id	sid	tid	token	lemma	pos
fcis-25108	1	1	frontiers	frontier	NOUN
fcis-25108	1	2	in	in	ADP
fcis-25108	1	3	computing	computing	NOUN
fcis-25108	1	4	and	and	CCONJ
fcis-25108	1	5	intelligent	intelligent	ADJ
fcis-25108	1	6	systems	system	NOUN
fcis-25108	1	7	issn	issn	VERB
fcis-25108	1	8	:	:	PUNCT
fcis-25108	1	9	2832	2832	NUM
fcis-25108	1	10	-	-	SYM
fcis-25108	1	11	6024	6024	NUM
fcis-25108	1	12	|	|	NOUN
fcis-25108	1	13	vol	vol	NOUN
fcis-25108	1	14	.	.	PROPN
fcis-25108	2	1	9	9	NUM
fcis-25108	2	2	,	,	PUNCT
fcis-25108	2	3	no	no	INTJ
fcis-25108	2	4	.	.	NOUN
fcis-25108	2	5	2	2	NUM
fcis-25108	2	6	,	,	PUNCT
fcis-25108	2	7	2024	2024	NUM
fcis-25108	2	8	56	56	NUM
fcis-25108	2	9	research	research	NOUN
fcis-25108	2	10	on	on	ADP
fcis-25108	2	11	the	the	DET
fcis-25108	2	12	application	application	NOUN
fcis-25108	2	13	of	of	ADP
fcis-25108	2	14	deep	deep	ADJ
fcis-25108	2	15	learning	learning	NOUN
fcis-25108	2	16	in	in	ADP
fcis-25108	2	17	natural	natural	ADJ
fcis-25108	2	18	language	language	NOUN
fcis-25108	2	19	processing	processing	NOUN
fcis-25108	2	20	fangxu	fangxu	PROPN
fcis-25108	2	21	yan	yan	PROPN
fcis-25108	2	22	,	,	PUNCT
fcis-25108	2	23	jianhui	jianhui	PROPN
fcis-25108	2	24	wang	wang	PROPN
fcis-25108	2	25	,	,	PUNCT
fcis-25108	2	26	wei	wei	PROPN
fcis-25108	2	27	li	li	PROPN
fcis-25108	2	28	school	school	PROPN
fcis-25108	2	29	of	of	ADP
fcis-25108	2	30	mathematics	mathematic	NOUN
fcis-25108	2	31	and	and	CCONJ
fcis-25108	2	32	systems	system	NOUN
fcis-25108	2	33	science	science	NOUN
fcis-25108	2	34	,	,	PUNCT
fcis-25108	2	35	shenyang	shenyang	PROPN
fcis-25108	2	36	normal	normal	PROPN
fcis-25108	2	37	university	university	PROPN
fcis-25108	2	38	,	,	PUNCT
fcis-25108	2	39	shenyang	shenyang	PROPN
fcis-25108	2	40	liaoning	liaoning	PROPN
fcis-25108	2	41	,	,	PUNCT
fcis-25108	2	42	110034	110034	NUM
fcis-25108	2	43	,	,	PUNCT
fcis-25108	2	44	china	china	PROPN
fcis-25108	2	45	abstract	abstract	NOUN
fcis-25108	2	46	:	:	PUNCT
fcis-25108	2	47	deep	deep	ADJ
fcis-25108	2	48	learning	learning	NOUN
fcis-25108	2	49	is	be	AUX
fcis-25108	2	50	a	a	DET
fcis-25108	2	51	kind	kind	NOUN
fcis-25108	2	52	of	of	ADP
fcis-25108	2	53	machine	machine	NOUN
fcis-25108	2	54	learning	learning	NOUN
fcis-25108	2	55	,	,	PUNCT
fcis-25108	2	56	which	which	PRON
fcis-25108	2	57	is	be	AUX
fcis-25108	2	58	the	the	DET
fcis-25108	2	59	necessary	necessary	ADJ
fcis-25108	2	60	path	path	NOUN
fcis-25108	2	61	to	to	PART
fcis-25108	2	62	realize	realize	VERB
fcis-25108	2	63	artificial	artificial	ADJ
fcis-25108	2	64	intelligence	intelligence	NOUN
fcis-25108	2	65	.	.	PUNCT
fcis-25108	3	1	the	the	DET
fcis-25108	3	2	application	application	NOUN
fcis-25108	3	3	of	of	ADP
fcis-25108	3	4	deep	deep	ADJ
fcis-25108	3	5	learning	learning	NOUN
fcis-25108	3	6	in	in	ADP
fcis-25108	3	7	the	the	DET
fcis-25108	3	8	field	field	NOUN
fcis-25108	3	9	of	of	ADP
fcis-25108	3	10	natural	natural	ADJ
fcis-25108	3	11	language	language	NOUN
fcis-25108	3	12	processing	processing	NOUN
fcis-25108	3	13	has	have	AUX
fcis-25108	3	14	gone	go	VERB
fcis-25108	3	15	far	far	ADV
fcis-25108	3	16	beyond	beyond	ADP
fcis-25108	3	17	the	the	DET
fcis-25108	3	18	limitations	limitation	NOUN
fcis-25108	3	19	of	of	ADP
fcis-25108	3	20	traditional	traditional	ADJ
fcis-25108	3	21	methods	method	NOUN
fcis-25108	3	22	,	,	PUNCT
fcis-25108	3	23	showing	show	VERB
fcis-25108	3	24	unparalleled	unparalleled	ADJ
fcis-25108	3	25	advantages	advantage	NOUN
fcis-25108	3	26	such	such	ADJ
fcis-25108	3	27	as	as	ADP
fcis-25108	3	28	automatically	automatically	ADV
fcis-25108	3	29	learning	learn	VERB
fcis-25108	3	30	abstract	abstract	ADJ
fcis-25108	3	31	features	feature	NOUN
fcis-25108	3	32	from	from	ADP
fcis-25108	3	33	the	the	DET
fcis-25108	3	34	original	original	ADJ
fcis-25108	3	35	data	datum	NOUN
fcis-25108	3	36	to	to	PART
fcis-25108	3	37	form	form	VERB
fcis-25108	3	38	preciser	preciser	NOUN
fcis-25108	3	39	representation	representation	NOUN
fcis-25108	3	40	.	.	PUNCT
fcis-25108	4	1	this	this	DET
fcis-25108	4	2	paper	paper	NOUN
fcis-25108	4	3	aims	aim	VERB
fcis-25108	4	4	to	to	PART
fcis-25108	4	5	analyze	analyze	VERB
fcis-25108	4	6	the	the	DET
fcis-25108	4	7	application	application	NOUN
fcis-25108	4	8	of	of	ADP
fcis-25108	4	9	deep	deep	ADJ
fcis-25108	4	10	learning	learning	NOUN
fcis-25108	4	11	in	in	ADP
fcis-25108	4	12	natural	natural	ADJ
fcis-25108	4	13	language	language	NOUN
fcis-25108	4	14	processing	processing	NOUN
fcis-25108	4	15	,	,	PUNCT
fcis-25108	4	16	and	and	CCONJ
fcis-25108	4	17	provides	provide	VERB
fcis-25108	4	18	reference	reference	NOUN
fcis-25108	4	19	for	for	ADP
fcis-25108	4	20	the	the	DET
fcis-25108	4	21	follow	follow	VERB
fcis-25108	4	22	-	-	PUNCT
fcis-25108	4	23	up	up	ADP
fcis-25108	4	24	research	research	NOUN
fcis-25108	4	25	and	and	CCONJ
fcis-25108	4	26	development	development	NOUN
fcis-25108	4	27	in	in	ADP
fcis-25108	4	28	this	this	DET
fcis-25108	4	29	field	field	NOUN
fcis-25108	4	30	.	.	PUNCT
fcis-25108	5	1	keywords	keyword	NOUN
fcis-25108	5	2	:	:	PUNCT
fcis-25108	5	3	deep	deep	ADJ
fcis-25108	5	4	learning	learning	NOUN
fcis-25108	5	5	;	;	PUNCT
fcis-25108	5	6	natural	natural	ADJ
fcis-25108	5	7	language	language	NOUN
fcis-25108	5	8	processing	processing	NOUN
fcis-25108	5	9	;	;	PUNCT
fcis-25108	5	10	neural	neural	ADJ
fcis-25108	5	11	network	network	NOUN
fcis-25108	5	12	.	.	PUNCT
fcis-25108	6	1	1	1	X
fcis-25108	6	2	.	.	X
fcis-25108	6	3	introduction	introduction	NOUN
fcis-25108	6	4	in	in	ADP
fcis-25108	6	5	recent	recent	ADJ
fcis-25108	6	6	years	year	NOUN
fcis-25108	6	7	,	,	PUNCT
fcis-25108	6	8	as	as	ADP
fcis-25108	6	9	a	a	DET
fcis-25108	6	10	hot	hot	ADJ
fcis-25108	6	11	field	field	NOUN
fcis-25108	6	12	of	of	ADP
fcis-25108	6	13	artificial	artificial	ADJ
fcis-25108	6	14	intelligence	intelligence	NOUN
fcis-25108	6	15	research	research	NOUN
fcis-25108	6	16	,	,	PUNCT
fcis-25108	6	17	deep	deep	ADJ
fcis-25108	6	18	learning	learning	NOUN
fcis-25108	6	19	(	(	PUNCT
fcis-25108	6	20	dl	dl	INTJ
fcis-25108	6	21	)	)	PUNCT
fcis-25108	6	22	has	have	AUX
fcis-25108	6	23	had	have	VERB
fcis-25108	6	24	an	an	DET
fcis-25108	6	25	increasingly	increasingly	ADV
fcis-25108	6	26	important	important	ADJ
fcis-25108	6	27	impact	impact	NOUN
fcis-25108	6	28	on	on	ADP
fcis-25108	6	29	many	many	ADJ
fcis-25108	6	30	tasks	task	NOUN
fcis-25108	6	31	due	due	ADP
fcis-25108	6	32	to	to	ADP
fcis-25108	6	33	its	its	PRON
fcis-25108	6	34	excellent	excellent	ADJ
fcis-25108	6	35	feature	feature	NOUN
fcis-25108	6	36	extraction	extraction	NOUN
fcis-25108	6	37	and	and	CCONJ
fcis-25108	6	38	learning	learn	VERB
fcis-25108	6	39	capabilities	capability	NOUN
fcis-25108	6	40	.	.	PUNCT
fcis-25108	7	1	the	the	DET
fcis-25108	7	2	concept	concept	NOUN
fcis-25108	7	3	of	of	ADP
fcis-25108	7	4	dl	dl	PROPN
fcis-25108	7	5	was	be	AUX
fcis-25108	7	6	first	first	ADV
fcis-25108	7	7	proposed	propose	VERB
fcis-25108	7	8	by	by	ADP
fcis-25108	7	9	hinton	hinton	PROPN
fcis-25108	7	10	in	in	ADP
fcis-25108	7	11	2006[1	2006[1	NUM
fcis-25108	7	12	]	]	PUNCT
fcis-25108	7	13	.	.	PUNCT
fcis-25108	8	1	it	it	PRON
fcis-25108	8	2	aims	aim	VERB
fcis-25108	8	3	to	to	PART
fcis-25108	8	4	learn	learn	VERB
fcis-25108	8	5	the	the	DET
fcis-25108	8	6	inherent	inherent	ADJ
fcis-25108	8	7	laws	law	NOUN
