id	sid	tid	token	lemma	pos
ajst-13993	1	1	academic	academic	ADJ
ajst-13993	1	2	journal	journal	NOUN
ajst-13993	1	3	of	of	ADP
ajst-13993	1	4	science	science	NOUN
ajst-13993	1	5	and	and	CCONJ
ajst-13993	1	6	technology	technology	NOUN
ajst-13993	1	7	issn	issn	NOUN
ajst-13993	1	8	:	:	PUNCT
ajst-13993	1	9	2771	2771	NUM
ajst-13993	1	10	-	-	SYM
ajst-13993	1	11	3032	3032	NUM
ajst-13993	1	12	|	|	NOUN
ajst-13993	1	13	vol	vol	NOUN
ajst-13993	1	14	.	.	PROPN
ajst-13993	1	15	8	8	NUM
ajst-13993	1	16	,	,	PUNCT
ajst-13993	1	17	no	no	INTJ
ajst-13993	1	18	.	.	NOUN
ajst-13993	1	19	1	1	NUM
ajst-13993	1	20	,	,	PUNCT
ajst-13993	1	21	2023	2023	NUM
ajst-13993	1	22	74	74	NUM
ajst-13993	1	23	the	the	DET
ajst-13993	1	24	source	source	NOUN
ajst-13993	1	25	code	code	NOUN
ajst-13993	1	26	comment	comment	NOUN
ajst-13993	1	27	generation	generation	NOUN
ajst-13993	1	28	based	base	VERB
ajst-13993	1	29	on	on	ADP
ajst-13993	1	30	deep	deep	ADJ
ajst-13993	1	31	reinforcement	reinforcement	NOUN
ajst-13993	1	32	learning	learning	NOUN
ajst-13993	1	33	and	and	CCONJ
ajst-13993	1	34	hierarchical	hierarchical	ADJ
ajst-13993	1	35	attention	attention	NOUN
ajst-13993	1	36	daoyang	daoyang	NOUN
ajst-13993	1	37	ming1	ming1	PROPN
ajst-13993	1	38	,	,	PUNCT
ajst-13993	1	39	weicheng	weicheng	PROPN
ajst-13993	1	40	xiong2	xiong2	PROPN
ajst-13993	1	41	,	,	PUNCT
ajst-13993	1	42	*	*	PUNCT
ajst-13993	2	1	1school	1school	NUM
ajst-13993	2	2	of	of	ADP
ajst-13993	2	3	foreign	foreign	ADJ
ajst-13993	2	4	languages	language	NOUN
ajst-13993	2	5	,	,	PUNCT
ajst-13993	2	6	baoshan	baoshan	PROPN
ajst-13993	2	7	university	university	NOUN
ajst-13993	2	8	,	,	PUNCT
ajst-13993	2	9	bao	bao	PROPN
ajst-13993	2	10	shan	shan	PROPN
ajst-13993	2	11	,	,	PUNCT
ajst-13993	2	12	678000	678000	NUM
ajst-13993	2	13	,	,	PUNCT
ajst-13993	2	14	china	china	PROPN
ajst-13993	2	15	2school	2school	NUM
ajst-13993	2	16	of	of	ADP
ajst-13993	2	17	mathematics	mathematic	NOUN
ajst-13993	2	18	and	and	CCONJ
ajst-13993	2	19	big	big	ADJ
ajst-13993	2	20	data	datum	NOUN
ajst-13993	2	21	,	,	PUNCT
ajst-13993	2	22	guizhou	guizhou	PROPN
ajst-13993	2	23	normal	normal	ADJ
ajst-13993	2	24	university	university	PROPN
ajst-13993	2	25	,	,	PUNCT
ajst-13993	2	26	gui	gui	PROPN
ajst-13993	2	27	yang	yang	PROPN
ajst-13993	2	28	,	,	PUNCT
ajst-13993	2	29	550018	550018	NUM
ajst-13993	2	30	,	,	PUNCT
ajst-13993	2	31	china	china	PROPN
ajst-13993	2	32	*	*	PUNCT
ajst-13993	2	33	corresponding	correspond	VERB
ajst-13993	2	34	author	author	NOUN
ajst-13993	2	35	abstract	abstract	NOUN
ajst-13993	2	36	:	:	PUNCT
ajst-13993	2	37	code	code	NOUN
ajst-13993	2	38	summarization	summarization	NOUN
ajst-13993	2	39	provides	provide	VERB
ajst-13993	2	40	the	the	DET
ajst-13993	2	41	main	main	ADJ
ajst-13993	2	42	aim	aim	NOUN
ajst-13993	2	43	described	describe	VERB
ajst-13993	2	44	in	in	ADP
ajst-13993	2	45	natural	natural	ADJ
ajst-13993	2	46	language	language	NOUN
ajst-13993	2	47	of	of	ADP
ajst-13993	2	48	the	the	DET
ajst-13993	2	49	given	give	VERB
ajst-13993	2	50	function	function	NOUN
ajst-13993	2	51	,	,	PUNCT
ajst-13993	2	52	it	it	PRON
ajst-13993	2	53	can	can	AUX
ajst-13993	2	54	benefit	benefit	VERB
ajst-13993	2	55	many	many	ADJ
ajst-13993	2	56	tasks	task	NOUN
ajst-13993	2	57	in	in	ADP
ajst-13993	2	58	software	software	NOUN
ajst-13993	2	59	engineering	engineering	NOUN
ajst-13993	2	60	.	.	PUNCT
ajst-13993	3	1	due	due	ADP
ajst-13993	3	2	to	to	ADP
ajst-13993	3	3	the	the	DET
ajst-13993	3	4	special	special	ADJ
ajst-13993	3	5	grammar	grammar	NOUN
ajst-13993	3	6	and	and	CCONJ
ajst-13993	3	7	syntax	syntax	NOUN
ajst-13993	3	8	structure	structure	NOUN
ajst-13993	3	9	of	of	ADP
ajst-13993	3	10	programming	programming	NOUN
ajst-13993	3	11	languages	language	NOUN
ajst-13993	3	12	and	and	CCONJ
ajst-13993	3	13	various	various	ADJ
ajst-13993	3	14	shortcomings	shortcoming	NOUN
ajst-13993	3	15	of	of	ADP
ajst-13993	3	16	different	different	ADJ
ajst-13993	3	17	deep	deep	ADJ
ajst-13993	3	18	neural	neural	ADJ
ajst-13993	3	19	networks	network	NOUN
ajst-13993	3	20	,	,	PUNCT
ajst-13993	3	21	the	the	DET
ajst-13993	3	22	accuracy	accuracy	NOUN
ajst-13993	3	23	of	of	ADP
ajst-13993	3	24	existing	exist	VERB
ajst-13993	3	25	code	code	NOUN
ajst-13993	3	26	summarization	summarization	NOUN
ajst-13993	3	27	approaches	approach	NOUN
ajst-13993	3	28	is	be	AUX
ajst-13993	3	29	not	not	PART
ajst-13993	3	30	good	good	ADJ
ajst-13993	3	31	enough	enough	ADV
ajst-13993	3	32	.	.	PUNCT
ajst-13993	4	1	this	this	DET
ajst-13993	4	2	work	work	NOUN
ajst-13993	4	3	proposes	propose	VERB
ajst-13993	4	4	to	to	PART
ajst-13993	4	5	adopt	adopt	VERB
ajst-13993	4	6	the	the	DET
ajst-13993	4	7	hierarchical	hierarchical	ADJ
ajst-13993	4	8	attention	attention	NOUN
ajst-13993	4	9	mechanism	mechanism	NOUN
ajst-13993	4	10	to	to	PART
ajst-13993	4	11	enable	enable	VERB
ajst-13993	4	12	the	the	DET
ajst-13993	4	13	code	code	NOUN
ajst-13993	4	14	summarization	summarization	NOUN
ajst-13993	4	15	framework	framework	NOUN
ajst-13993	4	16	to	to	PART
ajst-13993	4	17	translate	translate	VERB
ajst-13993	4	18	three	three	NUM
ajst-13993	4	19	representations	representation	NOUN
ajst-13993	4	20	of	of	ADP
ajst-13993	4	21	source	source	NOUN
ajst-13993	4	22	code	code	NOUN
ajst-13993	4	23	to	to	ADP
ajst-13993	4	24	the	the	DET
ajst-13993	4	25	hidden	hide	VERB
ajst-13993	4	26	spce	spce	NOUN
ajst-13993	4	27	and	and	CCONJ
ajst-13993	4	28	then	then	ADV
ajst-13993	4	29	it	it	PRON
ajst-13993	4	30	injects	inject	VERB
ajst-13993	4	31	them	they	PRON
ajst-13993	4	32	into	into	ADP
ajst-13993	4	33	a	a	DET
ajst-13993	4	34	deep	deep	ADJ
ajst-13993	4	35	reinforcement	reinforcement	NOUN
ajst-13993	4	36	learning	learning	NOUN
ajst-13993	4	37	model	model	NOUN
ajst-13993	4	38	to	to	PART
ajst-13993	4	39	enhance	enhance	VERB
ajst-13993	4	40	the	the	DET
ajst-13993	4	41	performance	performance	NOUN
ajst-13993	4	42	of	of	ADP
ajst-13993	4	43	code	code	NOUN
ajst-13993	4	44	summarization	summarization	NOUN
ajst-13993	4	45	.	.	PUNCT
ajst-13993	5	1	we	we	PRON
ajst-13993	5	2	conduct	conduct	VERB
ajst-13993	5	3	a	a	DET
ajst-13993	5	4	few	few	ADJ
ajst-13993	5	5	of	of	ADP
ajst-13993	5	6	experiments	experiment	NOUN
ajst-13993	5	7	,	,	PUNCT
ajst-13993	5	8	and	and	CCONJ
ajst-13993	5	9	the	the	DET
ajst-13993	5	10	results	result	NOUN
ajst-13993	5	11	of	of	ADP
ajst-13993	5	12	which	which	PRON
ajst-13993	5	13	prove	prove	VERB
ajst-13993	5	14	that	that	SCONJ
ajst-13993	5	15	the	the	DET
ajst-13993	5	16	proposed	propose	VERB
ajst-13993	5	17	approaches	approach	NOUN
ajst-13993	5	18	can	can	AUX
ajst-13993	5	19	obtain	obtain	VERB
ajst-13993	5	20	better	well	ADJ
ajst-13993	5	21	accuracy	accuracy	NOUN
ajst-13993	5	22	compared	compare	VERB
ajst-13993	5	23	with	with	ADP
ajst-13993	5	24	the	the	DET
ajst-13993	5	25	baseline	baseline	NOUN
ajst-13993	5	26	approaches	approach	NOUN
ajst-13993	5	27	.	.	PUNCT
ajst-13993	6	1	keywords	keyword	NOUN
ajst-13993	6	2	:	:	PUNCT
ajst-13993	6	3	deep	deep	ADJ
ajst-13993	6	4	learning	learning	NOUN
ajst-13993	6	5	,	,	PUNCT
ajst-13993	6	6	source	source	NOUN
ajst-13993	6	7	code	code	NOUN
ajst-13993	6	8	comment	comment	NOUN
ajst-13993	6	9	generation	generation	NOUN
ajst-13993	6	10	,	,	PUNCT
ajst-13993	6	11	deep	deep	ADJ
ajst-13993	6	12	reinforcement	reinforcement	NOUN
ajst-13993	6	13	learning	learning	NOUN
ajst-13993	6	14	,	,	PUNCT
ajst-13993	6	15	hierarchical	hierarchical	ADJ
ajst-13993	6	16	attention	attention	NOUN
ajst-13993	6	17	.	.	PUNCT
ajst-13993	7	1	1	1	X
ajst-13993	7	2	.	.	X
ajst-13993	7	3	introduction	introduction	NOUN
ajst-13993	7	4	source	source	NOUN
ajst-13993	7	5	code	code	NOUN
ajst-13993	7	6	summarization	summarization	NOUN
ajst-13993	7	7	is	be	AUX
ajst-13993	7	8	the	the	DET
ajst-13993	7	9	task	task	NOUN
ajst-13993	7	10	of	of	ADP
ajst-13993	7	11	writing	write	VERB
ajst-13993	7	12	brief	brief	ADJ
ajst-13993	7	13	natural	natural	ADJ
ajst-13993	7	14	language	language	NOUN
ajst-13993	7	15	descriptions	description	NOUN
ajst-13993	7	16	of	of	ADP
ajst-13993	7	17	code	code	NOUN
ajst-13993	7	18	[	[	X
ajst-13993	7	19	1	1	NUM
ajst-13993	7	20	,	,	PUNCT
ajst-13993	7	21	2	2	NUM
ajst-13993	7	22	,	,	PUNCT
ajst-13993	7	23	3	3	NUM
ajst-13993	7	24	,	,	PUNCT
ajst-13993	7	25	4	4	NUM
ajst-13993	7	26	]	]	PUNCT
ajst-13993	7	27	.	.	PUNCT
ajst-13993	8	1	these	these	DET
ajst-13993	8	2	descriptions	description	NOUN
ajst-13993	8	3	have	have	AUX
ajst-13993	8	4	long	long	ADV
ajst-13993	8	5	been	be	AUX
ajst-13993	8	6	the	the	DET
ajst-13993	8	7	backbone	backbone	NOUN
ajst-13993	8	8	of	of	ADP
ajst-13993	8	9	developer	developer	NOUN
ajst-13993	8	10	documentation	documentation	NOUN
ajst-13993	8	11	such	such	ADJ
ajst-13993	8	12	as	as	ADP
ajst-13993	8	13	javadocs	javadoc	NOUN
ajst-13993	8	14	[	[	X
ajst-13993	8	15	5	5	NUM
ajst-13993	8	16	]	]	PUNCT
ajst-13993	8	17	.	.	PUNCT
ajst-13993	9	1	the	the	DET
ajst-13993	9	2	idea	idea	NOUN
ajst-13993	9	3	is	be	AUX
ajst-13993	9	4	that	that	SCONJ
ajst-13993	9	5	a	a	DET
ajst-13993	9	6	short	short	ADJ
ajst-13993	9	7	description	description	NOUN
ajst-13993	9	8	allows	allow	VERB
ajst-13993	9	9	a	a	DET
ajst-13993	9	10	programmer	programmer	NOUN
ajst-13993	9	11	to	to	PART
ajst-13993	9	12	understand	understand	VERB
ajst-13993	9	13	what	what	PRON
ajst-13993	9	14	a	a	DET
ajst-13993	9	15	section	section	NOUN
ajst-13993	9	16	of	of	ADP
ajst-13993	9	17	code	code	NOUN
ajst-13993	9	18	does	do	VERB
ajst-13993	9	19	and	and	CCONJ
ajst-13993	9	20	that	that	DET
ajst-13993	9	21	code	code	NOUN
ajst-13993	9	22	’s	’s	PART
ajst-13993	9	23	purpose	purpose	NOUN
ajst-13993	9	24	in	in	ADP
ajst-13993	9	25	the	the	DET
ajst-13993	9	26	overall	overall	ADJ
ajst-13993	9	27	program	program	NOUN
ajst-13993	9	28	,	,	PUNCT
ajst-13993	9	29	without	without	ADP
ajst-13993	9	30	requiring	require	VERB
ajst-13993	9	31	the	the	DET
ajst-13993	9	32	programmer	programmer	NOUN
ajst-13993	9	33	to	to	PART
ajst-13993	9	34	read	read	VERB
ajst-13993	9	35	the	the	DET
ajst-13993	9	36	code	code	NOUN
ajst-13993	9	37	itself	itself	PRON
ajst-13993	9	38	.	.	PUNCT
ajst-13993	10	1	summaries	summary	NOUN
ajst-13993	10	2	like	like	ADP
ajst-13993	10	3	“	"	PUNCT
ajst-13993	10	4	uploads	upload	VERB
ajst-13993	10	5	log	log	NOUN
ajst-13993	10	6	files	file	NOUN
ajst-13993	10	7	to	to	ADP
ajst-13993	10	8	the	the	DET
ajst-13993	10	9	backup	backup	ADJ
ajst-13993	10	10	server	server	NOUN
ajst-13993	10	11	”	"	PUNCT
ajst-13993	10	12	or	or	CCONJ
ajst-13993	10	13	“	"	PUNCT
ajst-13993	10	14	formats	format	NOUN
ajst-13993	10	15	decimal	decimal	ADJ
ajst-13993	10	16	values	value	NOUN
ajst-13993	10	17	as	as	ADP
ajst-13993	10	18	scientific	scientific	ADJ
ajst-13993	10	19	notation	notation	NOUN
ajst-13993	10	20	”	"	PUNCT
ajst-13993	10	21	can	can	AUX
ajst-13993	10	22	give	give	VERB
ajst-13993	10	23	programmers	programmer	NOUN
ajst-13993	10	24	a	a	DET
ajst-13993	10	25	clear	clear	ADJ
ajst-13993	10	26	picture	picture	NOUN
ajst-13993	10	27	of	of	ADP
ajst-13993	10	28	what	what	PRON
ajst-13993	10	29	code	code	NOUN
ajst-13993	10	30	does	do	AUX
ajst-13993	10	31	,	,	PUNCT
ajst-13993	10	32	saving	save	VERB
ajst-13993	10	33	them	they	PRON
ajst-13993	10	34	time	time	NOUN
ajst-13993	10	35	from	from	ADP
ajst-13993	10	36	comprehending	comprehend	VERB
ajst-13993	10	37	the	the	DET
ajst-13993	10	38	details	detail	NOUN
ajst-13993	10	39	of	of	ADP
ajst-13993	10	40	that	that	DET
ajst-13993	10	41	code	code	NOUN
ajst-13993	10	42	.	.	PUNCT
ajst-13993	11	1	trace	trace	VERB
ajst-13993	11	2	its	its	PRON
ajst-13993	11	3	technological	technological	ADJ
ajst-13993	11	4	development	development	NOUN
ajst-13993	11	5	,	,	PUNCT
ajst-13993	11	6	at	at	ADP
ajst-13993	11	7	first	first	ADV
ajst-13993	11	8	,	,	PUNCT
ajst-13993	11	9	the	the	DET
ajst-13993	11	10	dominant	dominant	ADJ
ajst-13993	11	11	strategy	strategy	NOUN
ajst-13993	11	12	was	be	AUX
ajst-13993	11	13	based	base	VERB
ajst-13993	11	14	on	on	ADP
ajst-13993	11	15	sentence	sentence	NOUN
ajst-13993	11	16	templates	template	NOUN
ajst-13993	11	17	and	and	CCONJ
ajst-13993	11	18	heuristics	heuristic	NOUN
ajst-13993	11	19	derived	derive	VERB
ajst-13993	11	20	from	from	ADP
ajst-13993	11	21	empirical	empirical	ADJ
ajst-13993	11	22	studies	study	NOUN
ajst-13993	12	1	[	[	X
ajst-13993	12	2	6	6	NUM
ajst-13993	12	3	-	-	SYM
ajst-13993	12	4	10	10	NUM
ajst-13993	12	5	]	]	PUNCT
ajst-13993	12	6	.	.	PUNCT
ajst-13993	13	1	starting	start	VERB
ajst-13993	13	2	around	around	ADP
ajst-13993	13	3	2016	2016	NUM
ajst-13993	13	4	,	,	PUNCT
ajst-13993	13	5	data	data	NOUN
ajst-13993	13	6	-	-	PUNCT
ajst-13993	13	7	driven	drive	VERB
ajst-13993	13	8	strategies	strategy	NOUN
ajst-13993	13	9	based	base	VERB
ajst-13993	13	10	on	on	ADP
ajst-13993	13	11	neural	neural	ADJ
ajst-13993	13	12	networks	network	NOUN
ajst-13993	13	13	came	come	VERB
ajst-13993	13	14	to	to	ADP
ajst-13993	13	15	the	the	DET
ajst-13993	13	16	forefront	forefront	NOUN
ajst-13993	13	17	,	,	PUNCT
ajst-13993	13	18	leveraging	leverage	VERB
ajst-13993	13	19	gains	gain	NOUN
ajst-13993	13	20	from	from	ADP
ajst-13993	13	21	both	both	CCONJ
ajst-13993	13	22	the	the	DET
ajst-13993	13	23	ai	ai	ADJ
ajst-13993	13	24	/	/	SYM
ajst-13993	13	25	nlp	nlp	NOUN
ajst-13993	13	26	and	and	CCONJ
ajst-13993	13	27	mining	mining	NOUN
ajst-13993	13	28	software	software	NOUN
ajst-13993	13	29	repositories	repository	NOUN
ajst-13993	13	30	research	research	NOUN
ajst-13993	13	31	communities	community	NOUN
ajst-13993	13	32	[	[	X
ajst-13993	13	33	11	11	NUM
ajst-13993	13	34	-	-	SYM
ajst-13993	13	35	14].as	14].as	NUM
ajst-13993	13	36	far	far	ADV
ajst-13993	13	37	as	as	SCONJ
ajst-13993	13	38	we	we	PRON
ajst-13993	13	39	know	know	VERB
ajst-13993	13	40	,	,	PUNCT
ajst-13993	13	41	the	the	DET
ajst-13993	13	42	existing	exist	VERB
ajst-13993	13	43	deep	deep	ADJ
ajst-13993	13	44	learning	learning	NOUN
ajst-13993	13	45	based	base	VERB
ajst-13993	13	46	comment	comment	NOUN
ajst-13993	13	47	generation	generation	NOUN
ajst-13993	13	48	approaches	approach	NOUN
ajst-13993	13	49	mainly	mainly	ADV
ajst-13993	13	50	utilize	utilize	VERB
ajst-13993	13	51	the	the	DET
ajst-13993	13	52	seq2seq	seq2seq	NOUN
ajst-13993	13	53	model	model	NOUN
ajst-13993	13	54	in	in	ADP
ajst-13993	13	55	which	which	PRON
ajst-13993	13	56	the	the	DET
ajst-13993	13	57	program	program	NOUN
ajst-13993	13	58	code	code	NOUN
ajst-13993	13	59	is	be	AUX
ajst-13993	13	60	encoded	encode	VERB
ajst-13993	13	61	into	into	ADP
ajst-13993	13	62	hidden	hide	VERB
ajst-13993	13	63	space	space	NOUN
ajst-13993	13	64	first	first	ADV
ajst-13993	13	65	and	and	CCONJ
ajst-13993	13	66	then	then	ADV
ajst-13993	13	67	decode	decode	VERB
ajst-13993	13	68	it	it	PRON
ajst-13993	13	69	to	to	PART
ajst-13993	13	70	produce	produce	VERB
ajst-13993	13	71	the	the	DET
ajst-13993	13	72	target	target	NOUN
ajst-13993	13	73	comment.however	comment.however	NOUN
ajst-13993	13	74	,	,	PUNCT
ajst-13993	13	75	these	these	DET
ajst-13993	13	76	kind	kind	NOUN
ajst-13993	13	77	of	of	ADP
ajst-13993	13	78	approaches	approach	NOUN
ajst-13993	13	79	have	have	VERB
ajst-13993	13	80	the	the	DET
ajst-13993	13	81	following	follow	VERB
ajst-13993	13	82	drawbacks	drawback	NOUN
ajst-13993	13	83	:	:	PUNCT
ajst-13993	13	84	(	(	PUNCT
ajst-13993	13	85	1	1	X
ajst-13993	13	86	)	)	PUNCT
ajst-13993	13	87	they	they	PRON
ajst-13993	13	88	mainly	mainly	ADV
ajst-13993	13	89	take	take	VERB
ajst-13993	13	90	the	the	DET
ajst-13993	13	91	source	source	NOUN
ajst-13993	13	92	code	code	NOUN
ajst-13993	13	93	as	as	ADP
ajst-13993	13	94	plain	plain	ADJ
ajst-13993	13	95	text	text	NOUN
ajst-13993	13	96	and	and	CCONJ
ajst-13993	13	97	ignore	ignore	VERB
ajst-13993	13	98	the	the	DET
ajst-13993	13	99	hierarchical	hierarchical	ADJ
ajst-13993	13	100	structure	structure	NOUN
ajst-13993	13	101	of	of	ADP
ajst-13993	13	102	the	the	DET
ajst-13993	13	103	source	source	NOUN
ajst-13993	13	104	code	code	NOUN
ajst-13993	13	105	;	;	PUNCT
ajst-13993	13	106	(	(	PUNCT
ajst-13993	13	107	2	2	X
ajst-13993	13	108	)	)	PUNCT
ajst-13993	13	109	most	most	ADJ
ajst-13993	13	110	of	of	ADP
ajst-13993	13	111	the	the	DET
ajst-13993	13	112	approaches	approach	NOUN
ajst-13993	13	113	only	only	ADV
ajst-13993	13	114	consider	consider	VERB
ajst-13993	13	115	simple	simple	ADJ
ajst-13993	13	116	features	feature	NOUN
ajst-13993	13	117	,	,	PUNCT
ajst-13993	13	118	such	such	ADJ
ajst-13993	13	119	as	as	ADP
ajst-13993	13	120	,	,	PUNCT
ajst-13993	13	121	tokens	token	NOUN
ajst-13993	13	122	,	,	PUNCT
ajst-13993	13	123	which	which	PRON
ajst-13993	13	124	overlooking	overlook	VERB
ajst-13993	13	125	the	the	DET
ajst-13993	13	126	hidden	hidden	ADJ
ajst-13993	13	127	information	information	NOUN
ajst-13993	13	128	that	that	PRON
ajst-13993	13	129	can	can	AUX
ajst-13993	13	130	help	help	VERB
ajst-13993	13	131	grab	grab	VERB
ajst-13993	13	132	the	the	DET
ajst-13993	13	133	relationships	relationship	NOUN
ajst-13993	13	134	between	between	ADP
ajst-13993	13	135	source	source	NOUN
ajst-13993	13	136	code	code	NOUN
ajst-13993	13	137	and	and	CCONJ
ajst-13993	13	138	comments	comment	NOUN
ajst-13993	13	139	;	;	PUNCT
ajst-13993	13	140	(	(	PUNCT
ajst-13993	13	141	3	3	X
ajst-13993	13	142	)	)	PUNCT
ajst-13993	13	143	they	they	PRON
ajst-13993	13	144	typically	typically	ADV
ajst-13993	13	145	train	train	VERB
ajst-13993	13	146	the	the	DET
ajst-13993	13	147	decoder	decoder	NOUN
ajst-13993	13	148	to	to	PART
ajst-13993	13	149	produce	produce	VERB
ajst-13993	13	150	the	the	DET
ajst-13993	13	151	code	code	NOUN
ajst-13993	13	152	annotation	annotation	NOUN
ajst-13993	13	153	by	by	ADP
ajst-13993	13	154	calculating	calculate	VERB
ajst-13993	13	155	and	and	CCONJ
ajst-13993	13	156	maximizing	maximize	VERB
ajst-13993	13	157	the	the	DET
ajst-13993	13	158	odds	odd	NOUN
ajst-13993	13	159	based	base	VERB
ajst-13993	13	160	on	on	ADP
ajst-13993	13	161	the	the	DET
ajst-13993	13	162	subsequent	subsequent	ADJ
ajst-13993	13	163	natural	natural	ADJ
ajst-13993	13	164	language	language	NOUN
ajst-13993	13	165	words	word	NOUN
ajst-13993	13	166	,	,	PUNCT
ajst-13993	13	167	however	however	ADV
ajst-13993	13	168	in	in	ADP
ajst-13993	13	169	fact	fact	NOUN
ajst-13993	13	170	,	,	PUNCT
ajst-13993	13	171	they	they	PRON
ajst-13993	13	172	mainly	mainly	ADV
ajst-13993	13	173	produce	produce	VERB
ajst-13993	13	174	the	the	DET
ajst-13993	13	175	code	code	NOUN
ajst-13993	13	176	annotation	annotation	NOUN
ajst-13993	13	177	from	from	ADP
ajst-13993	13	178	scratch	scratch	NOUN
ajst-13993	13	179	.	.	PUNCT
ajst-13993	14	1	therefore	therefore	ADV
ajst-13993	14	2	,	,	PUNCT
ajst-13993	14	3	these	these	DET
ajst-13993	14	4	drawbacks	drawback	NOUN
ajst-13993	14	5	result	result	VERB
ajst-13993	14	6	in	in	ADP
ajst-13993	14	7	inferior	inferior	ADJ
ajst-13993	14	8	comment	comment	NOUN
ajst-13993	14	9	generation	generation	NOUN
ajst-13993	14	10	accuracy	accuracy	NOUN
ajst-13993	14	11	and	and	CCONJ
ajst-13993	14	12	inconsistent	inconsistent	NOUN
ajst-13993	14	13	of	of	ADP
ajst-13993	14	14	the	the	DET
ajst-13993	14	15	generated	generate	VERB
ajst-13993	14	16	comment	comment	NOUN
ajst-13993	14	17	.	.	PUNCT
ajst-13993	15	1	to	to	PART
ajst-13993	15	2	solve	solve	VERB
ajst-13993	15	3	the	the	DET
ajst-13993	15	4	limitations	limitation	NOUN
ajst-13993	15	5	described	describe	VERB
ajst-13993	15	6	above	above	ADV
ajst-13993	15	7	,	,	PUNCT
ajst-13993	15	8	this	this	DET
ajst-13993	15	9	work	work	NOUN
ajst-13993	15	10	proposes	propose	VERB
ajst-13993	15	11	to	to	PART
ajst-13993	15	12	utilize	utilize	VERB
ajst-13993	15	13	the	the	DET
ajst-13993	15	14	hierarchical	hierarchical	ADJ
ajst-13993	15	15	attention	attention	NOUN
ajst-13993	15	16	mechanism	mechanism	NOUN
ajst-13993	15	17	to	to	PART
ajst-13993	15	18	combine	combine	VERB
ajst-13993	15	19	several	several	ADJ
ajst-13993	15	20	source	source	NOUN
ajst-13993	15	21	code	code	NOUN
ajst-13993	15	22	features	feature	VERB
ajst-13993	15	23	for	for	ADP
ajst-13993	15	24	code	code	NOUN
ajst-13993	15	25	summarization.the	summarization.the	DET
ajst-13993	15	26	representation	representation	NOUN
ajst-13993	15	27	of	of	ADP
ajst-13993	15	28	these	these	DET
ajst-13993	15	29	features	feature	NOUN
ajst-13993	15	30	are	be	AUX
ajst-13993	15	31	input	input	NOUN
ajst-13993	15	32	to	to	ADP
ajst-13993	15	33	the	the	DET
ajst-13993	15	34	deep	deep	ADJ
ajst-13993	15	35	reinforcement	reinforcement	NOUN
ajst-13993	15	36	learning	learning	NOUN
ajst-13993	15	37	framework	framework	NOUN
ajst-13993	15	38	for	for	ADP
ajst-13993	15	39	the	the	DET
ajst-13993	15	40	comment	comment	NOUN
ajst-13993	15	41	generation	generation	NOUN
ajst-13993	15	42	task	task	NOUN
ajst-13993	15	43	.	.	PUNCT
ajst-13993	16	1	to	to	PART
ajst-13993	16	2	present	present	VERB
ajst-13993	16	3	the	the	DET
ajst-13993	16	4	hierarchical	hierarchical	ADJ
ajst-13993	16	5	structure	structure	NOUN
ajst-13993	16	6	of	of	ADP
ajst-13993	16	7	the	the	DET
ajst-13993	16	8	source	source	NOUN
ajst-13993	16	9	code	code	NOUN
ajst-13993	16	10	under	under	ADP
ajst-13993	16	11	different	different	ADJ
ajst-13993	16	12	contexts	context	NOUN
ajst-13993	16	13	considering	consider	VERB
ajst-13993	16	14	the	the	DET
ajst-13993	16	15	code	code	NOUN
ajst-13993	16	16	features	feature	VERB
ajst-13993	16	17	,	,	PUNCT
ajst-13993	16	18	the	the	DET
ajst-13993	16	19	proposed	propose	VERB
ajst-13993	16	20	approach	approach	NOUN
ajst-13993	16	21	allocate	allocate	VERB
ajst-13993	16	22	weights	weight	NOUN
ajst-13993	16	23	to	to	ADP
ajst-13993	16	24	different	different	ADJ
ajst-13993	16	25	statements	statement	NOUN
ajst-13993	16	26	and	and	CCONJ
ajst-13993	16	27	tokens	token	NOUN
ajst-13993	16	28	respectively	respectively	ADV
ajst-13993	16	29	when	when	SCONJ
ajst-13993	16	30	forming	form	VERB
ajst-13993	16	31	the	the	DET
ajst-13993	16	32	representation	representation	NOUN
ajst-13993	16	33	of	of	ADP
ajst-13993	16	34	the	the	DET
ajst-13993	16	35	source	source	NOUN
ajst-13993	16	36	code.furthermore	code.furthermore	NOUN
ajst-13993	16	37	,	,	PUNCT
ajst-13993	16	38	the	the	DET
ajst-13993	16	39	reinforcement	reinforcement	NOUN
ajst-13993	16	40	learning	learning	NOUN
ajst-13993	16	41	model	model	NOUN
ajst-13993	16	42	improves	improve	VERB
ajst-13993	16	43	the	the	DET
ajst-13993	16	44	generated	generate	VERB
ajst-13993	16	45	comment	comment	NOUN
ajst-13993	16	46	through	through	ADP
ajst-13993	16	47	the	the	DET
ajst-13993	16	48	actor	actor	NOUN
ajst-13993	16	49	and	and	CCONJ
ajst-13993	16	50	critic	critic	NOUN
ajst-13993	16	51	network	network	NOUN
ajst-13993	16	52	,	,	PUNCT
ajst-13993	16	53	in	in	ADP
ajst-13993	16	54	which	which	PRON
ajst-13993	16	55	given	give	VERB
ajst-13993	16	56	the	the	DET
ajst-13993	16	57	present	present	ADJ
ajst-13993	16	58	state	state	NOUN
ajst-13993	16	59	the	the	DET
ajst-13993	16	60	actor	actor	NOUN
ajst-13993	16	61	furnishes	furnish	VERB
ajst-13993	16	62	the	the	DET
ajst-13993	16	63	confidence	confidence	NOUN
ajst-13993	16	64	of	of	ADP
ajst-13993	16	65	generating	generate	VERB
ajst-13993	16	66	the	the	DET
ajst-13993	16	67	next	next	ADJ
ajst-13993	16	68	words	word	NOUN
ajst-13993	16	69	,	,	PUNCT
ajst-13993	16	70	and	and	CCONJ
ajst-13993	16	71	then	then	ADV
ajst-13993	16	72	the	the	DET
ajst-13993	16	73	critic	critic	ADJ
ajst-13993	16	74	network	network	NOUN
ajst-13993	16	75	calculates	calculate	VERB
ajst-13993	16	76	the	the	DET
ajst-13993	16	77	reward	reward	NOUN
ajst-13993	16	78	values	value	NOUN
ajst-13993	16	79	of	of	ADP
ajst-13993	16	80	the	the	DET
ajst-13993	16	81	potential	potential	ADJ
ajst-13993	16	82	next	next	ADJ
ajst-13993	16	83	word	word	NOUN
ajst-13993	16	84	to	to	PART
ajst-13993	16	85	provide	provide	VERB
ajst-13993	16	86	clues	clue	NOUN
ajst-13993	16	87	for	for	ADP
ajst-13993	16	88	the	the	DET
ajst-13993	16	89	generation	generation	NOUN
ajst-13993	16	90	explorations	exploration	NOUN
ajst-13993	16	91	.	.	PUNCT
ajst-13993	17	1	at	at	ADP
ajst-13993	17	2	last	last	ADV
ajst-13993	17	3	,	,	PUNCT
ajst-13993	17	4	we	we	PRON
ajst-13993	17	5	train	train	VERB
ajst-13993	17	6	the	the	DET
ajst-13993	17	7	framework	framework	NOUN
ajst-13993	17	8	based	base	VERB
ajst-13993	17	9	on	on	ADP
ajst-13993	17	10	the	the	DET
ajst-13993	17	11	reward	reward	NOUN
ajst-13993	17	12	and	and	CCONJ
ajst-13993	17	13	perform	perform	VERB
ajst-13993	17	14	experiments	experiment	NOUN
ajst-13993	17	15	bases	basis	NOUN
ajst-13993	17	16	on	on	ADP
ajst-13993	17	17	the	the	DET
ajst-13993	17	18	dataset	dataset	NOUN
ajst-13993	17	19	collected	collect	VERB
ajst-13993	17	20	from	from	ADP
ajst-13993	17	21	real	real	ADJ
ajst-13993	17	22	projects	project	NOUN
ajst-13993	17	23	in	in	ADP
ajst-13993	17	24	github	github	NOUN
ajst-13993	17	25	.	.	PUNCT
ajst-13993	18	1	2	2	X
ajst-13993	18	2	.	.	X
ajst-13993	18	3	related	relate	VERB
ajst-13993	18	4	work	work	NOUN
ajst-13993	18	5	automatic	automatic	ADJ
ajst-13993	18	6	comment	comment	NOUN
ajst-13993	18	7	generation	generation	NOUN
ajst-13993	18	8	approaches	approach	NOUN
ajst-13993	18	9	vary	vary	VERB
ajst-13993	18	10	from	from	ADP
ajst-13993	18	11	manually	manually	ADV
ajst-13993	18	12	-	-	PUNCT
ajst-13993	18	13	crafted	craft	VERB
ajst-13993	18	14	templates	template	NOUN
ajst-13993	18	15	[	[	X
ajst-13993	18	16	26,27	26,27	NUM
ajst-13993	18	17	,	,	PUNCT
ajst-13993	18	18	28	28	NUM
ajst-13993	18	19	]	]	PUNCT
ajst-13993	18	20	,	,	PUNCT
ajst-13993	18	21	ir	ir	PROPN
ajst-13993	19	1	[	[	X
ajst-13993	19	2	29,30	29,30	NUM
ajst-13993	19	3	,	,	PUNCT
ajst-13993	19	4	31	31	NUM
ajst-13993	19	5	,	,	PUNCT
ajst-13993	19	6	32	32	NUM
ajst-13993	19	7	]	]	PUNCT
ajst-13993	19	8	to	to	ADP
ajst-13993	19	9	neural	neural	ADJ
ajst-13993	19	10	models	model	NOUN
ajst-13993	19	11	[	[	X
ajst-13993	19	12	33,34,35	33,34,35	NUM
ajst-13993	19	13	]	]	PUNCT
ajst-13993	19	14	.	.	PUNCT
ajst-13993	20	1	comment	comment	NOUN
ajst-13993	20	2	generation	generation	NOUN
ajst-13993	20	3	based	base	VERB
ajst-13993	20	4	on	on	ADP
ajst-13993	20	5	manually	manually	ADJ
ajst-13993	20	6	-	-	PUNCT
ajst-13993	20	7	craft	craft	NOUN
ajst-13993	20	8	templates	template	NOUN
ajst-13993	20	9	was	be	AUX
ajst-13993	20	10	one	one	NUM
ajst-13993	20	11	of	of	ADP
ajst-13993	20	12	the	the	DET
ajst-13993	20	13	common	common	ADJ
ajst-13993	20	14	methods	method	NOUN
ajst-13993	20	15	for	for	ADP
ajst-13993	20	16	generating	generate	VERB
ajst-13993	20	17	comments	comment	NOUN
ajst-13993	20	18	.	.	PUNCT
ajst-13993	21	1	sridhara	sridhara	PROPN
ajst-13993	21	2	et	et	PROPN
ajst-13993	21	3	al	al	PROPN
ajst-13993	21	4	.	.	PUNCT
ajst-13993	22	1	[	[	X
ajst-13993	22	2	36	36	NUM
ajst-13993	22	3	]	]	PUNCT
ajst-13993	22	4	developed	develop	VERB
ajst-13993	22	5	the	the	DET
ajst-13993	22	6	software	software	NOUN
ajst-13993	22	7	word	word	NOUN
ajst-13993	22	8	usage	usage	NOUN
ajst-13993	22	9	model	model	NOUN
ajst-13993	22	10	(	(	PUNCT
ajst-13993	22	11	swum	swum	NOUN
ajst-13993	22	12	)	)	PUNCT
ajst-13993	22	13	to	to	PART
ajst-13993	22	14	capture	capture	VERB
ajst-13993	22	15	the	the	DET
ajst-13993	22	16	occurrences	occurrence	NOUN
ajst-13993	22	17	of	of	ADP
ajst-13993	22	18	terms	term	NOUN
ajst-13993	22	19	in	in	ADP
ajst-13993	22	20	source	source	NOUN
ajst-13993	22	21	code	code	NOUN
ajst-13993	22	22	and	and	CCONJ
ajst-13993	22	23	their	their	PRON
ajst-13993	22	24	linguistic	linguistic	ADJ
ajst-13993	22	25	and	and	CCONJ
ajst-13993	22	26	structural	structural	ADJ
ajst-13993	22	27	relationships	relationship	NOUN
ajst-13993	22	28	and	and	CCONJ
ajst-13993	22	29	then	then	ADV
ajst-13993	22	30	defined	define	VERB
ajst-13993	22	31	different	different	ADJ
ajst-13993	22	32	templates	template	NOUN
ajst-13993	22	33	for	for	ADP
ajst-13993	22	34	different	different	ADJ
ajst-13993	22	35	semantic	semantic	ADJ
ajst-13993	22	36	segments	segment	NOUN
ajst-13993	22	37	in	in	ADP
ajst-13993	22	38	source	source	NOUN
ajst-13993	22	39	code	code	NOUN
ajst-13993	22	40	to	to	PART
ajst-13993	22	41	generate	generate	VERB
ajst-13993	22	42	readable	readable	ADJ
ajst-13993	22	43	natural	natural	ADJ
ajst-13993	22	44	language	language	NOUN
ajst-13993	22	45	.	.	PUNCT
ajst-13993	23	1	moreno	moreno	PROPN
ajst-13993	23	2	et	et	PROPN
ajst-13993	23	3	al	al	PROPN
ajst-13993	23	4	.	.	PUNCT
ajst-13993	24	1	[	[	X
ajst-13993	24	2	37]defined	37]defined	NUM
ajst-13993	24	3	heuristic	heuristic	ADJ
ajst-13993	24	4	rules	rule	NOUN
ajst-13993	24	5	to	to	PART
ajst-13993	24	6	select	select	VERB
ajst-13993	24	7	relevant	relevant	ADJ
ajst-13993	24	8	information	information	NOUN
ajst-13993	24	9	in	in	ADP
ajst-13993	24	10	the	the	DET
ajst-13993	24	11	source	source	NOUN
ajst-13993	24	12	code	code	NOUN
ajst-13993	24	13	,	,	PUNCT
ajst-13993	24	14	and	and	CCONJ
ajst-13993	24	15	then	then	ADV
ajst-13993	24	16	divided	divide	VERB
ajst-13993	24	17	the	the	DET
ajst-13993	24	18	comments	comment	NOUN
ajst-13993	24	19	into	into	ADP
ajst-13993	24	20	four	four	NUM
ajst-13993	24	21	parts	part	NOUN
ajst-13993	24	22	,	,	PUNCT
ajst-13993	24	23	and	and	CCONJ
ajst-13993	24	24	defined	define	VERB
ajst-13993	24	25	different	different	ADJ
ajst-13993	24	26	text	text	NOUN
ajst-13993	24	27	templates	template	NOUN
ajst-13993	24	28	for	for	ADP
ajst-13993	24	29	each	each	DET
ajst-13993	24	30	part	part	NOUN
ajst-13993	24	31	to	to	PART
ajst-13993	24	32	generate	generate	VERB
ajst-13993	24	33	natural	natural	ADJ
ajst-13993	24	34	language	language	NOUN
ajst-13993	24	35	descriptions	description	NOUN
ajst-13993	24	36	.	.	PUNCT
ajst-13993	25	1	mcburney	mcburney	PROPN
ajst-13993	25	2	et	et	PROPN
ajst-13993	25	3	al	al	PROPN
ajst-13993	25	4	.	.	PUNCT
ajst-13993	26	1	[	[	X
ajst-13993	26	2	38	38	NUM
