz	INTJ
dad-10670	620	1	ed	ed	NOUN
dad-10670	620	2	sy	sy	INTJ
dad-10670	620	3	nt	not	PART
dad-10670	620	4	ax	ax	NOUN
dad-10670	620	5	tr	tr	ADV
dad-10670	620	6	ee	ee	NOUN
dad-10670	620	7	(	(	PUNCT
dad-10670	620	8	l	l	X
dad-10670	620	9	ex	ex	X
dad-10670	621	1	ic	ic	PROPN
dad-10670	621	2	al	al	PROPN
dad-10670	621	3	he	he	PRON
dad-10670	621	4	ad	ad	NOUN
dad-10670	621	5	s	s	PART
dad-10670	621	6	ar	ar	PROPN
dad-10670	621	7	e	e	PROPN
dad-10670	621	8	in	in	ADP
dad-10670	621	9	di	di	X
dad-10670	621	10	ca	ca	NOUN
dad-10670	621	11	te	te	PROPN
dad-10670	621	12	d	d	NOUN
dad-10670	621	13	be	be	AUX
dad-10670	621	14	tw	tw	INTJ
dad-10670	621	15	ee	ee	ADP
dad-10670	621	16	n	n	PROPN
dad-10670	621	17	pa	pa	PROPN
dad-10670	621	18	re	re	PROPN
dad-10670	621	19	nt	not	PART
dad-10670	621	20	he	he	PRON
dad-10670	621	21	se	se	PROPN
dad-10670	621	22	s	s	VERB
dad-10670	621	23	)	)	PUNCT
dad-10670	621	24	31	31	NUM
dad-10670	621	25	hernault	hernault	ADJ
dad-10670	621	26	,	,	PUNCT
dad-10670	621	27	prendinger	prendinger	NOUN
dad-10670	621	28	,	,	PUNCT
dad-10670	621	29	duverle	duverle	NOUN
dad-10670	621	30	,	,	PUNCT
dad-10670	621	31	and	and	CCONJ
dad-10670	621	32	ishizuka	ishizuka	PROPN
dad-10670	621	33	manner	manner	NOUN
dad-10670	621	34	-	-	PUNCT
dad-10670	621	35	means	means	NOUN
dad-10670	621	36	enablement	enablement	NOUN
dad-10670	621	37	1a	1a	NOUN
dad-10670	621	38	1b	1b	NUM
dad-10670	621	39	1c	1c	NUM
dad-10670	621	40	enablement	enablement	PROPN
dad-10670	621	41	1a	1a	NOUN
dad-10670	621	42	manner	manner	NOUN
dad-10670	621	43	-	-	PUNCT
dad-10670	621	44	means	mean	VERB
dad-10670	621	45	1b	1b	NUM
dad-10670	621	46	1c	1c	NUM
dad-10670	621	47	figure	figure	NOUN
dad-10670	621	48	8	8	NUM
dad-10670	621	49	:	:	PUNCT
dad-10670	621	50	two	two	NUM
dad-10670	621	51	possible	possible	ADJ
dad-10670	621	52	attachments	attachment	NOUN
dad-10670	621	53	in	in	ADP
dad-10670	621	54	the	the	DET
dad-10670	621	55	rs	rs	NOUN
dad-10670	621	56	-	-	PUNCT
dad-10670	621	57	tree	tree	NOUN
dad-10670	621	58	of	of	ADP
dad-10670	621	59	the	the	DET
dad-10670	621	60	text	text	NOUN
dad-10670	621	61	in	in	ADP
dad-10670	621	62	figure	figure	NOUN
dad-10670	621	63	1	1	NUM
dad-10670	621	64	.	.	X
dad-10670	621	65	32	32	NUM
dad-10670	621	66	hilda	hilda	VERB
dad-10670	621	67	:	:	PUNCT
dad-10670	621	68	a	a	DET
dad-10670	621	69	discourse	discourse	NOUN
dad-10670	621	70	parser	parser	NOUN
dad-10670	621	71	using	use	VERB
dad-10670	621	72	support	support	NOUN
dad-10670	621	73	vector	vector	NOUN
dad-10670	621	74	machine	machine	NOUN
dad-10670	621	75	classification	classification	NOUN
dad-10670	621	76	s	s	PART
dad-10670	621	77	np	np	INTJ
dad-10670	621	78	nn	nn	INTJ
dad-10670	621	79	fa	fa	PROPN
dad-10670	621	80	rm	rm	PROPN
dad-10670	621	81	nn	nn	PROPN
dad-10670	621	82	le	le	PROPN
dad-10670	621	83	nd	nd	PROPN
dad-10670	621	84	in	in	ADP
dad-10670	621	85	g	g	PROPN
dad-10670	621	86	vp	vp	PROPN
dad-10670	621	87	vb	vb	PROPN
dad-10670	621	88	d	d	X
dad-10670	621	89	w	w	PROPN
dad-10670	621	90	as	as	ADP
dad-10670	621	91	vp	vp	PROPN
dad-10670	621	92	vb	vb	PROPN
dad-10670	621	93	n	n	PROPN
dad-10670	621	94	en	en	X
dad-10670	621	95	ac	ac	PROPN
dad-10670	622	1	te	te	PROPN
dad-10670	623	1	d	d	X
dad-10670	623	2	s	s	PROPN
dad-10670	623	3	vp	vp	NOUN
dad-10670	623	4	to	to	PART
dad-10670	623	5	to	to	ADP
dad-10670	623	6	vp	vp	PROPN