fcis-25108	8	8	and	and	CCONJ
fcis-25108	8	9	representation	representation	NOUN
fcis-25108	8	10	levels	level	NOUN
fcis-25108	8	11	of	of	ADP
fcis-25108	8	12	sample	sample	NOUN
fcis-25108	8	13	data	datum	NOUN
fcis-25108	8	14	,	,	PUNCT
fcis-25108	8	15	so	so	SCONJ
fcis-25108	8	16	that	that	SCONJ
fcis-25108	8	17	machines	machine	NOUN
fcis-25108	8	18	can	can	AUX
fcis-25108	8	19	have	have	VERB
fcis-25108	8	20	analytical	analytical	ADJ
fcis-25108	8	21	learning	learning	NOUN
fcis-25108	8	22	capabilities	capability	NOUN
fcis-25108	8	23	like	like	ADP
fcis-25108	8	24	humans	human	NOUN
fcis-25108	8	25	.	.	PUNCT
fcis-25108	9	1	at	at	ADP
fcis-25108	9	2	the	the	DET
fcis-25108	9	3	same	same	ADJ
fcis-25108	9	4	time	time	NOUN
fcis-25108	9	5	,	,	PUNCT
fcis-25108	9	6	dl	dl	PROPN
fcis-25108	9	7	has	have	AUX
fcis-25108	9	8	been	be	AUX
fcis-25108	9	9	widely	widely	ADV
fcis-25108	9	10	used	use	VERB
fcis-25108	9	11	in	in	ADP
fcis-25108	9	12	natural	natural	ADJ
fcis-25108	9	13	language	language	NOUN
fcis-25108	9	14	processing	processing	NOUN
fcis-25108	9	15	(	(	PUNCT
fcis-25108	9	16	nlp	nlp	NOUN
fcis-25108	9	17	)	)	PUNCT
fcis-25108	9	18	,	,	PUNCT
fcis-25108	9	19	computer	computer	NOUN
fcis-25108	9	20	vision	vision	NOUN
fcis-25108	9	21	(	(	PUNCT
fcis-25108	9	22	cv	cv	PROPN
fcis-25108	9	23	)	)	PUNCT
fcis-25108	9	24	,	,	PUNCT
fcis-25108	9	25	speech	speech	NOUN
fcis-25108	9	26	recognition	recognition	NOUN
fcis-25108	9	27	(	(	PUNCT
fcis-25108	9	28	sr	sr	PROPN
fcis-25108	9	29	)	)	PUNCT
fcis-25108	9	30	and	and	CCONJ
fcis-25108	9	31	other	other	ADJ
fcis-25108	9	32	fields	field	NOUN
fcis-25108	9	33	,	,	PUNCT
fcis-25108	9	34	which	which	PRON
fcis-25108	9	35	has	have	AUX
fcis-25108	9	36	attracted	attract	VERB
fcis-25108	9	37	much	much	ADJ
fcis-25108	9	38	attention	attention	NOUN
fcis-25108	9	39	.	.	PUNCT
fcis-25108	10	1	nlp	nlp	NOUN
fcis-25108	10	2	is	be	AUX
fcis-25108	10	3	regarded	regard	VERB
fcis-25108	10	4	as	as	ADP
fcis-25108	10	5	the	the	DET
fcis-25108	10	6	pearl	pearl	NOUN
fcis-25108	10	7	on	on	ADP
fcis-25108	10	8	the	the	DET
fcis-25108	10	9	crown	crown	NOUN
fcis-25108	10	10	of	of	ADP
fcis-25108	10	11	artificial	artificial	ADJ
fcis-25108	10	12	intelligence	intelligence	NOUN
fcis-25108	10	13	.	.	PUNCT
fcis-25108	11	1	the	the	DET
fcis-25108	11	2	goal	goal	NOUN
fcis-25108	11	3	of	of	ADP
fcis-25108	11	4	nlp	nlp	NOUN
fcis-25108	11	5	is	be	AUX
fcis-25108	11	6	to	to	PART
fcis-25108	11	7	enable	enable	VERB
fcis-25108	11	8	computers	computer	NOUN
fcis-25108	11	9	to	to	PART
fcis-25108	11	10	understand	understand	VERB
fcis-25108	11	11	,	,	PUNCT
fcis-25108	11	12	process	process	NOUN
fcis-25108	11	13	and	and	CCONJ
fcis-25108	11	14	generate	generate	VERB
fcis-25108	11	15	natural	natural	ADJ
fcis-25108	11	16	language	language	NOUN
fcis-25108	11	17	,	,	PUNCT
fcis-25108	11	18	so	so	SCONJ
fcis-25108	11	19	as	as	SCONJ
fcis-25108	11	20	to	to	PART
fcis-25108	11	21	provide	provide	VERB
fcis-25108	11	22	users	user	NOUN
fcis-25108	11	23	with	with	ADP
fcis-25108	11	24	better	well	ADJ
fcis-25108	11	25	interactive	interactive	ADJ
fcis-25108	11	26	experience	experience	NOUN
fcis-25108	11	27	and	and	CCONJ
fcis-25108	11	28	more	more	ADV
fcis-25108	11	29	efficient	efficient	ADJ
fcis-25108	11	30	output	output	NOUN
fcis-25108	11	31	.	.	PUNCT
fcis-25108	12	1	therefore	therefore	ADV
fcis-25108	12	2	,	,	PUNCT
fcis-25108	12	3	how	how	SCONJ
fcis-25108	12	4	to	to	PART
fcis-25108	12	5	use	use	VERB
fcis-25108	12	6	dl	dl	PROPN
fcis-25108	12	7	to	to	PART
fcis-25108	12	8	promote	promote	VERB
fcis-25108	12	9	the	the	DET
fcis-25108	12	10	development	development	NOUN
fcis-25108	12	11	of	of	ADP
fcis-25108	12	12	nlp	nlp	NOUN
fcis-25108	12	13	is	be	AUX
fcis-25108	12	14	the	the	DET
fcis-25108	12	15	hot	hot	ADJ
fcis-25108	12	16	spot	spot	NOUN
fcis-25108	12	17	and	and	CCONJ
fcis-25108	12	18	difficulty	difficulty	NOUN
fcis-25108	12	19	in	in	ADP
fcis-25108	12	20	current	current	ADJ
fcis-25108	12	21	research	research	NOUN
fcis-25108	12	22	.	.	PUNCT
fcis-25108	13	1	at	at	ADP
fcis-25108	13	2	present	present	ADJ
fcis-25108	13	3	,	,	PUNCT
fcis-25108	13	4	the	the	DET
fcis-25108	13	5	breakthrough	breakthrough	NOUN
fcis-25108	13	6	of	of	ADP
fcis-25108	13	7	deep	deep	ADJ
fcis-25108	13	8	learning	learning	NOUN
fcis-25108	13	9	technology	technology	NOUN
fcis-25108	13	10	has	have	AUX
fcis-25108	13	11	brought	bring	VERB
fcis-25108	13	12	great	great	ADJ
fcis-25108	13	13	changes	change	NOUN
fcis-25108	13	14	to	to	ADP
fcis-25108	13	15	the	the	DET
fcis-25108	13	16	development	development	NOUN
fcis-25108	13	17	of	of	ADP
fcis-25108	13	18	nlp	nlp	NOUN
fcis-25108	13	19	,	,	PUNCT
fcis-25108	13	20	which	which	PRON
fcis-25108	13	21	has	have	AUX
fcis-25108	13	22	greatly	greatly	ADV
fcis-25108	13	23	promoted	promote	VERB
fcis-25108	13	24	the	the	DET
fcis-25108	13	25	frontier	frontier	NOUN
fcis-25108	13	26	research	research	NOUN
fcis-25108	13	27	and	and	CCONJ
fcis-25108	13	28	practical	practical	ADJ
fcis-25108	13	29	application	application	NOUN
fcis-25108	13	30	in	in	ADP
fcis-25108	13	31	this	this	DET
fcis-25108	13	32	field	field	NOUN
fcis-25108	13	33	.	.	PUNCT
fcis-25108	14	1	dl	dl	PROPN
fcis-25108	14	2	has	have	AUX
fcis-25108	14	3	dominated	dominate	VERB
fcis-25108	14	4	especially	especially	ADV
fcis-25108	14	5	in	in	ADP
fcis-25108	14	6	key	key	ADJ
fcis-25108	14	7	applications	application	NOUN
fcis-25108	14	8	such	such	ADJ
fcis-25108	14	9	as	as	ADP
fcis-25108	14	10	machine	machine	NOUN
fcis-25108	14	11	translation	translation	NOUN
fcis-25108	14	12	,	,	PUNCT
fcis-25108	14	13	text	text	NOUN
fcis-25108	14	14	sentiment	sentiment	NOUN
fcis-25108	14	15	analysis	analysis	NOUN
fcis-25108	14	16	,	,	PUNCT
fcis-25108	14	17	and	and	CCONJ
fcis-25108	14	18	question	question	VERB
fcis-25108	14	19	answering	answering	NOUN
fcis-25108	14	20	systems	system	NOUN
fcis-25108	14	21	.	.	PUNCT
fcis-25108	15	1	in	in	ADP
fcis-25108	15	2	addition	addition	NOUN
fcis-25108	15	3	to	to	ADP
fcis-25108	15	4	these	these	DET
fcis-25108	15	5	main	main	ADJ
fcis-25108	15	6	application	application	NOUN
fcis-25108	15	7	fields	field	NOUN
fcis-25108	15	8	,	,	PUNCT
fcis-25108	15	9	dl	dl	PROPN
fcis-25108	15	10	also	also	ADV
fcis-25108	15	11	shows	show	VERB
fcis-25108	15	12	strong	strong	ADJ
fcis-25108	15	13	potential	potential	ADJ
fcis-25108	15	14	and	and	CCONJ
fcis-25108	15	15	development	development	NOUN
fcis-25108	15	16	prospects	prospect	NOUN
fcis-25108	15	17	in	in	ADP
fcis-25108	15	18	many	many	ADJ
fcis-25108	15	19	directions	direction	NOUN
fcis-25108	15	20	such	such	ADJ
fcis-25108	15	21	as	as	ADP
fcis-25108	15	22	text	text	NOUN
fcis-25108	15	23	generation	generation	NOUN
fcis-25108	15	24	,	,	PUNCT
fcis-25108	15	25	named	name	VERB
fcis-25108	15	26	entity	entity	NOUN
fcis-25108	15	27	recognition	recognition	NOUN
fcis-25108	15	28	and	and	CCONJ
fcis-25108	15	29	relation	relation	NOUN
fcis-25108	15	30	extraction	extraction	NOUN
fcis-25108	15	31	.	.	PUNCT
fcis-25108	16	1	these	these	PRON
fcis-25108	16	2	not	not	PART
fcis-25108	16	3	only	only	ADV
fcis-25108	16	4	promote	promote	VERB
fcis-25108	16	5	the	the	DET
fcis-25108	16	6	rapid	rapid	ADJ
fcis-25108	16	7	development	development	NOUN
fcis-25108	16	8	of	of	ADP