ajst-13993	26	3	]	]	PUNCT
ajst-13993	26	4	also	also	ADV
ajst-13993	26	5	used	use	VERB
ajst-13993	26	6	the	the	DET
ajst-13993	26	7	swum	swum	ADJ
ajst-13993	26	8	model	model	NOUN
ajst-13993	26	9	to	to	PART
ajst-13993	26	10	extract	extract	VERB
ajst-13993	26	11	the	the	DET
ajst-13993	26	12	keywords	keyword	NOUN
ajst-13993	26	13	in	in	ADP
ajst-13993	26	14	the	the	DET
ajst-13993	26	15	java	java	PROPN
ajst-13993	26	16	method	method	NOUN
ajst-13993	26	17	,	,	PUNCT
ajst-13993	26	18	employed	employ	VERB
ajst-13993	26	19	the	the	DET
ajst-13993	26	20	pagerank	pagerank	NOUN
ajst-13993	26	21	algorithm	algorithm	NOUN
ajst-13993	26	22	to	to	PART
ajst-13993	26	23	select	select	VERB
ajst-13993	26	24	the	the	DET
ajst-13993	26	25	important	important	ADJ
ajst-13993	26	26	methods	method	NOUN
ajst-13993	26	27	in	in	ADP
ajst-13993	26	28	the	the	DET
ajst-13993	26	29	given	give	VERB
ajst-13993	26	30	method	method	NOUN
ajst-13993	26	31	’s	’s	PART
ajst-13993	26	32	context	context	NOUN
ajst-13993	26	33	,	,	PUNCT
ajst-13993	26	34	and	and	CCONJ
ajst-13993	26	35	used	use	VERB
ajst-13993	26	36	a	a	DET
ajst-13993	26	37	template	template	NOUN
ajst-13993	26	38	-	-	PUNCT
ajst-13993	26	39	based	base	VERB
ajst-13993	26	40	text	text	NOUN
ajst-13993	26	41	generation	generation	NOUN
ajst-13993	26	42	system	system	NOUN
ajst-13993	26	43	to	to	PART
ajst-13993	26	44	generate	generate	VERB
ajst-13993	26	45	comments	comment	NOUN
ajst-13993	26	46	.	.	PUNCT
ajst-13993	27	1	these	these	DET
ajst-13993	27	2	frameworks	framework	NOUN
ajst-13993	27	3	have	have	AUX
ajst-13993	27	4	achieved	achieve	VERB
ajst-13993	27	5	good	good	ADJ
ajst-13993	27	6	results	result	NOUN
ajst-13993	27	7	on	on	ADP
ajst-13993	27	8	java	java	PROPN
ajst-13993	27	9	classes	class	NOUN
ajst-13993	27	10	and	and	CCONJ
ajst-13993	27	11	methods	method	NOUN
ajst-13993	27	12	.	.	PUNCT
ajst-13993	28	1	ir	ir	PROPN
ajst-13993	28	2	techniques	technique	NOUN
ajst-13993	28	3	have	have	AUX
ajst-13993	28	4	been	be	AUX
ajst-13993	28	5	widely	widely	ADV
ajst-13993	28	6	used	use	VERB
ajst-13993	28	7	in	in	ADP
ajst-13993	28	8	comment	comment	NOUN
ajst-13993	28	9	generation	generation	NOUN
ajst-13993	28	10	task	task	NOUN
ajst-13993	28	11	.	.	PUNCT
ajst-13993	29	1	haiduc	haiduc	PROPN
ajst-13993	29	2	et	et	PROPN
ajst-13993	29	3	al	al	PROPN
ajst-13993	29	4	.	.	PUNCT
ajst-13993	30	1	[	[	X
ajst-13993	30	2	39	39	NUM
ajst-13993	30	3	]	]	PUNCT
ajst-13993	30	4	used	use	VERB
ajst-13993	30	5	two	two	NUM
ajst-13993	30	6	ir	ir	PROPN
ajst-13993	30	7	techniques	technique	NOUN
ajst-13993	30	8	,	,	PUNCT
ajst-13993	30	9	vector	vector	NOUN
ajst-13993	30	10	space	space	NOUN
ajst-13993	30	11	model	model	NOUN
ajst-13993	30	12	and	and	CCONJ
ajst-13993	30	13	latent	latent	ADJ
ajst-13993	30	14	semantic	semantic	ADJ
ajst-13993	30	15	indexing	indexing	NOUN
ajst-13993	30	16	,	,	PUNCT
ajst-13993	30	17	to	to	PART
ajst-13993	30	18	retrieve	retrieve	VERB
ajst-13993	30	19	relevant	relevant	ADJ
ajst-13993	30	20	terms	term	NOUN
ajst-13993	30	21	from	from	ADP
ajst-13993	30	22	a	a	DET
ajst-13993	30	23	software	software	NOUN
ajst-13993	30	24	corpus	corpus	NOUN
ajst-13993	30	25	,	,	PUNCT
ajst-13993	30	26	and	and	CCONJ
ajst-13993	30	27	then	then	ADV
ajst-13993	30	28	75	75	NUM
ajst-13993	30	29	organized	organize	VERB
ajst-13993	30	30	these	these	DET
ajst-13993	30	31	terms	term	NOUN
ajst-13993	30	32	into	into	ADP
ajst-13993	30	33	comments	comment	NOUN
ajst-13993	30	34	.	.	PUNCT
ajst-13993	31	1	eddy	eddy	PROPN
ajst-13993	31	2	et	et	PROPN
ajst-13993	31	3	al	al	PROPN
ajst-13993	31	4	.	.	PUNCT
ajst-13993	32	1	[	[	X
ajst-13993	32	2	40	40	NUM
ajst-13993	32	3	]	]	PUNCT
ajst-13993	32	4	used	use	VERB
ajst-13993	32	5	hierarchical	hierarchical	ADJ
ajst-13993	32	6	pam	pam	NOUN
ajst-13993	32	7	,	,	PUNCT
ajst-13993	32	8	a	a	DET
ajst-13993	32	9	probabilistic	probabilistic	ADJ
ajst-13993	32	10	model	model	NOUN
ajst-13993	32	11	that	that	PRON
ajst-13993	32	12	selected	select	VERB
ajst-13993	32	13	relevant	relevant	ADJ
ajst-13993	32	14	terms	term	NOUN
ajst-13993	32	15	from	from	ADP
ajst-13993	32	16	the	the	DET
ajst-13993	32	17	corpus	corpus	NOUN
ajst-13993	32	18	and	and	CCONJ
ajst-13993	32	19	included	include	VERB
ajst-13993	32	20	them	they	PRON
ajst-13993	32	21	to	to	ADP
ajst-13993	32	22	the	the	DET
ajst-13993	32	23	comments	comment	NOUN
ajst-13993	32	24	.	.	PUNCT
ajst-13993	33	1	unlike	unlike	ADP
ajst-13993	33	2	the	the	DET
ajst-13993	33	3	first	first	ADJ
ajst-13993	33	4	two	two	NUM
ajst-13993	33	5	research	research	NOUN
ajst-13993	33	6	works	work	NOUN
ajst-13993	33	7	,	,	PUNCT
ajst-13993	33	8	wong	wong	PROPN
ajst-13993	33	9	et	et	PROPN
ajst-13993	33	10	al	al	PROPN
ajst-13993	33	11	.	.	PUNCT
ajst-13993	34	1	[	[	X
ajst-13993	34	2	43	43	NUM
ajst-13993	34	3	]	]	PUNCT
ajst-13993	34	4	proposed	propose	VERB
ajst-13993	34	5	that	that	SCONJ
ajst-13993	34	6	code	code	NOUN
ajst-13993	34	7	snippets	snippet	NOUN
ajst-13993	34	8	and	and	CCONJ
ajst-13993	34	9	their	their	PRON
ajst-13993	34	10	descriptions	description	NOUN
ajst-13993	34	11	on	on	ADP
ajst-13993	34	12	the	the	DET
ajst-13993	34	13	q&a	q&a	PROPN
ajst-13993	34	14	sites	site	NOUN
ajst-13993	34	15	can	can	AUX
ajst-13993	34	16	be	be	AUX
ajst-13993	34	17	used	use	VERB
ajst-13993	34	18	to	to	PART
ajst-13993	34	19	generate	generate	VERB
ajst-13993	34	20	comments	comment	NOUN
ajst-13993	34	21	for	for	ADP
ajst-13993	34	22	a	a	DET
ajst-13993	34	23	piece	piece	NOUN
ajst-13993	34	24	of	of	ADP
ajst-13993	34	25	code	code	NOUN
ajst-13993	34	26	.	.	PUNCT
ajst-13993	35	1	they	they	PRON
ajst-13993	35	2	used	use	VERB
ajst-13993	35	3	a	a	DET
ajst-13993	35	4	token	token	VERB
ajst-13993	35	5	-	-	PUNCT
ajst-13993	35	6	based	base	VERB
ajst-13993	35	7	code	code	NOUN
ajst-13993	35	8	clone	clone	NOUN
ajst-13993	35	9	detection	detection	NOUN
ajst-13993	35	10	tool	tool	NOUN
ajst-13993	35	11	sim	sim	ADJ
ajst-13993	35	12	to	to	PART
ajst-13993	35	13	detect	detect	VERB
ajst-13993	35	14	similar	similar	ADJ
ajst-13993	35	15	code	code	NOUN
ajst-13993	35	16	snippets	snippet	NOUN
ajst-13993	35	17	and	and	CCONJ
ajst-13993	35	18	used	use	VERB
ajst-13993	35	19	their	their	PRON
ajst-13993	35	20	comments	comment	NOUN
ajst-13993	35	21	as	as	ADP
ajst-13993	35	22	target	target	NOUN
ajst-13993	35	23	comments	comment	NOUN
ajst-13993	35	24	.	.	PUNCT
ajst-13993	36	1	wong	wong	PROPN
ajst-13993	36	2	et	et	PROPN
ajst-13993	36	3	al	al	PROPN
ajst-13993	36	4	.	.	PUNCT
ajst-13993	37	1	[	[	X
ajst-13993	37	2	42	42	NUM
ajst-13993	37	3	]	]	PUNCT
ajst-13993	37	4	further	far	ADV
ajst-13993	37	5	thought	think	VERB
ajst-13993	37	6	that	that	SCONJ
ajst-13993	37	7	the	the	DET
ajst-13993	37	8	resources	resource	NOUN
ajst-13993	37	9	of	of	ADP
ajst-13993	37	10	the	the	DET
ajst-13993	37	11	q&a	q&a	PROPN
ajst-13993	37	12	sites	site	NOUN
ajst-13993	37	13	were	be	AUX
ajst-13993	37	14	limited	limit	VERB
ajst-13993	37	15	and	and	CCONJ
ajst-13993	37	16	proposed	propose	VERB
ajst-13993	37	17	to	to	PART
ajst-13993	37	18	use	use	VERB
ajst-13993	37	19	token	token	VERB
ajst-13993	37	20	-	-	PUNCT
ajst-13993	37	21	based	base	VERB
ajst-13993	37	22	code	code	NOUN
ajst-13993	37	23	clone	clone	NOUN
ajst-13993	37	24	detection	detection	NOUN
ajst-13993	37	25	tools	tool	NOUN
ajst-13993	37	26	to	to	PART
ajst-13993	37	27	retrieve	retrieve	VERB
ajst-13993	37	28	similar	similar	ADJ
ajst-13993	37	29	code	code	NOUN
ajst-13993	37	30	snippets	snippet	NOUN
ajst-13993	37	31	from	from	ADP
ajst-13993	37	32	github	github	NOUN
ajst-13993	37	33	and	and	CCONJ
ajst-13993	37	34	leverage	leverage	VERB
ajst-13993	37	35	the	the	DET
ajst-13993	37	36	information	information	NOUN
ajst-13993	37	37	obtained	obtain	VERB
ajst-13993	37	38	from	from	ADP
ajst-13993	37	39	their	their	PRON
ajst-13993	37	40	comments	comment	NOUN
ajst-13993	37	41	to	to	PART
ajst-13993	37	42	generate	generate	VERB
ajst-13993	37	43	comments	comment	NOUN
ajst-13993	37	44	.	.	PUNCT
ajst-13993	38	1	recently	recently	ADV
ajst-13993	38	2	many	many	ADJ
ajst-13993	38	3	neural	neural	ADJ
ajst-13993	38	4	networks	network	NOUN
ajst-13993	38	5	have	have	AUX
ajst-13993	38	6	been	be	AUX
ajst-13993	38	7	proposed	propose	VERB
ajst-13993	38	8	for	for	ADP
ajst-13993	38	9	comment	comment	NOUN
ajst-13993	38	10	generation	generation	NOUN
ajst-13993	38	11	.	.	PUNCT
ajst-13993	39	1	with	with	ADP
ajst-13993	39	2	large	large	ADJ
ajst-13993	39	3	-	-	PUNCT
ajst-13993	39	4	scale	scale	NOUN
ajst-13993	39	5	corpora	corpora	NOUN
ajst-13993	39	6	for	for	ADP
ajst-13993	39	7	training	training	NOUN
ajst-13993	39	8	,	,	PUNCT
ajst-13993	39	9	neuralbased	neuralbase	VERB
ajst-13993	39	10	approaches	approach	NOUN
ajst-13993	39	11	quickly	quickly	ADV
ajst-13993	39	12	became	become	VERB
ajst-13993	39	13	state	state	NOUN
ajst-13993	39	14	-	-	PUNCT
ajst-13993	39	15	of	of	ADP
ajst-13993	39	16	-	-	PUNCT
ajst-13993	39	17	the	the	DET
ajst-13993	39	18	-	-	PUNCT
ajst-13993	39	19	art	art	NOUN
ajst-13993	39	20	models	model	NOUN
ajst-13993	39	21	on	on	ADP
ajst-13993	39	22	this	this	DET
ajst-13993	39	23	task	task	NOUN
ajst-13993	39	24	.	.	PUNCT
ajst-13993	40	1	iyer	iyer	PROPN
ajst-13993	40	2	et	et	PROPN
ajst-13993	40	3	al	al	PROPN
ajst-13993	40	4	.	.	PUNCT
ajst-13993	41	1	[	[	X
ajst-13993	41	2	16	16	NUM
ajst-13993	41	3	]	]	PUNCT
ajst-13993	41	4	first	first	ADV
ajst-13993	41	5	introduced	introduce	VERB
ajst-13993	41	6	the	the	DET
ajst-13993	41	7	seq2seq	seq2seq	NOUN
ajst-13993	41	8	model	model	NOUN
ajst-13993	41	9	from	from	ADP
ajst-13993	41	10	neural	neural	ADJ
ajst-13993	41	11	machine	machine	NOUN
ajst-13993	41	12	translation	translation	NOUN
ajst-13993	41	13	into	into	ADP
ajst-13993	41	14	comment	comment	NOUN
ajst-13993	41	15	generation	generation	NOUN
ajst-13993	41	16	,	,	PUNCT
ajst-13993	41	17	whose	whose	DET
ajst-13993	41	18	encoder	encoder	NOUN
ajst-13993	41	19	is	be	AUX
ajst-13993	41	20	the	the	DET
ajst-13993	41	21	token	token	ADJ
ajst-13993	41	22	embedding	embedding	NOUN
ajst-13993	41	23	and	and	CCONJ
ajst-13993	41	24	decoder	decoder	NOUN
ajst-13993	41	25	is	be	AUX
ajst-13993	41	26	an	an	DET
ajst-13993	41	27	lstm	lstm	NOUN
ajst-13993	41	28	.	.	PUNCT
ajst-13993	42	1	their	their	PRON
ajst-13993	42	2	model	model	NOUN
ajst-13993	42	3	outperforms	outperform	VERB
ajst-13993	42	4	traditional	traditional	ADJ
ajst-13993	42	5	methods	method	NOUN
ajst-13993	42	6	on	on	ADP
ajst-13993	42	7	c	c	NOUN
ajst-13993	42	8	#	#	NOUN
ajst-13993	42	9	and	and	CCONJ
ajst-13993	42	10	sql	sql	NOUN
ajst-13993	42	11	summaries	summary	NOUN
ajst-13993	42	12	.	.	PUNCT
ajst-13993	43	1	inspired	inspire	VERB
ajst-13993	43	2	by	by	ADP
ajst-13993	43	3	the	the	DET
ajst-13993	43	4	difference	difference	NOUN
ajst-13993	43	5	between	between	ADP
ajst-13993	43	6	natural	natural	ADJ
ajst-13993	43	7	language	language	NOUN
ajst-13993	43	8	and	and	CCONJ
ajst-13993	43	9	programming	programming	NOUN
ajst-13993	43	10	language	language	NOUN
ajst-13993	43	11	,	,	PUNCT
ajst-13993	43	12	hu	hu	PROPN
ajst-13993	43	13	et	et	PROPN
ajst-13993	43	14	al	al	PROPN
ajst-13993	43	15	.	.	PUNCT
ajst-13993	44	1	[	[	X
ajst-13993	44	2	15	15	NUM
ajst-13993	44	3	]	]	PUNCT
ajst-13993	44	4	proposed	propose	VERB
ajst-13993	44	5	a	a	DET
ajst-13993	44	6	neural	neural	ADJ
ajst-13993	44	7	model	model	NOUN
ajst-13993	44	8	named	name	VERB
ajst-13993	44	9	deepcom	deepcom	NOUN
ajst-13993	44	10	to	to	PART
ajst-13993	44	11	capture	capture	VERB
ajst-13993	44	12	the	the	DET
ajst-13993	44	13	structural	structural	ADJ
ajst-13993	44	14	information	information	NOUN
ajst-13993	44	15	of	of	ADP
ajst-13993	44	16	source	source	NOUN
ajst-13993	44	17	code	code	NOUN
ajst-13993	44	18	.	.	PUNCT
ajst-13993	45	1	they	they	PRON
ajst-13993	45	2	proposed	propose	VERB
ajst-13993	45	3	a	a	DET
ajst-13993	45	4	structure	structure	NOUN
ajst-13993	45	5	-	-	PUNCT
ajst-13993	45	6	based	base	VERB
ajst-13993	45	7	traversal	traversal	NOUN
ajst-13993	45	8	method	method	NOUN
ajst-13993	45	9	,	,	PUNCT
ajst-13993	45	10	using	use	VERB
ajst-13993	45	11	one	one	NUM
ajst-13993	45	12	lstm	lstm	NOUN
ajst-13993	45	13	to	to	PART
ajst-13993	45	14	process	process	VERB
ajst-13993	45	15	the	the	DET
ajst-13993	45	16	ast	ast	NOUN
ajst-13993	45	17	’s	’s	PART
ajst-13993	45	18	traversal	traversal	NOUN
ajst-13993	45	19	sequence	sequence	NOUN
ajst-13993	45	20	,	,	PUNCT
ajst-13993	45	21	and	and	CCONJ
ajst-13993	45	22	the	the	DET
ajst-13993	45	23	other	other	ADJ
ajst-13993	45	24	lstm	lstm	NOUN
ajst-13993	45	25	to	to	PART
ajst-13993	45	26	generate	generate	VERB
ajst-13993	45	27	comments	comment	NOUN
ajst-13993	45	28	for	for	ADP
ajst-13993	45	29	java	java	PROPN
ajst-13993	45	30	methods	method	NOUN
ajst-13993	45	31	.	.	PUNCT
ajst-13993	46	1	leclair	leclair	PROPN
ajst-13993	46	2	et	et	PROPN
ajst-13993	46	3	al	al	PROPN
ajst-13993	46	4	.	.	PUNCT
ajst-13993	47	1	[	[	X
ajst-13993	47	2	25	25	NUM
ajst-13993	47	3	]	]	PUNCT
ajst-13993	47	4	proposed	propose	VERB
ajst-13993	47	5	a	a	DET
ajst-13993	47	6	neural	neural	ADJ
ajst-13993	47	7	method	method	NOUN
ajst-13993	47	8	to	to	PART
ajst-13993	47	9	predict	predict	VERB
ajst-13993	47	10	the	the	DET
ajst-13993	47	11	comment	comment	NOUN
ajst-13993	47	12	by	by	ADP
ajst-13993	47	13	combining	combine	VERB
ajst-13993	47	14	the	the	DET
ajst-13993	47	15	sequence	sequence	NOUN
ajst-13993	47	16	information	information	NOUN
ajst-13993	47	17	and	and	CCONJ
ajst-13993	47	18	structure	structure	NOUN
ajst-13993	47	19	information	information	NOUN
ajst-13993	47	20	of	of	ADP
ajst-13993	47	21	the	the	DET
ajst-13993	47	22	source	source	NOUN
ajst-13993	47	23	code	code	NOUN
ajst-13993	47	24	with	with	ADP
ajst-13993	47	25	two	two	NUM
ajst-13993	47	26	gru	gru	NOUN
ajst-13993	47	27	encoders	encoder	NOUN
ajst-13993	47	28	.	.	PUNCT
ajst-13993	48	1	in	in	ADP
ajst-13993	48	2	addition	addition	NOUN
ajst-13993	48	3	,	,	PUNCT
ajst-13993	48	4	they	they	PRON
ajst-13993	48	5	reconstructed	reconstruct	VERB
ajst-13993	48	6	the	the	DET
ajst-13993	48	7	benchmark	benchmark	NOUN
ajst-13993	48	8	dataset	dataset	NOUN
ajst-13993	48	9	for	for	ADP
ajst-13993	48	10	this	this	DET
ajst-13993	48	11	task	task	NOUN
ajst-13993	48	12	,	,	PUNCT
ajst-13993	48	13	removed	remove	VERB
ajst-13993	48	14	duplicate	duplicate	NOUN
ajst-13993	48	15	and	and	CCONJ
ajst-13993	48	16	auto	auto	NOUN
ajst-13993	48	17	-	-	PUNCT
ajst-13993	48	18	generated	generate	VERB
ajst-13993	48	19	code	code	NOUN
ajst-13993	48	20	in	in	ADP
ajst-13993	48	21	the	the	DET
ajst-13993	48	22	dataset	dataset	NOUN
ajst-13993	48	23	,	,	PUNCT
ajst-13993	48	24	and	and	CCONJ
ajst-13993	48	25	divided	divide	VERB
ajst-13993	48	26	the	the	DET
ajst-13993	48	27	dataset	dataset	NOUN
ajst-13993	48	28	into	into	ADP
ajst-13993	48	29	training	training	NOUN
ajst-13993	48	30	,	,	PUNCT
ajst-13993	48	31	validation	validation	NOUN
ajst-13993	48	32	,	,	PUNCT
ajst-13993	48	33	and	and	CCONJ
ajst-13993	48	34	test	test	NOUN
ajst-13993	48	35	by	by	ADP
ajst-13993	48	36	project	project	NOUN
ajst-13993	48	37	.	.	PUNCT
ajst-13993	49	1	this	this	DET
ajst-13993	49	2	work	work	NOUN
ajst-13993	49	3	proposes	propose	VERB
ajst-13993	49	4	to	to	PART
ajst-13993	49	5	adopt	adopt	VERB
ajst-13993	49	6	the	the	DET
ajst-13993	49	7	hierarchical	hierarchical	ADJ
ajst-13993	49	8	attention	attention	NOUN
ajst-13993	49	9	mechanism	mechanism	NOUN
ajst-13993	49	10	to	to	PART
ajst-13993	49	11	enable	enable	VERB
ajst-13993	49	12	the	the	DET
ajst-13993	49	13	code	code	NOUN
ajst-13993	49	14	summarization	summarization	NOUN
ajst-13993	49	15	framework	framework	NOUN
ajst-13993	49	16	to	to	PART
ajst-13993	49	17	translate	translate	VERB
ajst-13993	49	18	three	three	NUM
ajst-13993	49	19	representations	representation	NOUN
ajst-13993	49	20	of	of	ADP
ajst-13993	49	21	source	source	NOUN
ajst-13993	49	22	code	code	NOUN
ajst-13993	49	23	to	to	ADP
ajst-13993	49	24	the	the	DET
ajst-13993	49	25	hidden	hide	VERB
ajst-13993	49	26	spce	spce	NOUN
ajst-13993	49	27	and	and	CCONJ
ajst-13993	49	28	then	then	ADV
ajst-13993	49	29	it	it	PRON
ajst-13993	49	30	injects	inject	VERB
ajst-13993	49	31	them	they	PRON
ajst-13993	49	32	into	into	ADP
ajst-13993	49	33	a	a	DET
ajst-13993	49	34	deep	deep	ADJ
ajst-13993	49	35	reinforcement	reinforcement	NOUN
ajst-13993	49	36	learning	learning	NOUN
ajst-13993	49	37	model	model	NOUN
ajst-13993	49	38	to	to	PART
ajst-13993	49	39	enhance	enhance	VERB
ajst-13993	49	40	the	the	DET
ajst-13993	49	41	performance	performance	NOUN
ajst-13993	49	42	of	of	ADP
ajst-13993	49	43	code	code	NOUN
ajst-13993	49	44	summarization	summarization	NOUN
ajst-13993	49	45	.	.	PUNCT
ajst-13993	50	1	3	3	X
ajst-13993	50	2	.	.	X
ajst-13993	50	3	the	the	DET
ajst-13993	50	4	comment	comment	NOUN
ajst-13993	50	5	generation	generation	NOUN
ajst-13993	50	6	based	base	VERB
ajst-13993	50	7	on	on	ADP
ajst-13993	50	8	drl	drl	PROPN
ajst-13993	50	9	and	and	CCONJ
ajst-13993	50	10	hierarchical	hierarchical	ADJ
ajst-13993	50	11	attention	attention	NOUN
ajst-13993	50	12	figure	figure	NOUN
ajst-13993	50	13	1	1	NUM
ajst-13993	50	14	presents	present	VERB
ajst-13993	50	15	the	the	DET
ajst-13993	50	16	overview	overview	NOUN
ajst-13993	50	17	framework	framework	NOUN
ajst-13993	50	18	of	of	ADP
ajst-13993	50	19	the	the	DET
ajst-13993	50	20	proposed	propose	VERB
ajst-13993	50	21	comment	comment	NOUN
ajst-13993	50	22	generation	generation	NOUN
ajst-13993	50	23	approach	approach	NOUN
ajst-13993	50	24	through	through	ADP
ajst-13993	50	25	the	the	DET
ajst-13993	50	26	hierarchical	hierarchical	ADJ
ajst-13993	50	27	attention	attention	NOUN
ajst-13993	50	28	mechanism	mechanism	NOUN
ajst-13993	50	29	,	,	PUNCT
ajst-13993	50	30	which	which	PRON
ajst-13993	50	31	mainly	mainly	ADV
ajst-13993	50	32	follows	follow	VERB
ajst-13993	50	33	the	the	DET
ajst-13993	50	34	actor	actor	NOUN
ajst-13993	50	35	-	-	PUNCT
ajst-13993	50	36	critic	critic	NOUN
ajst-13993	50	37	framework.in	framework.in	NUM
ajst-13993	50	38	specific	specific	ADJ
ajst-13993	50	39	,	,	PUNCT
ajst-13993	50	40	four	four	NUM
ajst-13993	50	41	submodules	submodule	NOUN
ajst-13993	50	42	are	be	AUX
ajst-13993	50	43	included	include	VERB
ajst-13993	50	44	in	in	ADP
ajst-13993	50	45	the	the	DET
ajst-13993	50	46	proposed	propose	VERB
ajst-13993	50	47	framework	framework	NOUN
ajst-13993	50	48	,	,	PUNCT
ajst-13993	50	49	namely	namely	ADV
ajst-13993	50	50	(	(	PUNCT
ajst-13993	50	51	a	a	X
ajst-13993	50	52	)	)	PUNCT
ajst-13993	50	53	the	the	DET
ajst-13993	50	54	representation	representation	NOUN
ajst-13993	50	55	module	module	NOUN
ajst-13993	50	56	of	of	ADP
ajst-13993	50	57	the	the	DET
ajst-13993	50	58	source	source	NOUN
ajst-13993	50	59	code	code	NOUN
ajst-13993	50	60	which	which	PRON
ajst-13993	50	61	is	be	AUX
ajst-13993	50	62	used	use	VERB
ajst-13993	50	63	to	to	PART
ajst-13993	50	64	explain	explain	VERB
ajst-13993	50	65	the	the	DET
ajst-13993	50	66	structural	structural	ADJ
ajst-13993	50	67	and	and	CCONJ
ajst-13993	50	68	unstructured	unstructured	ADJ
ajst-13993	50	69	syntax	syntax	NOUN
ajst-13993	50	70	of	of	ADP
ajst-13993	50	71	the	the	DET
ajst-13993	50	72	code	code	NOUN
ajst-13993	50	73	snippet	snippet	NOUN
ajst-13993	50	74	;	;	PUNCT
ajst-13993	50	75	(	(	PUNCT
ajst-13993	50	76	b	b	X
ajst-13993	50	77	)	)	PUNCT
ajst-13993	50	78	the	the	DET
ajst-13993	50	79	module	module	NOUN
ajst-13993	50	80	of	of	ADP
ajst-13993	50	81	hierarchical	hierarchical	ADJ
ajst-13993	50	82	attention	attention	NOUN
ajst-13993	50	83	is	be	AUX
ajst-13993	50	84	utilized	utilize	VERB
ajst-13993	50	85	to	to	PART
ajst-13993	50	86	translate	translate	VERB
ajst-13993	50	87	the	the	DET
ajst-13993	50	88	representations	representation	NOUN
ajst-13993	50	89	of	of	ADP
ajst-13993	50	90	the	the	DET
ajst-13993	50	91	source	source	NOUN
ajst-13993	50	92	code	code	NOUN
ajst-13993	50	93	into	into	ADP
ajst-13993	50	94	vectors	vector	NOUN
ajst-13993	50	95	;	;	PUNCT
ajst-13993	50	96	(	(	PUNCT
ajst-13993	50	97	c	c	X
ajst-13993	50	98	)	)	PUNCT
ajst-13993	50	99	the	the	DET
ajst-13993	50	100	lstmbased	lstmbase	VERB
ajst-13993	50	101	text	text	NOUN
ajst-13993	50	102	generation	generation	NOUN
ajst-13993	50	103	module	module	NOUN
ajst-13993	50	104	which	which	PRON
ajst-13993	50	105	used	use	VERB
ajst-13993	50	106	to	to	PART
ajst-13993	50	107	generate	generate	VERB
ajst-13993	50	108	the	the	DET
ajst-13993	50	109	next	next	ADJ
ajst-13993	50	110	word	word	NOUN
ajst-13993	50	111	according	accord	VERB
ajst-13993	50	112	to	to	ADP
ajst-13993	50	113	the	the	DET
ajst-13993	50	114	words	word	NOUN
ajst-13993	50	115	before	before	ADV
ajst-13993	50	116	;	;	PUNCT
ajst-13993	50	117	and	and	CCONJ
ajst-13993	50	118	(	(	PUNCT
ajst-13993	50	119	d	d	X
ajst-13993	50	120	)	)	PUNCT
ajst-13993	50	121	the	the	DET
ajst-13993	50	122	critic	critic	NOUN
ajst-13993	50	123	module	module	NOUN
ajst-13993	50	124	is	be	AUX
ajst-13993	50	125	adopted	adopt	VERB
ajst-13993	50	126	to	to	PART
ajst-13993	50	127	estimate	estimate	VERB
ajst-13993	50	128	the	the	DET
ajst-13993	50	129	accuracy	accuracy	NOUN
ajst-13993	50	130	of	of	ADP
ajst-13993	50	131	the	the	DET
ajst-13993	50	132	generated	generate	VERB
ajst-13993	50	133	annotation	annotation	NOUN
ajst-13993	50	134	and	and	CCONJ
ajst-13993	50	135	supply	supply	NOUN
ajst-13993	50	136	feedback	feedback	VERB
ajst-13993	50	137	the	the	DET
ajst-13993	50	138	the	the	DET
ajst-13993	50	139	above	above	ADJ
ajst-13993	50	140	modules	module	NOUN
ajst-13993	50	141	.	.	PUNCT
ajst-13993	51	1	figure	figure	NOUN
ajst-13993	51	2	1	1	NUM
ajst-13993	51	3	.	.	PUNCT
ajst-13993	52	1	the	the	DET
ajst-13993	52	2	overview	overview	NOUN
ajst-13993	52	3	of	of	ADP
ajst-13993	52	4	the	the	DET
ajst-13993	52	5	proposed	propose	VERB
ajst-13993	52	6	code	code	NOUN
ajst-13993	52	7	summarization	summarization	NOUN
ajst-13993	52	8	via	via	ADP
ajst-13993	52	9	hierarchical	hierarchical	ADJ
ajst-13993	52	10	attention	attention	NOUN
ajst-13993	52	11	mechanism	mechanism	NOUN
ajst-13993	52	12	:	:	PUNCT
ajst-13993	52	13	(	(	PUNCT
ajst-13993	52	14	a	a	X
ajst-13993	52	15	)	)	PUNCT
ajst-13993	52	16	the	the	DET
ajst-13993	52	17	three	three	NUM
ajst-13993	52	18	representations	representation	NOUN
ajst-13993	52	19	of	of	ADP
ajst-13993	52	20	source	source	NOUN
ajst-13993	52	21	code	code	NOUN
ajst-13993	52	22	are	be	AUX
ajst-13993	52	23	signed	sign	VERB
ajst-13993	52	24	as	as	ADP
ajst-13993	52	25	txt	txt	NOUN
ajst-13993	52	26	ijx	ijx	NOUN
ajst-13993	52	27	,	,	PUNCT
ajst-13993	52	28	ast	ast	NOUN
ajst-13993	52	29	ijx	ijx	NOUN
ajst-13993	52	30	and	and	CCONJ
ajst-13993	52	31	cft	cft	VERB
ajst-13993	52	32	ijx	ijx	NOUN
ajst-13993	52	33	where	where	SCONJ
ajst-13993	52	34	i	i	PRON
ajst-13993	52	35	and	and	CCONJ
ajst-13993	52	36	j	j	PROPN
ajst-13993	52	37	represent	represent	VERB
ajst-13993	52	38	the	the	DET
ajst-13993	52	39	jth	jth	PROPN
ajst-13993	52	40	token	token	VERB
ajst-13993	52	41	in	in	ADP
ajst-13993	52	42	the	the	DET
ajst-13993	52	43	ith	ith	PROPN
ajst-13993	52	44	statement	statement	NOUN
ajst-13993	52	45	;	;	PUNCT
ajst-13993	52	46	(	(	PUNCT
ajst-13993	52	47	b	b	X
ajst-13993	52	48	)	)	PUNCT
ajst-13993	52	49	dtxt	dtxt	ADJ
ajst-13993	52	50	,	,	PUNCT
ajst-13993	52	51	dast	dast	PROPN
ajst-13993	52	52	and	and	CCONJ
ajst-13993	52	53	dcfg	dcfg	PROPN
ajst-13993	52	54	denote	denote	VERB
ajst-13993	52	55	the	the	DET
ajst-13993	52	56	vectors	vector	NOUN
ajst-13993	52	57	encoded	encode	VERB
ajst-13993	52	58	of	of	ADP
ajst-13993	52	59	the	the	DET
ajst-13993	52	60	three	three	NUM
ajst-13993	52	61	code	code	NOUN
ajst-13993	52	62	representations	representation	NOUN
ajst-13993	52	63	by	by	ADP
ajst-13993	52	64	the	the	DET
ajst-13993	52	65	hierarchical	hierarchical	ADJ
ajst-13993	52	66	attention	attention	NOUN
ajst-13993	52	67	mechanism	mechanism	NOUN
ajst-13993	52	68	respectively	respectively	ADV
ajst-13993	52	69	;	;	PUNCT
ajst-13993	52	70	(	(	PUNCT
ajst-13993	52	71	c	c	X
ajst-13993	52	72	)	)	PUNCT
ajst-13993	52	73	the	the	DET
ajst-13993	52	74	annotation	annotation	NOUN
ajst-13993	52	75	of	of	ADP
ajst-13993	52	76	the	the	DET
ajst-13993	52	77	source	source	NOUN
ajst-13993	52	78	code	code	NOUN
ajst-13993	52	79	is	be	AUX
ajst-13993	52	80	produced	produce	VERB
ajst-13993	52	81	by	by	ADP
ajst-13993	52	82	the	the	DET
ajst-13993	52	83	lstm	lstm	NOUN
ajst-13993	52	84	-	-	PUNCT
ajst-13993	52	85	based	base	VERB
ajst-13993	52	86	decoder	decoder	NOUN
ajst-13993	52	87	;	;	PUNCT
ajst-13993	52	88	(	(	PUNCT
ajst-13993	52	89	d	d	X
ajst-13993	52	90	)	)	PUNCT
ajst-13993	52	91	given	give	VERB
ajst-13993	52	92	the	the	DET
ajst-13993	52	93	state	state	NOUN
ajst-13993	52	94	st	st	PROPN
ajst-13993	52	95	,	,	PUNCT
ajst-13993	52	96	the	the	DET
ajst-13993	52	97	critic	critic	ADJ
ajst-13993	52	98	network	network	NOUN
ajst-13993	52	99	estimates	estimate	VERB
ajst-13993	52	100	its	its	PRON
ajst-13993	52	101	value	value	NOUN
ajst-13993	52	102	according	accord	VERB
ajst-13993	52	103	to|baseline	to|baseline	NOUN
ajst-13993	52	104	-	-	PUNCT
ajst-13993	52	105	reward|	reward|	NOUN
ajst-13993	52	106	.	.	PUNCT
ajst-13993	53	1	3.1	3.1	NUM
ajst-13993	53	2	.	.	PUNCT
ajst-13993	53	3	representations	representation	NOUN
ajst-13993	53	4	of	of	ADP
ajst-13993	53	5	the	the	DET
ajst-13993	53	6	source	source	NOUN
ajst-13993	53	7	code	code	NOUN
ajst-13993	53	8	considering	consider	VERB
ajst-13993	53	9	the	the	DET
ajst-13993	53	10	characteristic	characteristic	NOUN
ajst-13993	53	11	of	of	ADP
ajst-13993	53	12	the	the	DET
ajst-13993	53	13	source	source	NOUN
ajst-13993	53	14	code	code	NOUN
ajst-13993	53	15	,	,	PUNCT
ajst-13993	53	16	we	we	PRON
ajst-13993	53	17	adopt	adopt	VERB
ajst-13993	53	18	a	a	DET
ajst-13993	53	19	set	set	NOUN
ajst-13993	53	20	of	of	ADP
ajst-13993	53	21	symbols	symbol	NOUN
ajst-13993	53	22	,	,	PUNCT
ajst-13993	53	23	i.e.	i.e.	X
ajst-13993	53	24	,	,	PUNCT
ajst-13993	53	25	{	{	PUNCT
ajst-13993	53	26	.	.	PUNCT
ajst-13993	53	27	,	,	PUNCT
ajst-13993	53	28	"	"	PUNCT
ajst-13993	53	29	’	'	PUNCT
ajst-13993	53	30	_	_	PRON
ajst-13993	53	31	(	(	PUNCT
ajst-13993	53	32	)	)	PUNCT
ajst-13993	53	33	{	{	PUNCT
ajst-13993	53	34	}	}	PUNCT
ajst-13993	53	35	:	:	PUNCT
ajst-13993	53	36	!	!	PUNCT
ajst-13993	54	1	(	(	PUNCT
ajst-13993	54	2	space	space	NOUN
ajst-13993	54	3	)	)	PUNCT
ajst-13993	54	4	}	}	PUNCT
ajst-13993	54	5	to	to	PART
ajst-13993	54	6	tokenize	tokenize	VERB
ajst-13993	54	7	and	and	CCONJ
ajst-13993	54	8	split	split	VERB
ajst-13993	54	9	the	the	DET
ajst-13993	54	10	code	code	NOUN
ajst-13993	54	11	into	into	ADP
ajst-13993	54	12	different	different	ADJ
ajst-13993	54	13	identifiers	identifier	NOUN
ajst-13993	54	14	and	and	CCONJ
ajst-13993	54	15	then	then	ADV
ajst-13993	54	16	translate	translate	VERB
ajst-13993	54	17	them	they	PRON
ajst-13993	54	18	to	to	ADP
ajst-13993	54	19	lowercase	lowercase	NOUN
ajst-13993	54	20	letter	letter	NOUN
ajst-13993	54	21	.	.	PUNCT
ajst-13993	55	1	next	next	ADV
ajst-13993	55	2	,	,	PUNCT
ajst-13993	55	3	the	the	DET
ajst-13993	55	4	gotten	get	VERB
ajst-13993	55	5	words	word	NOUN
ajst-13993	55	6	are	be	AUX
ajst-13993	55	7	embedded	embed	VERB
ajst-13993	55	8	into	into	ADP
ajst-13993	55	9	vectors	vector	NOUN
ajst-13993	55	10	according	accord	VERB
ajst-13993	55	11	to	to	ADP
ajst-13993	55	12	the	the	DET
ajst-13993	55	13	module	module	NOUN
ajst-13993	55	14	genism	genism	NOUN
ajst-13993	55	15	in	in	ADP
ajst-13993	55	16	python	python	PROPN
ajst-13993	55	17	.	.	PUNCT
ajst-13993	56	1	the	the	DET
ajst-13993	56	2	tokens	token	NOUN
ajst-13993	56	3	have	have	AUX
ajst-13993	56	4	not	not	PART
ajst-13993	56	5	appeared	appear	VERB
ajst-13993	56	6	are	be	AUX
ajst-13993	56	7	regarded	regard	VERB
ajst-13993	56	8	as	as	ADP
ajst-13993	56	9	unknown	unknown	ADJ
ajst-13993	56	10	words	word	NOUN
ajst-13993	56	11	which	which	PRON
ajst-13993	56	12	is	be	AUX
ajst-13993	56	13	similar	similar	ADJ
ajst-13993	56	14	to	to	ADP
ajst-13993	56	15	[	[	X
ajst-13993	56	16	16	16	NUM
ajst-13993	56	17	,	,	PUNCT
ajst-13993	56	18	17	17	NUM
ajst-13993	56	19	,	,	PUNCT
ajst-13993	56	20	18	18	NUM
ajst-13993	56	21	]	]	PUNCT
ajst-13993	56	22	.	.	PUNCT
ajst-13993	57	1	in	in	ADP
ajst-13993	57	2	this	this	DET
ajst-13993	57	3	chapter	chapter	NOUN
ajst-13993	57	4	,	,	PUNCT
ajst-13993	57	5	we	we	PRON
ajst-13993	57	6	utilize	utilize	VERB
ajst-13993	57	7	the	the	DET
ajst-13993	57	8	following	follow	VERB
ajst-13993	57	9	three	three	NUM
ajst-13993	57	10	representations	representation	NOUN
ajst-13993	57	11	of	of	ADP
ajst-13993	57	12	source	source	NOUN
ajst-13993	57	13	code	code	NOUN
ajst-13993	57	14	:	:	PUNCT
ajst-13993	57	15	plain	plain	ADJ
ajst-13993	57	16	text	text	NOUN
ajst-13993	57	17	sequence	sequence	NOUN
ajst-13993	57	18	,	,	PUNCT
ajst-13993	57	19	the	the	DET
ajst-13993	57	20	sequenced	sequenced	ADJ
ajst-13993	57	21	abstract	abstract	ADJ
ajst-13993	57	22	syntax	syntax	NOUN
ajst-13993	57	23	tree	tree	NOUN
ajst-13993	57	24	,	,	PUNCT
ajst-13993	57	25	and	and	CCONJ
ajst-13993	57	26	the	the	DET
ajst-13993	57	27	sequenced	sequence	VERB
ajst-13993	57	28	control	control	NOUN
ajst-13993	57	29	flow	flow	NOUN
ajst-13993	57	30	graph	graph	NOUN
ajst-13993	57	31	.	.	PUNCT
ajst-13993	58	1	3.1.1	3.1.1	X
ajst-13993	58	2	.	.	PUNCT
ajst-13993	58	3	plain	plain	ADJ
ajst-13993	58	4	text	text	NOUN
ajst-13993	58	5	representation	representation	NOUN
ajst-13993	58	6	consider	consider	VERB
ajst-13993	58	7	that	that	SCONJ
ajst-13993	58	8	the	the	DET
ajst-13993	58	9	comments	comment	NOUN
ajst-13993	58	10	are	be	AUX
ajst-13993	58	11	usually	usually	ADV
ajst-13993	58	12	generated	generate	VERB
ajst-13993	58	13	by	by	ADP
ajst-13993	58	14	choosing	choose	VERB
ajst-13993	58	15	the	the	DET