dad-10670	623	7	vb	vb	PROPN
dad-10670	623	8	co	co	X
dad-10670	623	9	rre	rre	PROPN
dad-10670	624	1	ct	ct	INTJ
dad-10670	624	2	np	np	INTJ
dad-10670	625	1	dt	dt	X
dad-10670	625	2	th	th	X
dad-10670	625	3	is	be	AUX
dad-10670	625	4	nn	nn	INTJ
dad-10670	625	5	pr	pr	NOUN
dad-10670	625	6	ob	ob	INTJ
dad-10670	625	7	le	le	X
dad-10670	625	8	m	m	VERB
dad-10670	625	9	pp	pp	ADV
dad-10670	625	10	in	in	ADP
dad-10670	625	11	by	by	ADP
dad-10670	625	12	s	s	PROPN
dad-10670	625	13	vp	vp	PROPN
dad-10670	625	14	vb	vb	PROPN
dad-10670	625	15	g	g	PROPN
dad-10670	625	16	pr	pr	INTJ
dad-10670	626	1	ov	ov	INTJ
dad-10670	626	2	i	i	PROPN
dad-10670	626	3	d	d	PROPN
dad-10670	626	4	in	in	ADP
dad-10670	626	5	g	g	PROPN
dad-10670	626	6	np	np	INTJ
dad-10670	626	7	np	np	INTJ
dad-10670	626	8	dt	dt	PROPN
dad-10670	626	9	a	a	DET
dad-10670	626	10	jj	jj	X
dad-10670	626	11	re	re	X
dad-10670	626	12	lia	lia	PROPN
dad-10670	626	13	bl	bl	PROPN
dad-10670	626	14	e	e	PROPN
dad-10670	626	15	nn	nn	PROPN
dad-10670	626	16	flo	flo	PROPN
dad-10670	627	1	w	w	ADV
dad-10670	627	2	pp	pp	ADV
dad-10670	627	3	in	in	ADP
dad-10670	627	4	of	of	ADP
dad-10670	627	5	np	np	INTJ
dad-10670	627	6	jj	jj	PROPN
dad-10670	627	7	le	le	PROPN
dad-10670	627	8	nd	nd	PROPN
dad-10670	627	9	ab	ab	PROPN
dad-10670	627	10	le	le	X
dad-10670	627	11	nn	nn	PROPN
dad-10670	627	12	s	s	PROPN
dad-10670	627	13	fu	fu	PROPN
dad-10670	627	14	nd	nd	NOUN
dad-10670	627	15	s	s	PROPN
dad-10670	627	16	.	.	PUNCT
dad-10670	627	17	.	.	PUNCT
dad-10670	628	1	(	(	PUNCT
dad-10670	628	2	e	e	NOUN
dad-10670	628	3	na	na	ADP
dad-10670	628	4	ct	ct	PROPN
dad-10670	628	5	ed	ed	NOUN
dad-10670	628	6	)	)	PUNCT
dad-10670	628	7	(	(	PUNCT
dad-10670	628	8	c	c	NOUN
dad-10670	628	9	or	or	CCONJ
dad-10670	628	10	re	re	NOUN
dad-10670	628	11	ct	ct	NUM
dad-10670	628	12	)	)	PUNCT
dad-10670	628	13	(	(	PUNCT
dad-10670	628	14	w	w	ADP
dad-10670	628	15	as	as	ADP
dad-10670	628	16	)	)	PUNCT
dad-10670	628	17	(	(	PUNCT
dad-10670	628	18	w	w	ADP
dad-10670	628	19	as	as	ADP
dad-10670	628	20	)	)	PUNCT
dad-10670	628	21	(	(	PUNCT
dad-10670	628	22	c	c	NOUN
dad-10670	628	23	or	or	CCONJ
dad-10670	628	24	re	re	NOUN
dad-10670	628	25	ct	ct	NUM
dad-10670	628	26	)	)	PUNCT
dad-10670	628	27	(	(	PUNCT
dad-10670	628	28	b	b	X
dad-10670	628	29	y	y	X
dad-10670	628	30	)	)	PUNCT
dad-10670	628	31	1a	1a	PROPN
dad-10670	628	32	1b	1b	NUM
dad-10670	628	33	1c	1c	NUM
dad-10670	628	34	fi	fi	X
dad-10670	628	35	gu	gu	NOUN
dad-10670	628	36	re	re	ADP
dad-10670	628	37	9	9	NUM
dad-10670	628	38	:	:	PUNCT
dad-10670	628	39	pa	pa	PROPN
dad-10670	628	40	rt	rt	PROPN
dad-10670	628	41	ia	ia	PROPN
dad-10670	628	42	lly	lly	ADV
dad-10670	628	43	-l	-l	PRON
dad-10670	628	44	ex	ex	INTJ
dad-10670	629	1	ic	ic	PROPN
dad-10670	629	2	al	al	PROPN
dad-10670	629	3	iz	iz	INTJ
dad-10670	630	1	ed	ed	NOUN
dad-10670	630	2	sy	sy	INTJ
dad-10670	630	3	nt	not	PART
dad-10670	630	4	ax	ax	NOUN
dad-10670	630	5	tr	tr	NOUN
dad-10670	630	6	ee	ee	PROPN
dad-10670	630	7	,	,	PUNCT
dad-10670	630	8	s	s	VERB
dad-10670	630	9	ho	ho	PROPN
dad-10670	630	10	w	w	PROPN
dad-10670	630	11	in	in	ADP
dad-10670	630	12	g	g	PROPN
dad-10670	630	13	do	do	VERB
dad-10670	630	14	m	m	PRON
dad-10670	630	15	in	in	ADP
dad-10670	630	16	an	an	DET
dad-10670	630	17	ce	ce	PROPN
dad-10670	630	18	se	se	PROPN
dad-10670	630	19	ts	ts	ADP
dad-10670	630	20	33	33	NUM