fcis-25108	16	9	nlp	nlp	NOUN
fcis-25108	16	10	technology	technology	NOUN
fcis-25108	16	11	,	,	PUNCT
fcis-25108	16	12	but	but	CCONJ
fcis-25108	16	13	also	also	ADV
fcis-25108	16	14	have	have	VERB
fcis-25108	16	15	great	great	ADJ
fcis-25108	16	16	economic	economic	ADJ
fcis-25108	16	17	value	value	NOUN
fcis-25108	16	18	in	in	ADP
fcis-25108	16	19	many	many	ADJ
fcis-25108	16	20	practical	practical	ADJ
fcis-25108	16	21	application	application	NOUN
fcis-25108	16	22	scenarios	scenario	NOUN
fcis-25108	16	23	such	such	ADJ
fcis-25108	16	24	as	as	ADP
fcis-25108	16	25	search	search	NOUN
fcis-25108	16	26	engine	engine	NOUN
fcis-25108	16	27	optimization	optimization	NOUN
fcis-25108	16	28	,	,	PUNCT
fcis-25108	16	29	knowledge	knowledge	NOUN
fcis-25108	16	30	graph	graph	NOUN
fcis-25108	16	31	construction	construction	NOUN
fcis-25108	16	32	,	,	PUNCT
fcis-25108	16	33	intelligent	intelligent	ADJ
fcis-25108	16	34	customer	customer	NOUN
fcis-25108	16	35	service	service	NOUN
fcis-25108	16	36	and	and	CCONJ
fcis-25108	16	37	so	so	ADV
fcis-25108	16	38	on	on	ADV
fcis-25108	16	39	.	.	PUNCT
fcis-25108	17	1	in	in	ADP
fcis-25108	17	2	summary	summary	NOUN
fcis-25108	17	3	,	,	PUNCT
fcis-25108	17	4	in	in	ADP
fcis-25108	17	5	-	-	PUNCT
fcis-25108	17	6	depth	depth	NOUN
fcis-25108	17	7	study	study	NOUN
fcis-25108	17	8	of	of	ADP
fcis-25108	17	9	the	the	DET
fcis-25108	17	10	application	application	NOUN
fcis-25108	17	11	development	development	NOUN
fcis-25108	17	12	and	and	CCONJ
fcis-25108	17	13	problems	problem	NOUN
fcis-25108	17	14	of	of	ADP
fcis-25108	17	15	dl	dl	PROPN
fcis-25108	17	16	in	in	ADP
fcis-25108	17	17	the	the	DET
fcis-25108	17	18	field	field	NOUN
fcis-25108	17	19	of	of	ADP
fcis-25108	17	20	nlp	nlp	NOUN
fcis-25108	17	21	is	be	AUX
fcis-25108	17	22	of	of	ADP
fcis-25108	17	23	great	great	ADJ
fcis-25108	17	24	practical	practical	ADJ
fcis-25108	17	25	significance	significance	NOUN
fcis-25108	17	26	for	for	ADP
fcis-25108	17	27	promoting	promote	VERB
fcis-25108	17	28	the	the	DET
fcis-25108	17	29	deep	deep	ADJ
fcis-25108	17	30	integration	integration	NOUN
fcis-25108	17	31	of	of	ADP
fcis-25108	17	32	artificial	artificial	ADJ
fcis-25108	17	33	intelligence	intelligence	NOUN
fcis-25108	17	34	and	and	CCONJ
fcis-25108	17	35	various	various	ADJ
fcis-25108	17	36	fields	field	NOUN
fcis-25108	17	37	of	of	ADP
fcis-25108	17	38	economic	economic	ADJ
fcis-25108	17	39	society	society	NOUN
fcis-25108	17	40	,	,	PUNCT
fcis-25108	17	41	stimulating	stimulate	VERB
fcis-25108	17	42	high	high	ADJ
fcis-25108	17	43	-	-	PUNCT
fcis-25108	17	44	quality	quality	NOUN
fcis-25108	17	45	economic	economic	ADJ
fcis-25108	17	46	development	development	NOUN
fcis-25108	17	47	,	,	PUNCT
fcis-25108	17	48	and	and	CCONJ
fcis-25108	17	49	motivating	motivate	VERB
fcis-25108	17	50	the	the	DET
fcis-25108	17	51	high	high	ADJ
fcis-25108	17	52	-	-	PUNCT
fcis-25108	17	53	level	level	NOUN
fcis-25108	17	54	application	application	NOUN
fcis-25108	17	55	of	of	ADP
fcis-25108	17	56	intelligent	intelligent	ADJ
fcis-25108	17	57	manufacturing	manufacturing	NOUN
fcis-25108	17	58	[	[	X
fcis-25108	17	59	2	2	NUM
fcis-25108	17	60	]	]	PUNCT
fcis-25108	17	61	.	.	PUNCT
fcis-25108	18	1	2	2	X
fcis-25108	18	2	.	.	X
fcis-25108	18	3	principles	principle	NOUN
fcis-25108	18	4	of	of	ADP
fcis-25108	18	5	deep	deep	ADJ
fcis-25108	18	6	learning	learn	VERB
fcis-25108	18	7	the	the	DET
fcis-25108	18	8	concept	concept	NOUN
fcis-25108	18	9	of	of	ADP
fcis-25108	18	10	deep	deep	ADJ
fcis-25108	18	11	learning	learning	NOUN
fcis-25108	18	12	originates	originate	NOUN
fcis-25108	18	13	from	from	ADP
fcis-25108	18	14	the	the	DET
fcis-25108	18	15	research	research	NOUN
fcis-25108	18	16	of	of	ADP
fcis-25108	18	17	artificial	artificial	ADJ
fcis-25108	18	18	neural	neural	ADJ
fcis-25108	18	19	network	network	NOUN
fcis-25108	18	20	.	.	PUNCT
fcis-25108	19	1	dl	dl	PROPN
fcis-25108	19	2	combines	combine	VERB
fcis-25108	19	3	low	low	ADJ
fcis-25108	19	4	-	-	PUNCT
fcis-25108	19	5	level	level	NOUN
fcis-25108	19	6	features	feature	NOUN
fcis-25108	19	7	to	to	PART
fcis-25108	19	8	form	form	VERB
fcis-25108	19	9	more	more	ADJ
fcis-25108	19	10	abstract	abstract	ADJ
fcis-25108	19	11	high	high	ADJ
fcis-25108	19	12	-	-	PUNCT
fcis-25108	19	13	level	level	NOUN
fcis-25108	19	14	representation	representation	NOUN
fcis-25108	19	15	attribute	attribute	NOUN
fcis-25108	19	16	categories	category	NOUN
fcis-25108	19	17	or	or	CCONJ
fcis-25108	19	18	features	feature	NOUN
fcis-25108	19	19	to	to	PART
fcis-25108	19	20	discover	discover	VERB
fcis-25108	19	21	distributed	distribute	VERB
fcis-25108	19	22	feature	feature	NOUN
fcis-25108	19	23	representations	representation	NOUN
fcis-25108	19	24	of	of	ADP
fcis-25108	19	25	data	datum	NOUN
fcis-25108	19	26	.	.	PUNCT
fcis-25108	20	1	the	the	DET
fcis-25108	20	2	motivation	motivation	NOUN
fcis-25108	20	3	of	of	ADP
fcis-25108	20	4	studying	study	VERB
fcis-25108	20	5	deep	deep	ADJ
fcis-25108	20	6	learning	learning	NOUN
fcis-25108	20	7	is	be	AUX
fcis-25108	20	8	to	to	PART
fcis-25108	20	9	establish	establish	VERB
fcis-25108	20	10	a	a	DET
fcis-25108	20	11	neural	neural	ADJ
fcis-25108	20	12	network	network	NOUN
fcis-25108	20	13	that	that	PRON
fcis-25108	20	14	simulates	simulate	VERB
fcis-25108	20	15	the	the	DET
fcis-25108	20	16	human	human	ADJ
fcis-25108	20	17	brain	brain	NOUN
fcis-25108	20	18	for	for	ADP
fcis-25108	20	19	analytical	analytical	ADJ
fcis-25108	20	20	learning	learning	NOUN
fcis-25108	20	21	.	.	PUNCT
fcis-25108	21	1	it	it	PRON
fcis-25108	21	2	imitates	imitate	VERB
fcis-25108	21	3	the	the	DET
fcis-25108	21	4	mechanism	mechanism	NOUN
fcis-25108	21	5	of	of	ADP
fcis-25108	21	6	the	the	DET
fcis-25108	21	7	human	human	ADJ
fcis-25108	21	8	brain	brain	NOUN
fcis-25108	21	9	to	to	PART
fcis-25108	21	10	interpret	interpret	VERB
fcis-25108	21	11	data	datum	NOUN
fcis-25108	21	12	,	,	PUNCT
fcis-25108	21	13	such	such	ADJ
fcis-25108	21	14	as	as	ADP
fcis-25108	21	15	images	image	NOUN
fcis-25108	21	16	,	,	PUNCT
fcis-25108	21	17	sounds	sound	NOUN
fcis-25108	21	18	and	and	CCONJ
fcis-25108	21	19	texts[3	texts[3	NUM
fcis-25108	21	20	]	]	X
fcis-25108	21	21	.	.	PUNCT
fcis-25108	22	1	a	a	DET
fcis-25108	22	2	basic	basic	ADJ
fcis-25108	22	3	neural	neural	ADJ
fcis-25108	22	4	network	network	NOUN
fcis-25108	22	5	includes	include	VERB
fcis-25108	22	6	input	input	NOUN
fcis-25108	22	7	layer	layer	NOUN
fcis-25108	22	8	,	,	PUNCT
fcis-25108	22	9	hidden	hide	VERB
fcis-25108	22	10	layer	layer	NOUN
fcis-25108	22	11	and	and	CCONJ
fcis-25108	22	12	output	output	NOUN
fcis-25108	22	13	layer	layer	NOUN
fcis-25108	22	14	.	.	PUNCT
fcis-25108	23	1	the	the	DET
fcis-25108	23	2	input	input	NOUN
fcis-25108	23	3	layer	layer	NOUN
fcis-25108	23	4	is	be	AUX
fcis-25108	23	5	responsible	responsible	ADJ
fcis-25108	23	6	for	for	ADP
fcis-25108	23	7	receiving	receive	VERB
fcis-25108	23	8	the	the	DET
fcis-25108	23	9	original	original	ADJ
fcis-25108	23	10	data	datum	NOUN
fcis-25108	23	11	set	set	NOUN
fcis-25108	23	12	,	,	PUNCT
fcis-25108	23	13	and	and	CCONJ
fcis-25108	23	14	the	the	DET
fcis-25108	23	15	output	output	NOUN
fcis-25108	23	16	layer	layer	NOUN
fcis-25108	23	17	delivers	deliver	VERB
fcis-25108	23	18	the	the	DET