ajst-13993	58	16	lexical	lexical	ADJ
ajst-13993	58	17	tokens	token	NOUN
ajst-13993	58	18	of	of	ADP
ajst-13993	58	19	the	the	DET
ajst-13993	58	20	program	program	NOUN
ajst-13993	58	21	code	code	NOUN
ajst-13993	58	22	,	,	PUNCT
ajst-13993	58	23	for	for	ADP
ajst-13993	58	24	example	example	NOUN
ajst-13993	58	25	the	the	DET
ajst-13993	58	26	function	function	NOUN
ajst-13993	58	27	name	name	NOUN
ajst-13993	58	28	,	,	PUNCT
ajst-13993	58	29	operator	operator	NOUN
ajst-13993	58	30	name	name	NOUN
ajst-13993	58	31	etc	etc	X
ajst-13993	58	32	.	.	X
ajst-13993	58	33	,	,	PUNCT
ajst-13993	58	34	thus	thus	ADV
ajst-13993	58	35	one	one	NUM
ajst-13993	58	36	main	main	ADJ
ajst-13993	58	37	representation	representation	NOUN
ajst-13993	58	38	of	of	ADP
ajst-13993	58	39	the	the	DET
ajst-13993	58	40	source	source	NOUN
ajst-13993	58	41	code	code	NOUN
ajst-13993	58	42	is	be	AUX
ajst-13993	58	43	the	the	DET
ajst-13993	58	44	plain	plain	ADJ
ajst-13993	58	45	text	text	NOUN
ajst-13993	58	46	.	.	PUNCT
ajst-13993	59	1	3.1.2	3.1.2	X
ajst-13993	59	2	.	.	PUNCT
ajst-13993	59	3	structural	structural	ADJ
ajst-13993	59	4	representation	representation	NOUN
ajst-13993	59	5	when	when	SCONJ
ajst-13993	59	6	executing	execute	VERB
ajst-13993	59	7	the	the	DET
ajst-13993	59	8	program	program	NOUN
ajst-13993	59	9	,	,	PUNCT
ajst-13993	59	10	the	the	DET
ajst-13993	59	11	compiler	compiler	NOUN
ajst-13993	59	12	transforms	transform	VERB
ajst-13993	59	13	the	the	DET
ajst-13993	59	14	source	source	NOUN
ajst-13993	59	15	code	code	NOUN
ajst-13993	59	16	into	into	ADP
ajst-13993	59	17	intermediate	intermediate	ADJ
ajst-13993	59	18	code	code	NOUN
ajst-13993	59	19	by	by	ADP
ajst-13993	59	20	constructing	construct	VERB
ajst-13993	59	21	the	the	DET
ajst-13993	59	22	abstract	abstract	ADJ
ajst-13993	59	23	syntax	syntax	NOUN
ajst-13993	59	24	tree	tree	NOUN
ajst-13993	59	25	[	[	X
ajst-13993	59	26	19	19	NUM
ajst-13993	59	27	]	]	PUNCT
ajst-13993	59	28	and	and	CCONJ
ajst-13993	59	29	then	then	ADV
ajst-13993	59	30	translating	translate	VERB
ajst-13993	59	31	to	to	PART
ajst-13993	59	32	control	control	VERB
ajst-13993	59	33	flow	flow	NOUN
ajst-13993	59	34	graph	graph	NOUN
ajst-13993	59	35	which	which	PRON
ajst-13993	59	36	reflects	reflect	VERB
ajst-13993	59	37	the	the	DET
ajst-13993	59	38	vital	vital	ADJ
ajst-13993	59	39	information	information	NOUN
ajst-13993	59	40	of	of	ADP
ajst-13993	59	41	the	the	DET
ajst-13993	59	42	source	source	NOUN
ajst-13993	59	43	code	code	NOUN
ajst-13993	59	44	.	.	PUNCT
ajst-13993	60	1	therefore	therefore	ADV
ajst-13993	60	2	,	,	PUNCT
ajst-13993	60	3	the	the	DET
ajst-13993	60	4	abstract	abstract	ADJ
ajst-13993	60	5	syntax	syntax	NOUN
ajst-13993	60	6	tree	tree	NOUN
ajst-13993	60	7	and	and	CCONJ
ajst-13993	60	8	control	control	NOUN
ajst-13993	60	9	flow	flow	NOUN
ajst-13993	60	10	graph	graph	NOUN
ajst-13993	60	11	are	be	AUX
ajst-13993	60	12	also	also	ADV
ajst-13993	60	13	adopted	adopt	VERB
ajst-13993	60	14	as	as	ADP
ajst-13993	60	15	the	the	DET
ajst-13993	60	16	structural	structural	ADJ
ajst-13993	60	17	representations	representation	NOUN
ajst-13993	60	18	of	of	ADP
ajst-13993	60	19	the	the	DET
ajst-13993	60	20	source	source	NOUN
ajst-13993	60	21	code	code	NOUN
ajst-13993	60	22	.	.	PUNCT
ajst-13993	61	1	based	base	VERB
ajst-13993	61	2	on	on	ADP
ajst-13993	61	3	the	the	DET
ajst-13993	61	4	ast	ast	NOUN
ajst-13993	61	5	module	module	NOUN
ajst-13993	61	6	[	[	X
ajst-13993	61	7	20	20	NUM
ajst-13993	61	8	]	]	PUNCT
ajst-13993	61	9	in	in	ADP
ajst-13993	61	10	python	python	PROPN
ajst-13993	61	11	,	,	PUNCT
ajst-13993	61	12	the	the	DET
ajst-13993	61	13	sequenced	sequenced	ADJ
ajst-13993	61	14	abstract	abstract	ADJ
ajst-13993	61	15	syntax	syntax	NOUN
ajst-13993	61	16	tree	tree	NOUN
ajst-13993	61	17	can	can	AUX
ajst-13993	61	18	be	be	AUX
ajst-13993	61	19	obtained	obtain	VERB
ajst-13993	61	20	.	.	PUNCT
ajst-13993	62	1	for	for	ADP
ajst-13993	62	2	the	the	DET
ajst-13993	62	3	control	control	NOUN
ajst-13993	62	4	flow	flow	NOUN
ajst-13993	62	5	graph	graph	NOUN
ajst-13993	62	6	representation	representation	NOUN
ajst-13993	62	7	of	of	ADP
ajst-13993	62	8	the	the	DET
ajst-13993	62	9	source	source	NOUN
ajst-13993	62	10	code	code	NOUN
ajst-13993	62	11	,	,	PUNCT
ajst-13993	62	12	each	each	DET
ajst-13993	62	13	node	node	NOUN
ajst-13993	62	14	includes	include	VERB
ajst-13993	62	15	a	a	DET
ajst-13993	62	16	sequence	sequence	NOUN
ajst-13993	62	17	of	of	ADP
ajst-13993	62	18	tokens	token	NOUN
ajst-13993	62	19	which	which	PRON
ajst-13993	62	20	composes	compose	VERB
ajst-13993	62	21	the	the	DET
ajst-13993	62	22	statement	statement	NOUN
ajst-13993	62	23	and	and	CCONJ
ajst-13993	62	24	each	each	DET
ajst-13993	62	25	edge	edge	NOUN
ajst-13993	62	26	which	which	PRON
ajst-13993	62	27	connects	connect	VERB
ajst-13993	62	28	two	two	NUM
ajst-13993	62	29	nodes	node	NOUN
ajst-13993	62	30	represents	represent	VERB
ajst-13993	62	31	the	the	DET
ajst-13993	62	32	flow	flow	NOUN
ajst-13993	62	33	of	of	ADP
ajst-13993	62	34	statements	statement	NOUN
ajst-13993	62	35	in	in	ADP
ajst-13993	62	36	the	the	DET
ajst-13993	62	37	source	source	NOUN
ajst-13993	62	38	code	code	NOUN
ajst-13993	62	39	.	.	PUNCT
ajst-13993	63	1	following	follow	VERB
ajst-13993	63	2	the	the	DET
ajst-13993	63	3	ast	ast	NOUN
ajst-13993	63	4	module	module	NOUN
ajst-13993	63	5	[	[	X
ajst-13993	63	6	1	1	NUM
ajst-13993	63	7	]	]	PUNCT
ajst-13993	63	8	and	and	CCONJ
ajst-13993	63	9	[	[	X
ajst-13993	63	10	15	15	NUM
ajst-13993	63	11	]	]	PUNCT
ajst-13993	63	12	,	,	PUNCT
ajst-13993	63	13	the	the	DET
ajst-13993	63	14	control	control	NOUN
ajst-13993	63	15	flow	flow	NOUN
ajst-13993	63	16	graph	graph	NOUN
ajst-13993	63	17	can	can	AUX
ajst-13993	63	18	be	be	AUX
ajst-13993	63	19	obtained	obtain	VERB
ajst-13993	63	20	and	and	CCONJ
ajst-13993	63	21	then	then	ADV
ajst-13993	63	22	it	it	PRON
ajst-13993	63	23	is	be	AUX
ajst-13993	63	24	76	76	NUM
ajst-13993	63	25	transformed	transform	VERB
ajst-13993	63	26	to	to	PART
ajst-13993	63	27	get	get	VERB
ajst-13993	63	28	the	the	DET
ajst-13993	63	29	control	control	NOUN
ajst-13993	63	30	flow	flow	NOUN
ajst-13993	63	31	sequence	sequence	NOUN
ajst-13993	63	32	in	in	ADP
ajst-13993	63	33	depth	depth	NOUN
ajst-13993	63	34	-	-	PUNCT
ajst-13993	63	35	first	first	ADJ
ajst-13993	63	36	order	order	NOUN
ajst-13993	63	37	.	.	PUNCT
ajst-13993	64	1	3.1.3	3.1.3	NUM
ajst-13993	64	2	.	.	PUNCT
ajst-13993	64	3	hybrid	hybrid	ADJ
ajst-13993	64	4	hierarchical	hierarchical	ADJ
ajst-13993	64	5	attention	attention	NOUN
ajst-13993	64	6	network	network	NOUN
ajst-13993	64	7	different	different	ADJ
ajst-13993	64	8	part	part	NOUN
ajst-13993	64	9	of	of	ADP
ajst-13993	64	10	the	the	DET
ajst-13993	64	11	source	source	NOUN
ajst-13993	64	12	code	code	NOUN
ajst-13993	64	13	contributes	contribute	VERB
ajst-13993	64	14	differently	differently	ADV
ajst-13993	64	15	to	to	ADP
ajst-13993	64	16	the	the	DET
ajst-13993	64	17	final	final	ADJ
ajst-13993	64	18	output	output	NOUN
ajst-13993	64	19	of	of	ADP
ajst-13993	64	20	the	the	DET
ajst-13993	64	21	comment	comment	NOUN
ajst-13993	64	22	.	.	PUNCT
ajst-13993	65	1	in	in	ADP
ajst-13993	65	2	specific	specific	ADJ
ajst-13993	65	3	,	,	PUNCT
ajst-13993	65	4	in	in	ADP
ajst-13993	65	5	different	different	ADJ
ajst-13993	65	6	code	code	NOUN
ajst-13993	65	7	snippet	snippet	NOUN
ajst-13993	65	8	,	,	PUNCT
ajst-13993	65	9	the	the	DET
ajst-13993	65	10	same	same	ADJ
ajst-13993	65	11	token	token	NOUN
ajst-13993	65	12	or	or	CCONJ
ajst-13993	65	13	statement	statement	NOUN
ajst-13993	65	14	is	be	AUX
ajst-13993	65	15	deferentially	deferentially	ADV
ajst-13993	65	16	important	important	ADJ
ajst-13993	65	17	.	.	PUNCT
ajst-13993	66	1	besides	besides	ADV
ajst-13993	66	2	,	,	PUNCT
ajst-13993	66	3	the	the	DET
ajst-13993	66	4	source	source	NOUN
ajst-13993	66	5	code	code	NOUN
ajst-13993	66	6	has	have	VERB
ajst-13993	66	7	the	the	DET
ajst-13993	66	8	hierarchical	hierarchical	ADJ
ajst-13993	66	9	structure	structure	NOUN
ajst-13993	66	10	essentially	essentially	ADV
ajst-13993	66	11	,	,	PUNCT
ajst-13993	66	12	for	for	ADP
ajst-13993	66	13	example	example	NOUN
ajst-13993	66	14	statements	statement	NOUN
ajst-13993	66	15	are	be	AUX
ajst-13993	66	16	consisted	consist	VERB
ajst-13993	66	17	of	of	ADP
ajst-13993	66	18	tokens	token	NOUN
ajst-13993	66	19	and	and	CCONJ
ajst-13993	66	20	the	the	DET
ajst-13993	66	21	functions	function	NOUN
ajst-13993	66	22	are	be	AUX
ajst-13993	66	23	consisted	consist	VERB
ajst-13993	66	24	of	of	ADP
ajst-13993	66	25	statements	statement	NOUN
ajst-13993	66	26	respectively	respectively	ADV
ajst-13993	66	27	.	.	PUNCT
ajst-13993	67	1	thus	thus	ADV
ajst-13993	67	2	,	,	PUNCT
ajst-13993	67	3	we	we	PRON
ajst-13993	67	4	introduce	introduce	VERB
ajst-13993	67	5	the	the	DET
ajst-13993	67	6	hierarchical	hierarchical	ADJ
ajst-13993	67	7	attention	attention	NOUN
ajst-13993	67	8	mechanism	mechanism	NOUN
ajst-13993	67	9	[	[	X
ajst-13993	67	10	21	21	NUM
ajst-13993	67	11	]	]	X
ajst-13993	67	12	,	,	PUNCT
ajst-13993	67	13	which	which	PRON
ajst-13993	67	14	is	be	AUX
ajst-13993	67	15	proposed	propose	VERB
ajst-13993	67	16	in	in	ADP
ajst-13993	67	17	natural	natural	ADJ
ajst-13993	67	18	language	language	NOUN
ajst-13993	67	19	processing	processing	NOUN
ajst-13993	67	20	,	,	PUNCT
ajst-13993	67	21	to	to	PART
ajst-13993	67	22	enable	enable	VERB
ajst-13993	67	23	the	the	DET
ajst-13993	67	24	proposed	propose	VERB
ajst-13993	67	25	comment	comment	NOUN
ajst-13993	67	26	generation	generation	NOUN
ajst-13993	67	27	approach	approach	NOUN
ajst-13993	67	28	to	to	PART
ajst-13993	67	29	pay	pay	VERB
ajst-13993	67	30	attention	attention	NOUN
ajst-13993	67	31	deferentially	deferentially	ADV
ajst-13993	67	32	to	to	ADP
ajst-13993	67	33	different	different	ADJ
ajst-13993	67	34	statements	statement	NOUN
ajst-13993	67	35	and	and	CCONJ
ajst-13993	67	36	tokens	token	NOUN
ajst-13993	67	37	respectively	respectively	ADV
ajst-13993	67	38	when	when	SCONJ
ajst-13993	67	39	forming	form	VERB
ajst-13993	67	40	the	the	DET
ajst-13993	67	41	representations	representation	NOUN
ajst-13993	67	42	of	of	ADP
ajst-13993	67	43	the	the	DET
ajst-13993	67	44	source	source	NOUN
ajst-13993	67	45	code	code	NOUN
ajst-13993	67	46	.	.	PUNCT
ajst-13993	68	1	as	as	SCONJ
ajst-13993	68	2	shown	show	VERB
ajst-13993	68	3	in	in	ADP
ajst-13993	68	4	figure	figure	NOUN
ajst-13993	68	5	1	1	NUM
ajst-13993	68	6	(	(	PUNCT
ajst-13993	68	7	b	b	NOUN
ajst-13993	68	8	)	)	PUNCT
ajst-13993	68	9	,	,	PUNCT
ajst-13993	68	10	a	a	DET
ajst-13993	68	11	two	two	NUM
ajst-13993	68	12	-	-	PUNCT
ajst-13993	68	13	layer	layer	NOUN
ajst-13993	68	14	attention	attention	NOUN
ajst-13993	68	15	network	network	NOUN
ajst-13993	68	16	are	be	AUX
ajst-13993	68	17	utilized	utilize	VERB
ajst-13993	68	18	in	in	ADP
ajst-13993	68	19	the	the	DET
ajst-13993	68	20	proposed	propose	VERB
ajst-13993	68	21	approach	approach	NOUN
ajst-13993	68	22	,	,	PUNCT
ajst-13993	68	23	they	they	PRON
ajst-13993	68	24	are	be	AUX
ajst-13993	68	25	ate	eat	VERB
ajst-13993	68	26	the	the	DET
ajst-13993	68	27	token	token	ADJ
ajst-13993	68	28	layer	layer	NOUN
ajst-13993	68	29	and	and	CCONJ
ajst-13993	68	30	the	the	DET
ajst-13993	68	31	statement	statement	NOUN
ajst-13993	68	32	layer	layer	NOUN
ajst-13993	68	33	respectively	respectively	ADV
ajst-13993	68	34	.	.	PUNCT
ajst-13993	69	1	the	the	DET
ajst-13993	69	2	utilized	utilize	VERB
ajst-13993	69	3	network	network	NOUN
ajst-13993	69	4	is	be	AUX
ajst-13993	69	5	consisted	consist	VERB
ajst-13993	69	6	of	of	ADP
ajst-13993	69	7	four	four	NUM
ajst-13993	69	8	parts	part	NOUN
ajst-13993	69	9	:	:	PUNCT
ajst-13993	69	10	the	the	DET
ajst-13993	69	11	token	token	ADJ
ajst-13993	69	12	encoder	encoder	NOUN
ajst-13993	69	13	,	,	PUNCT
ajst-13993	69	14	the	the	DET
ajst-13993	69	15	token	token	VERB
ajst-13993	69	16	-	-	PUNCT
ajst-13993	69	17	level	level	NOUN
ajst-13993	69	18	attention	attention	NOUN
ajst-13993	69	19	,	,	PUNCT
ajst-13993	69	20	the	the	DET
ajst-13993	69	21	statement	statement	NOUN
ajst-13993	69	22	encoder	encoder	NOUN
ajst-13993	69	23	,	,	PUNCT
ajst-13993	69	24	and	and	CCONJ
ajst-13993	69	25	the	the	DET
ajst-13993	69	26	statement	statement	NOUN
ajst-13993	69	27	-	-	PUNCT
ajst-13993	69	28	level	level	NOUN
ajst-13993	69	29	attention	attention	NOUN
ajst-13993	69	30	.	.	PUNCT
ajst-13993	70	1	assumed	assume	VERB
ajst-13993	70	2	that	that	SCONJ
ajst-13993	70	3	the	the	DET
ajst-13993	70	4	vectors	vector	NOUN
ajst-13993	70	5	of	of	ADP
ajst-13993	70	6	the	the	DET
ajst-13993	70	7	the	the	DET
ajst-13993	70	8	three	three	NUM
ajst-13993	70	9	code	code	NOUN
ajst-13993	70	10	representations	representation	NOUN
ajst-13993	70	11	are	be	AUX
ajst-13993	70	12	represented	represent	VERB
ajst-13993	70	13	as	as	ADP
ajst-13993	70	14	dtxt	dtxt	ADJ
ajst-13993	70	15	,	,	PUNCT
ajst-13993	70	16	dast	dast	INTJ
ajst-13993	70	17	,	,	PUNCT
ajst-13993	70	18	and	and	CCONJ
ajst-13993	70	19	dcfg	dcfg	PROPN
ajst-13993	70	20	respectively	respectively	ADV
ajst-13993	70	21	.	.	PUNCT
ajst-13993	71	1	then	then	ADV
ajst-13993	71	2	,	,	PUNCT
ajst-13993	71	3	they	they	PRON
ajst-13993	71	4	are	be	AUX
ajst-13993	71	5	integrated	integrate	VERB
ajst-13993	71	6	into	into	ADP
ajst-13993	71	7	one	one	NUM
ajst-13993	71	8	vector	vector	NOUN
ajst-13993	71	9	d	d	NOUN
ajst-13993	71	10	to	to	PART
ajst-13993	71	11	obtain	obtain	VERB
ajst-13993	71	12	the	the	DET
ajst-13993	71	13	final	final	ADJ
ajst-13993	71	14	representation	representation	NOUN
ajst-13993	71	15	of	of	ADP
ajst-13993	71	16	the	the	DET
ajst-13993	71	17	source	source	NOUN
ajst-13993	71	18	code	code	NOUN
ajst-13993	71	19	.	.	PUNCT
ajst-13993	72	1	the	the	DET
ajst-13993	72	2	detailed	detailed	ADJ
ajst-13993	72	3	information	information	NOUN
ajst-13993	72	4	of	of	ADP
ajst-13993	72	5	this	this	DET
ajst-13993	72	6	network	network	NOUN
ajst-13993	72	7	are	be	AUX
ajst-13993	72	8	described	describe	VERB
ajst-13993	72	9	as	as	SCONJ
ajst-13993	72	10	follows	follow	VERB
ajst-13993	72	11	.	.	PUNCT
ajst-13993	73	1	the	the	DET
ajst-13993	73	2	token	token	ADJ
ajst-13993	73	3	level	level	NOUN
ajst-13993	73	4	encoding	encoding	NOUN
ajst-13993	73	5	.	.	PUNCT
ajst-13993	74	1	assuming	assume	VERB
ajst-13993	74	2	that	that	SCONJ
ajst-13993	74	3	a	a	DET
ajst-13993	74	4	sequence	sequence	NOUN
ajst-13993	74	5	of	of	ADP
ajst-13993	74	6	tokens	token	NOUN
ajst-13993	74	7	xi0	xi0	PROPN
ajst-13993	74	8	,	,	PUNCT
ajst-13993	74	9	...	...	PUNCT
ajst-13993	74	10	,	,	PUNCT
ajst-13993	74	11	xiti−1	xiti−1	PROPN
ajst-13993	74	12	consists	consist	VERB
ajst-13993	74	13	a	a	DET
ajst-13993	74	14	statement	statement	NOUN
ajst-13993	74	15	si	si	NOUN
ajst-13993	74	16	,	,	PUNCT
ajst-13993	74	17	and	and	CCONJ
ajst-13993	74	18	ti	ti	NOUN
ajst-13993	74	19	means	mean	VERB
ajst-13993	74	20	the	the	DET
ajst-13993	74	21	length	length	NOUN
ajst-13993	74	22	of	of	ADP
ajst-13993	74	23	the	the	DET
ajst-13993	74	24	tokens	token	NOUN
ajst-13993	74	25	in	in	ADP
ajst-13993	74	26	the	the	DET
ajst-13993	74	27	statement	statement	NOUN
ajst-13993	74	28	sequence	sequence	NOUN
ajst-13993	74	29	.	.	PUNCT
ajst-13993	75	1	firstly	firstly	ADV
ajst-13993	75	2	,	,	PUNCT
ajst-13993	75	3	the	the	DET
ajst-13993	75	4	tokens	token	NOUN
ajst-13993	75	5	are	be	AUX
ajst-13993	75	6	embedded	embed	VERB
ajst-13993	75	7	into	into	ADP
ajst-13993	75	8	vectors	vector	NOUN
ajst-13993	75	9	by	by	ADP
ajst-13993	75	10	the	the	DET
ajst-13993	75	11	embedding	embed	VERB
ajst-13993	75	12	matrix	matrix	NOUN
ajst-13993	75	13	wi	wi	PROPN
ajst-13993	75	14	,	,	PUNCT
ajst-13993	75	15	namely	namely	ADV
ajst-13993	75	16	,	,	PUNCT
ajst-13993	75	17	vit	vit	NOUN
ajst-13993	75	18	=	=	NOUN
ajst-13993	75	19	wixit	wixit	NOUN
ajst-13993	75	20	.	.	PUNCT
ajst-13993	76	1	then	then	ADV
ajst-13993	76	2	,	,	PUNCT
ajst-13993	76	3	as	as	SCONJ
ajst-13993	76	4	shown	show	VERB
ajst-13993	76	5	in	in	ADP
ajst-13993	76	6	equation(1	equation(1	PROPN
ajst-13993	76	7	)	)	PUNCT
ajst-13993	76	8	,	,	PUNCT
ajst-13993	76	9	the	the	DET
ajst-13993	76	10	lstm	lstm	PROPN
ajst-13993	76	11	model	model	NOUN
ajst-13993	76	12	is	be	AUX
ajst-13993	76	13	utilized	utilize	VERB
ajst-13993	76	14	to	to	PART
ajst-13993	76	15	encode	encode	VERB
ajst-13993	76	16	the	the	DET
ajst-13993	76	17	tokens	token	NOUN
ajst-13993	76	18	from	from	ADP
ajst-13993	76	19	xi0	xi0	PROPN
ajst-13993	76	20	to	to	ADP
ajst-13993	76	21	xiti−1	xiti−1	PROPN
ajst-13993	76	22	in	in	ADP
ajst-13993	76	23	the	the	DET
ajst-13993	76	24	statement	statement	NOUN
ajst-13993	76	25	.	.	PUNCT
ajst-13993	77	1	(	(	PUNCT
ajst-13993	77	2	1	1	X
ajst-13993	77	3	)	)	PUNCT
ajst-13993	77	4	consider	consider	VERB
ajst-13993	77	5	that	that	SCONJ
ajst-13993	77	6	different	different	ADJ
ajst-13993	77	7	tokens	token	NOUN
ajst-13993	77	8	in	in	ADP
ajst-13993	77	9	the	the	DET
ajst-13993	77	10	statement	statement	NOUN
ajst-13993	77	11	supply	supply	VERB
ajst-13993	77	12	different	different	ADJ
ajst-13993	77	13	semantic	semantic	ADJ
ajst-13993	77	14	information	information	NOUN
ajst-13993	77	15	and	and	CCONJ
ajst-13993	77	16	contribute	contribute	VERB
ajst-13993	77	17	differently	differently	ADV
ajst-13993	77	18	to	to	ADP
ajst-13993	77	19	the	the	DET
ajst-13993	77	20	generated	generate	VERB
ajst-13993	77	21	comment	comment	NOUN
ajst-13993	77	22	.	.	PUNCT
ajst-13993	78	1	for	for	ADP
ajst-13993	78	2	example	example	NOUN
ajst-13993	78	3	,	,	PUNCT
ajst-13993	78	4	in	in	ADP
ajst-13993	78	5	figure	figure	NOUN
ajst-13993	78	6	1	1	NUM
ajst-13993	78	7	which	which	PRON
ajst-13993	78	8	shows	show	VERB
ajst-13993	78	9	the	the	DET
ajst-13993	78	10	representation	representation	NOUN
ajst-13993	78	11	of	of	ADP
ajst-13993	78	12	the	the	DET
ajst-13993	78	13	source	source	NOUN
ajst-13993	78	14	code	code	NOUN
ajst-13993	78	15	in	in	ADP
ajst-13993	78	16	the	the	DET
ajst-13993	78	17	editor	editor	NOUN
ajst-13993	78	18	,	,	PUNCT
ajst-13993	78	19	as	as	SCONJ
ajst-13993	78	20	the	the	DET
ajst-13993	78	21	words	word	NOUN
ajst-13993	78	22	“	"	PUNCT
ajst-13993	78	23	numbers	number	NOUN
ajst-13993	78	24	”	"	PUNCT
ajst-13993	78	25	and	and	CCONJ
ajst-13993	78	26	“	"	PUNCT
ajst-13993	78	27	string	string	NOUN
ajst-13993	78	28	”	"	PUNCT
ajst-13993	78	29	are	be	AUX
ajst-13993	78	30	contained	contain	VERB
ajst-13993	78	31	in	in	ADP
ajst-13993	78	32	the	the	DET
ajst-13993	78	33	given	give	VERB
ajst-13993	78	34	comment	comment	NOUN
ajst-13993	78	35	,	,	PUNCT
ajst-13993	78	36	thus	thus	ADV
ajst-13993	78	37	the	the	DET
ajst-13993	78	38	tokens	token	NOUN
ajst-13993	78	39	“	"	PUNCT
ajst-13993	78	40	number	number	NOUN
ajst-13993	78	41	”	"	PUNCT
ajst-13993	78	42	and	and	CCONJ
ajst-13993	78	43	“	"	PUNCT
ajst-13993	78	44	str	str	NOUN
ajst-13993	78	45	”	"	PUNCT
ajst-13993	78	46	can	can	AUX
ajst-13993	78	47	supply	supply	VERB
ajst-13993	78	48	more	more	ADJ
ajst-13993	78	49	information	information	NOUN
ajst-13993	78	50	than	than	ADP
ajst-13993	78	51	token	token	ADJ
ajst-13993	78	52	“	"	PUNCT
ajst-13993	78	53	def	def	ADJ
ajst-13993	78	54	”	"	PUNCT
ajst-13993	78	55	in	in	ADP
ajst-13993	78	56	statement	statement	NOUN
ajst-13993	78	57	“	"	PUNCT
ajst-13993	78	58	def	def	ADJ
ajst-13993	78	59	check_number_exist(str	check_number_exist(str	PROPN
ajst-13993	78	60	):	):	PUNCT
ajst-13993	78	61	”	"	PUNCT
ajst-13993	78	62	essentially	essentially	ADV
ajst-13993	78	63	.	.	PUNCT
ajst-13993	79	1	therefore	therefore	ADV
ajst-13993	79	2	,	,	PUNCT
ajst-13993	79	3	the	the	DET
ajst-13993	79	4	attention	attention	NOUN
ajst-13993	79	5	mechanism	mechanism	NOUN
ajst-13993	79	6	is	be	AUX
ajst-13993	79	7	adopted	adopt	VERB
ajst-13993	79	8	by	by	ADP
ajst-13993	79	9	the	the	DET
ajst-13993	79	10	proposed	propose	VERB
ajst-13993	79	11	approach	approach	NOUN
ajst-13993	79	12	to	to	PART
ajst-13993	79	13	collect	collect	VERB
ajst-13993	79	14	the	the	DET
ajst-13993	79	15	tokens	token	NOUN
ajst-13993	79	16	that	that	PRON
ajst-13993	79	17	are	be	AUX
ajst-13993	79	18	more	more	ADV
ajst-13993	79	19	important	important	ADJ
ajst-13993	79	20	for	for	ADP
ajst-13993	79	21	the	the	DET
ajst-13993	79	22	generating	generating	NOUN
ajst-13993	79	23	of	of	ADP
ajst-13993	79	24	the	the	DET
ajst-13993	79	25	comment	comment	NOUN
ajst-13993	79	26	and	and	CCONJ
ajst-13993	79	27	the	the	DET
ajst-13993	79	28	extracted	extract	VERB
ajst-13993	79	29	words	word	NOUN
ajst-13993	79	30	are	be	AUX
ajst-13993	79	31	merged	merge	VERB
ajst-13993	79	32	to	to	PART
ajst-13993	79	33	generate	generate	VERB
ajst-13993	79	34	the	the	DET
ajst-13993	79	35	representation	representation	NOUN
ajst-13993	79	36	of	of	ADP
ajst-13993	79	37	the	the	DET
ajst-13993	79	38	corresponding	corresponding	ADJ
ajst-13993	79	39	statement	statement	NOUN
ajst-13993	79	40	as	as	SCONJ
ajst-13993	79	41	shown	show	VERB
ajst-13993	79	42	in	in	ADP
ajst-13993	79	43	equation	equation	NOUN
ajst-13993	79	44	(	(	PUNCT
ajst-13993	79	45	2	2	NUM
ajst-13993	79	46	)	)	PUNCT
ajst-13993	79	47	.	.	PUNCT
ajst-13993	80	1	(	(	PUNCT
ajst-13993	80	2	2	2	X
ajst-13993	80	3	)	)	PUNCT
ajst-13993	80	4	figure	figure	NOUN
ajst-13993	80	5	2	2	NUM
ajst-13993	80	6	.	.	PUNCT
ajst-13993	81	1	the	the	DET
ajst-13993	81	2	source	source	NOUN
ajst-13993	81	3	code	code	NOUN
ajst-13993	81	4	example	example	NOUN
ajst-13993	81	5	that	that	SCONJ
ajst-13993	81	6	tokens	token	NOUN
ajst-13993	81	7	supply	supply	VERB
ajst-13993	81	8	information	information	NOUN
ajst-13993	81	9	differently	differently	ADV
ajst-13993	81	10	for	for	ADP
ajst-13993	81	11	comment	comment	NOUN
ajst-13993	81	12	generation	generation	NOUN
ajst-13993	81	13	.	.	PUNCT
ajst-13993	82	1	wx	wx	PROPN
ajst-13993	82	2	represents	represent	VERB
ajst-13993	82	3	the	the	DET
ajst-13993	82	4	weight	weight	NOUN
ajst-13993	82	5	matrix	matrix	NOUN
ajst-13993	82	6	,	,	PUNCT
ajst-13993	82	7	bx	bx	PROPN
ajst-13993	82	8	denotes	denote	VERB
ajst-13993	82	9	the	the	DET
ajst-13993	82	10	bias	bias	PROPN
ajst-13993	82	11	vector	vector	NOUN
ajst-13993	82	12	,	,	PUNCT
ajst-13993	82	13	the	the	DET
ajst-13993	82	14	attention	attention	NOUN
ajst-13993	82	15	from	from	ADP
ajst-13993	82	16	token	token	ADJ
ajst-13993	82	17	xit	xit	PROPN
ajst-13993	82	18	to	to	ADP
ajst-13993	82	19	the	the	DET
ajst-13993	82	20	statement	statement	NOUN
ajst-13993	82	21	si	si	PROPN
ajst-13993	82	22	is	be	AUX
ajst-13993	82	23	signed	sign	VERB
ajst-13993	82	24	as	as	ADP
ajst-13993	82	25	αit	αit	NOUN
ajst-13993	82	26	,	,	PUNCT
ajst-13993	82	27	and	and	CCONJ
ajst-13993	82	28	ux	ux	PROPN
ajst-13993	82	29	means	mean	VERB
ajst-13993	82	30	the	the	DET
ajst-13993	82	31	sequence	sequence	NOUN
ajst-13993	82	32	vector	vector	NOUN
ajst-13993	82	33	in	in	ADP
ajst-13993	82	34	the	the	DET
ajst-13993	82	35	token	token	ADJ
ajst-13993	82	36	level	level	NOUN
ajst-13993	82	37	.	.	PUNCT
ajst-13993	83	1	particularly	particularly	ADV
ajst-13993	83	2	,	,	PUNCT
ajst-13993	83	3	ux	ux	PROPN
ajst-13993	83	4	is	be	AUX
ajst-13993	83	5	initialized	initialize	VERB
ajst-13993	83	6	randomly	randomly	ADV
ajst-13993	83	7	and	and	CCONJ
ajst-13993	83	8	optimized	optimize	VERB
ajst-13993	83	9	during	during	ADP
ajst-13993	83	10	the	the	DET
ajst-13993	83	11	training	training	NOUN
ajst-13993	83	12	process	process	NOUN
ajst-13993	83	13	gradually	gradually	ADV
ajst-13993	83	14	.	.	PUNCT
ajst-13993	84	1	the	the	DET
ajst-13993	84	2	statement	statement	NOUN
ajst-13993	84	3	level	level	NOUN
ajst-13993	84	4	encoding	encoding	NOUN
ajst-13993	84	5	.	.	PUNCT
ajst-13993	85	1	as	as	SCONJ
ajst-13993	85	2	the	the	DET
ajst-13993	85	3	statement	statement	NOUN
ajst-13993	85	4	vector	vector	NOUN
ajst-13993	85	5	si	si	PROPN
ajst-13993	85	6	has	have	AUX
ajst-13993	85	7	been	be	AUX
ajst-13993	85	8	obtained	obtain	VERB
ajst-13993	85	9	,	,	PUNCT
ajst-13993	85	10	similar	similar	ADJ
ajst-13993	85	11	to	to	ADP
ajst-13993	85	12	the	the	DET
ajst-13993	85	13	encoding	encoding	NOUN
ajst-13993	85	14	of	of	ADP
ajst-13993	85	15	the	the	DET
ajst-13993	85	16	tokens	token	NOUN
ajst-13993	85	17	,	,	PUNCT
ajst-13993	85	18	the	the	DET
ajst-13993	85	19	function	function	NOUN
ajst-13993	85	20	vector	vector	NOUN
ajst-13993	85	21	can	can	AUX
ajst-13993	85	22	be	be	AUX
ajst-13993	85	23	generated	generate	VERB
ajst-13993	85	24	.	.	PUNCT
ajst-13993	86	1	firstly	firstly	ADV
ajst-13993	86	2	,	,	PUNCT
ajst-13993	86	3	the	the	DET
ajst-13993	86	4	statements	statement	NOUN
ajst-13993	86	5	are	be	AUX
ajst-13993	86	6	encoded	encode	VERB
ajst-13993	86	7	by	by	ADP
ajst-13993	86	8	lstm	lstm	NOUN
ajst-13993	86	9	according	accord	VERB
ajst-13993	86	10	to	to	ADP
ajst-13993	86	11	equation	equation	NOUN
ajst-13993	86	12	(	(	PUNCT
ajst-13993	86	13	3	3	NUM
ajst-13993	86	14	)	)	PUNCT
ajst-13993	86	15	.	.	PUNCT
ajst-13993	87	1	(	(	PUNCT
ajst-13993	87	2	3	3	X
ajst-13993	87	3	)	)	PUNCT
ajst-13993	87	4	where	where	SCONJ
ajst-13993	87	5	l	l	NOUN
ajst-13993	87	6	means	mean	VERB
ajst-13993	87	7	the	the	DET
ajst-13993	87	8	statements	statement	NOUN
ajst-13993	87	9	number	number	NOUN
ajst-13993	87	10	contained	contain	VERB
ajst-13993	87	11	in	in	ADP
ajst-13993	87	12	the	the	DET
ajst-13993	87	13	code	code	NOUN
ajst-13993	87	14	snippet	snippet	NOUN
ajst-13993	87	15	.	.	PUNCT
ajst-13993	88	1	to	to	PART
ajst-13993	88	2	extract	extract	VERB
ajst-13993	88	3	the	the	DET
ajst-13993	88	4	statements	statement	NOUN
ajst-13993	88	5	which	which	PRON
ajst-13993	88	6	contain	contain	VERB
ajst-13993	88	7	more	more	ADV
ajst-13993	88	8	important	important	ADJ
ajst-13993	88	9	information	information	NOUN
ajst-13993	88	10	to	to	ADP
ajst-13993	88	11	the	the	DET
ajst-13993	88	12	corresponding	correspond	VERB
ajst-13993	88	13	code	code	NOUN
ajst-13993	88	14	snippet	snippet	NOUN
ajst-13993	88	15	for	for	ADP
ajst-13993	88	16	the	the	DET
ajst-13993	88	17	comment	comment	NOUN
ajst-13993	88	18	generation	generation	NOUN
ajst-13993	88	19	task	task	NOUN
ajst-13993	88	20	,	,	PUNCT
ajst-13993	88	21	the	the	DET
ajst-13993	88	22	attention	attention	NOUN
ajst-13993	88	23	mechanism	mechanism	NOUN
ajst-13993	88	24	is	be	AUX
ajst-13993	88	25	adopted	adopt	VERB
ajst-13993	88	26	again	again	ADV
ajst-13993	88	27	to	to	PART
ajst-13993	88	28	generate	generate	VERB
ajst-13993	88	29	the	the	DET
ajst-13993	88	30	function	function	NOUN
ajst-13993	88	31	vector	vector	NOUN
ajst-13993	88	32	us	we	PRON
ajst-13993	88	33	in	in	ADP
ajst-13993	88	34	the	the	DET
ajst-13993	88	35	statement	statement	NOUN
ajst-13993	88	36	level	level	NOUN
ajst-13993	88	37	which	which	PRON
ajst-13993	88	38	enables	enable	VERB
ajst-13993	88	39	the	the	DET
ajst-13993	88	40	framework	framework	NOUN
ajst-13993	88	41	to	to	PART
ajst-13993	88	42	estimate	estimate	VERB
ajst-13993	88	43	the	the	DET
ajst-13993	88	44	importance	importance	NOUN
ajst-13993	88	45	of	of	ADP
ajst-13993	88	46	different	different	ADJ
ajst-13993	88	47	statements	statement	NOUN
ajst-13993	88	48	.	.	PUNCT
ajst-13993	89	1	(	(	PUNCT
ajst-13993	89	2	4	4	NUM
ajst-13993	89	3	)	)	PUNCT
ajst-13993	89	4	where	where	SCONJ
ajst-13993	89	5	,	,	PUNCT
ajst-13993	89	6	the	the	DET
ajst-13993	89	7	weight	weight	NOUN
ajst-13993	89	8	matrix	matrix	NOUN
ajst-13993	89	9	is	be	AUX
ajst-13993	89	10	signed	sign	VERB
ajst-13993	89	11	as	as	ADP
ajst-13993	89	12	ws	ws	NOUN
ajst-13993	89	13	,	,	PUNCT
ajst-13993	89	14	bs	bs	PROPN
ajst-13993	89	15	represents	represent	VERB
ajst-13993	89	16	the	the	DET
ajst-13993	89	17	bias	bias	NOUN
ajst-13993	89	18	vector	vector	NOUN
ajst-13993	89	19	,	,	PUNCT
ajst-13993	89	20	and	and	CCONJ
ajst-13993	89	21	the	the	DET
ajst-13993	89	22	attention	attention	NOUN
ajst-13993	89	23	from	from	ADP
ajst-13993	89	24	each	each	DET
ajst-13993	89	25	statement	statement	NOUN
ajst-13993	89	26	si	si	INTJ
ajst-13993	89	27	to	to	ADP
ajst-13993	89	28	the	the	DET
ajst-13993	89	29	final	final	ADJ
ajst-13993	89	30	representation	representation	NOUN
ajst-13993	89	31	vector	vector	NOUN
ajst-13993	89	32	dc	dc	PROPN
ajst-13993	89	33	is	be	AUX
ajst-13993	89	34	signed	sign	VERB
ajst-13993	89	35	αi	αi	PRON
ajst-13993	89	36	.	.	PUNCT
ajst-13993	90	1	as	as	ADP
ajst-13993	90	2	the	the	DET
ajst-13993	90	3	vectors	vector	NOUN
ajst-13993	90	4	of	of	ADP
ajst-13993	90	5	the	the	DET
ajst-13993	90	6	three	three	NUM
ajst-13993	90	7	representations	representation	NOUN
ajst-13993	90	8	of	of	ADP
ajst-13993	90	9	the	the	DET
ajst-13993	90	10	source	source	NOUN
ajst-13993	90	11	code	code	NOUN
ajst-13993	90	12	,	,	PUNCT
ajst-13993	90	13	namely	namely	ADV
ajst-13993	90	14	the	the	DET
ajst-13993	90	15	vector	vector	NOUN
ajst-13993	90	16	of	of	ADP
ajst-13993	90	17	the	the	DET
ajst-13993	90	18	plain	plain	ADJ
ajst-13993	90	19	text	text	NOUN
ajst-13993	90	20	sequence	sequence	NOUN
ajst-13993	90	21	,	,	PUNCT
ajst-13993	90	22	the	the	DET
ajst-13993	90	23	vector	vector	NOUN
ajst-13993	90	24	of	of	ADP
ajst-13993	90	25	the	the	DET
ajst-13993	90	26	sequenced	sequence	VERB
ajst-13993	90	27	abstract	abstract	ADJ
ajst-13993	90	28	syntax	syntax	NOUN
ajst-13993	90	29	tree	tree	NOUN
ajst-13993	90	30	and	and	CCONJ
ajst-13993	90	31	the	the	DET
ajst-13993	90	32	vector	vector	NOUN
ajst-13993	90	33	of	of	ADP
ajst-13993	90	34	the	the	DET
ajst-13993	90	35	sequenced	sequence	VERB
ajst-13993	90	36	control	control	NOUN
ajst-13993	90	37	flow	flow	NOUN
ajst-13993	90	38	graph	graph	NOUN
ajst-13993	90	39	have	have	AUX
ajst-13993	90	40	been	be	AUX
ajst-13993	90	41	generated	generate	VERB
ajst-13993	90	42	,	,	PUNCT
ajst-13993	90	43	they	they	PRON
ajst-13993	90	44	are	be	AUX
ajst-13993	90	45	concatenated	concatenate	VERB
ajst-13993	90	46	firstly	firstly	ADV
ajst-13993	90	47	and	and	CCONJ