fcis-25108	23	19	prediction	prediction	NOUN
fcis-25108	23	20	results	result	NOUN
fcis-25108	23	21	of	of	ADP
fcis-25108	23	22	the	the	DET
fcis-25108	23	23	model	model	NOUN
fcis-25108	23	24	.	.	PUNCT
fcis-25108	24	1	the	the	DET
fcis-25108	24	2	hidden	hide	VERB
fcis-25108	24	3	layer	layer	NOUN
fcis-25108	24	4	is	be	AUX
fcis-25108	24	5	located	locate	VERB
fcis-25108	24	6	between	between	ADP
fcis-25108	24	7	the	the	DET
fcis-25108	24	8	input	input	NOUN
fcis-25108	24	9	layer	layer	NOUN
fcis-25108	24	10	and	and	CCONJ
fcis-25108	24	11	the	the	DET
fcis-25108	24	12	output	output	NOUN
fcis-25108	24	13	layer	layer	NOUN
fcis-25108	24	14	,	,	PUNCT
fcis-25108	24	15	which	which	PRON
fcis-25108	24	16	usually	usually	ADV
fcis-25108	24	17	appears	appear	VERB
fcis-25108	24	18	as	as	ADP
fcis-25108	24	19	multiple	multiple	ADJ
fcis-25108	24	20	layers	layer	NOUN
fcis-25108	24	21	and	and	CCONJ
fcis-25108	24	22	contains	contain	VERB
fcis-25108	24	23	multiple	multiple	ADJ
fcis-25108	24	24	neurons	neuron	NOUN
fcis-25108	24	25	.	.	PUNCT
fcis-25108	25	1	each	each	DET
fcis-25108	25	2	neuron	neuron	NOUN
fcis-25108	25	3	and	and	CCONJ
fcis-25108	25	4	the	the	DET
fcis-25108	25	5	others	other	NOUN
fcis-25108	25	6	in	in	ADP
fcis-25108	25	7	the	the	DET
fcis-25108	25	8	anterior	anterior	ADJ
fcis-25108	25	9	and	and	CCONJ
fcis-25108	25	10	posterior	posterior	ADJ
fcis-25108	25	11	layers	layer	NOUN
fcis-25108	25	12	transmit	transmit	VERB
fcis-25108	25	13	information	information	NOUN
fcis-25108	25	14	through	through	ADP
fcis-25108	25	15	weight	weight	NOUN
fcis-25108	25	16	.	.	PUNCT
fcis-25108	26	1	by	by	ADP
fcis-25108	26	2	nonlinearly	nonlinearly	ADV
fcis-25108	26	3	converting	convert	VERB
fcis-25108	26	4	the	the	DET
fcis-25108	26	5	input	input	NOUN
fcis-25108	26	6	signals	signal	NOUN
fcis-25108	26	7	of	of	ADP
fcis-25108	26	8	neurons	neuron	NOUN
fcis-25108	26	9	,	,	PUNCT
fcis-25108	26	10	the	the	DET
fcis-25108	26	11	activation	activation	NOUN
fcis-25108	26	12	function	function	NOUN
fcis-25108	26	13	enables	enable	VERB
fcis-25108	26	14	the	the	DET
fcis-25108	26	15	network	network	NOUN
fcis-25108	26	16	to	to	PART
fcis-25108	26	17	learn	learn	VERB
fcis-25108	26	18	and	and	CCONJ
fcis-25108	26	19	perform	perform	VERB
fcis-25108	26	20	more	more	ADJ
fcis-25108	26	21	complex	complex	ADJ
fcis-25108	26	22	tasks	task	NOUN
fcis-25108	26	23	.	.	PUNCT
fcis-25108	27	1	common	common	ADJ
fcis-25108	27	2	activation	activation	NOUN
fcis-25108	27	3	functions	function	NOUN
fcis-25108	27	4	contain	contain	VERB
fcis-25108	27	5	,	,	PUNCT
fcis-25108	27	6	,	,	PUNCT
fcis-25108	27	7	,	,	PUNCT
fcis-25108	27	8	and	and	CCONJ
fcis-25108	27	9	so	so	ADV
fcis-25108	27	10	on	on	ADV
fcis-25108	27	11	.	.	PUNCT
fcis-25108	28	1	in	in	ADP
fcis-25108	28	2	addition	addition	NOUN
fcis-25108	28	3	,	,	PUNCT
fcis-25108	28	4	the	the	DET
fcis-25108	28	5	propagation	propagation	NOUN
fcis-25108	28	6	of	of	ADP
fcis-25108	28	7	data	datum	NOUN
fcis-25108	28	8	is	be	AUX
fcis-25108	28	9	divided	divide	VERB
fcis-25108	28	10	into	into	ADP
fcis-25108	28	11	forward	forward	ADJ
fcis-25108	28	12	propagation	propagation	NOUN
fcis-25108	28	13	and	and	CCONJ
fcis-25108	28	14	back	back	ADJ
fcis-25108	28	15	propagation	propagation	NOUN
fcis-25108	28	16	.	.	PUNCT
fcis-25108	29	1	in	in	ADP
fcis-25108	29	2	forward	forward	ADJ
fcis-25108	29	3	propagation	propagation	NOUN
fcis-25108	29	4	,	,	PUNCT
fcis-25108	29	5	the	the	DET
fcis-25108	29	6	input	input	NOUN
fcis-25108	29	7	data	data	NOUN
fcis-25108	29	8	is	be	AUX
fcis-25108	29	9	linearly	linearly	ADV
fcis-25108	29	10	transformed	transform	VERB
fcis-25108	29	11	by	by	ADP
fcis-25108	29	12	the	the	DET
fcis-25108	29	13	weight	weight	NOUN
fcis-25108	29	14	and	and	CCONJ
fcis-25108	29	15	bias	bias	NOUN
fcis-25108	29	16	of	of	ADP
fcis-25108	29	17	each	each	DET
fcis-25108	29	18	layer	layer	NOUN
fcis-25108	29	19	,	,	PUNCT
fcis-25108	29	20	and	and	CCONJ
fcis-25108	29	21	non	non	ADJ
fcis-25108	29	22	-	-	ADJ
fcis-25108	29	23	linearly	linearly	ADV
fcis-25108	29	24	transformed	transform	VERB
fcis-25108	29	25	by	by	ADP
fcis-25108	29	26	the	the	DET
fcis-25108	29	27	activation	activation	NOUN
fcis-25108	29	28	function	function	NOUN
fcis-25108	29	29	,	,	PUNCT
fcis-25108	29	30	and	and	CCONJ
fcis-25108	29	31	then	then	ADV
fcis-25108	29	32	output	output	VERB
fcis-25108	29	33	to	to	ADP
fcis-25108	29	34	the	the	DET
fcis-25108	29	35	next	next	ADJ
fcis-25108	29	36	layer	layer	NOUN
fcis-25108	29	37	until	until	ADP
fcis-25108	29	38	reaching	reach	VERB
fcis-25108	29	39	the	the	DET
fcis-25108	29	40	output	output	NOUN
fcis-25108	29	41	layer	layer	NOUN
fcis-25108	29	42	.	.	PUNCT
fcis-25108	30	1	in	in	ADP
fcis-25108	30	2	the	the	DET
fcis-25108	30	3	back	back	ADJ
fcis-25108	30	4	propagation	propagation	NOUN
fcis-25108	30	5	,	,	PUNCT
fcis-25108	30	6	the	the	DET
fcis-25108	30	7	error	error	NOUN
fcis-25108	30	8	of	of	ADP
fcis-25108	30	9	the	the	DET
fcis-25108	30	10	output	output	NOUN
fcis-25108	30	11	layer	layer	NOUN
fcis-25108	30	12	is	be	AUX
fcis-25108	30	13	first	first	ADV
fcis-25108	30	14	calculated	calculate	VERB
fcis-25108	30	15	,	,	PUNCT
fcis-25108	30	16	and	and	CCONJ
fcis-25108	30	17	then	then	ADV
fcis-25108	30	18	propagated	propagate	VERB
fcis-25108	30	19	from	from	ADP
fcis-25108	30	20	the	the	DET
fcis-25108	30	21	output	output	NOUN
fcis-25108	30	22	layer	layer	NOUN
fcis-25108	30	23	to	to	ADP
fcis-25108	30	24	the	the	DET
fcis-25108	30	25	hidden	hide	VERB
fcis-25108	30	26	layer	layer	NOUN
fcis-25108	30	27	until	until	SCONJ
fcis-25108	30	28	propagated	propagate	VERB
fcis-25108	30	29	to	to	ADP
fcis-25108	30	30	the	the	DET
fcis-25108	30	31	input	input	NOUN
fcis-25108	30	32	layer	layer	NOUN
fcis-25108	30	33	.	.	PUNCT
fcis-25108	31	1	through	through	ADP
fcis-25108	31	2	back	back	ADJ
fcis-25108	31	3	propagation	propagation	NOUN
fcis-25108	31	4	,	,	PUNCT
fcis-25108	31	5	the	the	DET
fcis-25108	31	6	gradient	gradient	ADJ
fcis-25108	31	7	information	information	NOUN
fcis-25108	31	8	of	of	ADP
fcis-25108	31	9	each	each	DET
fcis-25108	31	10	parameter	parameter	NOUN
fcis-25108	31	11	on	on	ADP
fcis-25108	31	12	the	the	DET
fcis-25108	31	13	loss	loss	NOUN
fcis-25108	31	14	function	function	NOUN
fcis-25108	31	15	can	can	AUX
fcis-25108	31	16	be	be	AUX
fcis-25108	31	17	obtained	obtain	VERB
fcis-25108	31	18	,	,	PUNCT
fcis-25108	31	19	so	so	SCONJ
fcis-25108	31	20	as	as	SCONJ
fcis-25108	31	21	to	to	PART
fcis-25108	31	22	optimize	optimize	VERB
fcis-25108	31	23	and	and	CCONJ
fcis-25108	31	24	update	update	VERB
fcis-25108	31	25	the	the	DET
fcis-25108	31	26	parameters	parameter	NOUN
fcis-25108	31	27	.	.	PUNCT
fcis-25108	32	1	different	different	ADJ
fcis-25108	32	2	from	from	ADP
fcis-25108	32	3	traditional	traditional	ADJ
fcis-25108	32	4	shallow	shallow	ADJ
fcis-25108	32	5	learning	learning	NOUN
fcis-25108	32	6	,	,	PUNCT
fcis-25108	32	7	dl	dl	PROPN
fcis-25108	32	8	emphasizes	emphasize	VERB
fcis-25108	32	9	the	the	DET
fcis-25108	32	10	depth	depth	NOUN
fcis-25108	32	11	of	of	ADP
fcis-25108	32	12	the	the	DET
fcis-25108	32	13	model	model	NOUN
fcis-25108	32	14	structure	structure	NOUN
fcis-25108	32	15	,	,	PUNCT
fcis-25108	32	16	which	which	PRON
fcis-25108	32	17	can	can	AUX
fcis-25108	32	18	be	be	AUX