ajst-13993	90	48	then	then	ADV
ajst-13993	90	49	are	be	AUX
ajst-13993	90	50	feed	feed	NOUN
ajst-13993	90	51	into	into	ADP
ajst-13993	90	52	the	the	DET
ajst-13993	90	53	linear	linear	ADJ
ajst-13993	90	54	network	network	NOUN
ajst-13993	90	55	:	:	PUNCT
ajst-13993	90	56	d	d	PUNCT
ajst-13993	90	57	=	=	X
ajst-13993	90	58	wd[dtxt;dast;d	wd[dtxt;dast;d	ADV
ajst-13993	90	59	cfg]+bd	cfg]+bd	NOUN
ajst-13993	90	60	,	,	PUNCT
ajst-13993	90	61	where	where	SCONJ
ajst-13993	90	62	d	d	PROPN
ajst-13993	90	63	denotes	denote	VERB
ajst-13993	90	64	the	the	DET
ajst-13993	90	65	final	final	ADJ
ajst-13993	90	66	representation	representation	NOUN
ajst-13993	90	67	of	of	ADP
ajst-13993	90	68	the	the	DET
ajst-13993	90	69	source	source	NOUN
ajst-13993	90	70	code	code	NOUN
ajst-13993	90	71	,	,	PUNCT
ajst-13993	90	72	[	[	X
ajst-13993	90	73	dtxt;dast;dcfg	dtxt;dast;dcfg	X
ajst-13993	90	74	]	]	PUNCT
ajst-13993	90	75	represents	represent	VERB
ajst-13993	90	76	the	the	DET
ajst-13993	90	77	concatenation	concatenation	NOUN
ajst-13993	90	78	of	of	ADP
ajst-13993	90	79	the	the	DET
ajst-13993	90	80	three	three	NUM
ajst-13993	90	81	representations	representation	NOUN
ajst-13993	90	82	of	of	ADP
ajst-13993	90	83	the	the	DET
ajst-13993	90	84	source	source	NOUN
ajst-13993	90	85	code	code	NOUN
ajst-13993	90	86	.	.	PUNCT
ajst-13993	91	1	at	at	ADP
ajst-13993	91	2	last	last	ADJ
ajst-13993	91	3	,	,	PUNCT
ajst-13993	91	4	an	an	DET
ajst-13993	91	5	additional	additional	ADJ
ajst-13993	91	6	hidden	hidden	ADJ
ajst-13993	91	7	layer	layer	NOUN
ajst-13993	91	8	is	be	AUX
ajst-13993	91	9	utilized	utilize	VERB
ajst-13993	91	10	for	for	ADP
ajst-13993	91	11	the	the	DET
ajst-13993	91	12	comment	comment	NOUN
ajst-13993	91	13	generation	generation	NOUN
ajst-13993	91	14	:	:	PUNCT
ajst-13993	91	15	st	st	PROPN
ajst-13993	91	16	=	=	PROPN
ajst-13993	91	17	tanh(wcst	tanh(wcst	PROPN
ajst-13993	91	18	+	+	PROPN
ajst-13993	91	19	bd	bd	PROPN
ajst-13993	91	20	)	)	PUNCT
ajst-13993	91	21	,	,	PUNCT
ajst-13993	91	22	where	where	SCONJ
ajst-13993	91	23	st	st	PROPN
ajst-13993	91	24	is	be	AUX
ajst-13993	91	25	hidden	hide	VERB
ajst-13993	91	26	state	state	NOUN
ajst-13993	91	27	and	and	CCONJ
ajst-13993	91	28	s0	s0	NOUN
ajst-13993	91	29	is	be	AUX
ajst-13993	91	30	initialled	initial	VERB
ajst-13993	91	31	as	as	ADP
ajst-13993	91	32	d.	d.	PROPN
ajst-13993	91	33	3.2	3.2	NUM
ajst-13993	91	34	.	.	PUNCT
ajst-13993	92	1	text	text	NOUN
ajst-13993	92	2	generation	generation	NOUN
ajst-13993	92	3	as	as	SCONJ
ajst-13993	92	4	the	the	DET
ajst-13993	92	5	representation	representation	NOUN
ajst-13993	92	6	of	of	ADP
ajst-13993	92	7	the	the	DET
ajst-13993	92	8	source	source	NOUN
ajst-13993	92	9	code	code	NOUN
ajst-13993	92	10	has	have	AUX
ajst-13993	92	11	been	be	AUX
ajst-13993	92	12	deduced	deduce	VERB
ajst-13993	92	13	,	,	PUNCT
ajst-13993	92	14	then	then	ADV
ajst-13993	92	15	comment	comment	NOUN
ajst-13993	92	16	can	can	AUX
ajst-13993	92	17	be	be	AUX
ajst-13993	92	18	generated	generate	VERB
ajst-13993	92	19	by	by	ADP
ajst-13993	92	20	a	a	DET
ajst-13993	92	21	softmax	softmax	NOUN
ajst-13993	92	22	function	function	NOUN
ajst-13993	92	23	.	.	PUNCT
ajst-13993	93	1	assume	assume	VERB
ajst-13993	93	2	that	that	SCONJ
ajst-13993	93	3	the	the	DET
ajst-13993	93	4	policy	policy	NOUN
ajst-13993	93	5	π	π	NOUN
ajst-13993	93	6	defined	define	VERB
ajst-13993	93	7	from	from	ADP
ajst-13993	93	8	the	the	DET
ajst-13993	93	9	actor	actor	NOUN
ajst-13993	93	10	network	network	NOUN
ajst-13993	93	11	is	be	AUX
ajst-13993	93	12	signed	sign	VERB
ajst-13993	93	13	as	as	ADP
ajst-13993	93	14	pπ	pπ	NOUN
ajst-13993	93	15	,	,	PUNCT
ajst-13993	93	16	and	and	CCONJ
ajst-13993	93	17	the	the	DET
ajst-13993	93	18	distribution	distribution	NOUN
ajst-13993	93	19	of	of	ADP
ajst-13993	93	20	the	the	DET
ajst-13993	93	21	probability	probability	NOUN
ajst-13993	93	22	of	of	ADP
ajst-13993	93	23	the	the	DET
ajst-13993	93	24	tth	tth	PROPN
ajst-13993	93	25	word	word	NOUN
ajst-13993	93	26	yt	yt	NOUN
ajst-13993	93	27	is	be	AUX
ajst-13993	93	28	signed	sign	VERB
ajst-13993	93	29	as	as	ADP
ajst-13993	93	30	pπ(yt	pπ(yt	NOUN
ajst-13993	93	31	|st	|st	NUM
ajst-13993	93	32	)	)	PUNCT
ajst-13993	93	33	,	,	PUNCT
ajst-13993	93	34	we	we	PRON
ajst-13993	93	35	can	can	AUX
ajst-13993	93	36	get	get	VERB
ajst-13993	93	37	the	the	DET
ajst-13993	93	38	comment	comment	NOUN
ajst-13993	93	39	generation	generation	NOUN
ajst-13993	93	40	equation	equation	NOUN
ajst-13993	93	41	:	:	PUNCT
ajst-13993	93	42	(	(	PUNCT
ajst-13993	93	43	5	5	X
ajst-13993	93	44	)	)	PUNCT
ajst-13993	93	45	3.3	3.3	NUM
ajst-13993	93	46	.	.	PUNCT
ajst-13993	94	1	critic	critic	ADJ
ajst-13993	94	2	network	network	NOUN
ajst-13993	94	3	unlike	unlike	ADP
ajst-13993	94	4	the	the	DET
ajst-13993	94	5	traditional	traditional	ADJ
ajst-13993	94	6	code	code	NOUN
ajst-13993	94	7	summarization	summarization	NOUN
ajst-13993	94	8	approaches	approach	NOUN
ajst-13993	94	9	which	which	PRON
ajst-13993	94	10	generate	generate	VERB
ajst-13993	94	11	comments	comment	NOUN
ajst-13993	94	12	by	by	ADP
ajst-13993	94	13	optimizing	optimize	VERB
ajst-13993	94	14	the	the	DET
ajst-13993	94	15	probability	probability	NOUN
ajst-13993	94	16	of	of	ADP
ajst-13993	94	17	next	next	ADJ
ajst-13993	94	18	words	word	NOUN
ajst-13993	94	19	based	base	VERB
ajst-13993	94	20	on	on	ADP
ajst-13993	94	21	the	the	DET
ajst-13993	94	22	ground	ground	NOUN
ajst-13993	94	23	truth	truth	NOUN
ajst-13993	94	24	,	,	PUNCT
ajst-13993	94	25	while	while	SCONJ
ajst-13993	94	26	the	the	DET
ajst-13993	94	27	comments	comment	NOUN
ajst-13993	94	28	are	be	AUX
ajst-13993	94	29	produced	produce	VERB
ajst-13993	94	30	by	by	ADP
ajst-13993	94	31	optimizing	optimize	VERB
ajst-13993	94	32	the	the	DET
ajst-13993	94	33	reward	reward	NOUN
ajst-13993	94	34	iteratively	iteratively	ADV
ajst-13993	94	35	in	in	ADP
ajst-13993	94	36	the	the	DET
ajst-13993	94	37	proposed	propose	VERB
ajst-13993	94	38	approach	approach	NOUN
ajst-13993	94	39	which	which	PRON
ajst-13993	94	40	is	be	AUX
ajst-13993	94	41	realized	realize	VERB
ajst-13993	94	42	by	by	ADP
ajst-13993	94	43	reinforcement	reinforcement	NOUN
ajst-13993	94	44	learning	learning	NOUN
ajst-13993	94	45	.	.	PUNCT
ajst-13993	95	1	in	in	ADP
ajst-13993	95	2	specific	specific	ADJ
ajst-13993	95	3	,	,	PUNCT
ajst-13993	95	4	the	the	DET
ajst-13993	95	5	critic	critic	ADJ
ajst-13993	95	6	network	network	NOUN
ajst-13993	95	7	is	be	AUX
ajst-13993	95	8	introduced	introduce	VERB
ajst-13993	95	9	to	to	PART
ajst-13993	95	10	calculate	calculate	VERB
ajst-13993	95	11	the	the	DET
ajst-13993	95	12	reward	reward	NOUN
ajst-13993	95	13	of	of	ADP
ajst-13993	95	14	the	the	DET
ajst-13993	95	15	summary	summary	NOUN
ajst-13993	95	16	action	action	NOUN
ajst-13993	95	17	to	to	PART
ajst-13993	95	18	supply	supply	VERB
ajst-13993	95	19	a	a	DET
ajst-13993	95	20	feedback	feedback	NOUN
ajst-13993	95	21	to	to	PART
ajst-13993	95	22	train	train	VERB
ajst-13993	95	23	the	the	DET
ajst-13993	95	24	network	network	NOUN
ajst-13993	95	25	iteratively	iteratively	ADV
ajst-13993	95	26	.	.	PUNCT
ajst-13993	96	1	77	77	NUM
ajst-13993	96	2	4	4	NUM
ajst-13993	96	3	.	.	PUNCT
ajst-13993	96	4	experiments	experiment	NOUN
ajst-13993	96	5	and	and	CCONJ
ajst-13993	96	6	analysis	analysis	NOUN
ajst-13993	96	7	to	to	PART
ajst-13993	96	8	estimate	estimate	VERB
ajst-13993	96	9	the	the	DET
ajst-13993	96	10	performance	performance	NOUN
ajst-13993	96	11	of	of	ADP
ajst-13993	96	12	the	the	DET
ajst-13993	96	13	proposed	propose	VERB
ajst-13993	96	14	approach	approach	NOUN
ajst-13993	96	15	,	,	PUNCT
ajst-13993	96	16	sufficient	sufficient	ADJ
ajst-13993	96	17	experiments	experiment	NOUN
ajst-13993	96	18	based	base	VERB
ajst-13993	96	19	on	on	ADP
ajst-13993	96	20	real	real	ADJ
ajst-13993	96	21	-	-	PUNCT
ajst-13993	96	22	world	world	NOUN
ajst-13993	96	23	dataset	dataset	NOUN
ajst-13993	96	24	is	be	AUX
ajst-13993	96	25	performed	perform	VERB
ajst-13993	96	26	.	.	PUNCT
ajst-13993	97	1	the	the	DET
ajst-13993	97	2	following	follow	VERB
ajst-13993	97	3	research	research	NOUN
ajst-13993	97	4	questions	question	NOUN
ajst-13993	97	5	are	be	AUX
ajst-13993	97	6	designed	design	VERB
ajst-13993	97	7	for	for	ADP
ajst-13993	97	8	the	the	DET
ajst-13993	97	9	estimation	estimation	NOUN
ajst-13993	97	10	in	in	ADP
ajst-13993	97	11	the	the	DET
ajst-13993	97	12	experiment	experiment	NOUN
ajst-13993	97	13	.	.	PUNCT
ajst-13993	98	1	rq1	rq1	NOUN
ajst-13993	98	2	.	.	PUNCT
ajst-13993	99	1	what	what	PRON
ajst-13993	99	2	is	be	AUX
ajst-13993	99	3	the	the	DET
ajst-13993	99	4	performance	performance	NOUN
ajst-13993	99	5	of	of	ADP
ajst-13993	99	6	different	different	ADJ
ajst-13993	99	7	components	component	NOUN
ajst-13993	99	8	in	in	ADP
ajst-13993	99	9	the	the	DET
ajst-13993	99	10	proposed	propose	VERB
ajst-13993	99	11	model	model	NOUN
ajst-13993	99	12	?	?	PUNCT
ajst-13993	100	1	for	for	ADP
ajst-13993	100	2	instance	instance	NOUN
ajst-13993	100	3	,	,	PUNCT
ajst-13993	100	4	whether	whether	SCONJ
ajst-13993	100	5	the	the	DET
ajst-13993	100	6	representations	representation	NOUN
ajst-13993	100	7	of	of	ADP
ajst-13993	100	8	the	the	DET
ajst-13993	100	9	source	source	NOUN
ajst-13993	100	10	code	code	NOUN
ajst-13993	100	11	,	,	PUNCT
ajst-13993	100	12	the	the	DET
ajst-13993	100	13	hierarchical	hierarchical	ADJ
ajst-13993	100	14	attention	attention	NOUN
ajst-13993	100	15	network	network	NOUN
ajst-13993	100	16	and	and	CCONJ
ajst-13993	100	17	the	the	DET
ajst-13993	100	18	reinforcement	reinforcement	NOUN
ajst-13993	100	19	learning	learning	NOUN
ajst-13993	100	20	can	can	AUX
ajst-13993	100	21	improve	improve	VERB
ajst-13993	100	22	the	the	DET
ajst-13993	100	23	performance	performance	NOUN
ajst-13993	100	24	respectively	respectively	ADV
ajst-13993	100	25	?	?	PUNCT
ajst-13993	101	1	rq2	rq2	NOUN
ajst-13993	101	2	.	.	PUNCT
ajst-13993	102	1	what	what	PRON
ajst-13993	102	2	is	be	AUX
ajst-13993	102	3	the	the	DET
ajst-13993	102	4	performance	performance	NOUN
ajst-13993	102	5	of	of	ADP
ajst-13993	102	6	the	the	DET
ajst-13993	102	7	proposed	propose	VERB
ajst-13993	102	8	code	code	NOUN
ajst-13993	102	9	summarization	summarization	NOUN
ajst-13993	102	10	approaches	approach	NOUN
ajst-13993	102	11	considering	consider	VERB
ajst-13993	102	12	different	different	ADJ
ajst-13993	102	13	code	code	NOUN
ajst-13993	102	14	or	or	CCONJ
ajst-13993	102	15	comment	comment	NOUN
ajst-13993	102	16	length	length	NOUN
ajst-13993	102	17	?	?	PUNCT
ajst-13993	103	1	rq3	rq3	NOUN
ajst-13993	103	2	.	.	PUNCT
ajst-13993	104	1	what	what	PRON
ajst-13993	104	2	is	be	AUX
ajst-13993	104	3	the	the	DET
ajst-13993	104	4	time	time	NOUN
ajst-13993	104	5	consumption	consumption	NOUN
ajst-13993	104	6	of	of	ADP
ajst-13993	104	7	different	different	ADJ
ajst-13993	104	8	parts	part	NOUN
ajst-13993	104	9	?	?	PUNCT
ajst-13993	105	1	the	the	DET
ajst-13993	105	2	research	research	NOUN
ajst-13993	105	3	question	question	NOUN
ajst-13993	105	4	1	1	NUM
ajst-13993	105	5	aims	aim	VERB
ajst-13993	105	6	to	to	PART
ajst-13993	105	7	estimate	estimate	VERB
ajst-13993	105	8	the	the	DET
ajst-13993	105	9	performance	performance	NOUN
ajst-13993	105	10	of	of	ADP
ajst-13993	105	11	each	each	DET
ajst-13993	105	12	component	component	NOUN
ajst-13993	105	13	in	in	ADP
ajst-13993	105	14	the	the	DET
ajst-13993	105	15	proposed	propose	VERB
ajst-13993	105	16	approach	approach	NOUN
ajst-13993	105	17	compared	compare	VERB
ajst-13993	105	18	with	with	ADP
ajst-13993	105	19	the	the	DET
ajst-13993	105	20	state	state	NOUN
ajst-13993	105	21	-	-	PUNCT
ajst-13993	105	22	of	of	ADP
ajst-13993	105	23	-	-	PUNCT
ajst-13993	105	24	the	the	DET
ajst-13993	105	25	-	-	PUNCT
ajst-13993	105	26	art	art	NOUN
ajst-13993	105	27	baselines	baseline	NOUN
ajst-13993	105	28	.	.	PUNCT
ajst-13993	106	1	the	the	DET
ajst-13993	106	2	research	research	NOUN
ajst-13993	106	3	question	question	NOUN
ajst-13993	106	4	2	2	NUM
ajst-13993	106	5	aims	aim	VERB
ajst-13993	106	6	to	to	PART
ajst-13993	106	7	estimate	estimate	VERB
ajst-13993	106	8	the	the	DET
ajst-13993	106	9	proposed	propose	VERB
ajst-13993	106	10	approach	approach	NOUN
ajst-13993	106	11	consider	consider	VERB
ajst-13993	106	12	the	the	DET
ajst-13993	106	13	range	range	NOUN
ajst-13993	106	14	of	of	ADP
ajst-13993	106	15	the	the	DET
ajst-13993	106	16	code	code	NOUN
ajst-13993	106	17	and	and	CCONJ
ajst-13993	106	18	comment	comment	NOUN
ajst-13993	106	19	length	length	NOUN
ajst-13993	106	20	.	.	PUNCT
ajst-13993	107	1	the	the	DET
ajst-13993	107	2	research	research	NOUN
ajst-13993	107	3	question	question	NOUN
ajst-13993	107	4	3	3	NUM
ajst-13993	107	5	is	be	AUX
ajst-13993	107	6	proposed	propose	VERB
ajst-13993	107	7	to	to	PART
ajst-13993	107	8	estimate	estimate	VERB
ajst-13993	107	9	the	the	DET
ajst-13993	107	10	time	time	NOUN
ajst-13993	107	11	complexity	complexity	NOUN
ajst-13993	107	12	of	of	ADP
ajst-13993	107	13	the	the	DET
ajst-13993	107	14	proposed	propose	VERB
ajst-13993	107	15	approach	approach	NOUN
ajst-13993	107	16	.	.	PUNCT
ajst-13993	108	1	4.1	4.1	NUM
ajst-13993	108	2	.	.	PUNCT
ajst-13993	109	1	dataset	dataset	ADJ
ajst-13993	109	2	preparation	preparation	NOUN
ajst-13993	109	3	to	to	PART
ajst-13993	109	4	estimate	estimate	VERB
ajst-13993	109	5	the	the	DET
ajst-13993	109	6	performance	performance	NOUN
ajst-13993	109	7	of	of	ADP
ajst-13993	109	8	the	the	DET
ajst-13993	109	9	proposed	propose	VERB
ajst-13993	109	10	approach	approach	NOUN
ajst-13993	109	11	,	,	PUNCT
ajst-13993	109	12	the	the	DET
ajst-13993	109	13	dataset	dataset	NOUN
ajst-13993	109	14	collected	collect	VERB
ajst-13993	109	15	in	in	ADP
ajst-13993	109	16	[	[	X
ajst-13993	109	17	14	14	NUM
ajst-13993	109	18	]	]	PUNCT
ajst-13993	109	19	,	,	PUNCT
ajst-13993	109	20	which	which	PRON
ajst-13993	109	21	is	be	AUX
ajst-13993	109	22	collected	collect	VERB
ajst-13993	109	23	from	from	ADP
ajst-13993	109	24	github	github	PROPN
ajst-13993	109	25	[	[	X
ajst-13993	109	26	14	14	NUM
ajst-13993	109	27	]	]	PUNCT
ajst-13993	109	28	which	which	PRON
ajst-13993	109	29	is	be	AUX
ajst-13993	109	30	the	the	DET
ajst-13993	109	31	most	most	ADV
ajst-13993	109	32	popular	popular	ADJ
ajst-13993	109	33	open	open	ADJ
ajst-13993	109	34	source	source	NOUN
ajst-13993	109	35	projects	project	NOUN
ajst-13993	109	36	hosting	host	VERB
ajst-13993	109	37	platform	platform	NOUN
ajst-13993	109	38	,	,	PUNCT
ajst-13993	109	39	is	be	AUX
ajst-13993	109	40	utilized	utilize	VERB
ajst-13993	109	41	and	and	CCONJ
ajst-13993	109	42	it	it	PRON
ajst-13993	109	43	is	be	AUX
ajst-13993	109	44	a	a	DET
ajst-13993	109	45	commonly	commonly	ADV
ajst-13993	109	46	used	use	VERB
ajst-13993	109	47	dataset	dataset	NOUN
ajst-13993	109	48	these	these	DET
ajst-13993	109	49	years	year	NOUN
ajst-13993	109	50	.	.	PUNCT
ajst-13993	110	1	there	there	PRON
ajst-13993	110	2	are	be	VERB
ajst-13993	110	3	mainly	mainly	ADV
ajst-13993	110	4	about	about	ADP
ajst-13993	110	5	108,700	108,700	NUM
ajst-13993	110	6	<	<	NOUN
ajst-13993	110	7	code	code	NOUN
ajst-13993	110	8	,	,	PUNCT
ajst-13993	110	9	comment	comment	NOUN
ajst-13993	110	10	>	>	X
ajst-13993	110	11	pairs	pair	NOUN
ajst-13993	110	12	included	include	VERB
ajst-13993	110	13	in	in	ADP
ajst-13993	110	14	the	the	DET
ajst-13993	110	15	dataset	dataset	NOUN
ajst-13993	110	16	,	,	PUNCT
ajst-13993	110	17	and	and	CCONJ
ajst-13993	110	18	about	about	ADV
ajst-13993	110	19	50,400	50,400	NUM
ajst-13993	110	20	code	code	NOUN
ajst-13993	110	21	tokens	token	NOUN
ajst-13993	110	22	and	and	CCONJ
ajst-13993	110	23	31,300	31,300	NUM
ajst-13993	110	24	comment	comment	NOUN
ajst-13993	110	25	tokens	token	NOUN
ajst-13993	110	26	are	be	AUX
ajst-13993	110	27	contained	contain	VERB
ajst-13993	110	28	respectively	respectively	ADV
ajst-13993	110	29	in	in	ADP
ajst-13993	110	30	the	the	DET
ajst-13993	110	31	datset	datset	NOUN
ajst-13993	110	32	.	.	PUNCT
ajst-13993	111	1	the	the	DET
ajst-13993	111	2	dataset	dataset	NOUN
ajst-13993	111	3	is	be	AUX
ajst-13993	111	4	split	split	VERB
ajst-13993	111	5	into	into	ADP
ajst-13993	111	6	different	different	ADJ
ajst-13993	111	7	part	part	NOUN
ajst-13993	111	8	in	in	ADP
ajst-13993	111	9	a	a	DET
ajst-13993	111	10	random	random	ADJ
ajst-13993	111	11	way	way	NOUN
ajst-13993	111	12	,	,	PUNCT
ajst-13993	111	13	the	the	DET
ajst-13993	111	14	first	first	ADJ
ajst-13993	111	15	part	part	NOUN
ajst-13993	111	16	contains	contain	VERB
ajst-13993	111	17	60	60	NUM
ajst-13993	111	18	%	%	NOUN
ajst-13993	111	19	for	for	ADP
ajst-13993	111	20	training	training	NOUN
ajst-13993	111	21	,	,	PUNCT
ajst-13993	111	22	the	the	DET
ajst-13993	111	23	second	second	ADJ
ajst-13993	111	24	part	part	NOUN
ajst-13993	111	25	contains	contain	VERB
ajst-13993	111	26	20	20	NUM
ajst-13993	111	27	%	%	NOUN
ajst-13993	111	28	for	for	ADP
ajst-13993	111	29	validation	validation	NOUN
ajst-13993	111	30	and	and	CCONJ
ajst-13993	111	31	the	the	DET
ajst-13993	111	32	last	last	ADJ
ajst-13993	111	33	20	20	NUM
ajst-13993	111	34	%	%	NOUN
ajst-13993	111	35	is	be	AUX
ajst-13993	111	36	used	use	VERB
ajst-13993	111	37	for	for	ADP
ajst-13993	111	38	testing	testing	NOUN
ajst-13993	111	39	.	.	PUNCT
ajst-13993	112	1	4.2	4.2	NUM
ajst-13993	112	2	.	.	PUNCT
ajst-13993	112	3	evaluation	evaluation	NOUN
ajst-13993	112	4	metrics	metric	NOUN
ajst-13993	112	5	in	in	ADP
ajst-13993	112	6	this	this	DET
ajst-13993	112	7	work	work	NOUN
ajst-13993	112	8	,	,	PUNCT
ajst-13993	112	9	similar	similar	ADJ
ajst-13993	112	10	as	as	ADP
ajst-13993	112	11	[	[	X
ajst-13993	112	12	8	8	NUM
ajst-13993	112	13	,	,	PUNCT
ajst-13993	112	14	100	100	NUM
ajst-13993	112	15	,	,	PUNCT
ajst-13993	112	16	113	113	NUM
ajst-13993	112	17	]	]	PUNCT
ajst-13993	112	18	,	,	PUNCT
ajst-13993	112	19	we	we	PRON
ajst-13993	112	20	estimate	estimate	VERB
ajst-13993	112	21	the	the	DET
ajst-13993	112	22	accuracy	accuracy	NOUN
ajst-13993	112	23	of	of	ADP
ajst-13993	112	24	the	the	DET
ajst-13993	112	25	produced	produce	VERB
ajst-13993	112	26	annotations	annotation	NOUN
ajst-13993	112	27	from	from	ADP
ajst-13993	112	28	the	the	DET
ajst-13993	112	29	aspect	aspect	NOUN
ajst-13993	112	30	of	of	ADP
ajst-13993	112	31	their	their	PRON
ajst-13993	112	32	similarity	similarity	NOUN
ajst-13993	112	33	with	with	ADP
ajst-13993	112	34	the	the	DET
ajst-13993	112	35	corresponding	corresponding	ADJ
ajst-13993	112	36	ground	ground	NOUN
ajst-13993	112	37	-	-	PUNCT
ajst-13993	112	38	truth	truth	NOUN
ajst-13993	112	39	comments	comment	NOUN
ajst-13993	112	40	.	.	PUNCT
ajst-13993	113	1	in	in	ADP
ajst-13993	113	2	specific	specific	ADJ
ajst-13993	113	3	,	,	PUNCT
ajst-13993	113	4	the	the	DET
ajst-13993	113	5	three	three	NUM
ajst-13993	113	6	widely	widely	ADV
ajst-13993	113	7	-	-	PUNCT
ajst-13993	113	8	used	use	VERB
ajst-13993	113	9	evaluation	evaluation	NOUN
ajst-13993	113	10	metrics	metric	NOUN
ajst-13993	113	11	adopted	adopt	VERB
ajst-13993	113	12	in	in	ADP
ajst-13993	113	13	nlp	nlp	NOUN
ajst-13993	113	14	area	area	NOUN
ajst-13993	113	15	particularly	particularly	ADV
ajst-13993	113	16	in	in	ADP
ajst-13993	113	17	the	the	DET
ajst-13993	113	18	the	the	DET
ajst-13993	113	19	nmt	nmt	NOUN
ajst-13993	113	20	task	task	NOUN
ajst-13993	113	21	are	be	AUX
ajst-13993	113	22	utilized	utilize	VERB
ajst-13993	113	23	:	:	PUNCT
ajst-13993	113	24	bleu[17	bleu[17	PROPN
ajst-13993	113	25	]	]	X
ajst-13993	113	26	,	,	PUNCT
ajst-13993	113	27	meteor[22	meteor[22	NOUN
ajst-13993	113	28	]	]	PUNCT
ajst-13993	113	29	,	,	PUNCT
ajst-13993	113	30	and	and	CCONJ
ajst-13993	113	31	rouge[18].these	rouge[18].these	ADJ
ajst-13993	113	32	metrics	metric	NOUN
ajst-13993	113	33	mainly	mainly	ADV
ajst-13993	113	34	calculate	calculate	VERB
ajst-13993	113	35	the	the	DET
ajst-13993	113	36	similarity	similarity	NOUN
ajst-13993	113	37	degree	degree	NOUN
ajst-13993	113	38	between	between	ADP
ajst-13993	113	39	the	the	DET
ajst-13993	113	40	generated	generate	VERB
ajst-13993	113	41	natural	natural	ADJ
ajst-13993	113	42	language	language	NOUN
ajst-13993	113	43	text	text	NOUN
ajst-13993	113	44	and	and	CCONJ
ajst-13993	113	45	the	the	DET
ajst-13993	113	46	ground	ground	NOUN
ajst-13993	113	47	-	-	PUNCT
ajst-13993	113	48	truth	truth	NOUN
ajst-13993	113	49	by	by	ADP
ajst-13993	113	50	measuring	measure	VERB
ajst-13993	113	51	the	the	DET
ajst-13993	113	52	frequency	frequency	NOUN
ajst-13993	113	53	of	of	ADP
ajst-13993	113	54	the	the	DET
ajst-13993	113	55	tokens	token	NOUN
ajst-13993	113	56	occurrence	occurrence	NOUN
ajst-13993	113	57	in	in	ADP
ajst-13993	113	58	both	both	PRON
ajst-13993	113	59	of	of	ADP
ajst-13993	113	60	them	they	PRON
ajst-13993	113	61	from	from	ADP
ajst-13993	113	62	different	different	ADJ
ajst-13993	113	63	aspects	aspect	NOUN
ajst-13993	113	64	.	.	PUNCT
ajst-13993	114	1	as	as	SCONJ
ajst-13993	114	2	comment	comment	NOUN
ajst-13993	114	3	generation	generation	NOUN
ajst-13993	114	4	is	be	AUX
ajst-13993	114	5	also	also	ADV
ajst-13993	114	6	a	a	DET
ajst-13993	114	7	text	text	NOUN
ajst-13993	114	8	generation	generation	NOUN
ajst-13993	114	9	task	task	NOUN
ajst-13993	114	10	in	in	ADP
ajst-13993	114	11	which	which	PRON
ajst-13993	114	12	the	the	DET
ajst-13993	114	13	natural	natural	ADJ
ajst-13993	114	14	language	language	NOUN
ajst-13993	114	15	words	word	NOUN
ajst-13993	114	16	composed	compose	VERB
ajst-13993	114	17	the	the	DET
ajst-13993	114	18	out	out	ADP
ajst-13993	114	19	put	put	NOUN
ajst-13993	114	20	,	,	PUNCT
ajst-13993	114	21	thus	thus	ADV
ajst-13993	114	22	they	they	PRON
ajst-13993	114	23	are	be	AUX
ajst-13993	114	24	adopted	adopt	VERB
ajst-13993	114	25	to	to	PART
ajst-13993	114	26	assess	assess	VERB
ajst-13993	114	27	the	the	DET
ajst-13993	114	28	accuracy	accuracy	NOUN
ajst-13993	114	29	of	of	ADP
ajst-13993	114	30	the	the	DET
ajst-13993	114	31	produced	produce	VERB
ajst-13993	114	32	annotations	annotation	NOUN
ajst-13993	114	33	.	.	PUNCT
ajst-13993	115	1	particularly	particularly	ADV
ajst-13993	115	2	,	,	PUNCT
ajst-13993	115	3	bleu	bleu	NOUN
ajst-13993	115	4	is	be	AUX
ajst-13993	115	5	the	the	DET
ajst-13993	115	6	most	most	ADV
ajst-13993	115	7	common	common	ADJ
ajst-13993	115	8	evaluate	evaluate	NOUN
ajst-13993	115	9	metric	metric	ADJ
ajst-13993	115	10	adopted	adopt	VERB
ajst-13993	115	11	in	in	ADP
ajst-13993	115	12	the	the	DET
ajst-13993	115	13	natural	natural	ADJ
ajst-13993	115	14	language	language	NOUN
ajst-13993	115	15	generation	generation	NOUN
ajst-13993	115	16	task	task	NOUN
ajst-13993	115	17	[	[	X
ajst-13993	115	18	22	22	NUM
ajst-13993	115	19	-	-	SYM
ajst-13993	115	20	23	23	NUM
ajst-13993	115	21	]	]	PUNCT
ajst-13993	115	22	which	which	PRON
ajst-13993	115	23	estimates	estimate	VERB
ajst-13993	115	24	the	the	DET
ajst-13993	115	25	n	n	CCONJ
ajst-13993	115	26	-	-	PUNCT
ajst-13993	115	27	gram	gram	NOUN
ajst-13993	115	28	accuracy	accuracy	NOUN
ajst-13993	115	29	by	by	ADP
ajst-13993	115	30	comparing	compare	VERB
ajst-13993	115	31	with	with	ADP
ajst-13993	115	32	a	a	DET
ajst-13993	115	33	few	few	ADJ
ajst-13993	115	34	reference	reference	NOUN
ajst-13993	115	35	sentences	sentence	NOUN
ajst-13993	115	36	.	.	PUNCT
ajst-13993	116	1	table	table	NOUN
ajst-13993	116	2	1	1	NUM
ajst-13993	116	3	.	.	PUNCT
ajst-13993	117	1	performance	performance	NOUN
ajst-13993	117	2	of	of	ADP
ajst-13993	117	3	different	different	ADJ
ajst-13993	117	4	code	code	NOUN
ajst-13993	117	5	representations	representation	NOUN
ajst-13993	117	6	.	.	PUNCT
ajst-13993	118	1	table	table	NOUN
ajst-13993	118	2	2	2	NUM
ajst-13993	118	3	.	.	PUNCT
ajst-13993	118	4	performance	performance	NOUN
ajst-13993	118	5	of	of	ADP
ajst-13993	118	6	the	the	DET
ajst-13993	118	7	hierarchical	hierarchical	ADJ
ajst-13993	118	8	attention	attention	NOUN
ajst-13993	118	9	network	network	NOUN
ajst-13993	118	10	.	.	PUNCT
ajst-13993	119	1	table	table	NOUN
ajst-13993	119	2	3	3	NUM
ajst-13993	119	3	.	.	PUNCT
ajst-13993	120	1	performance	performance	NOUN
ajst-13993	120	2	of	of	ADP
ajst-13993	120	3	deep	deep	ADJ
ajst-13993	120	4	reinforcement	reinforcement	NOUN
ajst-13993	120	5	learning	learning	NOUN
ajst-13993	120	6	.	.	PUNCT
ajst-13993	121	1	4.3	4.3	NUM
ajst-13993	121	2	.	.	PUNCT
ajst-13993	122	1	experimental	experimental	ADJ
ajst-13993	122	2	settings	setting	NOUN
ajst-13993	122	3	in	in	ADP
ajst-13993	122	4	the	the	DET
ajst-13993	122	5	experiment	experiment	NOUN
ajst-13993	122	6	,	,	PUNCT
ajst-13993	122	7	we	we	PRON
ajst-13993	122	8	set	set	VERB
ajst-13993	122	9	vector	vector	NOUN
ajst-13993	122	10	size	size	NOUN
ajst-13993	122	11	of	of	ADP
ajst-13993	122	12	the	the	DET
ajst-13993	122	13	hidden	hidden	ADJ
ajst-13993	122	14	layers	layer	NOUN
ajst-13993	122	15	of	of	ADP
ajst-13993	122	16	lstm	lstm	NOUN
ajst-13993	122	17	to	to	PART
ajst-13993	122	18	be	be	AUX
ajst-13993	122	19	512	512	NUM
ajst-13993	122	20	for	for	ADP
ajst-13993	122	21	both	both	DET
ajst-13993	122	22	encoder	encoder	NOUN
ajst-13993	122	23	and	and	CCONJ
ajst-13993	122	24	decoder	decoder	NOUN
ajst-13993	122	25	.	.	PUNCT
ajst-13993	123	1	the	the	DET
ajst-13993	123	2	minibatch	minibatch	NOUN
ajst-13993	123	3	size	size	NOUN
ajst-13993	123	4	is	be	AUX
ajst-13993	123	5	initialized	initialize	VERB
ajst-13993	123	6	to	to	ADP
ajst-13993	123	7	32	32	NUM
ajst-13993	123	8	,	,	PUNCT
ajst-13993	123	9	and	and	CCONJ
ajst-13993	123	10	the	the	DET
ajst-13993	123	11	learning	learning	NOUN
ajst-13993	123	12	rate	rate	NOUN
ajst-13993	123	13	is	be	AUX
ajst-13993	123	14	initialized	initialize	VERB
ajst-13993	123	15	to	to	ADP
ajst-13993	123	16	0.001	0.001	NUM
ajst-13993	123	17	.	.	PUNCT
ajst-13993	124	1	firstly	firstly	ADV
ajst-13993	124	2	,	,	PUNCT
ajst-13993	124	3	the	the	DET
ajst-13993	124	4	the	the	DET
ajst-13993	124	5	actor	actor	NOUN
ajst-13993	124	6	and	and	CCONJ
ajst-13993	124	7	critic	critic	PROPN
ajst-13993	124	8	network	network	NOUN
ajst-13993	124	9	are	be	AUX
ajst-13993	124	10	pretrained	pretraine	VERB
ajst-13993	124	11	by	by	ADP
ajst-13993	124	12	10	10	NUM
ajst-13993	124	13	epochs	epoch	NOUN
ajst-13993	124	14	respectively	respectively	ADV
ajst-13993	124	15	,	,	PUNCT
ajst-13993	124	16	and	and	CCONJ
ajst-13993	124	17	then	then	ADV
ajst-13993	124	18	the	the	DET
ajst-13993	124	19	whole	whole	ADJ
ajst-13993	124	20	framework	framework	NOUN
ajst-13993	124	21	is	be	AUX
ajst-13993	124	22	trained	train	VERB
ajst-13993	124	23	by	by	ADP
ajst-13993	124	24	10	10	NUM
ajst-13993	124	25	epochs	epoch	NOUN
ajst-13993	124	26	simultaneously	simultaneously	ADV
ajst-13993	124	27	.	.	PUNCT
ajst-13993	125	1	all	all	DET
ajst-13993	125	2	the	the	DET
ajst-13993	125	3	experiments	experiment	NOUN
ajst-13993	125	4	are	be	AUX
ajst-13993	125	5	implemented	implement	VERB
ajst-13993	125	6	based	base	VERB
ajst-13993	125	7	on	on	ADP
ajst-13993	125	8	python	python	PROPN
ajst-13993	125	9	3.7	3.7	NUM
ajst-13993	125	10	.	.	PUNCT
ajst-13993	126	1	4.4	4.4	NUM
ajst-13993	126	2	.	.	PUNCT
ajst-13993	127	1	rq1	rq1	NOUN
ajst-13993	127	2	:	:	PUNCT
ajst-13993	127	3	the	the	DET
ajst-13993	127	4	performance	performance	NOUN
ajst-13993	127	5	analysis	analysis	NOUN
ajst-13993	127	6	of	of	ADP
ajst-13993	127	7	different	different	ADJ
ajst-13993	127	8	components	component	NOUN
ajst-13993	127	9	4.4.1	4.4.1	NUM
ajst-13993	127	10	.	.	PUNCT
ajst-13993	128	1	the	the	DET
ajst-13993	128	2	performance	performance	NOUN
ajst-13993	128	3	considering	consider	VERB
ajst-13993	128	4	different	different	ADJ
ajst-13993	128	5	code	code	NOUN
ajst-13993	128	6	representations	representation	NOUN
ajst-13993	128	7	we	we	PRON
ajst-13993	128	8	estimate	estimate	VERB
ajst-13993	128	9	the	the	DET
ajst-13993	128	10	proposed	propose	VERB
ajst-13993	128	11	approach	approach	NOUN
ajst-13993	128	12	by	by	ADP
ajst-13993	128	13	different	different	ADJ
ajst-13993	128	14	settings	setting	NOUN
ajst-13993	128	15	considering	consider	VERB
ajst-13993	128	16	different	different	ADJ
ajst-13993	128	17	components	component	NOUN
ajst-13993	128	18	.	.	PUNCT
ajst-13993	129	1	the	the	DET
ajst-13993	129	2	three	three	NUM
ajst-13993	129	3	different	different	ADJ
ajst-13993	129	4	code	code	NOUN
ajst-13993	129	5	representations	representation	NOUN
ajst-13993	129	6	are	be	AUX
ajst-13993	129	7	signed	sign	VERB
ajst-13993	129	8	as	as	ADP
ajst-13993	129	9	txt	txt	PROPN
ajst-13993	129	10	,	,	PUNCT
ajst-13993	129	11	ast	ast	NOUN
ajst-13993	129	12	and	and	CCONJ
ajst-13993	129	13	cfg	cfg	NOUN
ajst-13993	129	14	respectively	respectively	ADV
ajst-13993	129	15	.	.	PUNCT
ajst-13993	130	1	the	the	DET
ajst-13993	130	2	hierarchical	hierarchical	ADJ
ajst-13993	130	3	attention	attention	NOUN
ajst-13993	130	4	mechanism	mechanism	NOUN
ajst-13993	130	5	is	be	AUX
ajst-13993	130	6	denoted	denote	VERB
ajst-13993	130	7	as	as	SCONJ
ajst-13993	130	8	han	han	PROPN
ajst-13993	130	9	and	and	CCONJ
ajst-13993	130	10	drl	drl	PROPN
ajst-13993	130	11	refers	refer	VERB
ajst-13993	130	12	to	to	ADP
ajst-13993	130	13	deep	deep	ADJ
ajst-13993	130	14	reinforcement	reinforcement	NOUN
ajst-13993	130	15	learning	learning	NOUN
ajst-13993	130	16	.	.	PUNCT
ajst-13993	131	1	•	•	NUM
ajst-13993	131	2	txt+han+drl	txt+han+drl	PROPN
ajst-13993	131	3	.	.	PUNCT
ajst-13993	132	1	this	this	DET
ajst-13993	132	2	baseline	baseline	PROPN
ajst-13993	132	3	regards	regard	VERB
ajst-13993	132	4	the	the	DET
ajst-13993	132	5	source	source	NOUN
ajst-13993	132	6	code	code	NOUN
ajst-13993	132	7	as	as	ADP
ajst-13993	132	8	plain	plain	ADJ
ajst-13993	132	9	text	text	NOUN
ajst-13993	132	10	and	and	CCONJ
ajst-13993	132	11	which	which	PRON
ajst-13993	132	12	is	be	AUX
ajst-13993	132	13	encoded	encode	VERB
ajst-13993	132	14	by	by	ADP
ajst-13993	132	15	the	the	DET
ajst-13993	132	16	lstm	lstm	NOUN
ajst-13993	132	17	-	-	PUNCT
ajst-13993	132	18	based	base	VERB
ajst-13993	132	19	hierarchical	hierarchical	ADJ
ajst-13993	132	20	attention	attention	NOUN
ajst-13993	132	21	network	network	NOUN
ajst-13993	132	22	and	and	CCONJ
ajst-13993	132	23	then	then	ADV
ajst-13993	132	24	train	train	VERB
ajst-13993	132	25	the	the	DET
ajst-13993	132	26	model	model	NOUN
ajst-13993	132	27	by	by	ADP
ajst-13993	132	28	drl	drl	PROPN
ajst-13993	132	29	.	.	PROPN
ajst-13993	133	1	•	•	PROPN
ajst-13993	133	2	ast+han+drl	ast+han+drl	PROPN
ajst-13993	133	3	.	.	PUNCT
ajst-13993	134	1	this	this	DET
ajst-13993	134	2	approach	approach	NOUN
ajst-13993	134	3	takes	take	VERB
ajst-13993	134	4	the	the	DET
ajst-13993	134	5	sequenced	sequence	VERB
ajst-13993	134	6	abstract	abstract	ADJ
ajst-13993	134	7	syntax	syntax	NOUN
ajst-13993	134	8	tree	tree	NOUN
ajst-13993	134	9	as	as	ADP
ajst-13993	134	10	the	the	DET
ajst-13993	134	11	representation	representation	NOUN
ajst-13993	134	12	of	of	ADP
ajst-13993	134	13	the	the	DET
ajst-13993	134	14	source	source	NOUN
ajst-13993	134	15	code	code	NOUN
ajst-13993	134	16	and	and	CCONJ
ajst-13993	134	17	then	then	ADV
ajst-13993	134	18	encodes	encode	VERB
ajst-13993	134	19	it	it	PRON
ajst-13993	134	20	by	by	ADP
ajst-13993	134	21	the	the	DET
ajst-13993	134	22	lstm	lstm	NOUN
ajst-13993	134	23	-	-	PUNCT
ajst-13993	134	24	based	base	VERB
ajst-13993	134	25	hierarchical	hierarchical	ADJ
ajst-13993	134	26	attention	attention	NOUN
ajst-13993	134	27	network	network	NOUN
ajst-13993	134	28	.	.	PUNCT
ajst-13993	135	1	•cfg+han+drl	•cfg+han+drl	PROPN
ajst-13993	135	2	.	.	PUNCT
ajst-13993	136	1	this	this	DET
ajst-13993	136	2	approach	approach	NOUN
ajst-13993	136	3	utilizes	utilize	VERB
ajst-13993	136	4	the	the	DET
ajst-13993	136	5	sequenced	sequenced	ADJ
ajst-13993	136	6	control	control	NOUN
ajst-13993	136	7	flow	flow	NOUN
ajst-13993	136	8	as	as	ADP
ajst-13993	136	9	the	the	DET
ajst-13993	136	10	source	source	NOUN
ajst-13993	136	11	code	code	NOUN
ajst-13993	136	12	representation	representation	NOUN
ajst-13993	136	13	and	and	CCONJ
ajst-13993	136	14	then	then	ADV
ajst-13993	136	15	use	use	VERB
ajst-13993	136	16	the	the	DET
ajst-13993	136	17	same	same	ADJ
ajst-13993	136	18	framework	framework	NOUN
ajst-13993	136	19	.	.	PUNCT
ajst-13993	137	1	•	•	NUM
ajst-13993	137	2	txt&ast+han+drl	txt&ast+han+drl	NOUN
ajst-13993	137	3	.	.	PUNCT
ajst-13993	138	1	in	in	ADP
ajst-13993	138	2	this	this	DET
ajst-13993	138	3	approach	approach	NOUN
ajst-13993	138	4	,	,	PUNCT
ajst-13993	138	5	the	the	DET
ajst-13993	138	6	plain	plain	ADJ
ajst-13993	138	7	text	text	NOUN
ajst-13993	138	8	sequence	sequence	NOUN
ajst-13993	138	9	and	and	CCONJ
ajst-13993	138	10	the	the	DET
ajst-13993	138	11	sequenced	sequence	VERB
ajst-13993	138	12	abstract	abstract	ADJ
ajst-13993	138	13	syntax	syntax	NOUN
ajst-13993	138	14	tree	tree	NOUN
ajst-13993	138	15	as	as	ADP
ajst-13993	138	16	the	the	DET
ajst-13993	138	17	representations	representation	NOUN
ajst-13993	138	18	of	of	ADP
ajst-13993	138	19	the	the	DET
ajst-13993	138	20	source	source	NOUN
ajst-13993	138	21	code	code	NOUN
ajst-13993	138	22	and	and	CCONJ
ajst-13993	138	23	the	the	DET
ajst-13993	138	24	framework	framework	NOUN
ajst-13993	138	25	concatenates	concatenate	VERB
ajst-13993	138	26	their	their	PRON
ajst-13993	138	27	encoded	encode	VERB
ajst-13993	138	28	vectors	vector	NOUN
ajst-13993	138	29	to	to	PART
ajst-13993	138	30	obtain	obtain	VERB
ajst-13993	138	31	the	the	DET
ajst-13993	138	32	hybrid	hybrid	ADJ
ajst-13993	138	33	code	code	NOUN
ajst-13993	138	34	representation	representation	NOUN
ajst-13993	138	35	for	for	ADP
ajst-13993	138	36	the	the	DET
ajst-13993	138	37	comment	comment	NOUN
ajst-13993	138	38	generation	generation	NOUN
ajst-13993	138	39	.	.	PUNCT
ajst-13993	139	1	•	•	NUM
ajst-13993	139	2	txt&cfg+han+drl	txt&cfg+han+drl	PROPN
ajst-13993	139	3	.	.	PROPN
ajst-13993	140	1	as	as	ADP
ajst-13993	140	2	the	the	DET
ajst-13993	140	3	same	same	ADJ
ajst-13993	140	4	,	,	PUNCT
ajst-13993	140	5	the	the	DET
ajst-13993	140	6	plain	plain	ADJ
ajst-13993	140	7	text	text	NOUN
ajst-13993	140	8	sequence	sequence	NOUN
ajst-13993	140	9	and	and	CCONJ
ajst-13993	140	10	the	the	DET
ajst-13993	140	11	sequenced	sequence	VERB
ajst-13993	140	12	control	control	NOUN
ajst-13993	140	13	flow	flow	NOUN
ajst-13993	140	14	are	be	AUX
ajst-13993	140	15	adopted	adopt	VERB
ajst-13993	140	16	as	as	ADP
ajst-13993	140	17	the	the	DET
ajst-13993	140	18	representations	representation	NOUN
ajst-13993	140	19	of	of	ADP
ajst-13993	140	20	the	the	DET
ajst-13993	140	21	source	source	NOUN
ajst-13993	140	22	code	code	NOUN
ajst-13993	140	23	and	and	CCONJ
ajst-13993	140	24	then	then	ADV
ajst-13993	140	25	follows	follow	VERB
ajst-13993	140	26	the	the	DET
ajst-13993	140	27	same	same	ADJ
ajst-13993	140	28	framework	framework	NOUN
ajst-13993	140	29	.	.	PUNCT
ajst-13993	141	1	•ast&cfg+han+drl.in	•ast&cfg+han+drl.in	PROPN
ajst-13993	141	2	this	this	DET
ajst-13993	141	3	approach	approach	NOUN
ajst-13993	141	4	,	,	PUNCT
ajst-13993	141	5	the	the	DET
ajst-13993	141	6	sequenced	sequence	VERB
ajst-13993	141	7	abstract	abstract	ADJ
ajst-13993	141	8	syntax	syntax	NOUN
ajst-13993	141	9	tree	tree	NOUN
ajst-13993	141	10	and	and	CCONJ
ajst-13993	141	11	the	the	DET
ajst-13993	141	12	sequenced	sequence	VERB
ajst-13993	141	13	control	control	NOUN
ajst-13993	141	14	flow	flow	NOUN
ajst-13993	141	15	are	be	AUX
ajst-13993	141	16	utilized	utilize	VERB
ajst-13993	141	17	as	as	ADP
ajst-13993	141	18	the	the	DET
ajst-13993	141	19	representations	representation	NOUN
ajst-13993	141	20	of	of	ADP
ajst-13993	141	21	the	the	DET
ajst-13993	141	22	source	source	NOUN
ajst-13993	141	23	code	code	NOUN
ajst-13993	141	24	and	and	CCONJ
ajst-13993	141	25	then	then	ADV
ajst-13993	141	26	follows	follow	VERB
ajst-13993	141	27	the	the	DET
ajst-13993	141	28	same	same	ADJ
ajst-13993	141	29	framework	framework	NOUN
ajst-13993	141	30	.	.	PUNCT
ajst-13993	142	1	•the	•the	PRON
ajst-13993	142	2	proposed	propose	VERB
ajst-13993	142	3	approach	approach	NOUN
ajst-13993	142	4	:	:	PUNCT
ajst-13993	142	5	txt&ast&cfg+han+drl	txt&ast&cfg+han+drl	X
ajst-13993	142	6	.	.	PUNCT
ajst-13993	143	1	this	this	PRON
ajst-13993	143	2	is	be	AUX
ajst-13993	143	3	the	the	DET
ajst-13993	143	4	proposed	propose	VERB
ajst-13993	143	5	approach	approach	NOUN
ajst-13993	143	6	in	in	ADP
ajst-13993	143	7	which	which	PRON
ajst-13993	143	8	the	the	DET
ajst-13993	143	9	plain	plain	ADJ
ajst-13993	143	10	text	text	NOUN
ajst-13993	143	11	sequence	sequence	NOUN
ajst-13993	143	12	,	,	PUNCT
ajst-13993	143	13	the	the	DET
ajst-13993	143	14	sequenced	sequenced	ADJ
ajst-13993	143	15	abstract	abstract	ADJ
ajst-13993	143	16	syntax	syntax	NOUN
ajst-13993	143	17	and	and	CCONJ
ajst-13993	143	18	the	the	DET
ajst-13993	143	19	sequenced	sequence	VERB
ajst-13993	143	20	control	control	NOUN
ajst-13993	143	21	flow	flow	NOUN
ajst-13993	143	22	are	be	AUX
ajst-13993	143	23	adopted	adopt	VERB
ajst-13993	143	24	as	as	ADP
ajst-13993	143	25	the	the	DET
ajst-13993	143	26	representations	representation	NOUN
ajst-13993	143	27	of	of	ADP
ajst-13993	143	28	the	the	DET
ajst-13993	143	29	source	source	NOUN
ajst-13993	143	30	code	code	NOUN
ajst-13993	143	31	.	.	PUNCT
ajst-13993	144	1	then	then	ADV
ajst-13993	144	2	,	,	PUNCT
ajst-13993	144	3	the	the	DET
ajst-13993	144	4	lstm	lstm	NOUN
ajst-13993	144	5	-	-	PUNCT
ajst-13993	144	6	based	base	VERB
ajst-13993	144	7	hierarchical	hierarchical	ADJ
ajst-13993	144	8	network	network	NOUN
ajst-13993	144	9	is	be	AUX
ajst-13993	144	10	utilized	utilize	VERB
ajst-13993	144	11	to	to	PART
ajst-13993	144	12	encode	encode	VERB
ajst-13993	144	13	them	they	PRON
ajst-13993	144	14	into	into	ADP
ajst-13993	144	15	vectors	vector	NOUN
ajst-13993	144	16	and	and	CCONJ
ajst-13993	144	17	then	then	ADV
ajst-13993	144	18	a	a	DET
ajst-13993	144	19	hybrid	hybrid	ADJ
ajst-13993	144	20	layer	layer	NOUN
ajst-13993	144	21	is	be	AUX
ajst-13993	144	22	utilized	utilize	VERB
ajst-13993	144	23	to	to	PART
ajst-13993	144	24	concatenated	concatenate	VERB
ajst-13993	144	25	them	they	PRON
ajst-13993	144	26	into	into	ADP
ajst-13993	144	27	vector	vector	NOUN
ajst-13993	144	28	.	.	PUNCT
ajst-13993	145	1	the	the	DET
ajst-13993	145	2	experimental	experimental	ADJ
ajst-13993	145	3	results	result	NOUN
ajst-13993	145	4	compared	compare	VERB
ajst-13993	145	5	between	between	ADP
ajst-13993	145	6	the	the	DET
ajst-13993	145	7	proposed	propose	VERB
ajst-13993	145	8	approach	approach	NOUN
ajst-13993	145	9	and	and	CCONJ
ajst-13993	145	10	the	the	DET
ajst-13993	145	11	baselines	baseline	NOUN
ajst-13993	145	12	described	describe	VERB
ajst-13993	145	13	above	above	ADV
ajst-13993	145	14	is	be	AUX
ajst-13993	145	15	shown	show	VERB
ajst-13993	145	16	in	in	ADP
ajst-13993	145	17	table	table	NOUN
ajst-13993	145	18	78	78	NUM
ajst-13993	145	19	1	1	NUM
ajst-13993	145	20	.	.	PUNCT
ajst-13993	145	21	from	from	ADP
ajst-13993	145	22	the	the	DET
ajst-13993	145	23	results	result	NOUN
ajst-13993	145	24	we	we	PRON
ajst-13993	145	25	can	can	AUX
ajst-13993	145	26	see	see	VERB
ajst-13993	145	27	that	that	SCONJ
ajst-13993	145	28	the	the	DET
ajst-13993	145	29	proposed	propose	VERB
ajst-13993	145	30	approach	approach	NOUN
ajst-13993	145	31	can	can	AUX
ajst-13993	145	32	outperform	outperform	VERB
ajst-13993	145	33	almost	almost	ADV
ajst-13993	145	34	all	all	DET
ajst-13993	145	35	baselines	baseline	NOUN
ajst-13993	145	36	considering	consider	VERB
ajst-13993	145	37	most	most	ADJ
ajst-13993	145	38	estimation	estimation	NOUN
ajst-13993	145	39	metrics	metric	NOUN
ajst-13993	145	40	.	.	PUNCT
ajst-13993	146	1	in	in	ADP
ajst-13993	146	2	specific	specific	ADJ
ajst-13993	146	3	,	,	PUNCT
ajst-13993	146	4	the	the	DET
ajst-13993	146	5	accuracy	accuracy	NOUN
ajst-13993	146	6	of	of	ADP
ajst-13993	146	7	generated	generate	VERB
ajst-13993	146	8	comment	comment	NOUN
ajst-13993	146	9	by	by	ADP
ajst-13993	146	10	the	the	DET
ajst-13993	146	11	proposed	propose	VERB
ajst-13993	146	12	approach	approach	NOUN
ajst-13993	146	13	can	can	AUX
ajst-13993	146	14	outperform	outperform	VERB
ajst-13993	146	15	the	the	DET
ajst-13993	146	16	baselines	baseline	NOUN
ajst-13993	146	17	which	which	PRON
ajst-13993	146	18	use	use	VERB
ajst-13993	146	19	the	the	DET
ajst-13993	146	20	plain	plain	ADJ
ajst-13993	146	21	text	text	NOUN
ajst-13993	146	22	,	,	PUNCT
ajst-13993	146	23	the	the	DET
ajst-13993	146	24	sequenced	sequence	VERB
ajst-13993	146	25	abstract	abstract	ADJ
ajst-13993	146	26	syntax	syntax	NOUN
ajst-13993	146	27	tree	tree	NOUN
ajst-13993	146	28	or	or	CCONJ
ajst-13993	146	29	the	the	DET
ajst-13993	146	30	sequenced	sequence	VERB
ajst-13993	146	31	control	control	NOUN
ajst-13993	146	32	flow	flow	NOUN
ajst-13993	146	33	by	by	ADP
ajst-13993	146	34	29.46	29.46	NUM
ajst-13993	146	35	%	%	NOUN
ajst-13993	146	36	to	to	ADP
ajst-13993	146	37	31.42	31.42	NUM
ajst-13993	146	38	%	%	NOUN
ajst-13993	146	39	in	in	ADP
ajst-13993	146	40	terms	term	NOUN
ajst-13993	146	41	of	of	ADP
ajst-13993	146	42	bleu-1	bleu-1	PROPN
ajst-13993	146	43	.	.	PUNCT
ajst-13993	147	1	besides	besides	SCONJ
ajst-13993	147	2	,	,	PUNCT
ajst-13993	147	3	the	the	DET
ajst-13993	147	4	propose	propose	ADJ
ajst-13993	147	5	approach	approach	NOUN
ajst-13993	147	6	improves	improve	VERB
ajst-13993	147	7	the	the	DET
ajst-13993	147	8	generated	generate	VERB
ajst-13993	147	9	comments	comment	NOUN
ajst-13993	147	10	accuracy	accuracy	NOUN
ajst-13993	147	11	by	by	ADP
ajst-13993	147	12	16.58	16.58	NUM
ajst-13993	147	13	%	%	NOUN
ajst-13993	147	14	to	to	ADP
ajst-13993	147	15	20.53	20.53	NUM
ajst-13993	147	16	%	%	NOUN
ajst-13993	147	17	in	in	ADP
ajst-13993	147	18	terms	term	NOUN
ajst-13993	147	19	of	of	ADP
ajst-13993	147	20	bleu-1	bleu-1	NUM
ajst-13993	147	21	compared	compare	VERB
ajst-13993	147	22	with	with	ADP
ajst-13993	147	23	approaches	approach	NOUN
ajst-13993	147	24	in	in	ADP
ajst-13993	147	25	which	which	PRON
ajst-13993	147	26	two	two	NUM
ajst-13993	147	27	source	source	NOUN
ajst-13993	147	28	code	code	NOUN
ajst-13993	147	29	representations	representation	NOUN
ajst-13993	147	30	are	be	AUX
ajst-13993	147	31	adopted	adopt	VERB
ajst-13993	147	32	.	.	PUNCT
ajst-13993	148	1	furthermore	furthermore	ADV
ajst-13993	148	2	,	,	PUNCT
ajst-13993	148	3	similar	similar	ADJ
ajst-13993	148	4	result	result	NOUN
ajst-13993	148	5	trends	trend	NOUN
ajst-13993	148	6	are	be	AUX
ajst-13993	148	7	obtained	obtain	VERB
ajst-13993	148	8	considering	consider	VERB
ajst-13993	148	9	the	the	DET
ajst-13993	148	10	other	other	ADJ
ajst-13993	148	11	estimation	estimation	NOUN
ajst-13993	148	12	metrics	metric	NOUN
ajst-13993	148	13	.	.	PUNCT
ajst-13993	149	1	to	to	PART
ajst-13993	149	2	sum	sum	VERB
ajst-13993	149	3	up	up	ADP
ajst-13993	149	4	,	,	PUNCT
ajst-13993	149	5	better	well	ADJ
ajst-13993	149	6	performance	performance	NOUN
ajst-13993	149	7	can	can	AUX
ajst-13993	149	8	be	be	AUX
ajst-13993	149	9	performed	perform	VERB
ajst-13993	149	10	by	by	ADP
ajst-13993	149	11	the	the	DET
ajst-13993	149	12	proposed	propose	VERB
ajst-13993	149	13	approach	approach	NOUN
ajst-13993	149	14	according	accord	VERB
ajst-13993	149	15	to	to	ADP
ajst-13993	149	16	the	the	DET
ajst-13993	149	17	finer	finer	ADV
ajst-13993	149	18	-	-	PUNCT
ajst-13993	149	19	grained	grain	VERB
ajst-13993	149	20	code	code	NOUN
ajst-13993	149	21	representations	representation	NOUN
ajst-13993	149	22	for	for	ADP
ajst-13993	149	23	code	code	NOUN
ajst-13993	149	24	summarization	summarization	NOUN
ajst-13993	149	25	.	.	PUNCT
ajst-13993	150	1	4.4.2	4.4.2	X
ajst-13993	150	2	.	.	PUNCT
ajst-13993	151	1	the	the	DET
ajst-13993	151	2	performance	performance	NOUN
ajst-13993	151	3	of	of	ADP
ajst-13993	151	4	hierarchical	hierarchical	ADJ
ajst-13993	151	5	attention	attention	NOUN
ajst-13993	151	6	mechanism	mechanism	NOUN
ajst-13993	151	7	to	to	PART
ajst-13993	151	8	estimate	estimate	VERB
ajst-13993	151	9	the	the	DET
ajst-13993	151	10	performance	performance	NOUN
ajst-13993	151	11	of	of	ADP
ajst-13993	151	12	the	the	DET
ajst-13993	151	13	hierarchical	hierarchical	ADJ
ajst-13993	151	14	attention	attention	NOUN
ajst-13993	151	15	mechanism	mechanism	NOUN
ajst-13993	151	16	,	,	PUNCT
ajst-13993	151	17	we	we	PRON
ajst-13993	151	18	design	design	VERB
ajst-13993	151	19	the	the	DET
ajst-13993	151	20	baselines	baseline	NOUN
ajst-13993	151	21	by	by	ADP
ajst-13993	151	22	encoding	encode	VERB
ajst-13993	151	23	the	the	DET
ajst-13993	151	24	code	code	NOUN
ajst-13993	151	25	representations	representation	NOUN
ajst-13993	151	26	with	with	ADP
ajst-13993	151	27	no	no	DET
ajst-13993	151	28	attention	attention	NOUN
ajst-13993	151	29	,	,	PUNCT
ajst-13993	151	30	with	with	ADP
ajst-13993	151	31	1	1	NUM
ajst-13993	151	32	-	-	PUNCT
ajst-13993	151	33	layer	layer	NOUN
ajst-13993	151	34	attention	attention	NOUN
ajst-13993	151	35	and	and	CCONJ
ajst-13993	151	36	with	with	ADP
ajst-13993	151	37	2	2	NUM
ajst-13993	151	38	-	-	PUNCT
ajst-13993	151	39	layer	layer	NOUN
ajst-13993	151	40	attention	attention	NOUN
ajst-13993	151	41	respectively	respectively	ADV
ajst-13993	151	42	.	.	PUNCT
ajst-13993	152	1	in	in	ADP
ajst-13993	152	2	the	the	DET
ajst-13993	152	3	no	no	DET
ajst-13993	152	4	attention	attention	NOUN
ajst-13993	152	5	baseline	baseline	NOUN
ajst-13993	152	6	,	,	PUNCT
ajst-13993	152	7	the	the	DET
ajst-13993	152	8	code	code	NOUN
ajst-13993	152	9	representations	representation	NOUN
ajst-13993	152	10	are	be	AUX
ajst-13993	152	11	encoded	encode	VERB
ajst-13993	152	12	by	by	ADP
ajst-13993	152	13	normal	normal	ADJ
ajst-13993	152	14	lstm	lstm	NOUN
ajst-13993	152	15	without	without	ADP
ajst-13993	152	16	attention	attention	NOUN
ajst-13993	152	17	.	.	PUNCT
ajst-13993	153	1	while	while	SCONJ
ajst-13993	153	2	the	the	DET
ajst-13993	153	3	code	code	NOUN
ajst-13993	153	4	representations	representation	NOUN
ajst-13993	153	5	of	of	ADP
ajst-13993	153	6	the	the	DET
ajst-13993	153	7	source	source	NOUN
ajst-13993	153	8	code	code	NOUN
ajst-13993	153	9	are	be	AUX
ajst-13993	153	10	encoded	encode	VERB
ajst-13993	153	11	from	from	ADP
ajst-13993	153	12	tokens	token	NOUN
ajst-13993	153	13	to	to	PART
ajst-13993	153	14	function	function	VERB
ajst-13993	153	15	directly	directly	ADV
ajst-13993	153	16	in	in	ADP
ajst-13993	153	17	the	the	DET
ajst-13993	153	18	1	1	NUM
ajst-13993	153	19	-	-	PUNCT
ajst-13993	153	20	layer	layer	NOUN
ajst-13993	153	21	attention	attention	NOUN
ajst-13993	153	22	baseline	baseline	NOUN
ajst-13993	153	23	.	.	PUNCT
ajst-13993	154	1	in	in	ADP
ajst-13993	154	2	the	the	DET
ajst-13993	154	3	proposed	propose	VERB
ajst-13993	154	4	approach	approach	NOUN
ajst-13993	154	5	,	,	PUNCT
ajst-13993	154	6	the	the	DET
ajst-13993	154	7	hierarchical	hierarchical	ADJ
ajst-13993	154	8	structure	structure	NOUN
ajst-13993	154	9	of	of	ADP
ajst-13993	154	10	the	the	DET
ajst-13993	154	11	code	code	NOUN
ajst-13993	154	12	representations	representation	NOUN
ajst-13993	154	13	is	be	AUX
ajst-13993	154	14	adopted	adopt	VERB
ajst-13993	154	15	and	and	CCONJ
ajst-13993	154	16	the	the	DET
ajst-13993	154	17	2	2	NUM
ajst-13993	154	18	-	-	PUNCT
ajst-13993	154	19	layer	layer	NOUN
ajst-13993	154	20	attention	attention	NOUN
ajst-13993	154	21	mechanism	mechanism	NOUN
ajst-13993	154	22	is	be	AUX
ajst-13993	154	23	utilized	utilize	VERB
ajst-13993	154	24	.	.	PUNCT
ajst-13993	155	1	the	the	DET
ajst-13993	155	2	results	result	NOUN
ajst-13993	155	3	is	be	AUX
ajst-13993	155	4	presented	present	VERB
ajst-13993	155	5	in	in	ADP
ajst-13993	155	6	table	table	NOUN
ajst-13993	155	7	3.2	3.2	NUM
ajst-13993	155	8	,	,	PUNCT
ajst-13993	155	9	from	from	ADP
ajst-13993	155	10	which	which	PRON
ajst-13993	155	11	wen	wen	NOUN
ajst-13993	155	12	can	can	AUX
ajst-13993	155	13	observe	observe	VERB
ajst-13993	155	14	that	that	SCONJ
ajst-13993	155	15	the	the	DET
ajst-13993	155	16	baseline	baseline	NOUN
ajst-13993	155	17	performed	perform	VERB
ajst-13993	155	18	by	by	ADP
ajst-13993	155	19	1	1	NUM
ajst-13993	155	20	-	-	PUNCT
ajst-13993	155	21	layer	layer	NOUN
ajst-13993	155	22	attention	attention	NOUN
ajst-13993	155	23	can	can	AUX
ajst-13993	155	24	achieve	achieve	VERB
ajst-13993	155	25	better	well	ADJ
ajst-13993	155	26	performance	performance	NOUN
ajst-13993	155	27	by	by	ADP
ajst-13993	155	28	2.92	2.92	NUM
ajst-13993	155	29	%	%	NOUN
ajst-13993	155	30	to	to	ADP
ajst-13993	155	31	37.47	37.47	NUM
ajst-13993	155	32	%	%	NOUN
ajst-13993	155	33	consider	consider	VERB
ajst-13993	155	34	the	the	DET
ajst-13993	155	35	estimation	estimation	NOUN
ajst-13993	155	36	metrics	metric	NOUN
ajst-13993	155	37	.	.	PUNCT
ajst-13993	156	1	while	while	SCONJ
ajst-13993	156	2	the	the	DET
ajst-13993	156	3	proposed	propose	VERB
ajst-13993	156	4	approach	approach	NOUN
ajst-13993	156	5	can	can	AUX
ajst-13993	156	6	improve	improve	VERB
ajst-13993	156	7	the	the	DET
ajst-13993	156	8	baseline	baseline	NOUN
ajst-13993	156	9	with	with	ADP
ajst-13993	156	10	1	1	NUM
ajst-13993	156	11	-	-	PUNCT
ajst-13993	156	12	layer	layer	NOUN
ajst-13993	156	13	attention	attention	NOUN
ajst-13993	156	14	by	by	ADP
ajst-13993	156	15	4.36	4.36	NUM
ajst-13993	156	16	%	%	NOUN
ajst-13993	156	17	to	to	ADP
ajst-13993	156	18	49.59	49.59	NUM
ajst-13993	156	19	%	%	NOUN
ajst-13993	156	20	considering	consider	VERB
ajst-13993	156	21	different	different	ADJ
ajst-13993	156	22	estimation	estimation	NOUN
ajst-13993	156	23	metrics	metric	NOUN
ajst-13993	156	24	.	.	PUNCT
ajst-13993	157	1	these	these	DET
ajst-13993	157	2	results	result	NOUN
ajst-13993	157	3	demonstrate	demonstrate	VERB
ajst-13993	157	4	that	that	SCONJ
ajst-13993	157	5	the	the	DET
ajst-13993	157	6	adopted	adopt	VERB
ajst-13993	157	7	hierarchical	hierarchical	ADJ
ajst-13993	157	8	attention	attention	NOUN
ajst-13993	157	9	can	can	AUX
ajst-13993	157	10	improve	improve	VERB
ajst-13993	157	11	the	the	DET
ajst-13993	157	12	performance	performance	NOUN
ajst-13993	157	13	dramatically	dramatically	ADV
ajst-13993	157	14	.	.	PUNCT
ajst-13993	158	1	4.4.3	4.4.3	X
ajst-13993	158	2	.	.	PUNCT
ajst-13993	159	1	the	the	DET
ajst-13993	159	2	performance	performance	NOUN
ajst-13993	159	3	of	of	ADP
ajst-13993	159	4	deep	deep	ADJ
ajst-13993	159	5	reinforcement	reinforcement	NOUN
ajst-13993	159	6	learning	learning	NOUN
ajst-13993	159	7	to	to	PART
ajst-13993	159	8	estimate	estimate	VERB
ajst-13993	159	9	the	the	DET
ajst-13993	159	10	performance	performance	NOUN
ajst-13993	159	11	of	of	ADP
ajst-13993	159	12	the	the	DET
ajst-13993	159	13	proposed	propose	VERB
ajst-13993	159	14	reinforcement	reinforcement	NOUN
ajst-13993	159	15	learning	learning	NOUN
ajst-13993	159	16	component	component	NOUN
ajst-13993	159	17	,	,	PUNCT
ajst-13993	159	18	we	we	PRON
ajst-13993	159	19	train	train	VERB
ajst-13993	159	20	the	the	DET
ajst-13993	159	21	model	model	NOUN
ajst-13993	159	22	by	by	ADP
ajst-13993	159	23	withing	withe	VERB
ajst-13993	159	24	and	and	CCONJ
ajst-13993	159	25	without	without	ADP
ajst-13993	159	26	the	the	DET
ajst-13993	159	27	component	component	NOUN
ajst-13993	159	28	of	of	ADP
ajst-13993	159	29	reinforcement	reinforcement	NOUN
ajst-13993	159	30	learning	learning	NOUN
ajst-13993	159	31	,	,	PUNCT
ajst-13993	159	32	and	and	CCONJ
ajst-13993	159	33	they	they	PRON
ajst-13993	159	34	are	be	AUX
ajst-13993	159	35	signed	sign	VERB
ajst-13993	159	36	as	as	ADP
ajst-13993	159	37	“	"	PUNCT
ajst-13993	159	38	approach	approach	NOUN
ajst-13993	159	39	with	with	ADP
ajst-13993	159	40	drl	drl	PROPN
ajst-13993	159	41	”	"	PUNCT
ajst-13993	159	42	and	and	CCONJ
ajst-13993	159	43	“	"	PUNCT
ajst-13993	159	44	approach	approach	NOUN
ajst-13993	159	45	without	without	ADP
ajst-13993	159	46	drl	drl	PROPN
ajst-13993	159	47	”	"	PUNCT
ajst-13993	159	48	.	.	PUNCT
ajst-13993	160	1	the	the	DET
ajst-13993	160	2	corresponding	corresponding	ADJ
ajst-13993	160	3	results	result	NOUN
ajst-13993	160	4	are	be	AUX
ajst-13993	160	5	shown	show	VERB
ajst-13993	160	6	in	in	ADP
ajst-13993	160	7	table	table	NOUN
ajst-13993	160	8	3.3	3.3	NUM
ajst-13993	160	9	,	,	PUNCT
ajst-13993	160	10	from	from	ADP
ajst-13993	160	11	which	which	PRON
ajst-13993	160	12	we	we	PRON
ajst-13993	160	13	can	can	AUX
ajst-13993	160	14	see	see	VERB
ajst-13993	160	15	that	that	SCONJ
ajst-13993	160	16	the	the	DET
ajst-13993	160	17	proposed	propose	VERB
ajst-13993	160	18	approach	approach	NOUN
ajst-13993	160	19	can	can	AUX
ajst-13993	160	20	achieve	achieve	VERB
ajst-13993	160	21	better	well	ADJ
ajst-13993	160	22	values	value	NOUN
ajst-13993	160	23	by	by	ADP
ajst-13993	160	24	14.13	14.13	NUM
ajst-13993	160	25	%	%	NOUN
ajst-13993	160	26	to	to	ADP
ajst-13993	160	27	130.30	130.30	NUM
ajst-13993	160	28	%	%	NOUN
ajst-13993	160	29	compared	compare	VERB
ajst-13993	160	30	with	with	ADP
ajst-13993	160	31	the	the	DET
ajst-13993	160	32	baseline	baseline	NOUN
ajst-13993	160	33	without	without	ADP
ajst-13993	160	34	reinforcement	reinforcement	NOUN
ajst-13993	160	35	learning	learning	NOUN
ajst-13993	160	36	.	.	PUNCT
ajst-13993	161	1	these	these	DET
ajst-13993	161	2	results	result	NOUN
ajst-13993	161	3	can	can	AUX
ajst-13993	161	4	prove	prove	VERB
ajst-13993	161	5	that	that	SCONJ
ajst-13993	161	6	the	the	DET
ajst-13993	161	7	proposed	propose	VERB
ajst-13993	161	8	approach	approach	NOUN
ajst-13993	161	9	can	can	AUX
ajst-13993	161	10	improve	improve	VERB
ajst-13993	161	11	the	the	DET
ajst-13993	161	12	comment	comment	NOUN
ajst-13993	161	13	generation	generation	NOUN
ajst-13993	161	14	performance	performance	NOUN
ajst-13993	161	15	significantly	significantly	ADV
ajst-13993	161	16	.	.	PUNCT
ajst-13993	162	1	figure	figure	VERB
ajst-13993	162	2	3	3	NUM
ajst-13993	162	3	.	.	PUNCT
ajst-13993	163	1	the	the	DET
ajst-13993	163	2	results	result	NOUN
ajst-13993	163	3	trend	trend	NOUN
ajst-13993	163	4	vs.	vs.	ADP
ajst-13993	163	5	the	the	DET
ajst-13993	163	6	code	code	NOUN
ajst-13993	163	7	length	length	NOUN
ajst-13993	163	8	.	.	PUNCT
ajst-13993	164	1	figure	figure	NOUN
ajst-13993	164	2	4	4	NUM
ajst-13993	164	3	.	.	PUNCT
ajst-13993	165	1	the	the	DET
ajst-13993	165	2	results	result	NOUN
ajst-13993	165	3	trend	trend	NOUN
ajst-13993	165	4	vs.	vs.	ADP
ajst-13993	165	5	the	the	DET
ajst-13993	165	6	comment	comment	NOUN
ajst-13993	165	7	length	length	NOUN
ajst-13993	165	8	.	.	PUNCT
ajst-13993	166	1	4.5	4.5	NUM
ajst-13993	166	2	.	.	PUNCT
ajst-13993	166	3	rq2	rq2	NOUN
ajst-13993	166	4	:	:	PUNCT
ajst-13993	166	5	performance	performance	NOUN
ajst-13993	166	6	considering	consider	VERB
ajst-13993	166	7	different	different	ADJ
ajst-13993	166	8	code	code	NOUN
ajst-13993	166	9	and	and	CCONJ
ajst-13993	166	10	comment	comment	NOUN
ajst-13993	166	11	length	length	NOUN
ajst-13993	166	12	to	to	PART
ajst-13993	166	13	measure	measure	VERB
ajst-13993	166	14	the	the	DET
ajst-13993	166	15	influence	influence	NOUN
ajst-13993	166	16	on	on	ADP
ajst-13993	166	17	the	the	DET
ajst-13993	166	18	code	code	NOUN
ajst-13993	166	19	summarization	summarization	NOUN
ajst-13993	166	20	performance	performance	NOUN
ajst-13993	166	21	considering	consider	VERB
ajst-13993	166	22	different	different	ADJ
ajst-13993	166	23	code	code	NOUN
ajst-13993	166	24	and	and	CCONJ
ajst-13993	166	25	comment	comment	NOUN
ajst-13993	166	26	lengths	length	NOUN
ajst-13993	166	27	,	,	PUNCT
ajst-13993	166	28	we	we	PRON
ajst-13993	166	29	vary	vary	VERB
ajst-13993	166	30	the	the	DET
ajst-13993	166	31	split	split	NOUN
ajst-13993	166	32	of	of	ADP
ajst-13993	166	33	the	the	DET
ajst-13993	166	34	dataset	dataset	NOUN
ajst-13993	166	35	by	by	ADP
ajst-13993	166	36	the	the	DET
ajst-13993	166	37	lengths	length	NOUN
ajst-13993	166	38	of	of	ADP
ajst-13993	166	39	code	code	NOUN
ajst-13993	166	40	and	and	CCONJ
ajst-13993	166	41	comment	comment	VERB
ajst-13993	166	42	respectively	respectively	ADV
ajst-13993	166	43	.	.	PUNCT
ajst-13993	167	1	figures	figure	NOUN
ajst-13993	167	2	3	3	NUM
ajst-13993	167	3	and	and	CCONJ
ajst-13993	167	4	4	4	NUM
ajst-13993	167	5	demonstrate	demonstrate	VERB
ajst-13993	167	6	the	the	DET
ajst-13993	167	7	results	result	NOUN
ajst-13993	167	8	,	,	PUNCT
ajst-13993	167	9	and	and	CCONJ
ajst-13993	167	10	from	from	ADP
ajst-13993	167	11	which	which	PRON
ajst-13993	167	12	we	we	PRON
ajst-13993	167	13	can	can	AUX
ajst-13993	167	14	see	see	VERB
ajst-13993	167	15	that	that	SCONJ
ajst-13993	167	16	the	the	DET
ajst-13993	167	17	best	good	ADJ
ajst-13993	167	18	performance	performance	NOUN
ajst-13993	167	19	can	can	AUX
ajst-13993	167	20	be	be	AUX
ajst-13993	167	21	achieved	achieve	VERB
ajst-13993	167	22	by	by	ADP
ajst-13993	167	23	the	the	DET
ajst-13993	167	24	proposed	propose	VERB
ajst-13993	167	25	approach	approach	NOUN
ajst-13993	167	26	compared	compare	VERB
ajst-13993	167	27	with	with	ADP
ajst-13993	167	28	the	the	DET
ajst-13993	167	29	baselines	baseline	NOUN
ajst-13993	167	30	from	from	ADP
ajst-13993	167	31	the	the	DET
ajst-13993	167	32	aspect	aspect	NOUN
ajst-13993	167	33	of	of	ADP
ajst-13993	167	34	all	all	DET
ajst-13993	167	35	the	the	DET
ajst-13993	167	36	utilized	utilize	VERB
ajst-13993	167	37	metrics	metric	NOUN
ajst-13993	167	38	.	.	PUNCT
ajst-13993	168	1	from	from	ADP
ajst-13993	168	2	the	the	DET
ajst-13993	168	3	aspect	aspect	NOUN
ajst-13993	168	4	of	of	ADP
ajst-13993	168	5	bleu	bleu	NOUN
ajst-13993	168	6	,	,	PUNCT
ajst-13993	168	7	the	the	DET
ajst-13993	168	8	proposed	propose	VERB
ajst-13993	168	9	approach	approach	NOUN
ajst-13993	168	10	improves	improve	VERB
ajst-13993	168	11	the	the	DET
ajst-13993	168	12	baselines	baseline	NOUN
ajst-13993	168	13	that	that	PRON
ajst-13993	168	14	adopted	adopt	VERB
ajst-13993	168	15	different	different	ADJ
ajst-13993	168	16	representations	representation	NOUN
ajst-13993	168	17	of	of	ADP
ajst-13993	168	18	source	source	NOUN
ajst-13993	168	19	code	code	NOUN
ajst-13993	168	20	by	by	ADP
ajst-13993	168	21	35.74	35.74	NUM
ajst-13993	168	22	%	%	NOUN
ajst-13993	168	23	,	,	PUNCT
ajst-13993	168	24	43.31	43.31	NUM
ajst-13993	168	25	%	%	NOUN
ajst-13993	168	26	,	,	PUNCT
ajst-13993	168	27	41.13	41.13	NUM
ajst-13993	168	28	%	%	NOUN
ajst-13993	168	29	,	,	PUNCT
ajst-13993	168	30	17.04	17.04	NUM
ajst-13993	168	31	%	%	NOUN
ajst-13993	168	32	,	,	PUNCT
ajst-13993	168	33	1l6.97	1l6.97	NUM
ajst-13993	168	34	%	%	NOUN
ajst-13993	168	35	,	,	PUNCT
ajst-13993	168	36	and	and	CCONJ
ajst-13993	168	37	12.66	12.66	NUM
ajst-13993	168	38	%	%	NOUN
ajst-13993	168	39	respectively	respectively	ADV
ajst-13993	168	40	when	when	SCONJ
ajst-13993	168	41	the	the	DET
ajst-13993	168	42	source	source	NOUN
ajst-13993	168	43	code	code	NOUN
ajst-13993	168	44	contains	contain	VERB
ajst-13993	168	45	40	40	NUM
ajst-13993	168	46	tokens	token	NOUN
ajst-13993	168	47	.	.	PUNCT
ajst-13993	169	1	in	in	ADP
ajst-13993	169	2	specific	specific	ADJ
ajst-13993	169	3	,	,	PUNCT
ajst-13993	169	4	the	the	DET
ajst-13993	169	5	baselines	baseline	NOUN
ajst-13993	169	6	that	that	PRON
ajst-13993	169	7	adopted	adopt	VERB
ajst-13993	169	8	two	two	NUM
ajst-13993	169	9	representations	representation	NOUN
ajst-13993	169	10	of	of	ADP
ajst-13993	169	11	the	the	DET
ajst-13993	169	12	source	source	NOUN
ajst-13993	169	13	code	code	NOUN
ajst-13993	169	14	can	can	AUX
ajst-13993	169	15	all	all	ADV
ajst-13993	169	16	achieve	achieve	VERB
ajst-13993	169	17	better	well	ADJ
ajst-13993	169	18	performance	performance	NOUN
ajst-13993	169	19	than	than	ADP
ajst-13993	169	20	the	the	DET
ajst-13993	169	21	baselines	baseline	NOUN
ajst-13993	169	22	which	which	PRON
ajst-13993	169	23	adopts	adopt	VERB
ajst-13993	169	24	only	only	ADV
ajst-13993	169	25	one	one	NUM