fcis-25108	32	19	added	add	VERB
fcis-25108	32	20	to	to	PART
fcis-25108	32	21	obtain	obtain	VERB
fcis-25108	32	22	deep	deep	ADJ
fcis-25108	32	23	meaning	meaning	NOUN
fcis-25108	32	24	.	.	PUNCT
fcis-25108	33	1	secondly	secondly	ADV
fcis-25108	33	2	,	,	PUNCT
fcis-25108	33	3	dl	dl	PROPN
fcis-25108	33	4	clarifies	clarify	VERB
fcis-25108	33	5	the	the	DET
fcis-25108	33	6	importance	importance	NOUN
fcis-25108	33	7	of	of	ADP
fcis-25108	33	8	feature	feature	NOUN
fcis-25108	33	9	learning	learning	NOUN
fcis-25108	33	10	.	.	PUNCT
fcis-25108	34	1	by	by	ADP
fcis-25108	34	2	layer	layer	NOUN
fcis-25108	34	3	-	-	PUNCT
fcis-25108	34	4	by	by	ADP
fcis-25108	34	5	-	-	PUNCT
fcis-25108	34	6	layer	layer	NOUN
fcis-25108	34	7	feature	feature	NOUN
fcis-25108	34	8	transformation	transformation	NOUN
fcis-25108	34	9	,	,	PUNCT
fcis-25108	34	10	the	the	DET
fcis-25108	34	11	feature	feature	NOUN
fcis-25108	34	12	representation	representation	NOUN
fcis-25108	34	13	of	of	ADP
fcis-25108	34	14	the	the	DET
fcis-25108	34	15	sample	sample	NOUN
fcis-25108	34	16	in	in	ADP
fcis-25108	34	17	the	the	DET
fcis-25108	34	18	original	original	ADJ
fcis-25108	34	19	space	space	NOUN
fcis-25108	34	20	is	be	AUX
fcis-25108	34	21	transformed	transform	VERB
fcis-25108	34	22	into	into	ADP
fcis-25108	34	23	a	a	DET
fcis-25108	34	24	new	new	ADJ
fcis-25108	34	25	feature	feature	NOUN
fcis-25108	34	26	space	space	NOUN
fcis-25108	34	27	,	,	PUNCT
fcis-25108	34	28	which	which	PRON
fcis-25108	34	29	makes	make	VERB
fcis-25108	34	30	classification	classification	NOUN
fcis-25108	34	31	or	or	CCONJ
fcis-25108	34	32	prediction	prediction	NOUN
fcis-25108	34	33	easier	easy	ADJ
fcis-25108	34	34	.	.	PUNCT
fcis-25108	35	1	the	the	DET
fcis-25108	35	2	most	most	ADV
fcis-25108	35	3	classic	classic	ADJ
fcis-25108	35	4	deep	deep	ADJ
fcis-25108	35	5	learning	learning	NOUN
fcis-25108	35	6	networks	network	NOUN
fcis-25108	35	7	include	include	VERB
fcis-25108	35	8	convolutional	convolutional	ADJ
fcis-25108	35	9	neural	neural	ADJ
fcis-25108	35	10	network	network	NOUN
fcis-25108	35	11	(	(	PUNCT
fcis-25108	35	12	cnn	cnn	PROPN
fcis-25108	35	13	)	)	PUNCT
fcis-25108	35	14	and	and	CCONJ
fcis-25108	35	15	recurrent	recurrent	ADJ
fcis-25108	35	16	neural	neural	ADJ
fcis-25108	35	17	network	network	NOUN
fcis-25108	35	18	(	(	PUNCT
fcis-25108	35	19	rnn	rnn	PROPN
fcis-25108	35	20	)	)	PUNCT
fcis-25108	35	21	.	.	PUNCT
fcis-25108	36	1	2.1	2.1	NUM
fcis-25108	36	2	.	.	PUNCT
fcis-25108	37	1	convolutional	convolutional	ADJ
fcis-25108	37	2	neural	neural	ADJ
fcis-25108	37	3	network	network	NOUN
fcis-25108	37	4	cnn	cnn	PROPN
fcis-25108	37	5	is	be	AUX
fcis-25108	37	6	a	a	DET
fcis-25108	37	7	kind	kind	NOUN
fcis-25108	37	8	of	of	ADP
fcis-25108	37	9	feedforward	feedforward	ADJ
fcis-25108	37	10	neural	neural	ADJ
fcis-25108	37	11	networks	network	NOUN
fcis-25108	37	12	with	with	ADP
fcis-25108	37	13	convolution	convolution	NOUN
fcis-25108	37	14	calculation	calculation	NOUN
fcis-25108	37	15	and	and	CCONJ
fcis-25108	37	16	deep	deep	ADJ
fcis-25108	37	17	structure	structure	NOUN
fcis-25108	37	18	,	,	PUNCT
fcis-25108	37	19	which	which	PRON
fcis-25108	37	20	is	be	AUX
fcis-25108	37	21	one	one	NUM
fcis-25108	37	22	of	of	ADP
fcis-25108	37	23	the	the	DET
fcis-25108	37	24	representative	representative	ADJ
fcis-25108	37	25	algorithms	algorithm	NOUN
fcis-25108	37	26	of	of	ADP
fcis-25108	37	27	deep	deep	ADJ
fcis-25108	37	28	learning	learning	NOUN
fcis-25108	37	29	.	.	PUNCT
fcis-25108	38	1	cnn	cnn	PROPN
fcis-25108	38	2	adopts	adopt	VERB
fcis-25108	38	3	the	the	DET
fcis-25108	38	4	method	method	NOUN
fcis-25108	38	5	of	of	ADP
fcis-25108	38	6	local	local	ADJ
fcis-25108	38	7	connection	connection	NOUN
fcis-25108	38	8	and	and	CCONJ
fcis-25108	38	9	weight	weight	NOUN
fcis-25108	38	10	sharing	sharing	NOUN
fcis-25108	38	11	.	.	PUNCT
fcis-25108	39	1	on	on	ADP
fcis-25108	39	2	the	the	DET
fcis-25108	39	3	57	57	NUM
fcis-25108	39	4	one	one	NUM
fcis-25108	39	5	hand	hand	NOUN
fcis-25108	39	6	,	,	PUNCT
fcis-25108	39	7	it	it	PRON
fcis-25108	39	8	reduces	reduce	VERB
fcis-25108	39	9	the	the	DET
fcis-25108	39	10	number	number	NOUN
fcis-25108	39	11	of	of	ADP
fcis-25108	39	12	weights	weight	NOUN
fcis-25108	39	13	and	and	CCONJ
fcis-25108	39	14	makes	make	VERB
fcis-25108	39	15	the	the	DET
fcis-25108	39	16	network	network	NOUN
fcis-25108	39	17	easy	easy	ADJ
fcis-25108	39	18	to	to	PART
fcis-25108	39	19	optimize	optimize	VERB
fcis-25108	39	20	.	.	PUNCT
fcis-25108	40	1	on	on	ADP
fcis-25108	40	2	the	the	DET
fcis-25108	40	3	other	other	ADJ
fcis-25108	40	4	hand	hand	NOUN
fcis-25108	40	5	,	,	PUNCT
fcis-25108	40	6	it	it	PRON
fcis-25108	40	7	reduces	reduce	VERB
fcis-25108	40	8	the	the	DET
fcis-25108	40	9	complexity	complexity	NOUN
fcis-25108	40	10	of	of	ADP
fcis-25108	40	11	the	the	DET
fcis-25108	40	12	model	model	NOUN
fcis-25108	40	13	and	and	CCONJ
fcis-25108	40	14	reduces	reduce	VERB
fcis-25108	40	15	the	the	DET
fcis-25108	40	16	risk	risk	NOUN
fcis-25108	40	17	of	of	ADP
fcis-25108	40	18	overfitting[4	overfitting[4	PROPN
fcis-25108	40	19	]	]	PUNCT
fcis-25108	40	20	.	.	PUNCT
fcis-25108	41	1	when	when	SCONJ
fcis-25108	41	2	the	the	DET
fcis-25108	41	3	input	input	NOUN
fcis-25108	41	4	of	of	ADP
fcis-25108	41	5	the	the	DET
fcis-25108	41	6	network	network	NOUN
fcis-25108	41	7	is	be	AUX
fcis-25108	41	8	an	an	DET
fcis-25108	41	9	image	image	NOUN
fcis-25108	41	10	,	,	PUNCT
fcis-25108	41	11	these	these	DET
fcis-25108	41	12	advantages	advantage	NOUN
fcis-25108	41	13	will	will	AUX
fcis-25108	41	14	be	be	AUX
fcis-25108	41	15	more	more	ADV
fcis-25108	41	16	obvious	obvious	ADJ
fcis-25108	41	17	.	.	PUNCT
fcis-25108	42	1	each	each	DET
fcis-25108	42	2	neuron	neuron	NOUN
fcis-25108	42	3	only	only	ADV
fcis-25108	42	4	needs	need	VERB
fcis-25108	42	5	to	to	PART
fcis-25108	42	6	perceive	perceive	VERB
fcis-25108	42	7	some	some	DET
fcis-25108	42	8	elements	element	NOUN
fcis-25108	42	9	in	in	ADP
fcis-25108	42	10	the	the	DET
fcis-25108	42	11	image	image	NOUN
fcis-25108	42	12	or	or	CCONJ
fcis-25108	42	13	text	text	NOUN
fcis-25108	42	14	,	,	PUNCT
fcis-25108	42	15	and	and	CCONJ
fcis-25108	42	16	the	the	DET
fcis-25108	42	17	final	final	ADJ
fcis-25108	42	18	neuron	neuron	NOUN
fcis-25108	42	19	integrates	integrate	VERB
fcis-25108	42	20	the	the	DET
fcis-25108	42	21	perceived	perceive	VERB
fcis-25108	42	22	local	local	ADJ
fcis-25108	42	23	information	information	NOUN
fcis-25108	42	24	to	to	PART
fcis-25108	42	25	obtain	obtain	VERB
fcis-25108	42	26	the	the	DET
fcis-25108	42	27	comprehensive	comprehensive	ADJ
fcis-25108	42	28	representation	representation	NOUN
fcis-25108	42	29	information	information	NOUN
fcis-25108	42	30	of	of	ADP
fcis-25108	42	31	the	the	DET
fcis-25108	42	32	image	image	NOUN
fcis-25108	42	33	or	or	CCONJ
fcis-25108	42	34	text	text	NOUN
fcis-25108	42	35	.	.	PUNCT
fcis-25108	43	1	as	as	SCONJ
fcis-25108	43	2	shown	show	VERB
fcis-25108	43	3	in	in	ADP
fcis-25108	43	4	figure	figure	NOUN
fcis-25108	43	5	1	1	NUM
fcis-25108	43	6	,	,	PUNCT
fcis-25108	43	7	cnn	cnn	PROPN
fcis-25108	43	8	is	be	AUX
fcis-25108	43	9	usually	usually	ADV
fcis-25108	43	10	composed	compose	VERB
fcis-25108	43	11	of	of	ADP
fcis-25108	43	12	input	input	NOUN
fcis-25108	43	13	layer	layer	NOUN
fcis-25108	43	14	,	,	PUNCT
fcis-25108	43	15	convolutional	convolutional	ADJ
fcis-25108	43	16	layer	layer	NOUN
fcis-25108	43	17	,	,	PUNCT
fcis-25108	43	18	activation	activation	NOUN
fcis-25108	43	19	fuction	fuction	NOUN
fcis-25108	43	20	,	,	PUNCT
fcis-25108	43	21	pooling	pool	VERB
fcis-25108	43	22	layer	layer	NOUN
fcis-25108	43	23	,	,	PUNCT
fcis-25108	43	24	fully	fully	ADV
fcis-25108	43	25	connected	connected	ADJ
fcis-25108	43	26	layer	layer	NOUN
fcis-25108	43	27	and	and	CCONJ
fcis-25108	43	28	output	output	NOUN
fcis-25108	43	29	layer	layer	NOUN
fcis-25108	43	30	.	.	PUNCT
fcis-25108	44	1	the	the	DET
fcis-25108	44	2	convolutional	convolutional	ADJ
fcis-25108	44	3	layer	layer	NOUN
fcis-25108	44	4	is	be	AUX
fcis-25108	44	5	an	an	DET
fcis-25108	44	6	important	important	ADJ
fcis-25108	44	7	part	part	NOUN
fcis-25108	44	8	of	of	ADP
fcis-25108	44	9	cnn	cnn	PROPN
fcis-25108	44	10	.	.	PUNCT
fcis-25108	45	1	each	each	DET
fcis-25108	45	2	node	node	ADJ
fcis-25108	45	3	input	input	NOUN
fcis-25108	45	4	in	in	ADP
fcis-25108	45	5	the	the	DET
fcis-25108	45	6	convolutional	convolutional	ADJ
fcis-25108	45	7	layer	layer	NOUN
fcis-25108	45	8	is	be	AUX
fcis-25108	45	9	part	part	NOUN
fcis-25108	45	10	of	of	ADP
fcis-25108	45	11	the	the	DET
fcis-25108	45	12	previous	previous	ADJ
fcis-25108	45	13	neural	neural	ADJ
fcis-25108	45	14	network	network	NOUN
fcis-25108	45	15	layer	layer	NOUN
fcis-25108	45	16	,	,	PUNCT
fcis-25108	45	17	which	which	PRON
fcis-25108	45	18	aims	aim	VERB
fcis-25108	45	19	to	to	PART
fcis-25108	45	20	extract	extract	VERB
fcis-25108	45	21	different	different	ADJ
fcis-25108	45	22	features	feature	NOUN
fcis-25108	45	23	of	of	ADP
fcis-25108	45	24	the	the	DET
fcis-25108	45	25	input	input	NOUN
fcis-25108	45	26	picture	picture	NOUN
fcis-25108	45	27	or	or	CCONJ
fcis-25108	45	28	text	text	NOUN
fcis-25108	45	29	.	.	PUNCT
fcis-25108	46	1	when	when	SCONJ
fcis-25108	46	2	dealing	deal	VERB
fcis-25108	46	3	with	with	ADP
fcis-25108	46	4	the	the	DET
fcis-25108	46	5	text	text	NOUN
fcis-25108	46	6	sequence	sequence	NOUN
fcis-25108	46	7	problem	problem	NOUN
fcis-25108	46	8	,	,	PUNCT
fcis-25108	46	9	the	the	DET
fcis-25108	46	10	convolution	convolution	NOUN
fcis-25108	46	11	layer	layer	NOUN
fcis-25108	46	12	usually	usually	ADV
fcis-25108	46	13	uses	use	VERB
fcis-25108	46	14	filters	filter	NOUN
fcis-25108	46	15	of	of	ADP
fcis-25108	46	16	different	different	ADJ
fcis-25108	46	17	sizes	size	NOUN
fcis-25108	46	18	to	to	PART
fcis-25108	46	19	extract	extract	VERB
fcis-25108	46	20	different	different	ADJ
fcis-25108	46	21	features	feature	NOUN
fcis-25108	46	22	in	in	ADP
fcis-25108	46	23	the	the	DET
fcis-25108	46	24	text	text	NOUN
fcis-25108	46	25	sequence	sequence	NOUN
fcis-25108	46	26	.	.	PUNCT
fcis-25108	47	1	then	then	ADV
fcis-25108	47	2	the	the	DET
fcis-25108	47	3	nonlinearity	nonlinearity	NOUN
fcis-25108	47	4	of	of	ADP
fcis-25108	47	5	cnn	cnn	PROPN
fcis-25108	47	6	is	be	AUX
fcis-25108	47	7	increased	increase	VERB
fcis-25108	47	8	by	by	ADP
fcis-25108	47	9	the	the	DET
fcis-25108	47	10	activation	activation	NOUN
fcis-25108	47	11	function	function	NOUN
fcis-25108	47	12	.	.	PUNCT
fcis-25108	48	1	the	the	DET
fcis-25108	48	2	pooling	pool	VERB
fcis-25108	48	3	layer	layer	NOUN
fcis-25108	48	4	is	be	AUX
fcis-25108	48	5	to	to	PART
fcis-25108	48	6	reduce	reduce	VERB
fcis-25108	48	7	the	the	DET
fcis-25108	48	8	input	input	NOUN
fcis-25108	48	9	dimension	dimension	NOUN
fcis-25108	48	10	of	of	ADP
fcis-25108	48	11	the	the	DET
fcis-25108	48	12	model	model	NOUN
fcis-25108	48	13	,	,	PUNCT
fcis-25108	48	14	thereby	thereby	ADV
fcis-25108	48	15	reducing	reduce	VERB
fcis-25108	48	16	the	the	DET
fcis-25108	48	17	complexity	complexity	NOUN
fcis-25108	48	18	and	and	CCONJ
fcis-25108	48	19	parameters	parameter	NOUN
fcis-25108	48	20	,	,	PUNCT
fcis-25108	48	21	improving	improve	VERB
fcis-25108	48	22	robustness	robustness	NOUN
fcis-25108	48	23	,	,	PUNCT
fcis-25108	48	24	and	and	CCONJ
fcis-25108	48	25	preventing	prevent	VERB
fcis-25108	48	26	model	model	NOUN
fcis-25108	48	27	from	from	ADP
fcis-25108	48	28	overfitting	overfitte	VERB
fcis-25108	48	29	to	to	ADP
fcis-25108	48	30	a	a	DET
fcis-25108	48	31	certain	certain	ADJ
fcis-25108	48	32	extent	extent	NOUN
fcis-25108	48	33	.	.	PUNCT
fcis-25108	49	1	the	the	DET
fcis-25108	49	2	most	most	ADV
fcis-25108	49	3	common	common	ADJ
fcis-25108	49	4	pooling	pooling	NOUN
fcis-25108	49	5	methods	method	NOUN
fcis-25108	49	6	are	be	AUX
fcis-25108	49	7	maximum	maximum	ADJ
fcis-25108	49	8	pooling	pooling	NOUN
fcis-25108	49	9	and	and	CCONJ
fcis-25108	49	10	average	average	ADJ
fcis-25108	49	11	pooling	pooling	NOUN
fcis-25108	49	12	.	.	PUNCT
fcis-25108	50	1	the	the	DET
fcis-25108	50	2	fully	fully	ADV
fcis-25108	50	3	connected	connect	VERB
fcis-25108	50	4	layer	layer	NOUN
fcis-25108	50	5	is	be	AUX
fcis-25108	50	6	mainly	mainly	ADV
fcis-25108	50	7	responsible	responsible	ADJ
fcis-25108	50	8	for	for	ADP
fcis-25108	50	9	dimensionality	dimensionality	NOUN
fcis-25108	50	10	reduction	reduction	NOUN
fcis-25108	50	11	while	while	SCONJ
fcis-25108	50	12	retaining	retain	VERB
fcis-25108	50	13	truly	truly	ADV
fcis-25108	50	14	effective	effective	ADJ
fcis-25108	50	15	information	information	NOUN
fcis-25108	50	16	.	.	PUNCT
fcis-25108	51	1	usually	usually	ADV
fcis-25108	51	2	,	,	PUNCT
fcis-25108	51	3	the	the	DET
fcis-25108	51	4	components	component	NOUN
fcis-25108	51	5	from	from	ADP
fcis-25108	51	6	the	the	DET
fcis-25108	51	7	convolutional	convolutional	ADJ
fcis-25108	51	8	layer	layer	NOUN
fcis-25108	51	9	to	to	ADP
fcis-25108	51	10	the	the	DET
fcis-25108	51	11	pooling	pooling	NOUN
fcis-25108	51	12	layer	layer	NOUN
fcis-25108	51	13	are	be	AUX
fcis-25108	51	14	regarded	regard	VERB
fcis-25108	51	15	as	as	ADP
fcis-25108	51	16	the	the	DET
fcis-25108	51	17	process	process	NOUN
fcis-25108	51	18	of	of	ADP
fcis-25108	51	19	automatically	automatically	ADV
fcis-25108	51	20	extracting	extract	VERB
fcis-25108	51	21	features	feature	NOUN
fcis-25108	51	22	.	.	PUNCT
fcis-25108	52	1	after	after	SCONJ
fcis-25108	52	2	the	the	DET
fcis-25108	52	3	feature	feature	NOUN
fcis-25108	52	4	extraction	extraction	NOUN
fcis-25108	52	5	is	be	AUX
fcis-25108	52	6	completed	complete	VERB
fcis-25108	52	7	,	,	PUNCT
fcis-25108	52	8	the	the	DET
fcis-25108	52	9	output	output	NOUN
fcis-25108	52	10	layer	layer	NOUN
fcis-25108	52	11	needs	need	VERB
fcis-25108	52	12	to	to	PART
fcis-25108	52	13	be	be	AUX
fcis-25108	52	14	used	use	VERB
fcis-25108	52	15	to	to	PART
fcis-25108	52	16	complete	complete	VERB
fcis-25108	52	17	the	the	DET
fcis-25108	52	18	classification	classification	NOUN
fcis-25108	52	19	or	or	CCONJ
fcis-25108	52	20	prediction	prediction	NOUN
fcis-25108	52	21	tasks	task	NOUN
fcis-25108	52	22	.	.	PUNCT
fcis-25108	53	1	generally	generally	ADV
fcis-25108	53	2	,	,	PUNCT
fcis-25108	53	3	the	the	DET
fcis-25108	53	4	learned	learn	VERB
fcis-25108	53	5	high	high	ADJ
fcis-25108	53	6	-	-	PUNCT
fcis-25108	53	7	dimensional	dimensional	ADJ
fcis-25108	53	8	feature	feature	NOUN
fcis-25108	53	9	representation	representation	NOUN
fcis-25108	53	10	is	be	AUX
fcis-25108	53	11	fed	feed	VERB
fcis-25108	53	12	to	to	ADP
fcis-25108	53	13	the	the	DET
fcis-25108	53	14	output	output	NOUN