ajst-13993	169	26	representation	representation	NOUN
ajst-13993	169	27	of	of	ADP
ajst-13993	169	28	the	the	DET
ajst-13993	169	29	source	source	NOUN
ajst-13993	169	30	code	code	NOUN
ajst-13993	169	31	.	.	PUNCT
ajst-13993	170	1	furthermore	furthermore	ADV
ajst-13993	170	2	,	,	PUNCT
ajst-13993	170	3	the	the	DET
ajst-13993	170	4	proposed	propose	VERB
ajst-13993	170	5	approach	approach	NOUN
ajst-13993	170	6	can	can	AUX
ajst-13993	170	7	always	always	ADV
ajst-13993	170	8	improve	improve	VERB
ajst-13993	170	9	the	the	DET
ajst-13993	170	10	performance	performance	NOUN
ajst-13993	170	11	compared	compare	VERB
ajst-13993	170	12	with	with	ADP
ajst-13993	170	13	the	the	DET
ajst-13993	170	14	baselines	baseline	NOUN
ajst-13993	170	15	which	which	PRON
ajst-13993	170	16	adopted	adopt	VERB
ajst-13993	170	17	two	two	NUM
ajst-13993	170	18	representations	representation	NOUN
ajst-13993	170	19	of	of	ADP
ajst-13993	170	20	the	the	DET
ajst-13993	170	21	source	source	NOUN
ajst-13993	170	22	code	code	NOUN
ajst-13993	170	23	.	.	PUNCT
ajst-13993	171	1	for	for	ADP
ajst-13993	171	2	other	other	ADJ
ajst-13993	171	3	estimation	estimation	NOUN
ajst-13993	171	4	metrics	metric	NOUN
ajst-13993	171	5	,	,	PUNCT
ajst-13993	171	6	the	the	DET
ajst-13993	171	7	similar	similar	ADJ
ajst-13993	171	8	results	result	NOUN
ajst-13993	171	9	can	can	AUX
ajst-13993	171	10	be	be	AUX
ajst-13993	171	11	achieved	achieve	VERB
ajst-13993	171	12	.	.	PUNCT
ajst-13993	172	1	these	these	DET
ajst-13993	172	2	phenomenons	phenomenon	NOUN
ajst-13993	172	3	can	can	AUX
ajst-13993	172	4	prove	prove	VERB
ajst-13993	172	5	the	the	DET
ajst-13993	172	6	effectiveness	effectiveness	NOUN
ajst-13993	172	7	of	of	ADP
ajst-13993	172	8	the	the	DET
ajst-13993	172	9	proposed	propose	VERB
ajst-13993	172	10	code	code	NOUN
ajst-13993	172	11	representation	representation	NOUN
ajst-13993	172	12	component	component	NOUN
ajst-13993	172	13	.	.	PUNCT
ajst-13993	172	14	table	table	NOUN
ajst-13993	172	15	4	4	NUM
ajst-13993	172	16	.	.	PUNCT
ajst-13993	173	1	the	the	DET
ajst-13993	173	2	time	time	NOUN
ajst-13993	173	3	cost	cost	NOUN
ajst-13993	173	4	to	to	PART
ajst-13993	173	5	train	train	VERB
ajst-13993	173	6	different	different	ADJ
ajst-13993	173	7	models	model	NOUN
ajst-13993	173	8	(	(	PUNCT
ajst-13993	173	9	mins	min	NOUN
ajst-13993	173	10	)	)	PUNCT
ajst-13993	173	11	.	.	PUNCT
ajst-13993	174	1	79	79	NUM
ajst-13993	175	1	the	the	DET
ajst-13993	175	2	code	code	NOUN
ajst-13993	175	3	summarization	summarization	NOUN
ajst-13993	175	4	result	result	NOUN
ajst-13993	175	5	considering	consider	VERB
ajst-13993	175	6	different	different	ADJ
ajst-13993	175	7	comment	comment	NOUN
ajst-13993	175	8	lengths	length	NOUN
ajst-13993	175	9	is	be	AUX
ajst-13993	175	10	presented	present	VERB
ajst-13993	175	11	in	in	ADP
ajst-13993	175	12	figure	figure	NOUN
ajst-13993	175	13	3.10	3.10	NUM
ajst-13993	175	14	,	,	PUNCT
ajst-13993	175	15	from	from	ADP
ajst-13993	175	16	which	which	PRON
ajst-13993	175	17	we	we	PRON
ajst-13993	175	18	can	can	AUX
ajst-13993	175	19	find	find	VERB
ajst-13993	175	20	that	that	SCONJ
ajst-13993	175	21	the	the	DET
ajst-13993	175	22	proposed	propose	VERB
ajst-13993	175	23	approach	approach	NOUN
ajst-13993	175	24	performs	perform	VERB
ajst-13993	175	25	better	well	ADV
ajst-13993	175	26	compared	compare	VERB
ajst-13993	175	27	with	with	ADP
ajst-13993	175	28	the	the	DET
ajst-13993	175	29	baselines	baseline	NOUN
ajst-13993	175	30	considering	consider	VERB
ajst-13993	175	31	different	different	ADJ
ajst-13993	175	32	lengths	length	NOUN
ajst-13993	175	33	of	of	ADP
ajst-13993	175	34	comment	comment	NOUN
ajst-13993	175	35	.	.	PUNCT
ajst-13993	176	1	in	in	ADP
ajst-13993	176	2	specific	specific	ADJ
ajst-13993	176	3	,	,	PUNCT
ajst-13993	176	4	when	when	SCONJ
ajst-13993	176	5	the	the	DET
ajst-13993	176	6	length	length	NOUN
ajst-13993	176	7	of	of	ADP
ajst-13993	176	8	the	the	DET
ajst-13993	176	9	comment	comment	NOUN
ajst-13993	176	10	is	be	AUX
ajst-13993	176	11	20	20	NUM
ajst-13993	176	12	,	,	PUNCT
ajst-13993	176	13	the	the	DET
ajst-13993	176	14	performance	performance	NOUN
ajst-13993	176	15	of	of	ADP
ajst-13993	176	16	the	the	DET
ajst-13993	176	17	proposed	propose	VERB
ajst-13993	176	18	approach	approach	NOUN
ajst-13993	176	19	improves	improve	VERB
ajst-13993	176	20	107.61	107.61	NUM
ajst-13993	176	21	%	%	NOUN
ajst-13993	176	22	,	,	PUNCT
ajst-13993	176	23	100.31	100.31	NUM
ajst-13993	176	24	%	%	NOUN
ajst-13993	176	25	,	,	PUNCT
ajst-13993	176	26	200.59	200.59	NUM
ajst-13993	176	27	%	%	NOUN
ajst-13993	176	28	,	,	PUNCT
ajst-13993	176	29	51.77	51.77	NUM
ajst-13993	176	30	%	%	NOUN
ajst-13993	176	31	,	,	PUNCT
ajst-13993	176	32	42.64	42.64	NUM
ajst-13993	176	33	%	%	NOUN
ajst-13993	176	34	,	,	PUNCT
ajst-13993	176	35	and	and	CCONJ
ajst-13993	176	36	47.77	47.77	NUM
ajst-13993	176	37	%	%	NOUN
ajst-13993	176	38	respectively	respectively	ADV
ajst-13993	176	39	.	.	PUNCT
ajst-13993	177	1	we	we	PRON
ajst-13993	177	2	can	can	AUX
ajst-13993	177	3	observe	observe	VERB
ajst-13993	177	4	that	that	SCONJ
ajst-13993	177	5	the	the	DET
ajst-13993	177	6	performance	performance	NOUN
ajst-13993	177	7	tends	tend	VERB
ajst-13993	177	8	to	to	PART
ajst-13993	177	9	be	be	AUX
ajst-13993	177	10	worse	bad	ADJ
ajst-13993	177	11	when	when	SCONJ
ajst-13993	177	12	the	the	DET
ajst-13993	177	13	length	length	NOUN
ajst-13993	177	14	of	of	ADP
ajst-13993	177	15	the	the	DET
ajst-13993	177	16	comment	comment	NOUN
ajst-13993	177	17	increased	increase	VERB
ajst-13993	177	18	which	which	PRON
ajst-13993	177	19	is	be	AUX
ajst-13993	177	20	similar	similar	ADJ
ajst-13993	177	21	to	to	ADP
ajst-13993	177	22	the	the	DET
ajst-13993	177	23	results	result	NOUN
ajst-13993	177	24	in	in	ADP
ajst-13993	177	25	the	the	DET
ajst-13993	177	26	neural	neural	ADJ
ajst-13993	177	27	machine	machine	NOUN
ajst-13993	177	28	translation	translation	NOUN
ajst-13993	177	29	task	task	NOUN
ajst-13993	178	1	[	[	X
ajst-13993	178	2	24	24	NUM
ajst-13993	178	3	-	-	SYM
ajst-13993	178	4	25	25	NUM
ajst-13993	178	5	]	]	PUNCT
ajst-13993	178	6	.	.	PUNCT
ajst-13993	179	1	4.6	4.6	NUM
ajst-13993	179	2	.	.	PUNCT
ajst-13993	180	1	rq3	rq3	PROPN
ajst-13993	180	2	:	:	PUNCT
ajst-13993	180	3	time	time	NOUN
ajst-13993	180	4	consumption	consumption	NOUN
ajst-13993	180	5	analysis	analysis	NOUN
ajst-13993	180	6	to	to	PART
ajst-13993	180	7	estimate	estimate	VERB
ajst-13993	180	8	the	the	DET
ajst-13993	180	9	time	time	NOUN
ajst-13993	180	10	complexity	complexity	NOUN
ajst-13993	180	11	of	of	ADP
ajst-13993	180	12	the	the	DET
ajst-13993	180	13	proposed	propose	VERB
ajst-13993	180	14	approach	approach	NOUN
ajst-13993	180	15	,	,	PUNCT
ajst-13993	180	16	the	the	DET
ajst-13993	180	17	average	average	ADJ
ajst-13993	180	18	training	training	NOUN
ajst-13993	180	19	time	time	NOUN
ajst-13993	180	20	of	of	ADP
ajst-13993	180	21	each	each	DET
ajst-13993	180	22	epoch	epoch	NOUN
ajst-13993	180	23	considering	consider	VERB
ajst-13993	180	24	different	different	ADJ
ajst-13993	180	25	representations	representation	NOUN
ajst-13993	180	26	of	of	ADP
ajst-13993	180	27	the	the	DET
ajst-13993	180	28	source	source	NOUN
ajst-13993	180	29	code	code	NOUN
ajst-13993	180	30	are	be	AUX
ajst-13993	180	31	recorded	record	VERB
ajst-13993	180	32	.	.	PUNCT
ajst-13993	181	1	the	the	DET
ajst-13993	181	2	result	result	NOUN
ajst-13993	181	3	is	be	AUX
ajst-13993	181	4	given	give	VERB
ajst-13993	181	5	in	in	ADP
ajst-13993	181	6	table	table	NOUN
ajst-13993	181	7	4	4	NUM
ajst-13993	181	8	,	,	PUNCT
ajst-13993	181	9	from	from	ADP
ajst-13993	181	10	which	which	PRON
ajst-13993	181	11	we	we	PRON
ajst-13993	181	12	can	can	AUX
ajst-13993	181	13	find	find	VERB
ajst-13993	181	14	that	that	SCONJ
ajst-13993	181	15	for	for	ADP
ajst-13993	181	16	all	all	DET
ajst-13993	181	17	the	the	DET
ajst-13993	181	18	three	three	NUM
ajst-13993	181	19	stages	stage	NOUN
ajst-13993	181	20	in	in	ADP
ajst-13993	181	21	the	the	DET
ajst-13993	181	22	proposed	propose	VERB
ajst-13993	181	23	approach	approach	NOUN
ajst-13993	181	24	,	,	PUNCT
ajst-13993	181	25	each	each	DET
ajst-13993	181	26	epoch	epoch	NOUN
ajst-13993	181	27	costs	cost	VERB
ajst-13993	181	28	less	less	ADJ
ajst-13993	181	29	than	than	ADP
ajst-13993	181	30	1	1	NUM
ajst-13993	181	31	hour	hour	NOUN
ajst-13993	181	32	for	for	ADP
ajst-13993	181	33	training	training	NOUN
ajst-13993	181	34	.	.	PUNCT
ajst-13993	182	1	this	this	DET
ajst-13993	182	2	result	result	NOUN
ajst-13993	182	3	shows	show	VERB
ajst-13993	182	4	that	that	SCONJ
ajst-13993	182	5	the	the	DET
ajst-13993	182	6	time	time	NOUN
ajst-13993	182	7	complexity	complexity	NOUN
ajst-13993	182	8	of	of	ADP
ajst-13993	182	9	the	the	DET
ajst-13993	182	10	proposed	propose	VERB
ajst-13993	182	11	approach	approach	NOUN
ajst-13993	182	12	is	be	AUX
ajst-13993	182	13	low	low	ADJ
ajst-13993	182	14	and	and	CCONJ
ajst-13993	182	15	the	the	DET
ajst-13993	182	16	approach	approach	NOUN
ajst-13993	182	17	is	be	AUX
ajst-13993	182	18	reasonable	reasonable	ADJ
ajst-13993	182	19	for	for	ADP
ajst-13993	182	20	practical	practical	ADJ
ajst-13993	182	21	usage	usage	NOUN
ajst-13993	182	22	.	.	PUNCT
ajst-13993	183	1	4.7	4.7	NUM
ajst-13993	183	2	.	.	PUNCT
ajst-13993	183	3	case	case	NOUN
ajst-13993	183	4	study	study	NOUN
ajst-13993	183	5	we	we	PRON
ajst-13993	183	6	demonstrate	demonstrate	VERB
ajst-13993	183	7	four	four	NUM
ajst-13993	183	8	real	real	ADJ
ajst-13993	183	9	-	-	PUNCT
ajst-13993	183	10	world	world	NOUN
ajst-13993	183	11	code	code	NOUN
ajst-13993	183	12	examples	example	NOUN
ajst-13993	183	13	for	for	ADP
ajst-13993	183	14	generating	generate	VERB
ajst-13993	183	15	their	their	PRON
ajst-13993	183	16	comments	comment	NOUN
ajst-13993	183	17	using	use	VERB
ajst-13993	183	18	our	our	PRON
ajst-13993	183	19	approach	approach	NOUN
ajst-13993	183	20	in	in	ADP
ajst-13993	183	21	table	table	NOUN
ajst-13993	183	22	3.5	3.5	NUM
ajst-13993	183	23	.	.	PUNCT
ajst-13993	184	1	in	in	ADP
ajst-13993	184	2	this	this	DET
ajst-13993	184	3	table	table	NOUN
ajst-13993	184	4	,	,	PUNCT
ajst-13993	184	5	we	we	PRON
ajst-13993	184	6	first	first	ADV
ajst-13993	184	7	show	show	VERB
ajst-13993	184	8	the	the	DET
ajst-13993	184	9	code	code	NOUN
ajst-13993	184	10	snippet	snippet	NOUN
ajst-13993	184	11	in	in	ADP
ajst-13993	184	12	the	the	DET
ajst-13993	184	13	second	second	ADJ
ajst-13993	184	14	line	line	NOUN
ajst-13993	184	15	and	and	CCONJ
ajst-13993	184	16	then	then	ADV
ajst-13993	184	17	give	give	VERB
ajst-13993	184	18	the	the	DET
ajst-13993	184	19	ground	ground	NOUN
ajst-13993	184	20	truth	truth	NOUN
ajst-13993	184	21	comment	comment	NOUN
ajst-13993	184	22	which	which	PRON
ajst-13993	184	23	is	be	AUX
ajst-13993	184	24	the	the	DET
ajst-13993	184	25	code	code	NOUN
ajst-13993	184	26	comment	comment	NOUN
ajst-13993	184	27	that	that	PRON
ajst-13993	184	28	is	be	AUX
ajst-13993	184	29	collected	collect	VERB
ajst-13993	184	30	together	together	ADV
ajst-13993	184	31	with	with	ADP
ajst-13993	184	32	the	the	DET
ajst-13993	184	33	code	code	NOUN
ajst-13993	184	34	snippet	snippet	NOUN
ajst-13993	184	35	from	from	ADP
ajst-13993	184	36	github	github	PROPN
ajst-13993	184	37	.	.	PUNCT
ajst-13993	185	1	next	next	ADV
ajst-13993	185	2	,	,	PUNCT
ajst-13993	185	3	the	the	DET
ajst-13993	185	4	generated	generate	VERB
ajst-13993	185	5	comments	comment	NOUN
ajst-13993	185	6	by	by	ADP
ajst-13993	185	7	different	different	ADJ
ajst-13993	185	8	approaches	approach	NOUN
ajst-13993	185	9	are	be	AUX
ajst-13993	185	10	given.for	given.for	X
ajst-13993	185	11	our	our	PRON
ajst-13993	185	12	approach	approach	NOUN
ajst-13993	185	13	,	,	PUNCT
ajst-13993	185	14	shown	show	VERB
ajst-13993	185	15	as	as	ADP
ajst-13993	185	16	2layer+drl	2layer+drl	PROPN
ajst-13993	185	17	,	,	PUNCT
ajst-13993	185	18	we	we	PRON
ajst-13993	185	19	have	have	AUX
ajst-13993	185	20	highlighted	highlight	VERB
ajst-13993	185	21	the	the	DET
ajst-13993	185	22	words	word	NOUN
ajst-13993	185	23	that	that	PRON
ajst-13993	185	24	are	be	AUX
ajst-13993	185	25	closer	close	ADJ
ajst-13993	185	26	to	to	ADP
ajst-13993	185	27	the	the	DET
ajst-13993	185	28	ground	ground	NOUN
ajst-13993	185	29	truth	truth	NOUN
ajst-13993	185	30	.	.	PUNCT
ajst-13993	186	1	it	it	PRON
ajst-13993	186	2	can	can	AUX
ajst-13993	186	3	be	be	AUX
ajst-13993	186	4	observed	observe	VERB
ajst-13993	186	5	that	that	SCONJ
ajst-13993	186	6	the	the	DET
ajst-13993	186	7	generated	generate	VERB
ajst-13993	186	8	comments	comment	NOUN
ajst-13993	186	9	by	by	ADP
ajst-13993	186	10	our	our	PRON
ajst-13993	186	11	approach	approach	NOUN
ajst-13993	186	12	are	be	AUX
ajst-13993	186	13	the	the	DET
ajst-13993	186	14	closest	close	ADJ
ajst-13993	186	15	to	to	ADP
ajst-13993	186	16	the	the	DET
ajst-13993	186	17	ground	ground	NOUN
ajst-13993	186	18	truth	truth	NOUN
ajst-13993	186	19	.	.	PUNCT
ajst-13993	187	1	although	although	SCONJ
ajst-13993	187	2	the	the	DET
ajst-13993	187	3	approaches	approach	NOUN
ajst-13993	187	4	with	with	ADP
ajst-13993	187	5	drl	drl	PROPN
ajst-13993	187	6	(	(	PUNCT
ajst-13993	187	7	1	1	NUM
ajst-13993	187	8	-	-	PUNCT
ajst-13993	187	9	layer+drl	layer+drl	NOUN
ajst-13993	187	10	)	)	PUNCT
ajst-13993	187	11	can	can	AUX
ajst-13993	187	12	generate	generate	VERB
ajst-13993	187	13	some	some	DET
ajst-13993	187	14	tokens	token	NOUN
ajst-13993	187	15	which	which	PRON
ajst-13993	187	16	are	be	AUX
ajst-13993	187	17	also	also	ADV
ajst-13993	187	18	in	in	ADP
ajst-13993	187	19	the	the	DET
ajst-13993	187	20	ground	ground	NOUN
ajst-13993	187	21	truth	truth	NOUN
ajst-13993	187	22	,	,	PUNCT
ajst-13993	187	23	they	they	PRON
ajst-13993	187	24	can	can	AUX
ajst-13993	187	25	not	not	PART
ajst-13993	187	26	predict	predict	VERB
ajst-13993	187	27	those	those	DET
ajst-13993	187	28	tokens	token	NOUN
ajst-13993	187	29	which	which	PRON
ajst-13993	187	30	do	do	AUX
ajst-13993	187	31	not	not	PART
ajst-13993	187	32	frequently	frequently	ADV
ajst-13993	187	33	appear	appear	VERB
ajst-13993	187	34	in	in	ADP
ajst-13993	187	35	the	the	DET
ajst-13993	187	36	training	training	NOUN
ajst-13993	187	37	data	datum	NOUN
ajst-13993	187	38	,	,	PUNCT
ajst-13993	187	39	i.e.	i.e.	X
ajst-13993	187	40	,	,	PUNCT
ajst-13993	187	41	object	object	NOUN
ajst-13993	187	42	in	in	ADP
ajst-13993	187	43	the	the	DET
ajst-13993	187	44	case	case	NOUN
ajst-13993	187	45	example.on	example.on	NUM
ajst-13993	187	46	the	the	DET
ajst-13993	187	47	contrary	contrary	NOUN
ajst-13993	187	48	,	,	PUNCT
ajst-13993	187	49	the	the	DET
ajst-13993	187	50	deep	deep	ADJ
ajst-13993	187	51	-	-	PUNCT
ajst-13993	187	52	reinforcement	reinforcement	NOUN
ajst-13993	187	53	-	-	PUNCT
ajst-13993	187	54	learning	learning	NOUN
ajst-13993	187	55	-	-	PUNCT
ajst-13993	187	56	based	base	VERB
ajst-13993	187	57	approach	approach	NOUN
ajst-13993	187	58	can	can	AUX
ajst-13993	187	59	generate	generate	VERB
ajst-13993	187	60	some	some	DET
ajst-13993	187	61	tokens	token	NOUN
ajst-13993	187	62	which	which	PRON
ajst-13993	187	63	are	be	AUX
ajst-13993	187	64	closer	close	ADJ
ajst-13993	187	65	to	to	ADP
ajst-13993	187	66	the	the	DET
ajst-13993	187	67	ground	ground	NOUN
ajst-13993	187	68	truth	truth	NOUN
ajst-13993	187	69	,	,	PUNCT
ajst-13993	187	70	such	such	ADJ
ajst-13993	187	71	as	as	ADP
ajst-13993	187	72	process	process	NOUN
ajst-13993	187	73	.	.	PUNCT
ajst-13993	188	1	this	this	PRON
ajst-13993	188	2	can	can	AUX
ajst-13993	188	3	be	be	AUX
ajst-13993	188	4	illustrated	illustrate	VERB
ajst-13993	188	5	by	by	ADP
ajst-13993	188	6	the	the	DET
ajst-13993	188	7	fact	fact	NOUN
ajst-13993	188	8	that	that	SCONJ
ajst-13993	188	9	our	our	PRON
ajst-13993	188	10	approach	approach	NOUN
ajst-13993	188	11	has	have	AUX
ajst-13993	188	12	a	a	DET
ajst-13993	188	13	more	more	ADV
ajst-13993	188	14	comprehensive	comprehensive	ADJ
ajst-13993	188	15	exploration	exploration	NOUN
ajst-13993	188	16	on	on	ADP
ajst-13993	188	17	the	the	DET
ajst-13993	188	18	word	word	NOUN
ajst-13993	188	19	space	space	NOUN
ajst-13993	188	20	and	and	CCONJ
ajst-13993	188	21	optimizes	optimize	VERB
ajst-13993	188	22	the	the	DET
ajst-13993	188	23	bleu	bleu	NOUN
ajst-13993	188	24	score	score	NOUN
ajst-13993	188	25	directly	directly	ADV
ajst-13993	188	26	.	.	PUNCT
ajst-13993	189	1	table	table	NOUN
ajst-13993	189	2	5	5	NUM
ajst-13993	189	3	.	.	PUNCT
ajst-13993	189	4	case	case	NOUN
ajst-13993	189	5	study	study	NOUN
ajst-13993	189	6	of	of	ADP
ajst-13993	189	7	code	code	PROPN
ajst-13993	189	8	summary	summary	NOUN
ajst-13993	189	9	generated	generate	VERB
ajst-13993	189	10	by	by	ADP
ajst-13993	189	11	each	each	DET
ajst-13993	189	12	approach	approach	NOUN
ajst-13993	189	13	.	.	PUNCT
ajst-13993	190	1	5	5	X
ajst-13993	190	2	.	.	X
ajst-13993	190	3	threats	threat	NOUN
ajst-13993	190	4	to	to	PART
ajst-13993	190	5	validity	validity	VERB
ajst-13993	190	6	the	the	DET
ajst-13993	190	7	threat	threat	NOUN
ajst-13993	190	8	to	to	ADP
ajst-13993	190	9	validity	validity	NOUN
ajst-13993	190	10	is	be	AUX
ajst-13993	190	11	that	that	SCONJ
ajst-13993	190	12	we	we	PRON
ajst-13993	190	13	evaluate	evaluate	VERB
ajst-13993	190	14	the	the	DET
ajst-13993	190	15	proposed	propose	VERB
ajst-13993	190	16	approach	approach	NOUN
ajst-13993	190	17	only	only	ADV
ajst-13993	190	18	based	base	VERB
ajst-13993	190	19	on	on	ADP
ajst-13993	190	20	the	the	DET
ajst-13993	190	21	dataset	dataset	NOUN
ajst-13993	190	22	which	which	PRON
ajst-13993	190	23	is	be	AUX
ajst-13993	190	24	consisted	consist	VERB
ajst-13993	190	25	of	of	ADP
ajst-13993	190	26	python	python	PROPN
ajst-13993	190	27	code	code	NOUN
ajst-13993	190	28	and	and	CCONJ
ajst-13993	190	29	comment	comment	NOUN
ajst-13993	190	30	pairs	pair	NOUN
ajst-13993	190	31	,	,	PUNCT
ajst-13993	190	32	therefore	therefore	ADV
ajst-13993	190	33	it	it	PRON
ajst-13993	190	34	may	may	AUX
ajst-13993	190	35	be	be	AUX
ajst-13993	190	36	unrepresentative	unrepresentative	ADJ
ajst-13993	190	37	of	of	ADP
ajst-13993	190	38	code	code	NOUN
ajst-13993	190	39	summarization	summarization	NOUN
ajst-13993	190	40	considering	consider	VERB
ajst-13993	190	41	other	other	ADJ
ajst-13993	190	42	programming	programming	NOUN
ajst-13993	190	43	languages	language	NOUN
ajst-13993	190	44	.	.	PUNCT
ajst-13993	191	1	however	however	ADV
ajst-13993	191	2	,	,	PUNCT
ajst-13993	191	3	as	as	SCONJ
ajst-13993	191	4	the	the	DET
ajst-13993	191	5	components	component	NOUN
ajst-13993	191	6	in	in	ADP
ajst-13993	191	7	the	the	DET
ajst-13993	191	8	proposed	propose	VERB
ajst-13993	191	9	approach	approach	NOUN
ajst-13993	191	10	are	be	AUX
ajst-13993	191	11	all	all	PRON
ajst-13993	191	12	general	general	ADJ
ajst-13993	191	13	models	model	NOUN
ajst-13993	191	14	which	which	PRON
ajst-13993	191	15	can	can	AUX
ajst-13993	191	16	be	be	AUX
ajst-13993	191	17	adopted	adopt	VERB
ajst-13993	191	18	to	to	PART
ajst-13993	191	19	realize	realize	VERB
ajst-13993	191	20	the	the	DET
ajst-13993	191	21	code	code	NOUN
ajst-13993	191	22	annotation	annotation	NOUN
ajst-13993	191	23	regarding	regard	VERB
ajst-13993	191	24	other	other	ADJ
ajst-13993	191	25	programming	programming	NOUN
ajst-13993	191	26	language	language	NOUN
ajst-13993	191	27	.	.	PUNCT
ajst-13993	192	1	besides	besides	SCONJ
ajst-13993	192	2	,	,	PUNCT
ajst-13993	192	3	this	this	DET
ajst-13993	192	4	approach	approach	NOUN
ajst-13993	192	5	is	be	AUX
ajst-13993	192	6	based	base	VERB
ajst-13993	192	7	on	on	ADP
ajst-13993	192	8	static	static	ADJ
ajst-13993	192	9	analysis	analysis	NOUN
ajst-13993	192	10	which	which	PRON
ajst-13993	192	11	could	could	AUX
ajst-13993	192	12	be	be	AUX
ajst-13993	192	13	a	a	DET
ajst-13993	192	14	barrier	barrier	NOUN
ajst-13993	192	15	to	to	PART
ajst-13993	192	16	adopt	adopt	VERB
ajst-13993	192	17	the	the	DET
ajst-13993	192	18	proposed	propose	VERB
ajst-13993	192	19	approach	approach	NOUN
ajst-13993	192	20	,	,	PUNCT
ajst-13993	192	21	for	for	ADP
ajst-13993	192	22	example	example	NOUN
ajst-13993	192	23	,	,	PUNCT
ajst-13993	192	24	to	to	PART
ajst-13993	192	25	obtain	obtain	VERB
ajst-13993	192	26	the	the	DET
ajst-13993	192	27	effective	effective	ADJ
ajst-13993	192	28	static	static	ADJ
ajst-13993	192	29	analysis	analysis	NOUN
ajst-13993	192	30	results	result	NOUN
ajst-13993	192	31	,	,	PUNCT
ajst-13993	192	32	significant	significant	ADJ
ajst-13993	192	33	effort	effort	NOUN
ajst-13993	192	34	should	should	AUX
ajst-13993	192	35	be	be	AUX
ajst-13993	192	36	made	make	VERB
ajst-13993	192	37	.	.	PUNCT
ajst-13993	193	1	6	6	X
ajst-13993	193	2	.	.	X
ajst-13993	193	3	conclusion	conclusion	NOUN
ajst-13993	193	4	in	in	ADP
ajst-13993	193	5	this	this	DET
ajst-13993	193	6	work	work	NOUN
ajst-13993	193	7	,	,	PUNCT
ajst-13993	193	8	we	we	PRON
ajst-13993	193	9	propose	propose	VERB
ajst-13993	193	10	to	to	PART
ajst-13993	193	11	utilize	utilize	VERB
ajst-13993	193	12	three	three	NUM
ajst-13993	193	13	representations	representation	NOUN
ajst-13993	193	14	of	of	ADP
ajst-13993	193	15	the	the	DET
ajst-13993	193	16	source	source	NOUN
ajst-13993	193	17	code	code	NOUN
ajst-13993	193	18	considering	consider	VERB
ajst-13993	193	19	both	both	CCONJ
ajst-13993	193	20	the	the	DET
ajst-13993	193	21	unstructured	unstructured	ADJ
ajst-13993	193	22	and	and	CCONJ
ajst-13993	193	23	structural	structural	ADJ
ajst-13993	193	24	features	feature	NOUN
ajst-13993	193	25	,	,	PUNCT
ajst-13993	193	26	namely	namely	ADV
ajst-13993	193	27	,	,	PUNCT
ajst-13993	193	28	the	the	DET
ajst-13993	193	29	plain	plain	ADJ
ajst-13993	193	30	text	text	NOUN
ajst-13993	193	31	,	,	PUNCT
ajst-13993	193	32	the	the	DET
ajst-13993	193	33	sequenced	sequence	VERB
ajst-13993	193	34	abstract	abstract	ADJ
ajst-13993	193	35	syntax	syntax	NOUN
ajst-13993	193	36	tree	tree	NOUN
ajst-13993	193	37	and	and	CCONJ
ajst-13993	193	38	the	the	DET
ajst-13993	193	39	sequenced	sequence	VERB
ajst-13993	193	40	control	control	NOUN
ajst-13993	193	41	flow	flow	NOUN
ajst-13993	193	42	to	to	PART
ajst-13993	193	43	represent	represent	VERB
ajst-13993	193	44	the	the	DET
ajst-13993	193	45	hierarchical	hierarchical	ADJ
ajst-13993	193	46	structure	structure	NOUN
ajst-13993	193	47	of	of	ADP
ajst-13993	193	48	the	the	DET
ajst-13993	193	49	source	source	NOUN
ajst-13993	193	50	code	code	NOUN
ajst-13993	193	51	.	.	PUNCT
ajst-13993	194	1	and	and	CCONJ
ajst-13993	194	2	the	the	DET
ajst-13993	194	3	two	two	NUM
ajst-13993	194	4	-	-	PUNCT
ajst-13993	194	5	layer	layer	NOUN
ajst-13993	194	6	hierarchical	hierarchical	ADJ
ajst-13993	194	7	attention	attention	NOUN
ajst-13993	194	8	mechanism	mechanism	NOUN
ajst-13993	194	9	is	be	AUX
ajst-13993	194	10	adopted	adopt	VERB
ajst-13993	194	11	to	to	PART
ajst-13993	194	12	encode	encode	VERB
ajst-13993	194	13	the	the	DET
ajst-13993	194	14	source	source	NOUN
ajst-13993	194	15	code	code	NOUN
ajst-13993	194	16	representations.to	representations.to	PRON
ajst-13993	194	17	estimate	estimate	VERB
ajst-13993	194	18	the	the	DET
ajst-13993	194	19	effectiveness	effectiveness	NOUN
ajst-13993	194	20	of	of	ADP
ajst-13993	194	21	the	the	DET
ajst-13993	194	22	proposed	propose	VERB
ajst-13993	194	23	approach	approach	NOUN
ajst-13993	194	24	,	,	PUNCT
ajst-13993	194	25	a	a	DET
ajst-13993	194	26	few	few	ADJ
ajst-13993	194	27	of	of	ADP
ajst-13993	194	28	experiments	experiment	NOUN
ajst-13993	194	29	based	base	VERB
ajst-13993	194	30	on	on	ADP
ajst-13993	194	31	the	the	DET
ajst-13993	194	32	dataset	dataset	NOUN
ajst-13993	194	33	grabbed	grab	VERB
ajst-13993	194	34	from	from	ADP
ajst-13993	194	35	real	real	ADJ
ajst-13993	194	36	world	world	NOUN
ajst-13993	194	37	projects	project	NOUN
ajst-13993	194	38	are	be	AUX
ajst-13993	194	39	performed	perform	VERB
ajst-13993	194	40	.	.	PUNCT
ajst-13993	195	1	the	the	DET
ajst-13993	195	2	results	result	NOUN
ajst-13993	195	3	prove	prove	VERB
ajst-13993	195	4	the	the	DET
ajst-13993	195	5	high	high	ADJ
ajst-13993	195	6	quality	quality	NOUN
ajst-13993	195	7	of	of	ADP
ajst-13993	195	8	the	the	DET
ajst-13993	195	9	generated	generate	VERB
ajst-13993	195	10	comment	comment	NOUN
ajst-13993	195	11	.	.	PUNCT
ajst-13993	196	1	acknowledgment	acknowledgment	NOUN
ajst-13993	196	2	the	the	DET
ajst-13993	196	3	authors	author	NOUN
ajst-13993	196	4	gratefully	gratefully	ADV
ajst-13993	196	5	acknowledge	acknowledge	VERB
ajst-13993	196	6	the	the	DET
ajst-13993	196	7	financial	financial	ADJ
ajst-13993	196	8	support	support	NOUN
ajst-13993	196	9	from	from	ADP
ajst-13993	196	10	science	science	NOUN
ajst-13993	196	11	and	and	CCONJ
ajst-13993	196	12	technology	technology	NOUN
ajst-13993	196	13	planning	planning	NOUN
ajst-13993	196	14	project	project	NOUN
ajst-13993	196	15	of	of	ADP
ajst-13993	196	16	yunnan	yunnan	PROPN
ajst-13993	196	17	province	province	PROPN
ajst-13993	196	18	(	(	PUNCT
ajst-13993	196	19	joint	joint	ADJ
ajst-13993	196	20	project	project	NOUN
ajst-13993	196	21	of	of	ADP
ajst-13993	196	22	local	local	ADJ
ajst-13993	196	23	universities	university	NOUN
ajst-13993	196	24	)	)	PUNCT
ajst-13993	196	25	(	(	PUNCT
ajst-13993	196	26	202001ba070001	202001ba070001	NOUN
ajst-13993	196	27	-	-	PUNCT
ajst-13993	196	28	096	096	NUM
ajst-13993	196	29	)	)	PUNCT
ajst-13993	196	30	and	and	CCONJ
ajst-13993	196	31	fund	fund	NOUN
ajst-13993	196	32	project	project	NOUN
ajst-13993	196	33	of	of	ADP
ajst-13993	196	34	yunnan	yunnan	PROPN
ajst-13993	196	35	provincial	provincial	ADJ
ajst-13993	196	36	education	education	NOUN
ajst-13993	196	37	department	department	NOUN
ajst-13993	196	38	.	.	PUNCT
ajst-13993	197	1	references	reference	NOUN
ajst-13993	197	2	[	[	X
ajst-13993	197	3	1	1	NUM
ajst-13993	197	4	]	]	PUNCT
ajst-13993	197	5	n.kalchbrenner	n.kalchbrenner	NOUN
ajst-13993	197	6	,	,	PUNCT
ajst-13993	197	7	l.	l.	PROPN
ajst-13993	197	8	espeholt	espeholt	PROPN
ajst-13993	197	9	,	,	PUNCT
ajst-13993	197	10	k.	k.	PROPN
ajst-13993	197	11	simonyan	simonyan	PROPN
ajst-13993	197	12	,	,	PUNCT
ajst-13993	197	13	a.	a.	PROPN
ajst-13993	197	14	v.	v.	PROPN
ajst-13993	197	15	d.	d.	PROPN
ajst-13993	197	16	oord	oord	PROPN
ajst-13993	197	17	,	,	PUNCT
ajst-13993	197	18	and	and	CCONJ
ajst-13993	197	19	k.	k.	PROPN
ajst-13993	197	20	kavukcuoglu	kavukcuoglu	PROPN
ajst-13993	197	21	,	,	PUNCT
ajst-13993	197	22	neural	neural	ADJ
ajst-13993	197	23	machine	machine	NOUN
ajst-13993	197	24	translation	translation	NOUN
ajst-13993	197	25	in	in	ADP
ajst-13993	197	26	linear	linear	PROPN
ajst-13993	197	27	time	time	NOUN
ajst-13993	197	28	,	,	PUNCT
ajst-13993	197	29	arxiv	arxiv	PROPN
ajst-13993	197	30	preprint	preprint	NOUN
ajst-13993	197	31	arxiv:1610.10099	arxiv:1610.10099	NUM
ajst-13993	197	32	,	,	PUNCT
ajst-13993	197	33	(	(	PUNCT
ajst-13993	197	34	2016	2016	NUM
ajst-13993	197	35	)	)	PUNCT
ajst-13993	197	36	.	.	PUNCT
ajst-13993	198	1	[	[	X
ajst-13993	198	2	2	2	X
ajst-13993	198	3	]	]	PUNCT
ajst-13993	198	4	a.graves	a.grave	NOUN
ajst-13993	198	5	,	,	PUNCT
ajst-13993	198	6	generating	generate	VERB
ajst-13993	198	7	sequences	sequence	NOUN
ajst-13993	198	8	with	with	ADP
ajst-13993	198	9	recurrent	recurrent	ADJ
ajst-13993	198	10	neural	neural	ADJ
ajst-13993	198	11	networks	network	NOUN
ajst-13993	198	12	,	,	PUNCT
ajst-13993	198	13	arxiv	arxiv	PROPN
ajst-13993	198	14	preprint	preprint	VERB
ajst-13993	198	15	arxiv:1308.0850	arxiv:1308.0850	NOUN
ajst-13993	198	16	,	,	PUNCT
ajst-13993	198	17	(	(	PUNCT
ajst-13993	198	18	2013	2013	NUM
ajst-13993	198	19	)	)	PUNCT
ajst-13993	198	20	.	.	PUNCT
ajst-13993	199	1	[	[	X
ajst-13993	199	2	3	3	X
ajst-13993	199	3	]	]	PUNCT
ajst-13993	199	4	a.louis	a.louis	ADJ
ajst-13993	199	5	,	,	PUNCT
ajst-13993	199	6	s.k	s.k	PROPN
ajst-13993	199	7	.	.	NOUN
ajst-13993	199	8	dash	dash	NOUN
ajst-13993	199	9	,	,	PUNCT
ajst-13993	199	10	e.	e.	PROPN
ajst-13993	199	11	t.	t.	PROPN
ajst-13993	199	12	barr	barr	PROPN
ajst-13993	199	13	,	,	PUNCT
ajst-13993	199	14	and	and	CCONJ
ajst-13993	199	15	c.	c.	PROPN
ajst-13993	199	16	sutton	sutton	PROPN
ajst-13993	199	17	,	,	PUNCT
ajst-13993	199	18	deep	deep	ADJ
ajst-13993	199	19	learning	learn	VERB
ajst-13993	199	20	to	to	PART
ajst-13993	199	21	detect	detect	VERB
ajst-13993	199	22	redundant	redundant	ADJ
ajst-13993	199	23	method	method	NOUN
ajst-13993	199	24	comments	comment	NOUN
ajst-13993	199	25	,	,	PUNCT
ajst-13993	199	26	arxiv	arxiv	PROPN
ajst-13993	199	27	preprint	preprint	NOUN
ajst-13993	199	28	arxiv:1806.04616	arxiv:1806.04616	ADV
ajst-13993	199	29	,	,	PUNCT
ajst-13993	199	30	(	(	PUNCT
ajst-13993	199	31	2018	2018	NUM
ajst-13993	199	32	)	)	PUNCT
ajst-13993	199	33	.	.	PUNCT
ajst-13993	200	1	[	[	X
ajst-13993	200	2	4	4	NUM
ajst-13993	200	3	]	]	SYM
ajst-13993	200	4	r.collobert	r.collobert	NOUN
ajst-13993	200	5	and	and	CCONJ
ajst-13993	200	6	j.weston	j.weston	PROPN
ajst-13993	200	7	,	,	PUNCT
ajst-13993	200	8	a	a	DET
ajst-13993	200	9	unified	unified	ADJ
ajst-13993	200	10	architecture	architecture	NOUN
ajst-13993	200	11	for	for	ADP
ajst-13993	200	12	natural	natural	ADJ
ajst-13993	200	13	language	language	NOUN
ajst-13993	200	14	processing	processing	NOUN
ajst-13993	200	15	:	:	PUNCT
ajst-13993	200	16	deep	deep	ADJ
ajst-13993	200	17	neural	neural	ADJ
ajst-13993	200	18	networks	network	NOUN
ajst-13993	200	19	with	with	ADP
ajst-13993	200	20	multitask	multitask	ADJ
ajst-13993	200	21	learning	learning	NOUN
ajst-13993	200	22	,	,	PUNCT
ajst-13993	200	23	in	in	ADP
ajst-13993	200	24	machine	machine	NOUN
ajst-13993	200	25	learning	learning	NOUN
ajst-13993	200	26	,	,	PUNCT
ajst-13993	200	27	proceedings	proceeding	NOUN
ajst-13993	200	28	of	of	ADP
ajst-13993	200	29	the	the	DET
ajst-13993	200	30	twenty	twenty	NUM
ajst-13993	200	31	-	-	PUNCT
ajst-13993	200	32	fifth	fifth	ADJ
ajst-13993	200	33	international	international	ADJ