fcis-25108	53	15	layer	layer	NOUN
fcis-25108	53	16	,	,	PUNCT
fcis-25108	53	17	and	and	CCONJ
fcis-25108	53	18	the	the	DET
fcis-25108	53	19	probability	probability	NOUN
fcis-25108	53	20	that	that	SCONJ
fcis-25108	53	21	the	the	DET
fcis-25108	53	22	current	current	ADJ
fcis-25108	53	23	sample	sample	NOUN
fcis-25108	53	24	belongs	belong	VERB
fcis-25108	53	25	to	to	ADP
fcis-25108	53	26	different	different	ADJ
fcis-25108	53	27	categories	category	NOUN
fcis-25108	53	28	is	be	AUX
fcis-25108	53	29	calculated	calculate	VERB
fcis-25108	53	30	by	by	ADP
fcis-25108	53	31	the	the	DET
fcis-25108	53	32	softmax	softmax	NOUN
fcis-25108	53	33	function	function	NOUN
fcis-25108	53	34	.	.	PUNCT
fcis-25108	54	1	fig	fig	NOUN
fcis-25108	54	2	1	1	NUM
fcis-25108	54	3	.	.	PUNCT
fcis-25108	55	1	overall	overall	ADJ
fcis-25108	55	2	structure	structure	NOUN
fcis-25108	55	3	of	of	ADP
fcis-25108	55	4	cnn	cnn	PROPN
fcis-25108	55	5	2.2	2.2	NUM
fcis-25108	55	6	.	.	PUNCT
fcis-25108	56	1	recurrent	recurrent	ADJ
fcis-25108	56	2	neural	neural	ADJ
fcis-25108	56	3	network	network	NOUN
fcis-25108	56	4	recurrent	recurrent	NOUN
fcis-25108	56	5	neural	neural	ADJ
fcis-25108	56	6	network	network	NOUN
fcis-25108	56	7	has	have	VERB
fcis-25108	56	8	the	the	DET
fcis-25108	56	9	structure	structure	NOUN
fcis-25108	56	10	of	of	ADP
fcis-25108	56	11	tree	tree	NOUN
fcis-25108	56	12	layer	layer	NOUN
fcis-25108	56	13	,	,	PUNCT
fcis-25108	56	14	which	which	PRON
fcis-25108	56	15	is	be	AUX
fcis-25108	56	16	an	an	DET
fcis-25108	56	17	artificial	artificial	ADJ
fcis-25108	56	18	neural	neural	ADJ
fcis-25108	56	19	network	network	NOUN
fcis-25108	56	20	in	in	ADP
fcis-25108	56	21	which	which	PRON
fcis-25108	56	22	the	the	DET
fcis-25108	56	23	network	network	NOUN
fcis-25108	56	24	nodes	nod	VERB
fcis-25108	56	25	recursive	recursive	VERB
fcis-25108	56	26	the	the	DET
fcis-25108	56	27	input	input	NOUN
fcis-25108	56	28	information	information	NOUN
fcis-25108	56	29	according	accord	VERB
fcis-25108	56	30	to	to	ADP
fcis-25108	56	31	their	their	PRON
fcis-25108	56	32	connection	connection	NOUN
fcis-25108	56	33	order	order	NOUN
fcis-25108	56	34	.	.	PUNCT
fcis-25108	57	1	rnn	rnn	PROPN
fcis-25108	57	2	has	have	VERB
fcis-25108	57	3	a	a	DET
fcis-25108	57	4	variable	variable	ADJ
fcis-25108	57	5	topology	topology	NOUN
fcis-25108	57	6	and	and	CCONJ
fcis-25108	57	7	shares	share	VERB
fcis-25108	57	8	weights	weight	NOUN
fcis-25108	57	9	.	.	PUNCT
fcis-25108	58	1	it	it	PRON
fcis-25108	58	2	is	be	AUX
fcis-25108	58	3	mostly	mostly	ADV
fcis-25108	58	4	used	use	VERB
fcis-25108	58	5	in	in	ADP
fcis-25108	58	6	machine	machine	NOUN
fcis-25108	58	7	learning	learning	NOUN
fcis-25108	58	8	tasks	task	NOUN
fcis-25108	58	9	containing	contain	VERB
fcis-25108	58	10	structural	structural	ADJ
fcis-25108	58	11	relationships	relationship	NOUN
fcis-25108	58	12	and	and	CCONJ
fcis-25108	58	13	has	have	AUX
fcis-25108	58	14	attracted	attract	VERB
fcis-25108	58	15	the	the	DET
fcis-25108	58	16	attention	attention	NOUN
fcis-25108	58	17	of	of	ADP
fcis-25108	58	18	researchers	researcher	NOUN
fcis-25108	58	19	in	in	ADP
fcis-25108	58	20	the	the	DET
fcis-25108	58	21	field	field	NOUN
fcis-25108	58	22	of	of	ADP
fcis-25108	58	23	nlp	nlp	NOUN
fcis-25108	58	24	.	.	PUNCT
fcis-25108	59	1	the	the	DET
fcis-25108	59	2	basic	basic	ADJ
fcis-25108	59	3	structure	structure	NOUN
fcis-25108	59	4	of	of	ADP
fcis-25108	59	5	rnn	rnn	PROPN
fcis-25108	59	6	includes	include	VERB
fcis-25108	59	7	input	input	NOUN
fcis-25108	59	8	layer	layer	NOUN
fcis-25108	59	9	,	,	PUNCT
fcis-25108	59	10	hidden	hide	VERB
fcis-25108	59	11	layer	layer	NOUN
fcis-25108	59	12	and	and	CCONJ
fcis-25108	59	13	output	output	NOUN
fcis-25108	59	14	layer	layer	NOUN
fcis-25108	59	15	.	.	PUNCT
fcis-25108	60	1	the	the	DET
fcis-25108	60	2	biggest	big	ADJ
fcis-25108	60	3	difference	difference	NOUN
fcis-25108	60	4	between	between	ADP
fcis-25108	60	5	rnn	rnn	NOUN
fcis-25108	60	6	and	and	CCONJ
fcis-25108	60	7	traditional	traditional	ADJ
fcis-25108	60	8	neural	neural	ADJ
fcis-25108	60	9	network	network	NOUN
fcis-25108	60	10	is	be	AUX
fcis-25108	60	11	that	that	SCONJ
fcis-25108	60	12	rnn	rnn	NOUN
fcis-25108	60	13	will	will	AUX
fcis-25108	60	14	send	send	VERB
fcis-25108	60	15	the	the	DET
fcis-25108	60	16	output	output	NOUN
fcis-25108	60	17	of	of	ADP
fcis-25108	60	18	the	the	DET
fcis-25108	60	19	previous	previous	ADJ
fcis-25108	60	20	word	word	NOUN
fcis-25108	60	21	into	into	ADP
fcis-25108	60	22	the	the	DET
fcis-25108	60	23	hidden	hide	VERB
fcis-25108	60	24	layer	layer	NOUN
fcis-25108	60	25	of	of	ADP
fcis-25108	60	26	the	the	DET
fcis-25108	60	27	next	next	ADJ
fcis-25108	60	28	word	word	NOUN
fcis-25108	60	29	to	to	PART
fcis-25108	60	30	train	train	VERB
fcis-25108	60	31	together	together	ADV
fcis-25108	60	32	,	,	PUNCT
fcis-25108	60	33	and	and	CCONJ
fcis-25108	60	34	finally	finally	ADV
fcis-25108	60	35	only	only	ADV
fcis-25108	60	36	output	output	VERB
fcis-25108	60	37	the	the	DET
fcis-25108	60	38	calculation	calculation	NOUN
fcis-25108	60	39	result	result	NOUN
fcis-25108	60	40	of	of	ADP
fcis-25108	60	41	the	the	DET
fcis-25108	60	42	last	last	ADJ
fcis-25108	60	43	word	word	NOUN
fcis-25108	60	44	.	.	PUNCT
fcis-25108	61	1	at	at	ADP
fcis-25108	61	2	present	present	ADJ
fcis-25108	61	3	,	,	PUNCT
fcis-25108	61	4	rnn	rnn	PROPN
fcis-25108	61	5	has	have	VERB
fcis-25108	61	6	a	a	DET
fcis-25108	61	7	great	great	ADJ
fcis-25108	61	8	influence	influence	NOUN
fcis-25108	61	9	on	on	ADP
fcis-25108	61	10	short	short	ADJ
fcis-25108	61	11	-	-	PUNCT
fcis-25108	61	12	term	term	NOUN
fcis-25108	61	13	memory	memory	NOUN
fcis-25108	61	14	,	,	PUNCT
fcis-25108	61	15	but	but	CCONJ
fcis-25108	61	16	has	have	VERB
fcis-25108	61	17	little	little	ADJ
fcis-25108	61	18	effect	effect	NOUN
fcis-25108	61	19	on	on	ADP
fcis-25108	61	20	long	long	ADJ
fcis-25108	61	21	-	-	PUNCT
fcis-25108	61	22	term	term	NOUN
fcis-25108	61	23	memory	memory	NOUN
fcis-25108	61	24	,	,	PUNCT
fcis-25108	61	25	which	which	PRON
fcis-25108	61	26	decides	decide	VERB
fcis-25108	61	27	the	the	DET
fcis-25108	61	28	predicament	predicament	NOUN
fcis-25108	61	29	of	of	ADP
fcis-25108	61	30	the	the	DET
fcis-25108	61	31	incompetence	incompetence	NOUN
fcis-25108	61	32	in	in	ADP
fcis-25108	61	33	dealing	deal	VERB
fcis-25108	61	34	with	with	ADP
fcis-25108	61	35	input	input	NOUN
fcis-25108	61	36	sequences	sequence	NOUN
fcis-25108	61	37	with	with	ADP
fcis-25108	61	38	large	large	ADJ
fcis-25108	61	39	length	length	NOUN
fcis-25108	61	40	.	.	PUNCT
fcis-25108	62	1	and	and	CCONJ
fcis-25108	62	2	in	in	ADP
fcis-25108	62	3	the	the	DET
fcis-25108	62	4	process	process	NOUN
fcis-25108	62	5	of	of	ADP
fcis-25108	62	6	back	back	ADJ
fcis-25108	62	7	propagation	propagation	NOUN
fcis-25108	62	8	,	,	PUNCT
fcis-25108	62	9	the	the	DET
fcis-25108	62	10	gradient	gradient	ADJ
fcis-25108	62	11	multiplication	multiplication	NOUN
fcis-25108	62	12	is	be	AUX
fcis-25108	62	13	involved	involve	VERB
fcis-25108	62	14	in	in	ADP
fcis-25108	62	15	the	the	DET
fcis-25108	62	16	process	process	NOUN
fcis-25108	62	17	of	of	ADP
fcis-25108	62	18	finding	find	VERB
fcis-25108	62	19	the	the	DET
fcis-25108	62	20	parameter	parameter	NOUN
fcis-25108	62	21	gradient	gradient	NOUN