ajst-13993	200	34	conference	conference	NOUN
ajst-13993	200	35	(	(	PUNCT
ajst-13993	200	36	icml	icml	PROPN
ajst-13993	200	37	2008	2008	NUM
ajst-13993	200	38	)	)	PUNCT
ajst-13993	200	39	,	,	PUNCT
ajst-13993	200	40	2008	2008	NUM
ajst-13993	200	41	,	,	PUNCT
ajst-13993	200	42	pp	pp	ADJ
ajst-13993	200	43	.	.	PUNCT
ajst-13993	201	1	160–167	160–167	NUM
ajst-13993	201	2	.	.	PUNCT
ajst-13993	202	1	[	[	X
ajst-13993	202	2	5	5	NUM
ajst-13993	202	3	]	]	PUNCT
ajst-13993	202	4	s.iyer	s.iyer	NOUN
ajst-13993	202	5	,	,	PUNCT
ajst-13993	202	6	i.konstas	i.konsta	NOUN
ajst-13993	202	7	,	,	PUNCT
ajst-13993	202	8	a.cheung	a.cheung	NOUN
ajst-13993	202	9	,	,	PUNCT
ajst-13993	202	10	andl.zettlemoyer	andl.zettlemoyer	PROPN
ajst-13993	202	11	,	,	PUNCT
ajst-13993	202	12	summarizing	summarize	VERB
ajst-13993	202	13	source	source	NOUN
ajst-13993	202	14	code	code	NOUN
ajst-13993	202	15	using	use	VERB
ajst-13993	202	16	a	a	DET
ajst-13993	202	17	neural	neural	ADJ
ajst-13993	202	18	attention	attention	NOUN
ajst-13993	202	19	model	model	NOUN
ajst-13993	202	20	.	.	PUNCT
ajst-13993	202	21	,	,	PUNCT
ajst-13993	202	22	in	in	ADP
ajst-13993	202	23	proceedings	proceeding	NOUN
ajst-13993	202	24	of	of	ADP
ajst-13993	202	25	the	the	DET
ajst-13993	202	26	54th	54th	ADJ
ajst-13993	202	27	annual	annual	ADJ
ajst-13993	202	28	meeting	meeting	NOUN
ajst-13993	202	29	of	of	ADP
ajst-13993	202	30	the	the	DET
ajst-13993	202	31	association	association	NOUN
ajst-13993	202	32	for	for	ADP
ajst-13993	202	33	computational	computational	ADJ
ajst-13993	202	34	linguistics	linguistic	NOUN
ajst-13993	202	35	(	(	PUNCT
ajst-13993	202	36	1	1	NUM
ajst-13993	202	37	)	)	PUNCT
ajst-13993	202	38	,	,	PUNCT
ajst-13993	202	39	2016	2016	NUM
ajst-13993	202	40	,	,	PUNCT
ajst-13993	202	41	pp	pp	ADJ
ajst-13993	202	42	.	.	PUNCT
ajst-13993	203	1	2073–2083	2073–2083	X
ajst-13993	203	2	.	.	PUNCT
ajst-13993	204	1	[	[	X
ajst-13993	204	2	6	6	NUM
ajst-13993	204	3	]	]	PUNCT
ajst-13993	204	4	m.	m.	PROPN
ajst-13993	204	5	malhotra	malhotra	PROPN
ajst-13993	204	6	and	and	CCONJ
ajst-13993	204	7	j.	j.	PROPN
ajst-13993	204	8	k.	k.	PROPN
ajst-13993	204	9	chhabra	chhabra	PROPN
ajst-13993	204	10	,	,	PUNCT
ajst-13993	204	11	class	class	NOUN
ajst-13993	204	12	level	level	NOUN
ajst-13993	204	13	code	code	NOUN
ajst-13993	204	14	summarization	summarization	NOUN
ajst-13993	204	15	based	base	VERB
ajst-13993	204	16	on	on	ADP
ajst-13993	204	17	dependencies	dependency	NOUN
ajst-13993	204	18	and	and	CCONJ
ajst-13993	204	19	micro	micro	NOUN
ajst-13993	204	20	patterns	pattern	NOUN
ajst-13993	204	21	,	,	PUNCT
ajst-13993	204	22	in	in	ADP
ajst-13993	204	23	2018	2018	NUM
ajst-13993	204	24	second	second	ADJ
ajst-13993	204	25	international	international	ADJ
ajst-13993	204	26	conference	conference	NOUN
ajst-13993	204	27	on	on	ADP
ajst-13993	204	28	inventive	inventive	ADJ
ajst-13993	204	29	communication	communication	NOUN
ajst-13993	204	30	and	and	CCONJ
ajst-13993	204	31	computational	computational	ADJ
ajst-13993	204	32	technologies	technology	NOUN
ajst-13993	204	33	(	(	PUNCT
ajst-13993	204	34	icicct	icicct	ADJ
ajst-13993	204	35	)	)	PUNCT
ajst-13993	204	36	,	,	PUNCT
ajst-13993	204	37	ieee	ieee	NOUN
ajst-13993	204	38	,	,	PUNCT
ajst-13993	204	39	2018	2018	NUM
ajst-13993	204	40	,	,	PUNCT
ajst-13993	204	41	pp	pp	ADJ
ajst-13993	204	42	.	.	PUNCT
ajst-13993	205	1	1011–1016	1011–1016	NUM
ajst-13993	205	2	.	.	PUNCT
ajst-13993	206	1	[	[	X
ajst-13993	206	2	7	7	NUM
ajst-13993	206	3	]	]	PUNCT
ajst-13993	206	4	a.	a.	NOUN
ajst-13993	206	5	see	see	NOUN
ajst-13993	206	6	,	,	PUNCT
ajst-13993	206	7	p.	p.	PROPN
ajst-13993	206	8	j.	j.	PROPN
ajst-13993	206	9	liu	liu	PROPN
ajst-13993	206	10	,	,	PUNCT
ajst-13993	206	11	and	and	CCONJ
ajst-13993	206	12	c.	c.	PROPN
ajst-13993	206	13	d.	d.	PROPN
ajst-13993	206	14	manning	manning	PROPN
ajst-13993	206	15	,	,	PUNCT
ajst-13993	206	16	get	get	VERB
ajst-13993	206	17	to	to	ADP
ajst-13993	206	18	the	the	DET
ajst-13993	206	19	point	point	NOUN
ajst-13993	206	20	:	:	PUNCT
ajst-13993	206	21	summarization	summarization	NOUN
ajst-13993	206	22	with	with	ADP
ajst-13993	206	23	pointer	pointer	NOUN
ajst-13993	206	24	-	-	PUNCT
ajst-13993	206	25	generator	generator	NOUN
ajst-13993	206	26	networks	network	NOUN
ajst-13993	206	27	,	,	PUNCT
ajst-13993	206	28	in	in	ADP
ajst-13993	206	29	proceedings	proceeding	NOUN
ajst-13993	206	30	of	of	ADP
ajst-13993	206	31	the	the	DET
ajst-13993	206	32	55th	55th	ADJ
ajst-13993	206	33	annual	annual	ADJ
ajst-13993	206	34	meeting	meeting	NOUN
ajst-13993	206	35	of	of	ADP
ajst-13993	206	36	the	the	DET
ajst-13993	206	37	association	association	NOUN
ajst-13993	206	38	for	for	ADP
ajst-13993	206	39	computational	computational	ADJ
ajst-13993	206	40	linguistics	linguistic	NOUN
ajst-13993	206	41	(	(	PUNCT
ajst-13993	206	42	volume	volume	NOUN
ajst-13993	206	43	1	1	NUM
ajst-13993	206	44	:	:	PUNCT
ajst-13993	206	45	long	long	ADJ
ajst-13993	206	46	papers	paper	NOUN
ajst-13993	206	47	)	)	PUNCT
ajst-13993	206	48	,	,	PUNCT
ajst-13993	206	49	vancouver	vancouver	PROPN
ajst-13993	206	50	,	,	PUNCT
ajst-13993	206	51	canada	canada	PROPN
ajst-13993	206	52	,	,	PUNCT
ajst-13993	206	53	july	july	PROPN
ajst-13993	206	54	2017	2017	NUM
ajst-13993	206	55	,	,	PUNCT
ajst-13993	206	56	association	association	NOUN
ajst-13993	206	57	for	for	ADP
ajst-13993	206	58	computational	computational	ADJ
ajst-13993	206	59	linguistics	linguistic	NOUN
ajst-13993	206	60	,	,	PUNCT
ajst-13993	206	61	pp	pp	ADJ
ajst-13993	206	62	.	.	PUNCT
ajst-13993	207	1	1073–1083	1073–1083	NUM
ajst-13993	207	2	.	.	PUNCT
ajst-13993	208	1	[	[	X
ajst-13993	208	2	8	8	NUM
ajst-13993	208	3	]	]	X
ajst-13993	208	4	h.shi	h.shi	PROPN
ajst-13993	208	5	,	,	PUNCT
ajst-13993	208	6	h.	h.	PROPN
ajst-13993	208	7	zhou	zhou	PROPN
ajst-13993	208	8	,	,	PUNCT
ajst-13993	208	9	j.	j.	PROPN
ajst-13993	208	10	chen	chen	PROPN
ajst-13993	208	11	,	,	PUNCT
ajst-13993	208	12	and	and	CCONJ
ajst-13993	208	13	l.	l.	PROPN
ajst-13993	208	14	li	li	PROPN
ajst-13993	208	15	,	,	PUNCT
ajst-13993	208	16	on	on	ADP
ajst-13993	208	17	tree	tree	NOUN
ajst-13993	208	18	-	-	PUNCT
ajst-13993	208	19	based	base	VERB
ajst-13993	208	20	neural	neural	ADJ
ajst-13993	208	21	sentence	sentence	NOUN
ajst-13993	208	22	modeling	modeling	NOUN
ajst-13993	208	23	,	,	PUNCT
ajst-13993	208	24	arxiv	arxiv	PROPN
ajst-13993	208	25	preprint	preprint	NOUN
ajst-13993	208	26	arxiv:1808.09644	arxiv:1808.09644	NUM
ajst-13993	208	27	,	,	PUNCT
ajst-13993	208	28	(	(	PUNCT
ajst-13993	208	29	2018	2018	NUM
ajst-13993	208	30	)	)	PUNCT
ajst-13993	208	31	.	.	PUNCT
ajst-13993	209	1	80	80	NUM
ajst-13993	210	1	[	[	X
ajst-13993	210	2	9	9	NUM
ajst-13993	210	3	]	]	X
ajst-13993	210	4	k.papineni	k.papineni	NOUN
ajst-13993	210	5	,	,	PUNCT
ajst-13993	210	6	s.roukos	s.roukos	ADJ
ajst-13993	210	7	,	,	PUNCT
ajst-13993	210	8	t.	t.	NOUN
ajst-13993	210	9	ward	ward	NOUN
ajst-13993	210	10	,	,	PUNCT
ajst-13993	210	11	and	and	CCONJ
ajst-13993	210	12	w.	w.	PROPN
ajst-13993	210	13	j.	j.	PROPN
ajst-13993	210	14	zhu	zhu	PROPN
ajst-13993	210	15	,	,	PUNCT
ajst-13993	210	16	bleu	bleu	NOUN
ajst-13993	210	17	:	:	PUNCT
ajst-13993	210	18	a	a	DET
ajst-13993	210	19	method	method	NOUN
ajst-13993	210	20	for	for	ADP
ajst-13993	210	21	automatic	automatic	ADJ
ajst-13993	210	22	evaluation	evaluation	NOUN
ajst-13993	210	23	of	of	ADP
ajst-13993	210	24	machine	machine	NOUN
ajst-13993	210	25	translation	translation	NOUN
ajst-13993	210	26	,	,	PUNCT
ajst-13993	210	27	in	in	ADP
ajst-13993	210	28	proceedings	proceeding	NOUN
ajst-13993	210	29	of	of	ADP
ajst-13993	210	30	the	the	DET
ajst-13993	210	31	40th	40th	ADJ
ajst-13993	210	32	annual	annual	ADJ
ajst-13993	210	33	meeting	meeting	NOUN
ajst-13993	210	34	on	on	ADP
ajst-13993	210	35	association	association	NOUN
ajst-13993	210	36	for	for	ADP
ajst-13993	210	37	computational	computational	ADJ
ajst-13993	210	38	linguistics	linguistic	NOUN
ajst-13993	210	39	,	,	PUNCT
ajst-13993	210	40	association	association	NOUN
ajst-13993	210	41	for	for	ADP
ajst-13993	210	42	computational	computational	ADJ
ajst-13993	210	43	linguistics	linguistic	NOUN
ajst-13993	210	44	,	,	PUNCT
ajst-13993	210	45	2002	2002	NUM
ajst-13993	210	46	,	,	PUNCT
ajst-13993	210	47	pp	pp	ADV
ajst-13993	210	48	.	.	PUNCT
ajst-13993	211	1	311–318.k	311–318.k	NUM
ajst-13993	211	2	.	.	PUNCT
ajst-13993	212	1	[	[	X
ajst-13993	212	2	10	10	NUM
ajst-13993	212	3	]	]	SYM
ajst-13993	212	4	m.hammad	m.hammad	NOUN
ajst-13993	212	5	,	,	PUNCT
ajst-13993	212	6	a.abuljadayel	a.abuljadayel	NOUN
ajst-13993	212	7	,	,	PUNCT
ajst-13993	212	8	andm.khalaf	andm.khalaf	NUM
ajst-13993	212	9	,	,	PUNCT
ajst-13993	212	10	summarizing	summarize	VERB
ajst-13993	212	11	services	service	NOUN
ajst-13993	212	12	of	of	ADP
ajst-13993	212	13	java	java	PROPN
ajst-13993	212	14	packages	package	NOUN
ajst-13993	212	15	,	,	PUNCT
ajst-13993	212	16	lecture	lecture	NOUN
ajst-13993	212	17	notes	note	NOUN
ajst-13993	212	18	on	on	ADP
ajst-13993	212	19	software	software	NOUN
ajst-13993	212	20	engineering	engineering	NOUN
ajst-13993	212	21	,	,	PUNCT
ajst-13993	212	22	4	4	NUM
ajst-13993	212	23	(	(	PUNCT
ajst-13993	212	24	2016	2016	NUM
ajst-13993	212	25	)	)	PUNCT
ajst-13993	212	26	,	,	PUNCT
ajst-13993	212	27	pp	pp	ADP
ajst-13993	212	28	.	.	PUNCT
ajst-13993	213	1	129–132	129–132	NUM
ajst-13993	213	2	.	.	PUNCT
ajst-13993	214	1	[	[	X
ajst-13993	214	2	11	11	NUM
ajst-13993	214	3	]	]	X
ajst-13993	214	4	l.	l.	PROPN
ajst-13993	214	5	moreno	moreno	PROPN
ajst-13993	214	6	,	,	PUNCT
ajst-13993	214	7	j.	j.	PROPN
ajst-13993	214	8	aponte	aponte	PROPN
ajst-13993	214	9	,	,	PUNCT
ajst-13993	214	10	g.	g.	PROPN
ajst-13993	214	11	sridhara	sridhara	PROPN
ajst-13993	214	12	,	,	PUNCT
ajst-13993	214	13	a.	a.	PROPN
ajst-13993	214	14	marcus	marcus	PROPN
ajst-13993	214	15	,	,	PUNCT
ajst-13993	214	16	l.	l.	PROPN
ajst-13993	214	17	pollock	pollock	PROPN
ajst-13993	214	18	,	,	PUNCT
ajst-13993	214	19	and	and	CCONJ
ajst-13993	214	20	k.	k.	PROPN
ajst-13993	214	21	vijayshanker	vijayshanker	PROPN
ajst-13993	214	22	,	,	PUNCT
ajst-13993	214	23	automatic	automatic	ADJ
ajst-13993	214	24	generation	generation	NOUN
ajst-13993	214	25	of	of	ADP
ajst-13993	214	26	natural	natural	ADJ
ajst-13993	214	27	language	language	NOUN
ajst-13993	214	28	summaries	summary	NOUN
ajst-13993	214	29	for	for	ADP
ajst-13993	214	30	java	java	PROPN
ajst-13993	214	31	classes	class	NOUN
ajst-13993	214	32	,	,	PUNCT
ajst-13993	214	33	in	in	ADP
ajst-13993	214	34	2013	2013	NUM
ajst-13993	214	35	21st	21st	ADJ
ajst-13993	214	36	international	international	ADJ
ajst-13993	214	37	conference	conference	NOUN
ajst-13993	214	38	on	on	ADP
ajst-13993	214	39	program	program	NOUN
ajst-13993	214	40	comprehension	comprehension	NOUN
ajst-13993	214	41	(	(	PUNCT
ajst-13993	214	42	icpc	icpc	PROPN
ajst-13993	214	43	)	)	PUNCT
ajst-13993	214	44	,	,	PUNCT
ajst-13993	214	45	ieee	ieee	NOUN
ajst-13993	214	46	,	,	PUNCT
ajst-13993	214	47	2013	2013	NUM
ajst-13993	214	48	,	,	PUNCT
ajst-13993	214	49	pp	pp	ADJ
ajst-13993	214	50	.	.	PUNCT
ajst-13993	215	1	23–32	23–32	X
ajst-13993	215	2	.	.	PUNCT
ajst-13993	216	1	[	[	X
ajst-13993	216	2	12	12	NUM
ajst-13993	216	3	]	]	X
ajst-13993	216	4	r.	r.	PROPN
ajst-13993	216	5	collobert	collobert	PROPN
ajst-13993	216	6	and	and	CCONJ
ajst-13993	216	7	j.	j.	PROPN
ajst-13993	216	8	weston	weston	PROPN
ajst-13993	216	9	,	,	PUNCT
ajst-13993	216	10	a	a	DET
ajst-13993	216	11	unified	unified	ADJ
ajst-13993	216	12	architecture	architecture	NOUN
ajst-13993	216	13	for	for	ADP
ajst-13993	216	14	natural	natural	ADJ
ajst-13993	216	15	language	language	NOUN
ajst-13993	216	16	processing	processing	NOUN
ajst-13993	216	17	:	:	PUNCT
ajst-13993	216	18	deep	deep	ADJ
ajst-13993	216	19	neural	neural	ADJ
ajst-13993	216	20	networks	network	NOUN
ajst-13993	216	21	with	with	ADP
ajst-13993	216	22	multitask	multitask	ADJ
ajst-13993	216	23	learning	learning	NOUN
ajst-13993	216	24	,	,	PUNCT
ajst-13993	216	25	in	in	ADP
ajst-13993	216	26	machine	machine	NOUN
ajst-13993	216	27	learning	learning	NOUN
ajst-13993	216	28	,	,	PUNCT
ajst-13993	216	29	proceedings	proceeding	NOUN
ajst-13993	216	30	of	of	ADP
ajst-13993	216	31	the	the	DET
ajst-13993	216	32	twenty	twenty	NUM
ajst-13993	216	33	-	-	PUNCT
ajst-13993	216	34	fifth	fifth	ADJ
ajst-13993	216	35	international	international	ADJ
ajst-13993	216	36	conference	conference	NOUN
ajst-13993	216	37	(	(	PUNCT
ajst-13993	216	38	icml	icml	VERB
ajst-13993	216	39	200	200	NUM
ajst-13993	216	40	[	[	SYM
ajst-13993	216	41	13	13	NUM
ajst-13993	216	42	]	]	X
ajst-13993	216	43	n.	n.	NOUN
ajst-13993	216	44	kalchbrenner	kalchbrenner	PROPN
ajst-13993	216	45	,	,	PUNCT
ajst-13993	216	46	l.	l.	PROPN
ajst-13993	216	47	espeholt	espeholt	PROPN
ajst-13993	216	48	,	,	PUNCT
ajst-13993	216	49	k.	k.	PROPN
ajst-13993	216	50	simonyan	simonyan	PROPN
ajst-13993	216	51	,	,	PUNCT
ajst-13993	216	52	a.	a.	PROPN
ajst-13993	216	53	v.	v.	PROPN
ajst-13993	216	54	d.	d.	PROPN
ajst-13993	216	55	oord	oord	PROPN
ajst-13993	216	56	,	,	PUNCT
ajst-13993	216	57	and	and	CCONJ
ajst-13993	216	58	k.	k.	PROPN
ajst-13993	216	59	kavukcuoglu	kavukcuoglu	PROPN
ajst-13993	216	60	,	,	PUNCT
ajst-13993	216	61	neural	neural	ADJ
ajst-13993	216	62	machine	machine	NOUN
ajst-13993	216	63	translation	translation	NOUN
ajst-13993	216	64	in	in	ADP
ajst-13993	216	65	linear	linear	PROPN
ajst-13993	216	66	time	time	NOUN
ajst-13993	216	67	,	,	PUNCT
ajst-13993	216	68	arxiv	arxiv	PROPN
ajst-13993	216	69	preprint	preprint	NOUN
ajst-13993	216	70	arxiv:1610.10099	arxiv:1610.10099	NUM
ajst-13993	216	71	,	,	PUNCT
ajst-13993	216	72	(	(	PUNCT
ajst-13993	216	73	2016).8	2016).8	NOUN
ajst-13993	216	74	)	)	PUNCT
ajst-13993	216	75	,	,	PUNCT
ajst-13993	216	76	2008	2008	NUM
ajst-13993	216	77	,	,	PUNCT
ajst-13993	216	78	pp	pp	ADJ
ajst-13993	216	79	.	.	PUNCT
ajst-13993	217	1	160–167	160–167	NUM
ajst-13993	217	2	.	.	PUNCT
ajst-13993	218	1	[	[	X
ajst-13993	218	2	14	14	NUM
ajst-13993	218	3	]	]	X
ajst-13993	218	4	n.	n.	PROPN
ajst-13993	218	5	j.	j.	PROPN
ajst-13993	218	6	abid	abid	PROPN
ajst-13993	218	7	,	,	PUNCT
ajst-13993	218	8	n.	n.	NOUN
ajst-13993	218	9	dragan	dragan	PROPN
ajst-13993	218	10	,	,	PUNCT
ajst-13993	218	11	m.	m.	NOUN
ajst-13993	218	12	l.	l.	PROPN
ajst-13993	218	13	collard	collard	PROPN
ajst-13993	218	14	,	,	PUNCT
ajst-13993	218	15	and	and	CCONJ
ajst-13993	218	16	j.	j.	PROPN
ajst-13993	218	17	i.	i.	PROPN
ajst-13993	218	18	maletic	maletic	PROPN
ajst-13993	218	19	,	,	PUNCT
ajst-13993	218	20	using	use	VERB
ajst-13993	218	21	stereotypes	stereotype	NOUN
ajst-13993	218	22	in	in	ADP
ajst-13993	218	23	the	the	DET
ajst-13993	218	24	automatic	automatic	ADJ
ajst-13993	218	25	generation	generation	NOUN
ajst-13993	218	26	of	of	ADP
ajst-13993	218	27	natural	natural	ADJ
ajst-13993	218	28	language	language	NOUN
ajst-13993	218	29	summaries	summary	NOUN
ajst-13993	218	30	for	for	ADP
ajst-13993	218	31	c++	c++	NOUN
ajst-13993	218	32	methods	method	NOUN
ajst-13993	218	33	,	,	PUNCT
ajst-13993	218	34	in	in	ADP
ajst-13993	218	35	2015	2015	NUM
ajst-13993	218	36	ieee	ieee	NOUN
ajst-13993	218	37	international	international	ADJ
ajst-13993	218	38	conference	conference	NOUN
ajst-13993	218	39	on	on	ADP
ajst-13993	218	40	software	software	NOUN
ajst-13993	218	41	maintenance	maintenance	NOUN
ajst-13993	218	42	and	and	CCONJ
ajst-13993	218	43	evolution	evolution	NOUN
ajst-13993	218	44	(	(	PUNCT
ajst-13993	218	45	icsme	icsme	NOUN
ajst-13993	218	46	)	)	PUNCT
ajst-13993	218	47	,	,	PUNCT
ajst-13993	218	48	ieee	ieee	NOUN
ajst-13993	218	49	,	,	PUNCT
ajst-13993	218	50	2015	2015	NUM
ajst-13993	218	51	,	,	PUNCT
ajst-13993	218	52	pp	pp	ADJ
ajst-13993	218	53	.	.	PUNCT
ajst-13993	219	1	561–565	561–565	NUM
ajst-13993	219	2	.	.	PUNCT
ajst-13993	220	1	[	[	X
ajst-13993	220	2	15	15	NUM
ajst-13993	220	3	]	]	PUNCT
ajst-13993	220	4	x.	x.	NOUN
ajst-13993	220	5	hu	hu	PROPN
ajst-13993	220	6	,	,	PUNCT
ajst-13993	220	7	g.	g.	PROPN
ajst-13993	220	8	li	li	PROPN
ajst-13993	220	9	,	,	PUNCT
ajst-13993	220	10	x.	x.	PROPN
ajst-13993	220	11	xia	xia	PROPN
ajst-13993	220	12	,	,	PUNCT
ajst-13993	220	13	d.	d.	PROPN
ajst-13993	220	14	lo	lo	PROPN
ajst-13993	220	15	,	,	PUNCT
ajst-13993	220	16	and	and	CCONJ
ajst-13993	220	17	z.	z.	PROPN
ajst-13993	220	18	jin	jin	PROPN
ajst-13993	220	19	.	.	PUNCT
ajst-13993	221	1	deep	deep	ADJ
ajst-13993	221	2	code	code	NOUN
ajst-13993	221	3	comment	comment	NOUN
ajst-13993	221	4	generation	generation	NOUN
ajst-13993	221	5	.	.	PUNCT
ajst-13993	222	1	in	in	ADP
ajst-13993	222	2	proceedings	proceeding	NOUN
ajst-13993	222	3	of	of	ADP
ajst-13993	222	4	the	the	DET
ajst-13993	222	5	26th	26th	ADJ
ajst-13993	222	6	international	international	ADJ
ajst-13993	222	7	conference	conference	NOUN
ajst-13993	222	8	on	on	ADP
ajst-13993	222	9	program	program	NOUN
ajst-13993	222	10	comprehension	comprehension	NOUN
ajst-13993	222	11	,	,	PUNCT
ajst-13993	222	12	pages	page	NOUN
ajst-13993	222	13	200–210	200–210	NUM
ajst-13993	222	14	.	.	PUNCT
ajst-13993	222	15	acm	acm	PROPN
ajst-13993	222	16	,	,	PUNCT
ajst-13993	222	17	2018	2018	NUM
ajst-13993	222	18	.	.	PUNCT
ajst-13993	223	1	[	[	X
ajst-13993	223	2	16	16	NUM
ajst-13993	223	3	]	]	PUNCT
ajst-13993	223	4	github	github	NOUN
ajst-13993	223	5	,	,	PUNCT
ajst-13993	223	6	https://github.com/.	https://github.com/.	PROPN
ajst-13993	223	7	[	[	X
ajst-13993	223	8	17	17	NUM
ajst-13993	223	9	]	]	PUNCT
ajst-13993	223	10	k.	k.	PROPN
ajst-13993	223	11	papineni	papineni	PROPN
ajst-13993	223	12	,	,	PUNCT
ajst-13993	223	13	s.	s.	PROPN
ajst-13993	223	14	roukos	roukos	PROPN
ajst-13993	223	15	,	,	PUNCT
ajst-13993	223	16	t.	t.	NOUN
ajst-13993	223	17	ward	ward	NOUN
ajst-13993	223	18	,	,	PUNCT
ajst-13993	223	19	and	and	CCONJ
ajst-13993	223	20	w.	w.	PROPN
ajst-13993	223	21	j.	j.	PROPN
ajst-13993	223	22	zhu	zhu	PROPN
ajst-13993	223	23	,	,	PUNCT
ajst-13993	223	24	bleu	bleu	NOUN
ajst-13993	223	25	:	:	PUNCT
ajst-13993	223	26	a	a	DET
ajst-13993	223	27	method	method	NOUN
ajst-13993	223	28	for	for	ADP
ajst-13993	223	29	automatic	automatic	ADJ
ajst-13993	223	30	evaluation	evaluation	NOUN
ajst-13993	223	31	of	of	ADP
ajst-13993	223	32	machine	machine	NOUN
ajst-13993	223	33	translation	translation	NOUN
ajst-13993	223	34	,	,	PUNCT
ajst-13993	223	35	in	in	ADP
ajst-13993	223	36	proceedings	proceeding	NOUN
ajst-13993	223	37	of	of	ADP
ajst-13993	223	38	the	the	DET
ajst-13993	223	39	40th	40th	ADJ
ajst-13993	223	40	annual	annual	ADJ
ajst-13993	223	41	meeting	meeting	NOUN
ajst-13993	223	42	on	on	ADP
ajst-13993	223	43	association	association	NOUN
ajst-13993	223	44	for	for	ADP
ajst-13993	223	45	computational	computational	ADJ
ajst-13993	223	46	linguistics	linguistic	NOUN
ajst-13993	223	47	,	,	PUNCT
ajst-13993	223	48	association	association	NOUN
ajst-13993	223	49	for	for	ADP
ajst-13993	223	50	computational	computational	ADJ
ajst-13993	223	51	linguistics	linguistic	NOUN
ajst-13993	223	52	,	,	PUNCT
ajst-13993	223	53	2002	2002	NUM
ajst-13993	223	54	,	,	PUNCT
ajst-13993	223	55	pp	pp	ADV
ajst-13993	223	56	.	.	PUNCT
ajst-13993	224	1	311–318	311–318	NUM
ajst-13993	224	2	.	.	PUNCT
ajst-13993	225	1	[	[	X
ajst-13993	225	2	18	18	NUM
ajst-13993	225	3	]	]	X
ajst-13993	225	4	v.	v.	PROPN
ajst-13993	225	5	r.	r.	PROPN
ajst-13993	225	6	konda	konda	PROPN
ajst-13993	225	7	and	and	CCONJ
ajst-13993	225	8	j.	j.	PROPN
ajst-13993	225	9	n.	n.	PROPN
ajst-13993	225	10	tsitsiklis	tsitsiklis	PROPN
ajst-13993	225	11	,	,	PUNCT
ajst-13993	225	12	actor	actor	NOUN
ajst-13993	225	13	-	-	PUNCT
ajst-13993	225	14	critic	critic	NOUN
ajst-13993	225	15	algorithms	algorithm	NOUN
ajst-13993	225	16	,	,	PUNCT
ajst-13993	225	17	in	in	ADP
ajst-13993	225	18	advances	advance	NOUN
ajst-13993	225	19	in	in	ADP
ajst-13993	225	20	neural	neural	ADJ
ajst-13993	225	21	information	information	NOUN
ajst-13993	225	22	processing	processing	NOUN
ajst-13993	225	23	systems	system	NOUN
ajst-13993	225	24	,	,	PUNCT
ajst-13993	225	25	2000	2000	NUM
ajst-13993	225	26	,	,	PUNCT
ajst-13993	225	27	pp	pp	PROPN
ajst-13993	225	28	.	.	PUNCT
ajst-13993	226	1	1008–1014	1008–1014	NUM
ajst-13993	226	2	.	.	PUNCT
ajst-13993	227	1	[	[	X
ajst-13993	227	2	19	19	NUM
ajst-13993	227	3	]	]	X
ajst-13993	227	4	e.	e.	PROPN
ajst-13993	227	5	m.	m.	PROPN
ajst-13993	227	6	ponti	ponti	PROPN
ajst-13993	227	7	,	,	PUNCT
ajst-13993	227	8	r.	r.	PROPN
ajst-13993	227	9	reichart	reichart	PROPN
ajst-13993	227	10	,	,	PUNCT
ajst-13993	227	11	a.	a.	PROPN
ajst-13993	227	12	korhonen	korhonen	PROPN
ajst-13993	227	13	,	,	PUNCT
ajst-13993	227	14	and	and	CCONJ
ajst-13993	227	15	i.	i.	PROPN
ajst-13993	227	16	vulic	vulic	PROPN
ajst-13993	227	17	,	,	PUNCT
ajst-13993	227	18	isomorphic	isomorphic	ADJ
ajst-13993	227	19	transfer	transfer	NOUN
ajst-13993	227	20	of	of	ADP
ajst-13993	227	21	syntactic	syntactic	ADJ
ajst-13993	227	22	structures	structure	NOUN
ajst-13993	227	23	in	in	ADP
ajst-13993	227	24	crosslingual	crosslingual	ADJ
ajst-13993	227	25	nlp	nlp	NOUN
ajst-13993	227	26	,	,	PUNCT
ajst-13993	227	27	in	in	ADP
ajst-13993	227	28	the	the	DET
ajst-13993	227	29	56th	56th	ADJ
ajst-13993	227	30	annual	annual	ADJ
ajst-13993	227	31	meeting	meeting	NOUN
ajst-13993	227	32	of	of	ADP
ajst-13993	227	33	the	the	DET
ajst-13993	227	34	association	association	NOUN
ajst-13993	227	35	for	for	ADP
ajst-13993	227	36	computational	computational	ADJ
ajst-13993	227	37	linguistics	linguistic	NOUN
ajst-13993	227	38	,	,	PUNCT
ajst-13993	227	39	2018	2018	NUM
ajst-13993	227	40	,	,	PUNCT
ajst-13993	227	41	pp	pp	ADJ
ajst-13993	227	42	.	.	PUNCT
ajst-13993	228	1	1531–1542	1531–1542	NUM
ajst-13993	228	2	.	.	PUNCT
ajst-13993	229	1	[	[	X
ajst-13993	229	2	20	20	NUM
ajst-13993	229	3	]	]	PUNCT
ajst-13993	229	4	j.	j.	PROPN
ajst-13993	229	5	pouget	pouget	PROPN
ajst-13993	229	6	-	-	PUNCT
ajst-13993	229	7	abadie	abadie	PROPN
ajst-13993	229	8	,	,	PUNCT
ajst-13993	229	9	d.	d.	PROPN
ajst-13993	229	10	bahdanau	bahdanau	PROPN
ajst-13993	229	11	,	,	PUNCT
ajst-13993	229	12	b.	b.	PROPN
ajst-13993	230	1	v.	v.	PROPN
ajst-13993	230	2	merrienboer	merrienboer	PROPN
ajst-13993	230	3	,	,	PUNCT
ajst-13993	230	4	k.	k.	PROPN
ajst-13993	230	5	cho	cho	PROPN
ajst-13993	230	6	,	,	PUNCT
ajst-13993	230	7	and	and	CCONJ
ajst-13993	230	8	y.	y.	PROPN
ajst-13993	230	9	bengio	bengio	PROPN
ajst-13993	230	10	,	,	PUNCT
ajst-13993	230	11	overcoming	overcome	VERB
ajst-13993	230	12	the	the	DET
ajst-13993	230	13	curse	curse	NOUN
ajst-13993	230	14	of	of	ADP
ajst-13993	230	15	sentence	sentence	NOUN
ajst-13993	230	16	length	length	NOUN
ajst-13993	230	17	for	for	ADP
ajst-13993	230	18	neural	neural	ADJ
ajst-13993	230	19	machine	machine	NOUN
ajst-13993	230	20	translation	translation	NOUN
ajst-13993	230	21	using	use	VERB
ajst-13993	230	22	automatic	automatic	ADJ
ajst-13993	230	23	segmentation	segmentation	NOUN
ajst-13993	230	24	,	,	PUNCT
ajst-13993	230	25	arxiv	arxiv	PROPN
ajst-13993	230	26	preprint	preprint	NOUN
ajst-13993	230	27	arxiv:1409.1257	arxiv:1409.1257	NOUN
ajst-13993	230	28	,	,	PUNCT
ajst-13993	230	29	(	(	PUNCT
ajst-13993	230	30	2014	2014	NUM
ajst-13993	230	31	)	)	PUNCT
ajst-13993	230	32	.	.	PUNCT
ajst-13993	231	1	[	[	X
ajst-13993	231	2	21	21	NUM
ajst-13993	231	3	]	]	X
ajst-13993	231	4	h.	h.	PROPN
ajst-13993	231	5	shi	shi	PROPN
ajst-13993	231	6	,	,	PUNCT
ajst-13993	231	7	h.	h.	PROPN
ajst-13993	231	8	zhou	zhou	PROPN
ajst-13993	231	9	,	,	PUNCT
ajst-13993	231	10	j.	j.	PROPN
ajst-13993	231	11	chen	chen	PROPN
ajst-13993	231	12	,	,	PUNCT
ajst-13993	231	13	and	and	CCONJ
ajst-13993	231	14	l.	l.	PROPN
ajst-13993	231	15	li	li	PROPN
ajst-13993	231	16	,	,	PUNCT
ajst-13993	231	17	on	on	ADP
ajst-13993	231	18	tree	tree	NOUN
ajst-13993	231	19	-	-	PUNCT
ajst-13993	231	20	based	base	VERB
ajst-13993	231	21	neural	neural	ADJ
ajst-13993	231	22	sentence	sentence	NOUN
ajst-13993	231	23	modeling	modeling	NOUN
ajst-13993	231	24	,	,	PUNCT
ajst-13993	231	25	arxiv	arxiv	PROPN
ajst-13993	231	26	preprint	preprint	NOUN
ajst-13993	231	27	arxiv:1808.09644	arxiv:1808.09644	NUM
ajst-13993	231	28	,	,	PUNCT
ajst-13993	231	29	(	(	PUNCT
ajst-13993	231	30	2018	2018	NUM
ajst-13993	231	31	)	)	PUNCT
ajst-13993	231	32	.	.	PUNCT
ajst-13993	232	1	[	[	X
ajst-13993	232	2	22	22	NUM
ajst-13993	232	3	]	]	X
ajst-13993	232	4	s.	s.	PROPN
ajst-13993	232	5	banerjee	banerjee	PROPN
ajst-13993	232	6	and	and	CCONJ
ajst-13993	232	7	a.	a.	PROPN
ajst-13993	232	8	lavie	lavie	PROPN
ajst-13993	232	9	,	,	PUNCT
ajst-13993	232	10	meteor	meteor	NOUN
ajst-13993	232	11	:	:	PUNCT
ajst-13993	232	12	an	an	DET
ajst-13993	232	13	automatic	automatic	ADJ
ajst-13993	232	14	metric	metric	NOUN
ajst-13993	232	15	for	for	ADP
ajst-13993	232	16	mt	mt	PROPN
ajst-13993	232	17	evaluation	evaluation	NOUN
ajst-13993	232	18	with	with	ADP
ajst-13993	232	19	improved	improved	ADJ
ajst-13993	232	20	correlation	correlation	NOUN
ajst-13993	232	21	with	with	ADP
ajst-13993	232	22	human	human	ADJ
ajst-13993	232	23	judgments	judgment	NOUN
ajst-13993	232	24	,	,	PUNCT
ajst-13993	232	25	in	in	ADP
ajst-13993	232	26	proceedings	proceeding	NOUN
ajst-13993	232	27	of	of	ADP
ajst-13993	232	28	the	the	DET
ajst-13993	232	29	acl	acl	PROPN
ajst-13993	232	30	workshop	workshop	NOUN
ajst-13993	232	31	on	on	ADP
ajst-13993	232	32	intrinsic	intrinsic	ADJ
ajst-13993	232	33	and	and	CCONJ
ajst-13993	232	34	extrinsic	extrinsic	ADJ
ajst-13993	232	35	evaluation	evaluation	NOUN
ajst-13993	232	36	measures	measure	NOUN
ajst-13993	232	37	for	for	ADP
ajst-13993	232	38	machine	machine	NOUN
ajst-13993	232	39	translation	translation	NOUN
ajst-13993	232	40	and/or	and/or	CCONJ
ajst-13993	232	41	summarization	summarization	NOUN
ajst-13993	232	42	,	,	PUNCT
ajst-13993	232	43	vol	vol	NOUN
ajst-13993	232	44	.	.	PROPN
ajst-13993	232	45	29	29	NUM
ajst-13993	232	46	,	,	PUNCT
ajst-13993	232	47	2005	2005	NUM
ajst-13993	232	48	,	,	PUNCT
ajst-13993	232	49	pp	pp	ADJ
ajst-13993	232	50	.	.	PUNCT
ajst-13993	233	1	65–72	65–72	NUM
ajst-13993	233	2	.	.	PUNCT
ajst-13993	234	1	[	[	X
ajst-13993	234	2	23	23	NUM
ajst-13993	234	3	]	]	PUNCT
ajst-13993	234	4	z.	z.	PROPN
ajst-13993	234	5	yang	yang	PROPN
ajst-13993	234	6	,	,	PUNCT
ajst-13993	234	7	d.	d.	PROPN
ajst-13993	234	8	yang	yang	PROPN
ajst-13993	234	9	,	,	PUNCT
ajst-13993	234	10	c.	c.	PROPN
ajst-13993	234	11	dyer	dyer	PROPN
ajst-13993	234	12	,	,	PUNCT
ajst-13993	234	13	x.	x.	NOUN
ajst-13993	235	1	he	he	PRON
ajst-13993	235	2	,	,	PUNCT
ajst-13993	235	3	a.	a.	PROPN
ajst-13993	235	4	smola	smola	PROPN
ajst-13993	235	5	,	,	PUNCT
ajst-13993	235	6	and	and	CCONJ
ajst-13993	235	7	e.	e.	PROPN
ajst-13993	235	8	hovy	hovy	PROPN
ajst-13993	235	9	,	,	PUNCT
ajst-13993	235	10	hierarchical	hierarchical	ADJ
ajst-13993	235	11	attention	attention	NOUN
ajst-13993	235	12	networks	network	NOUN
ajst-13993	235	13	for	for	ADP
ajst-13993	235	14	document	document	NOUN
ajst-13993	235	15	classification	classification	NOUN
ajst-13993	235	16	,	,	PUNCT
ajst-13993	235	17	in	in	ADP
ajst-13993	235	18	proceedings	proceeding	NOUN
ajst-13993	235	19	of	of	ADP
ajst-13993	235	20	the	the	DET
ajst-13993	235	21	2016	2016	NUM
ajst-13993	235	22	conference	conference	NOUN
ajst-13993	235	23	of	of	ADP
ajst-13993	235	24	the	the	DET
ajst-13993	235	25	north	north	ADJ
ajst-13993	235	26	american	american	ADJ
ajst-13993	235	27	chapter	chapter	NOUN
ajst-13993	235	28	of	of	ADP
ajst-13993	235	29	the	the	DET
ajst-13993	235	30	association	association	NOUN
ajst-13993	235	31	for	for	ADP
ajst-13993	235	32	computational	computational	ADJ
ajst-13993	235	33	linguistics	linguistic	NOUN
ajst-13993	235	34	:	:	PUNCT
ajst-13993	235	35	human	human	ADJ
ajst-13993	235	36	language	language	NOUN
ajst-13993	235	37	technologies	technology	NOUN
ajst-13993	235	38	,	,	PUNCT
ajst-13993	235	39	naacl	naacl	NOUN
ajst-13993	235	40	,	,	PUNCT
ajst-13993	235	41	2016	2016	NUM
ajst-13993	235	42	,	,	PUNCT
ajst-13993	235	43	pp	pp	ADJ
ajst-13993	235	44	.	.	PUNCT
ajst-13993	235	45	1480–1489	1480–1489	NUM
ajst-13993	235	46	.	.	PUNCT
ajst-13993	236	1	[	[	X
ajst-13993	236	2	24	24	NUM
ajst-13993	236	3	]	]	PUNCT
ajst-13993	236	4	m.	m.	NOUN
ajst-13993	236	5	allamanis	allamanis	PROPN
ajst-13993	236	6	,	,	PUNCT
ajst-13993	236	7	h.	h.	PROPN
ajst-13993	236	8	peng	peng	PROPN
ajst-13993	236	9	,	,	PUNCT
ajst-13993	236	10	and	and	CCONJ
ajst-13993	236	11	c.	c.	PROPN
ajst-13993	236	12	sutton	sutton	PROPN
ajst-13993	236	13	,	,	PUNCT
ajst-13993	236	14	a	a	DET
ajst-13993	236	15	convolutional	convolutional	ADJ
ajst-13993	236	16	attention	attention	NOUN
ajst-13993	236	17	network	network	NOUN
ajst-13993	236	18	for	for	ADP
ajst-13993	236	19	extreme	extreme	ADJ
ajst-13993	236	20	summarization	summarization	NOUN
ajst-13993	236	21	of	of	ADP
ajst-13993	236	22	source	source	NOUN
ajst-13993	236	23	code	code	NOUN
ajst-13993	236	24	,	,	PUNCT
ajst-13993	236	25	in	in	ADP
ajst-13993	236	26	international	international	ADJ
ajst-13993	236	27	conference	conference	NOUN
ajst-13993	236	28	on	on	ADP
ajst-13993	236	29	machine	machine	NOUN
ajst-13993	236	30	learning	learning	NOUN
ajst-13993	236	31	,	,	PUNCT
ajst-13993	236	32	2016	2016	NUM
ajst-13993	236	33	,	,	PUNCT
ajst-13993	236	34	pp	pp	ADJ
ajst-13993	236	35	.	.	PUNCT
ajst-13993	236	36	2091–2100	2091–2100	NUM
ajst-13993	236	37	.	.	PUNCT
ajst-13993	237	1	[	[	X
ajst-13993	237	2	25	25	NUM
ajst-13993	237	3	]	]	X
ajst-13993	237	4	y.	y.	PROPN
ajst-13993	237	5	wan	wan	PROPN
ajst-13993	237	6	,	,	PUNCT
ajst-13993	237	7	z.	z.	PROPN
ajst-13993	237	8	zhao	zhao	PROPN
ajst-13993	237	9	,	,	PUNCT
ajst-13993	237	10	m.	m.	PROPN
ajst-13993	237	11	yang	yang	PROPN
ajst-13993	237	12	,	,	PUNCT
ajst-13993	237	13	g.	g.	PROPN
ajst-13993	237	14	xu	xu	PROPN
ajst-13993	237	15	,	,	PUNCT
ajst-13993	237	16	h.	h.	PROPN
ajst-13993	237	17	ying	ying	PROPN
ajst-13993	237	18	,	,	PUNCT
ajst-13993	237	19	j.	j.	PROPN
ajst-13993	237	20	wu	wu	PROPN
ajst-13993	237	21	,	,	PUNCT
ajst-13993	237	22	and	and	CCONJ
ajst-13993	237	23	p.	p.	PROPN
ajst-13993	237	24	s.	s.	PROPN
ajst-13993	237	25	yu	yu	PROPN
ajst-13993	237	26	,	,	PUNCT
ajst-13993	237	27	improving	improve	VERB
ajst-13993	237	28	automatic	automatic	ADJ
ajst-13993	237	29	source	source	NOUN
ajst-13993	237	30	code	code	NOUN
ajst-13993	237	31	summarization	summarization	NOUN
ajst-13993	237	32	via	via	ADP
ajst-13993	237	33	deep	deep	ADJ
ajst-13993	237	34	reinforcement	reinforcement	NOUN
ajst-13993	237	35	learning	learning	NOUN
ajst-13993	237	36	,	,	PUNCT
ajst-13993	237	37	in	in	ADP
ajst-13993	237	38	proceedings	proceeding	NOUN
ajst-13993	237	39	of	of	ADP
ajst-13993	237	40	the	the	DET
ajst-13993	237	41	33rd	33rd	PROPN
ajst-13993	237	42	acm	acm	PROPN
ajst-13993	237	43	/	/	SYM
ajst-13993	237	44	ieee	ieee	PROPN
ajst-13993	237	45	international	international	ADJ
ajst-13993	237	46	conference	conference	NOUN
ajst-13993	237	47	on	on	ADP
ajst-13993	237	48	automated	automate	VERB
ajst-13993	237	49	software	software	NOUN
ajst-13993	237	50	engineering	engineering	NOUN
ajst-13993	237	51	,	,	PUNCT
ajst-13993	237	52	acm	acm	PROPN
ajst-13993	237	53	,	,	PUNCT
ajst-13993	237	54	2018	2018	NUM
ajst-13993	237	55	,	,	PUNCT
ajst-13993	237	56	pp	pp	ADJ
ajst-13993	237	57	.	.	PUNCT
ajst-13993	238	1	397–407	397–407	NUM
ajst-13993	238	2	.	.	PUNCT
ajst-13993	239	1	[	[	X
ajst-13993	239	2	26	26	NUM
ajst-13993	239	3	]	]	X
ajst-13993	239	4	paul	paul	PROPN
ajst-13993	239	5	w.	w.	PROPN
ajst-13993	239	6	mcburney	mcburney	PROPN
ajst-13993	239	7	and	and	CCONJ
ajst-13993	239	8	collin	collin	PROPN
ajst-13993	239	9	mcmillan	mcmillan	PROPN
ajst-13993	239	10	.	.	PUNCT
ajst-13993	240	1	2016	2016	NUM
ajst-13993	240	2	.	.	PUNCT
ajst-13993	241	1	automatic	automatic	ADJ
ajst-13993	241	2	source	source	NOUN
ajst-13993	241	3	code	code	NOUN
ajst-13993	241	4	summarization	summarization	NOUN
ajst-13993	241	5	of	of	ADP
ajst-13993	241	6	context	context	NOUN
ajst-13993	241	7	for	for	ADP
ajst-13993	241	8	java	java	PROPN
ajst-13993	241	9	methods	method	NOUN
ajst-13993	241	10	.	.	PUNCT
ajst-13993	242	1	ieee	ieee	PROPN
ajst-13993	242	2	trans	trans	PROPN
ajst-13993	242	3	.	.	PUNCT
ajst-13993	243	1	software	software	PROPN
ajst-13993	243	2	eng	eng	PROPN
ajst-13993	243	3	.	.	PROPN
ajst-13993	244	1	42	42	NUM
ajst-13993	244	2	,	,	PUNCT
ajst-13993	244	3	2	2	NUM
ajst-13993	244	4	(	(	PUNCT
ajst-13993	244	5	2016	2016	NUM
ajst-13993	244	6	)	)	PUNCT
ajst-13993	244	7	,	,	PUNCT
ajst-13993	244	8	103–119	103–119	NUM
ajst-13993	244	9	.	.	PUNCT
ajst-13993	245	1	https://doi.org/10.1109/tse.2015.2465386	https://doi.org/10.1109/tse.2015.2465386	NOUN
ajst-13993	245	2	[	[	X
ajst-13993	245	3	27	27	NUM
ajst-13993	245	4	]	]	X
ajst-13993	245	5	laura	laura	PROPN
ajst-13993	245	6	moreno	moreno	PROPN
ajst-13993	245	7	,	,	PUNCT
ajst-13993	245	8	jairo	jairo	PROPN
ajst-13993	245	9	aponte	aponte	PROPN
ajst-13993	245	10	,	,	PUNCT
ajst-13993	245	11	giriprasad	giriprasad	PROPN
ajst-13993	245	12	sridhara	sridhara	NOUN
ajst-13993	245	13	,	,	PUNCT
ajst-13993	245	14	andrian	andrian	PROPN
ajst-13993	245	15	marcus	marcus	PROPN
ajst-13993	245	16	,	,	PUNCT
ajst-13993	245	17	lori	lori	PROPN
ajst-13993	245	18	l.	l.	PROPN
ajst-13993	245	19	pollock	pollock	PROPN
ajst-13993	245	20	,	,	PUNCT
ajst-13993	245	21	and	and	CCONJ
ajst-13993	245	22	k.	k.	PROPN
ajst-13993	245	23	vijay	vijay	PROPN
ajst-13993	245	24	-	-	PUNCT
ajst-13993	245	25	shanker	shanker	NOUN
ajst-13993	245	26	.	.	PUNCT
ajst-13993	246	1	2013	2013	NUM
ajst-13993	246	2	.	.	PUNCT
ajst-13993	247	1	automatic	automatic	ADJ
ajst-13993	247	2	generation	generation	NOUN
ajst-13993	247	3	of	of	ADP
ajst-13993	247	4	natural	natural	ADJ
ajst-13993	247	5	language	language	NOUN
ajst-13993	247	6	summaries	summary	NOUN
ajst-13993	247	7	for	for	ADP
ajst-13993	247	8	java	java	PROPN
ajst-13993	247	9	classes	class	NOUN
ajst-13993	247	10	.	.	PUNCT
ajst-13993	248	1	in	in	ADP
ajst-13993	248	2	ieee	ieee	PROPN
ajst-13993	248	3	21st	21st	PROPN
ajst-13993	248	4	international	international	ADJ
ajst-13993	248	5	conference	conference	NOUN
ajst-13993	248	6	on	on	ADP
ajst-13993	248	7	program	program	NOUN
ajst-13993	248	8	comprehension	comprehension	NOUN
ajst-13993	248	9	,	,	PUNCT
ajst-13993	248	10	icpc	icpc	PROPN
ajst-13993	248	11	2013	2013	NUM
ajst-13993	248	12	,	,	PUNCT
ajst-13993	248	13	san	san	PROPN
ajst-13993	248	14	francisco	francisco	PROPN
ajst-13993	248	15	,	,	PUNCT
ajst-13993	248	16	ca	ca	PROPN
ajst-13993	248	17	,	,	PUNCT
ajst-13993	248	18	usa	usa	PROPN
ajst-13993	248	19	,	,	PUNCT
ajst-13993	248	20	20	20	NUM
ajst-13993	248	21	-	-	SYM
ajst-13993	248	22	21	21	NUM
ajst-13993	248	23	may	may	NOUN
ajst-13993	248	24	,	,	PUNCT
ajst-13993	248	25	2013	2013	NUM
ajst-13993	248	26	.	.	PUNCT
ajst-13993	249	1	23–32	23–32	NUM
ajst-13993	249	2	.	.	PUNCT
ajst-13993	249	3	https://doi.org/10	https://doi.org/10	PROPN
ajst-13993	249	4	.	.	PUNCT
ajst-13993	250	1	1109	1109	NUM
ajst-13993	250	2	/	/	SYM
ajst-13993	250	3	icpc.2013.6613830	icpc.2013.6613830	PROPN
ajst-13993	250	4	.	.	PUNCT
ajst-13993	251	1	[	[	X
ajst-13993	251	2	28	28	NUM
ajst-13993	251	3	]	]	X
ajst-13993	251	4	giriprasad	giriprasad	PROPN
ajst-13993	251	5	sridhara	sridhara	NOUN
ajst-13993	251	6	,	,	PUNCT
ajst-13993	251	7	emily	emily	PROPN
ajst-13993	251	8	hill	hill	PROPN
ajst-13993	251	9	,	,	PUNCT
ajst-13993	251	10	divya	divya	PROPN
ajst-13993	251	11	muppaneni	muppaneni	PROPN
ajst-13993	251	12	,	,	PUNCT
ajst-13993	251	13	lori	lori	PROPN
ajst-13993	251	14	l.	l.	PROPN
ajst-13993	251	15	pollock	pollock	PROPN
ajst-13993	251	16	,	,	PUNCT
ajst-13993	251	17	and	and	CCONJ
ajst-13993	251	18	k.	k.	PROPN
ajst-13993	251	19	vijayshanker	vijayshanker	PROPN
ajst-13993	251	20	.	.	PUNCT
ajst-13993	252	1	2010	2010	NUM
ajst-13993	252	2	.	.	PUNCT
ajst-13993	253	1	towards	towards	ADP
ajst-13993	253	2	automatically	automatically	ADV
ajst-13993	253	3	generating	generate	VERB
ajst-13993	253	4	summary	summary	NOUN
ajst-13993	253	5	comments	comment	NOUN
ajst-13993	253	6	for	for	ADP
ajst-13993	253	7	java	java	NOUN
ajst-13993	253	8	methods	method	NOUN
ajst-13993	253	9	.	.	PUNCT
ajst-13993	254	1	in	in	ADP
ajst-13993	254	2	ase	ase	PROPN
ajst-13993	254	3	.	.	PROPN
ajst-13993	254	4	acm	acm	PROPN
ajst-13993	254	5	,	,	PUNCT
ajst-13993	254	6	43–52	43–52	NUM
ajst-13993	254	7	.	.	PUNCT
ajst-13993	255	1	[	[	X
ajst-13993	255	2	29	29	NUM
ajst-13993	255	3	]	]	X
ajst-13993	255	4	brian	brian	PROPN
ajst-13993	255	5	p.	p.	PROPN
ajst-13993	255	6	eddy	eddy	PROPN
ajst-13993	255	7	,	,	PUNCT
ajst-13993	255	8	jeffrey	jeffrey	PROPN
ajst-13993	255	9	a.	a.	PROPN
ajst-13993	255	10	robinson	robinson	PROPN
ajst-13993	255	11	,	,	PUNCT
ajst-13993	255	12	nicholas	nicholas	PROPN
ajst-13993	255	13	a.	a.	NOUN
ajst-13993	255	14	kraft	kraft	PROPN
ajst-13993	255	15	,	,	PUNCT
ajst-13993	255	16	and	and	CCONJ
ajst-13993	255	17	jeffrey	jeffrey	PROPN
ajst-13993	255	18	c.	c.	PROPN
ajst-13993	255	19	carver	carver	PROPN
ajst-13993	255	20	.	.	PUNCT
ajst-13993	256	1	2013	2013	NUM
ajst-13993	256	2	.	.	PUNCT
ajst-13993	257	1	evaluating	evaluate	VERB
ajst-13993	257	2	source	source	NOUN
ajst-13993	257	3	code	code	NOUN
ajst-13993	257	4	summarization	summarization	NOUN
ajst-13993	257	5	techniques	technique	NOUN
ajst-13993	257	6	:	:	PUNCT
ajst-13993	257	7	replication	replication	NOUN
ajst-13993	257	8	and	and	CCONJ
ajst-13993	257	9	expansion	expansion	NOUN
ajst-13993	257	10	.	.	PUNCT
ajst-13993	258	1	in	in	ADP
ajst-13993	258	2	ieee	ieee	PROPN
ajst-13993	258	3	21st	21st	PROPN
ajst-13993	258	4	international	international	ADJ
ajst-13993	258	5	conference	conference	NOUN
ajst-13993	258	6	on	on	ADP
ajst-13993	258	7	program	program	NOUN
ajst-13993	258	8	comprehension	comprehension	NOUN
ajst-13993	258	9	,	,	PUNCT
ajst-13993	258	10	icpc	icpc	PROPN
ajst-13993	258	11	2013	2013	NUM
ajst-13993	258	12	,	,	PUNCT
ajst-13993	258	13	san	san	PROPN
ajst-13993	258	14	francisco	francisco	PROPN
ajst-13993	258	15	,	,	PUNCT
ajst-13993	258	16	ca	ca	PROPN
ajst-13993	258	17	,	,	PUNCT
ajst-13993	258	18	usa	usa	PROPN
ajst-13993	258	19	,	,	PUNCT
ajst-13993	258	20	20	20	NUM
ajst-13993	258	21	-	-	SYM
ajst-13993	258	22	21	21	NUM
ajst-13993	258	23	may	may	NOUN
ajst-13993	258	24	,	,	PUNCT
ajst-13993	258	25	2013	2013	NUM
ajst-13993	258	26	.	.	PUNCT
ajst-13993	259	1	13–22	13–22	NUM
ajst-13993	259	2	.	.	PUNCT
ajst-13993	260	1	https://doi.org/10.1109/icpc.2013	https://doi.org/10.1109/icpc.2013	ADJ
ajst-13993	260	2	.	.	PUNCT
ajst-13993	261	1	6613829	6613829	NUM
ajst-13993	261	2	.	.	PUNCT
ajst-13993	262	1	[	[	X
ajst-13993	262	2	30	30	NUM
ajst-13993	262	3	]	]	X
ajst-13993	262	4	sonia	sonia	NOUN
ajst-13993	262	5	haiduc	haiduc	PROPN
ajst-13993	262	6	,	,	PUNCT
ajst-13993	262	7	jairo	jairo	PROPN
ajst-13993	262	8	aponte	aponte	PROPN
ajst-13993	262	9	,	,	PUNCT
ajst-13993	262	10	laura	laura	PROPN
ajst-13993	262	11	moreno	moreno	PROPN
ajst-13993	262	12	,	,	PUNCT
ajst-13993	262	13	and	and	CCONJ
ajst-13993	262	14	andrian	andrian	PROPN
ajst-13993	262	15	marcus	marcus	PROPN
ajst-13993	262	16	.	.	PUNCT
ajst-13993	262	17	2010	2010	NUM
ajst-13993	262	18	.	.	PUNCT
ajst-13993	263	1	on	on	ADP
ajst-13993	263	2	the	the	DET
ajst-13993	263	3	use	use	NOUN
ajst-13993	263	4	of	of	ADP
ajst-13993	263	5	automated	automate	VERB
ajst-13993	263	6	text	text	NOUN
ajst-13993	263	7	summarization	summarization	NOUN
ajst-13993	263	8	techniques	technique	NOUN
ajst-13993	263	9	for	for	ADP
ajst-13993	263	10	summarizing	summarize	VERB
ajst-13993	263	11	source	source	NOUN
ajst-13993	263	12	code	code	NOUN
ajst-13993	263	13	.	.	PUNCT
ajst-13993	264	1	in	in	ADP
ajst-13993	264	2	17th	17th	ADJ
ajst-13993	264	3	working	work	VERB
ajst-13993	264	4	conference	conference	NOUN
ajst-13993	264	5	on	on	ADP
ajst-13993	264	6	reverse	reverse	ADJ
ajst-13993	264	7	engineering	engineering	NOUN
ajst-13993	264	8	,	,	PUNCT
ajst-13993	264	9	wcre	wcre	PROPN
ajst-13993	264	10	2010	2010	NUM
ajst-13993	264	11	,	,	PUNCT
ajst-13993	264	12	13	13	NUM
ajst-13993	264	13	-	-	SYM
ajst-13993	264	14	16	16	NUM
ajst-13993	264	15	october	october	NOUN
ajst-13993	264	16	2010	2010	NUM
ajst-13993	264	17	,	,	PUNCT
ajst-13993	264	18	beverly	beverly	PROPN
ajst-13993	264	19	,	,	PUNCT
ajst-13993	264	20	ma	ma	PROPN
ajst-13993	264	21	,	,	PUNCT
ajst-13993	264	22	usa	usa	PROPN
ajst-13993	264	23	.	.	PROPN
ajst-13993	264	24	35–44	35–44	PROPN
ajst-13993	264	25	.	.	PUNCT
ajst-13993	264	26	https://doi.org/10.1109/wcre.2010.13	https://doi.org/10.1109/wcre.2010.13	PUNCT
ajst-13993	265	1	[	[	X
ajst-13993	265	2	31	31	NUM
ajst-13993	265	3	]	]	X
ajst-13993	265	4	martin	martin	PROPN
ajst-13993	265	5	white	white	PROPN
ajst-13993	265	6	,	,	PUNCT
ajst-13993	265	7	michele	michele	PROPN
ajst-13993	265	8	tufano	tufano	PROPN
ajst-13993	265	9	,	,	PUNCT
ajst-13993	265	10	christopher	christopher	PROPN
ajst-13993	265	11	vendome	vendome	PROPN
ajst-13993	265	12	,	,	PUNCT
ajst-13993	265	13	and	and	CCONJ
ajst-13993	265	14	denys	denys	PROPN
ajst-13993	265	15	poshyvanyk	poshyvanyk	NOUN
ajst-13993	265	16	.	.	PUNCT
ajst-13993	266	1	2016	2016	NUM
ajst-13993	266	2	.	.	PUNCT
ajst-13993	267	1	deep	deep	ADJ
ajst-13993	267	2	learning	learn	VERB
ajst-13993	267	3	code	code	NOUN
ajst-13993	267	4	fragments	fragment	NOUN
ajst-13993	267	5	for	for	ADP
ajst-13993	267	6	code	code	NOUN
ajst-13993	267	7	clone	clone	NOUN
ajst-13993	267	8	detection	detection	NOUN
ajst-13993	267	9	.	.	PUNCT
ajst-13993	268	1	in	in	ADP
ajst-13993	268	2	proceedings	proceeding	NOUN
ajst-13993	268	3	of	of	ADP
ajst-13993	268	4	the	the	DET
ajst-13993	268	5	31st	31st	ADJ
ajst-13993	268	6	ieee	ieee	PROPN
ajst-13993	268	7	/	/	SYM
ajst-13993	268	8	acm	acm	PROPN
ajst-13993	268	9	international	international	ADJ
ajst-13993	268	10	conference	conference	NOUN
ajst-13993	268	11	on	on	ADP
ajst-13993	268	12	automated	automate	VERB
ajst-13993	268	13	software	software	NOUN
ajst-13993	268	14	engineering	engineering	NOUN
ajst-13993	268	15	,	,	PUNCT
ajst-13993	268	16	ase	ase	NOUN
ajst-13993	268	17	2016	2016	NUM
ajst-13993	268	18	,	,	PUNCT
ajst-13993	268	19	singapore	singapore	PROPN
ajst-13993	268	20	,	,	PUNCT
ajst-13993	268	21	september	september	PROPN
ajst-13993	268	22	3	3	NUM
ajst-13993	268	23	-	-	SYM
ajst-13993	268	24	7	7	NUM
ajst-13993	268	25	,	,	PUNCT
ajst-13993	268	26	2016	2016	NUM
ajst-13993	268	27	.	.	PUNCT
ajst-13993	269	1	87–98	87–98	NUM
ajst-13993	269	2	.	.	PUNCT
ajst-13993	270	1	https://doi.org/10.1145/2970276	https://doi.org/10.1145/2970276	NOUN
ajst-13993	270	2	.	.	PUNCT
ajst-13993	271	1	2970326	2970326	NUM
ajst-13993	271	2	.	.	PUNCT
ajst-13993	272	1	[	[	X
ajst-13993	272	2	32	32	NUM
ajst-13993	272	3	]	]	X
ajst-13993	272	4	edmund	edmund	PROPN
ajst-13993	272	5	wong	wong	PROPN
ajst-13993	272	6	,	,	PUNCT
ajst-13993	272	7	taiyue	taiyue	PROPN
ajst-13993	272	8	liu	liu	PROPN
ajst-13993	272	9	,	,	PUNCT
ajst-13993	272	10	and	and	CCONJ
ajst-13993	272	11	lin	lin	PROPN
ajst-13993	272	12	tan	tan	PROPN
ajst-13993	272	13	.	.	PROPN
ajst-13993	272	14	2015	2015	NUM
ajst-13993	272	15	.	.	PUNCT
ajst-13993	273	1	clocom	clocom	NOUN
ajst-13993	273	2	:	:	PUNCT
ajst-13993	273	3	mining	mine	VERB
ajst-13993	273	4	existing	exist	VERB
ajst-13993	273	5	source	source	NOUN
ajst-13993	273	6	code	code	NOUN
ajst-13993	273	7	for	for	ADP
ajst-13993	273	8	automatic	automatic	ADJ
ajst-13993	273	9	comment	comment	NOUN
ajst-13993	273	10	generation	generation	NOUN
ajst-13993	273	11	.	.	PUNCT
ajst-13993	274	1	in	in	ADP
ajst-13993	274	2	22nd	22nd	PROPN
ajst-13993	274	3	ieee	ieee	PROPN
ajst-13993	274	4	international	international	ADJ
ajst-13993	274	5	conference	conference	NOUN
ajst-13993	274	6	on	on	ADP
ajst-13993	274	7	software	software	NOUN
ajst-13993	274	8	analysis	analysis	NOUN
ajst-13993	274	9	,	,	PUNCT
ajst-13993	274	10	evolution	evolution	NOUN
ajst-13993	274	11	,	,	PUNCT
ajst-13993	274	12	and	and	CCONJ
ajst-13993	274	13	reengineering	reengineering	NOUN
ajst-13993	274	14	,	,	PUNCT
ajst-13993	274	15	saner	sane	ADJ
ajst-13993	274	16	2015	2015	NUM
ajst-13993	274	17	,	,	PUNCT
ajst-13993	274	18	montreal	montreal	PROPN
ajst-13993	274	19	,	,	PUNCT
ajst-13993	274	20	qc	qc	PROPN
ajst-13993	274	21	,	,	PUNCT
ajst-13993	274	22	canada	canada	PROPN
ajst-13993	274	23	,	,	PUNCT
ajst-13993	274	24	march	march	PROPN
ajst-13993	274	25	2	2	NUM
ajst-13993	274	26	-	-	SYM
ajst-13993	274	27	6	6	NUM
ajst-13993	274	28	,	,	PUNCT
ajst-13993	274	29	2015	2015	NUM
ajst-13993	274	30	.	.	PUNCT
ajst-13993	275	1	380–389	380–389	NUM
ajst-13993	275	2	.	.	PUNCT
ajst-13993	276	1	https://doi.org/10.1109/saner.2015.7081848	https://doi.org/10.1109/saner.2015.7081848	PROPN
ajst-13993	276	2	.	.	PUNCT
ajst-13993	277	1	[	[	X
ajst-13993	277	2	33	33	NUM
ajst-13993	277	3	]	]	PUNCT
ajst-13993	277	4	xing	xing	PROPN
ajst-13993	277	5	hu	hu	PROPN
ajst-13993	277	6	,	,	PUNCT
ajst-13993	277	7	ge	ge	PROPN
ajst-13993	277	8	li	li	PROPN
ajst-13993	277	9	,	,	PUNCT
ajst-13993	277	10	xin	xin	PROPN
ajst-13993	277	11	xia	xia	PROPN
ajst-13993	277	12	,	,	PUNCT
ajst-13993	277	13	david	david	PROPN
ajst-13993	277	14	lo	lo	PROPN
ajst-13993	277	15	,	,	PUNCT
ajst-13993	277	16	and	and	CCONJ
ajst-13993	277	17	zhi	zhi	PROPN
ajst-13993	277	18	jin	jin	PROPN
ajst-13993	277	19	.	.	PUNCT
ajst-13993	278	1	2018	2018	NUM
ajst-13993	278	2	.	.	PUNCT
ajst-13993	279	1	deep	deep	ADJ
ajst-13993	279	2	code	code	NOUN
ajst-13993	279	3	comment	comment	NOUN
ajst-13993	279	4	generation	generation	NOUN
ajst-13993	279	5	.	.	PUNCT
ajst-13993	280	1	in	in	ADP
ajst-13993	280	2	proceedings	proceeding	NOUN
ajst-13993	280	3	of	of	ADP
ajst-13993	280	4	the	the	DET
ajst-13993	280	5	26th	26th	ADJ
ajst-13993	280	6	conference	conference	NOUN
ajst-13993	280	7	on	on	ADP
ajst-13993	280	8	program	program	NOUN
ajst-13993	280	9	comprehension	comprehension	NOUN
ajst-13993	280	10	[	[	X
ajst-13993	280	11	34	34	NUM
ajst-13993	280	12	]	]	X
ajst-13993	280	13	srinivasan	srinivasan	PROPN
ajst-13993	280	14	iyer	iyer	PROPN
ajst-13993	280	15	,	,	PUNCT
ajst-13993	280	16	ioannis	ioannis	PROPN
ajst-13993	280	17	konstas	konstas	PROPN
ajst-13993	280	18	,	,	PUNCT
ajst-13993	280	19	alvin	alvin	PROPN
ajst-13993	280	20	cheung	cheung	PROPN
ajst-13993	280	21	,	,	PUNCT
ajst-13993	280	22	and	and	CCONJ
ajst-13993	280	23	luke	luke	PROPN
ajst-13993	280	24	zettlemoyer	zettlemoyer	NOUN
ajst-13993	280	25	.	.	PUNCT
ajst-13993	281	1	2016	2016	NUM
ajst-13993	281	2	.	.	PUNCT
ajst-13993	282	1	summarizing	summarize	VERB
ajst-13993	282	2	source	source	NOUN
ajst-13993	282	3	code	code	NOUN
ajst-13993	282	4	using	use	VERB
ajst-13993	282	5	a	a	DET
ajst-13993	282	6	neural	neural	ADJ
ajst-13993	282	7	attention	attention	NOUN
ajst-13993	282	8	model	model	NOUN
ajst-13993	282	9	.	.	PUNCT
ajst-13993	283	1	in	in	ADP
ajst-13993	283	2	proceedings	proceeding	NOUN
ajst-13993	283	3	of	of	ADP
ajst-13993	283	4	the	the	DET
ajst-13993	283	5	54th	54th	ADJ
ajst-13993	283	6	annual	annual	ADJ
ajst-13993	283	7	meeting	meeting	NOUN
ajst-13993	283	8	of	of	ADP
ajst-13993	283	9	the	the	DET
ajst-13993	283	10	association	association	NOUN
ajst-13993	283	11	for	for	ADP
ajst-13993	283	12	computational	computational	ADJ
ajst-13993	283	13	linguistics	linguistic	NOUN
ajst-13993	283	14	,	,	PUNCT
ajst-13993	283	15	acl	acl	PROPN
ajst-13993	283	16	2016	2016	NUM
ajst-13993	283	17	,	,	PUNCT
ajst-13993	283	18	august	august	PROPN
ajst-13993	283	19	7	7	NUM
ajst-13993	283	20	-	-	SYM
ajst-13993	283	21	12	12	NUM
ajst-13993	283	22	,	,	PUNCT
ajst-13993	283	23	2016	2016	NUM
ajst-13993	283	24	,	,	PUNCT
ajst-13993	283	25	berlin	berlin	PROPN
ajst-13993	283	26	,	,	PUNCT
ajst-13993	283	27	germany	germany	PROPN
ajst-13993	283	28	,	,	PUNCT
ajst-13993	283	29	volume	volume	NOUN
ajst-13993	283	30	1	1	NUM
ajst-13993	283	31	:	:	PUNCT
ajst-13993	283	32	long	long	ADJ
ajst-13993	283	33	papers	paper	NOUN
ajst-13993	283	34	.	.	PUNCT
ajst-13993	284	1	https://www.aclweb	https://www.aclweb	PROPN
ajst-13993	284	2	.	.	PUNCT
ajst-13993	285	1	org	org	ADJ
ajst-13993	285	2	/	/	SYM
ajst-13993	285	3	anthology	anthology	NOUN
ajst-13993	285	4	/	/	SYM
ajst-13993	285	5	p16	p16	NOUN
ajst-13993	285	6	-	-	PUNCT
ajst-13993	285	7	1195/	1195/	NUM
ajst-13993	285	8	[	[	X
ajst-13993	285	9	35	35	NUM
ajst-13993	285	10	]	]	X
ajst-13993	285	11	alexander	alexander	PROPN
ajst-13993	285	12	leclair	leclair	PROPN
ajst-13993	285	13	,	,	PUNCT
ajst-13993	285	14	siyuan	siyuan	PROPN
ajst-13993	285	15	jiang	jiang	PROPN
ajst-13993	285	16	,	,	PUNCT
ajst-13993	285	17	and	and	CCONJ
ajst-13993	285	18	collin	collin	PROPN
ajst-13993	285	19	mcmillan	mcmillan	PROPN
ajst-13993	285	20	.	.	PUNCT
ajst-13993	286	1	2019	2019	NUM
ajst-13993	286	2	.	.	PUNCT
ajst-13993	287	1	a	a	DET
ajst-13993	287	2	neural	neural	ADJ
ajst-13993	287	3	model	model	NOUN
ajst-13993	287	4	for	for	ADP
ajst-13993	287	5	generating	generate	VERB
ajst-13993	287	6	natural	natural	ADJ
ajst-13993	287	7	language	language	NOUN
ajst-13993	287	8	summaries	summary	NOUN
ajst-13993	287	9	of	of	ADP
ajst-13993	287	10	program	program	NOUN
ajst-13993	287	11	subroutines	subroutine	NOUN
ajst-13993	287	12	.	.	PUNCT
ajst-13993	288	1	in	in	ADP
ajst-13993	288	2	proceedings	proceeding	NOUN
ajst-13993	288	3	of	of	ADP
ajst-13993	288	4	the	the	DET
ajst-13993	288	5	41st	41st	ADJ
ajst-13993	288	6	international	international	ADJ
ajst-13993	288	7	conference	conference	NOUN
ajst-13993	288	8	on	on	ADP
ajst-13993	288	9	software	software	NOUN
ajst-13993	288	10	engineering	engineering	NOUN
ajst-13993	288	11	,	,	PUNCT
ajst-13993	288	12	icse	icse	PROPN
ajst-13993	288	13	2019	2019	NUM
ajst-13993	288	14	,	,	PUNCT
ajst-13993	288	15	montreal	montreal	PROPN
ajst-13993	288	16	,	,	PUNCT
ajst-13993	288	17	qc	qc	PROPN
ajst-13993	288	18	,	,	PUNCT
ajst-13993	288	19	canada	canada	PROPN
ajst-13993	288	20	,	,	PUNCT
ajst-13993	288	21	may	may	AUX
ajst-13993	288	22	25	25	NUM
ajst-13993	288	23	-	-	SYM
ajst-13993	288	24	31	31	NUM
ajst-13993	288	25	,	,	PUNCT
ajst-13993	288	26	2019	2019	NUM
ajst-13993	288	27	.	.	PUNCT
ajst-13993	289	1	795–806	795–806	NUM
ajst-13993	289	2	.	.	PUNCT
ajst-13993	290	1	https://doi.org/10.1109/icse.2019.00087	https://doi.org/10.1109/icse.2019.00087	PROPN
ajst-13993	290	2	.	.	PUNCT
ajst-13993	291	1	[	[	X
ajst-13993	291	2	36	36	NUM
ajst-13993	291	3	]	]	X
ajst-13993	291	4	giriprasad	giriprasad	PROPN
ajst-13993	291	5	sridhara	sridhara	NOUN
ajst-13993	291	6	,	,	PUNCT
ajst-13993	291	7	emily	emily	PROPN
ajst-13993	291	8	hill	hill	PROPN
ajst-13993	291	9	,	,	PUNCT
ajst-13993	291	10	divya	divya	PROPN
ajst-13993	291	11	muppaneni	muppaneni	PROPN
ajst-13993	291	12	,	,	PUNCT
ajst-13993	291	13	lori	lori	PROPN
ajst-13993	291	14	l.	l.	PROPN
ajst-13993	291	15	pollock	pollock	PROPN
ajst-13993	291	16	,	,	PUNCT
ajst-13993	291	17	and	and	CCONJ
ajst-13993	291	18	k.	k.	PROPN
ajst-13993	291	19	vijayshanker	vijayshanker	PROPN
ajst-13993	291	20	.	.	PUNCT
ajst-13993	292	1	2010	2010	NUM
ajst-13993	292	2	.	.	PUNCT
ajst-13993	293	1	towards	towards	ADP
ajst-13993	293	2	automatically	automatically	ADV
ajst-13993	293	3	generating	generate	VERB
ajst-13993	293	4	summary	summary	NOUN
ajst-13993	293	5	comments	comment	NOUN
ajst-13993	293	6	for	for	ADP
ajst-13993	293	7	java	java	NOUN
ajst-13993	293	8	methods	method	NOUN
ajst-13993	293	9	.	.	PUNCT
ajst-13993	294	1	in	in	ADP
ajst-13993	294	2	ase	ase	PROPN
ajst-13993	294	3	.	.	PROPN
ajst-13993	294	4	acm	acm	PROPN
ajst-13993	294	5	,	,	PUNCT
ajst-13993	294	6	43–52	43–52	NUM
ajst-13993	294	7	.	.	PUNCT
ajst-13993	295	1	[	[	X
ajst-13993	295	2	37	37	NUM
ajst-13993	295	3	]	]	X
ajst-13993	295	4	laura	laura	PROPN
ajst-13993	295	5	moreno	moreno	PROPN
ajst-13993	295	6	,	,	PUNCT
ajst-13993	295	7	jairo	jairo	PROPN
ajst-13993	295	8	aponte	aponte	PROPN
ajst-13993	295	9	,	,	PUNCT
ajst-13993	295	10	giriprasad	giriprasad	PROPN
ajst-13993	295	11	sridhara	sridhara	NOUN
ajst-13993	295	12	,	,	PUNCT
ajst-13993	295	13	andrian	andrian	PROPN
ajst-13993	295	14	marcus	marcus	PROPN
ajst-13993	295	15	,	,	PUNCT
ajst-13993	295	16	lori	lori	PROPN
ajst-13993	295	17	l.	l.	PROPN
ajst-13993	295	18	pollock	pollock	PROPN
ajst-13993	295	19	,	,	PUNCT
ajst-13993	295	20	and	and	CCONJ
ajst-13993	295	21	k.	k.	PROPN
ajst-13993	295	22	vijay	vijay	PROPN
ajst-13993	295	23	-	-	PUNCT
ajst-13993	295	24	shanker	shanker	NOUN
ajst-13993	295	25	.	.	PUNCT
ajst-13993	296	1	2013	2013	NUM
ajst-13993	296	2	.	.	PUNCT
ajst-13993	297	1	automatic	automatic	ADJ
ajst-13993	297	2	generation	generation	NOUN
ajst-13993	297	3	of	of	ADP
ajst-13993	297	4	natural	natural	ADJ
ajst-13993	297	5	language	language	NOUN
ajst-13993	297	6	summaries	summary	NOUN
ajst-13993	297	7	for	for	ADP
ajst-13993	297	8	java	java	PROPN
ajst-13993	297	9	classes	class	NOUN
ajst-13993	297	10	.	.	PUNCT
ajst-13993	298	1	in	in	ADP
ajst-13993	298	2	ieee	ieee	PROPN
ajst-13993	298	3	21st	21st	PROPN
ajst-13993	298	4	international	international	ADJ
ajst-13993	298	5	conference	conference	NOUN
ajst-13993	298	6	on	on	ADP
ajst-13993	298	7	program	program	NOUN
ajst-13993	298	8	81	81	NUM
ajst-13993	298	9	comprehension	comprehension	NOUN
ajst-13993	298	10	,	,	PUNCT
ajst-13993	298	11	icpc	icpc	PROPN
ajst-13993	298	12	2013	2013	NUM
ajst-13993	298	13	,	,	PUNCT
ajst-13993	298	14	san	san	PROPN
ajst-13993	298	15	francisco	francisco	PROPN
ajst-13993	298	16	,	,	PUNCT
ajst-13993	298	17	ca	ca	PROPN
ajst-13993	298	18	,	,	PUNCT
ajst-13993	298	19	usa	usa	PROPN
ajst-13993	298	20	,	,	PUNCT
ajst-13993	298	21	20	20	NUM
ajst-13993	298	22	-	-	SYM
ajst-13993	298	23	21	21	NUM
ajst-13993	298	24	may	may	NOUN
ajst-13993	298	25	,	,	PUNCT
ajst-13993	298	26	2013	2013	NUM
ajst-13993	298	27	.	.	PUNCT
ajst-13993	299	1	23–32	23–32	NUM
ajst-13993	299	2	.	.	PUNCT
ajst-13993	299	3	https://doi.org/10	https://doi.org/10	PROPN
ajst-13993	299	4	.	.	PUNCT
ajst-13993	300	1	1109	1109	NUM
ajst-13993	300	2	/	/	SYM
ajst-13993	300	3	icpc.2013.6613830	icpc.2013.6613830	PROPN
ajst-13993	300	4	.	.	PUNCT
ajst-13993	301	1	[	[	X
ajst-13993	301	2	38	38	NUM
ajst-13993	301	3	]	]	PUNCT
ajst-13993	301	4	paul	paul	PROPN
ajst-13993	301	5	w.	w.	PROPN
ajst-13993	301	6	mcburney	mcburney	PROPN
ajst-13993	301	7	and	and	CCONJ
ajst-13993	301	8	collin	collin	PROPN
ajst-13993	301	9	mcmillan	mcmillan	PROPN
ajst-13993	301	10	.	.	PUNCT
ajst-13993	302	1	2016	2016	NUM
ajst-13993	302	2	.	.	PUNCT
ajst-13993	303	1	automatic	automatic	ADJ
ajst-13993	303	2	source	source	NOUN
ajst-13993	303	3	code	code	NOUN
ajst-13993	303	4	summarization	summarization	NOUN
ajst-13993	303	5	of	of	ADP
ajst-13993	303	6	context	context	NOUN
ajst-13993	303	7	for	for	ADP
ajst-13993	303	8	java	java	PROPN
ajst-13993	303	9	methods	method	NOUN
ajst-13993	303	10	.	.	PUNCT
ajst-13993	304	1	ieee	ieee	PROPN
ajst-13993	304	2	trans	trans	PROPN
ajst-13993	304	3	.	.	PUNCT
ajst-13993	305	1	software	software	PROPN
ajst-13993	305	2	eng	eng	PROPN
ajst-13993	305	3	.	.	PROPN
ajst-13993	306	1	42	42	NUM
ajst-13993	306	2	,	,	PUNCT
ajst-13993	306	3	2	2	NUM
ajst-13993	306	4	(	(	PUNCT
ajst-13993	306	5	2016	2016	NUM
ajst-13993	306	6	)	)	PUNCT
ajst-13993	306	7	,	,	PUNCT
ajst-13993	306	8	103–119	103–119	NUM
ajst-13993	306	9	.	.	PUNCT
ajst-13993	307	1	https://doi.org/10.1109/tse.2015.2465386	https://doi.org/10.1109/tse.2015.2465386	NOUN
ajst-13993	307	2	.	.	PUNCT
ajst-13993	308	1	[	[	X
ajst-13993	308	2	39	39	NUM
ajst-13993	308	3	]	]	PUNCT
ajst-13993	308	4	sonia	sonia	NOUN
ajst-13993	308	5	haiduc	haiduc	PROPN
ajst-13993	308	6	,	,	PUNCT
ajst-13993	308	7	jairo	jairo	PROPN
ajst-13993	308	8	aponte	aponte	PROPN
ajst-13993	308	9	,	,	PUNCT
ajst-13993	308	10	laura	laura	PROPN
ajst-13993	308	11	moreno	moreno	PROPN
ajst-13993	308	12	,	,	PUNCT
ajst-13993	308	13	and	and	CCONJ
ajst-13993	308	14	andrian	andrian	PROPN
ajst-13993	308	15	marcus	marcus	PROPN
ajst-13993	308	16	.	.	PUNCT
ajst-13993	308	17	2010	2010	NUM
ajst-13993	308	18	.	.	PUNCT
ajst-13993	309	1	on	on	ADP
ajst-13993	309	2	the	the	DET
ajst-13993	309	3	use	use	NOUN
ajst-13993	309	4	of	of	ADP
ajst-13993	309	5	automated	automate	VERB
ajst-13993	309	6	text	text	NOUN
ajst-13993	309	7	summarization	summarization	NOUN
ajst-13993	309	8	techniques	technique	NOUN
ajst-13993	309	9	for	for	ADP
ajst-13993	309	10	summarizing	summarize	VERB
ajst-13993	309	11	source	source	NOUN
ajst-13993	309	12	code	code	NOUN
ajst-13993	309	13	.	.	PUNCT
ajst-13993	310	1	in	in	ADP
ajst-13993	310	2	17th	17th	ADJ
ajst-13993	310	3	working	work	VERB
ajst-13993	310	4	conference	conference	NOUN
ajst-13993	310	5	on	on	ADP
ajst-13993	310	6	reverse	reverse	ADJ
ajst-13993	310	7	engineering	engineering	NOUN
ajst-13993	310	8	,	,	PUNCT
ajst-13993	310	9	wcre	wcre	PROPN
ajst-13993	310	10	2010	2010	NUM
ajst-13993	310	11	,	,	PUNCT
ajst-13993	310	12	13	13	NUM
ajst-13993	310	13	-	-	SYM
ajst-13993	310	14	16	16	NUM
ajst-13993	310	15	october	october	NOUN
ajst-13993	310	16	2010	2010	NUM
ajst-13993	310	17	,	,	PUNCT
ajst-13993	310	18	beverly	beverly	PROPN
ajst-13993	310	19	,	,	PUNCT
ajst-13993	310	20	ma	ma	PROPN
ajst-13993	310	21	,	,	PUNCT
ajst-13993	310	22	usa	usa	PROPN
ajst-13993	310	23	.	.	PROPN
ajst-13993	310	24	35–44	35–44	NUM
ajst-13993	310	25	.	.	PUNCT
ajst-13993	310	26	https://doi.org/10.1109/wcre.2010.13	https://doi.org/10.1109/wcre.2010.13	NOUN
ajst-13993	310	27	.	.	PUNCT
ajst-13993	311	1	[	[	X
ajst-13993	311	2	40	40	NUM
ajst-13993	311	3	]	]	X
ajst-13993	311	4	edmund	edmund	PROPN
ajst-13993	311	5	wong	wong	PROPN
ajst-13993	311	6	,	,	PUNCT
ajst-13993	311	7	jinqiu	jinqiu	PROPN
ajst-13993	311	8	yang	yang	PROPN
ajst-13993	311	9	,	,	PUNCT
ajst-13993	311	10	and	and	CCONJ
ajst-13993	311	11	lin	lin	PROPN
ajst-13993	311	12	tan	tan	PROPN
ajst-13993	311	13	.	.	PUNCT
ajst-13993	311	14	2013	2013	NUM
ajst-13993	311	15	.	.	PUNCT
ajst-13993	312	1	autocomment	autocomment	NOUN
ajst-13993	312	2	:	:	PUNCT
ajst-13993	312	3	mining	mining	NOUN
ajst-13993	312	4	question	question	NOUN
ajst-13993	312	5	and	and	CCONJ
ajst-13993	312	6	answer	answer	NOUN
ajst-13993	312	7	sites	site	NOUN
ajst-13993	312	8	for	for	ADP
ajst-13993	312	9	automatic	automatic	ADJ
ajst-13993	312	10	comment	comment	NOUN
ajst-13993	312	11	generation	generation	NOUN
ajst-13993	312	12	.	.	PUNCT
ajst-13993	313	1	in	in	ADP
ajst-13993	313	2	2013	2013	NUM
ajst-13993	313	3	28th	28th	ADJ
ajst-13993	313	4	ieee	ieee	NOUN
ajst-13993	313	5	/	/	SYM
ajst-13993	313	6	acm	acm	PROPN
ajst-13993	313	7	international	international	ADJ
ajst-13993	313	8	conference	conference	NOUN
ajst-13993	313	9	on	on	ADP
ajst-13993	313	10	automated	automate	VERB
ajst-13993	313	11	software	software	NOUN
ajst-13993	313	12	engineering	engineering	NOUN
ajst-13993	313	13	,	,	PUNCT
ajst-13993	313	14	ase	ase	NOUN
ajst-13993	313	15	2013	2013	NUM
ajst-13993	313	16	,	,	PUNCT
ajst-13993	313	17	silicon	silicon	NOUN
ajst-13993	313	18	valley	valley	NOUN
ajst-13993	313	19	,	,	PUNCT
ajst-13993	313	20	ca	ca	PROPN
ajst-13993	313	21	,	,	PUNCT
ajst-13993	313	22	usa	usa	PROPN
ajst-13993	313	23	,	,	PUNCT
ajst-13993	313	24	november	november	PROPN
ajst-13993	313	25	11	11	NUM
ajst-13993	313	26	-	-	SYM
ajst-13993	313	27	15	15	NUM
ajst-13993	313	28	,	,	PUNCT
ajst-13993	313	29	2013	2013	NUM
ajst-13993	313	30	.	.	PUNCT
ajst-13993	314	1	55–67	55–67	NUM
ajst-13993	314	2	.	.	PUNCT
ajst-13993	314	3	https://doi.org/10.1109/ase	https://doi.org/10.1109/ase	PROPN
ajst-13993	314	4	.	.	PUNCT
ajst-13993	315	1	2013.6693113	2013.6693113	NUM
ajst-13993	315	2	.	.	PUNCT