fcis-25108	62	22	of	of	ADP
fcis-25108	62	23	the	the	DET
fcis-25108	62	24	bottom	bottom	ADJ
fcis-25108	62	25	layer	layer	NOUN
fcis-25108	62	26	,	,	PUNCT
fcis-25108	62	27	which	which	PRON
fcis-25108	62	28	is	be	AUX
fcis-25108	62	29	prone	prone	ADJ
fcis-25108	62	30	to	to	ADP
fcis-25108	62	31	the	the	DET
fcis-25108	62	32	problem	problem	NOUN
fcis-25108	62	33	of	of	ADP
fcis-25108	62	34	gradient	gradient	ADJ
fcis-25108	62	35	disappearance	disappearance	NOUN
fcis-25108	62	36	or	or	CCONJ
fcis-25108	62	37	gradient	gradient	NOUN
fcis-25108	62	38	explosion	explosion	NOUN
fcis-25108	62	39	.	.	PUNCT
fcis-25108	63	1	long	long	ADJ
fcis-25108	63	2	short	short	ADJ
fcis-25108	63	3	-	-	PUNCT
fcis-25108	63	4	term	term	NOUN
fcis-25108	63	5	memory	memory	NOUN
fcis-25108	63	6	(	(	PUNCT
fcis-25108	63	7	lstm	lstm	NOUN
fcis-25108	63	8	)	)	PUNCT
fcis-25108	63	9	network	network	NOUN
fcis-25108	63	10	and	and	CCONJ
fcis-25108	63	11	gated	gate	VERB
fcis-25108	63	12	recurrent	recurrent	ADJ
fcis-25108	63	13	unit	unit	NOUN
fcis-25108	63	14	(	(	PUNCT
fcis-25108	63	15	gru	gru	PROPN
fcis-25108	63	16	)	)	PUNCT
fcis-25108	63	17	can	can	AUX
fcis-25108	63	18	solve	solve	VERB
fcis-25108	63	19	this	this	DET
fcis-25108	63	20	problem	problem	NOUN
fcis-25108	63	21	to	to	ADP
fcis-25108	63	22	a	a	DET
fcis-25108	63	23	certain	certain	ADJ
fcis-25108	63	24	extent	extent	NOUN
fcis-25108	63	25	.	.	PUNCT
fcis-25108	64	1	3	3	X
fcis-25108	64	2	.	.	X
fcis-25108	64	3	application	application	NOUN
fcis-25108	64	4	of	of	ADP
fcis-25108	64	5	deep	deep	ADJ
fcis-25108	64	6	learning	learning	NOUN
fcis-25108	64	7	in	in	ADP
fcis-25108	64	8	nlp	nlp	NOUN
fcis-25108	64	9	3.1	3.1	NUM
fcis-25108	64	10	.	.	PUNCT
fcis-25108	64	11	machine	machine	NOUN
fcis-25108	64	12	translation	translation	NOUN
fcis-25108	64	13	machine	machine	NOUN
fcis-25108	64	14	translation	translation	NOUN
fcis-25108	64	15	(	(	PUNCT
fcis-25108	64	16	mt	mt	PROPN
fcis-25108	64	17	)	)	PUNCT
fcis-25108	64	18	refers	refer	VERB
fcis-25108	64	19	to	to	ADP
fcis-25108	64	20	the	the	DET
fcis-25108	64	21	process	process	NOUN
fcis-25108	64	22	of	of	ADP
fcis-25108	64	23	using	use	VERB
fcis-25108	64	24	a	a	DET
fcis-25108	64	25	machine	machine	NOUN
fcis-25108	64	26	to	to	PART
fcis-25108	64	27	translate	translate	VERB
fcis-25108	64	28	a	a	DET
fcis-25108	64	29	natural	natural	ADJ
fcis-25108	64	30	language	language	NOUN
fcis-25108	64	31	in	in	ADP
fcis-25108	64	32	a	a	DET
fcis-25108	64	33	written	write	VERB
fcis-25108	64	34	form	form	NOUN
fcis-25108	64	35	or	or	CCONJ
fcis-25108	64	36	sound	sound	ADJ
fcis-25108	64	37	form	form	NOUN
fcis-25108	64	38	into	into	ADP
fcis-25108	64	39	a	a	DET
fcis-25108	64	40	natural	natural	ADJ
fcis-25108	64	41	language	language	NOUN
fcis-25108	64	42	in	in	ADP
fcis-25108	64	43	another	another	DET
fcis-25108	64	44	written	write	VERB
fcis-25108	64	45	form	form	NOUN
fcis-25108	64	46	or	or	CCONJ
fcis-25108	64	47	sound	sound	ADJ
fcis-25108	64	48	form	form	NOUN
fcis-25108	64	49	through	through	ADP
fcis-25108	64	50	a	a	DET
fcis-25108	64	51	specific	specific	ADJ
fcis-25108	64	52	computer	computer	NOUN
fcis-25108	64	53	program	program	NOUN
fcis-25108	64	54	.	.	PUNCT
fcis-25108	65	1	with	with	ADP
fcis-25108	65	2	the	the	DET
fcis-25108	65	3	development	development	NOUN
fcis-25108	65	4	of	of	ADP
fcis-25108	65	5	dl	dl	PROPN
fcis-25108	65	6	,	,	PUNCT
fcis-25108	65	7	the	the	DET
fcis-25108	65	8	quality	quality	NOUN
fcis-25108	65	9	of	of	ADP
fcis-25108	65	10	mt	mt	PROPN
fcis-25108	65	11	has	have	AUX
fcis-25108	65	12	been	be	AUX
fcis-25108	65	13	significantly	significantly	ADV
fcis-25108	65	14	improved	improve	VERB
fcis-25108	65	15	,	,	PUNCT
fcis-25108	65	16	and	and	CCONJ
fcis-25108	65	17	it	it	PRON
fcis-25108	65	18	has	have	AUX
fcis-25108	65	19	been	be	AUX
fcis-25108	65	20	able	able	ADJ
fcis-25108	65	21	to	to	PART
fcis-25108	65	22	handle	handle	VERB
fcis-25108	65	23	more	more	ADV
fcis-25108	65	24	complex	complex	ADJ
fcis-25108	65	25	and	and	CCONJ
fcis-25108	65	26	natural	natural	ADJ
fcis-25108	65	27	language	language	NOUN
fcis-25108	65	28	expressions	expression	NOUN
fcis-25108	65	29	.	.	PUNCT
fcis-25108	66	1	the	the	DET
fcis-25108	66	2	sequence	sequence	NOUN
fcis-25108	66	3	-	-	PUNCT
fcis-25108	66	4	to	to	ADP
fcis-25108	66	5	-	-	PUNCT
fcis-25108	66	6	sequence	sequence	NOUN
fcis-25108	66	7	(	(	PUNCT
fcis-25108	66	8	seq2seq	seq2seq	NOUN
fcis-25108	66	9	)	)	PUNCT
fcis-25108	66	10	model	model	NOUN
fcis-25108	66	11	is	be	AUX
fcis-25108	66	12	one	one	NUM
fcis-25108	66	13	of	of	ADP
fcis-25108	66	14	the	the	DET
fcis-25108	66	15	applications	application	NOUN
fcis-25108	66	16	of	of	ADP
fcis-25108	66	17	dl	dl	PROPN
fcis-25108	66	18	in	in	ADP
fcis-25108	66	19	mt	mt	PROPN
fcis-25108	66	20	.	.	PUNCT
fcis-25108	67	1	it	it	PRON
fcis-25108	67	2	is	be	AUX
fcis-25108	67	3	designed	design	VERB
fcis-25108	67	4	to	to	PART
fcis-25108	67	5	directly	directly	ADV
fcis-25108	67	6	establish	establish	VERB
fcis-25108	67	7	a	a	DET
fcis-25108	67	8	mapping	mapping	NOUN
fcis-25108	67	9	between	between	ADP
fcis-25108	67	10	the	the	DET
fcis-25108	67	11	input	input	NOUN
fcis-25108	67	12	and	and	CCONJ
fcis-25108	67	13	output	output	NOUN
fcis-25108	67	14	sequences	sequence	NOUN
fcis-25108	67	15	,	,	PUNCT
fcis-25108	67	16	without	without	ADP
fcis-25108	67	17	the	the	DET
fcis-25108	67	18	need	need	NOUN
fcis-25108	67	19	for	for	ADP
fcis-25108	67	20	complex	complex	ADJ
fcis-25108	67	21	feature	feature	NOUN
fcis-25108	67	22	engineering	engineering	NOUN
fcis-25108	67	23	or	or	CCONJ
fcis-25108	67	24	intermediate	intermediate	ADJ
fcis-25108	67	25	representation	representation	NOUN
fcis-25108	67	26	.	.	PUNCT
fcis-25108	68	1	seq2seq	seq2seq	NOUN
fcis-25108	68	2	includes	include	VERB
fcis-25108	68	3	an	an	DET
fcis-25108	68	4	encoder	encoder	NOUN
fcis-25108	68	5	responsible	responsible	ADJ
fcis-25108	68	6	for	for	ADP
fcis-25108	68	7	understanding	understand	VERB
fcis-25108	68	8	the	the	DET
fcis-25108	68	9	input	input	NOUN
fcis-25108	68	10	language	language	NOUN
fcis-25108	68	11	and	and	CCONJ
fcis-25108	68	12	a	a	DET
fcis-25108	68	13	decoder	decoder	NOUN
fcis-25108	68	14	responsible	responsible	ADJ
fcis-25108	68	15	for	for	ADP
fcis-25108	68	16	translating	translate	VERB
fcis-25108	68	17	the	the	DET
fcis-25108	68	18	target	target	NOUN
fcis-25108	68	19	language	language	NOUN
fcis-25108	68	20	.	.	PUNCT
fcis-25108	69	1	this	this	DET
fcis-25108	69	2	end	end	NOUN
fcis-25108	69	3	-	-	PUNCT
fcis-25108	69	4	to	to	ADP
fcis-25108	69	5	-	-	PUNCT
fcis-25108	69	6	end	end	NOUN
fcis-25108	69	7	training	training	NOUN
fcis-25108	69	8	method	method	NOUN
fcis-25108	69	9	enables	enable	VERB
fcis-25108	69	10	the	the	DET
fcis-25108	69	11	model	model	NOUN
fcis-25108	69	12	to	to	PART
fcis-25108	69	13	capture	capture	VERB
fcis-25108	69	14	rich	rich	ADJ
fcis-25108	69	15	language	language	NOUN
fcis-25108	69	16	patterns	pattern	NOUN
fcis-25108	69	17	and	and	CCONJ
fcis-25108	69	18	context	context	NOUN
fcis-25108	69	19	information	information	NOUN
fcis-25108	69	20	on	on	ADP
fcis-25108	69	21	a	a	DET
fcis-25108	69	22	large	large	ADJ
fcis-25108	69	23	number	number	NOUN
fcis-25108	69	24	of	of	ADP
fcis-25108	69	25	bilingual	bilingual	ADJ
fcis-25108	69	26	data	datum	NOUN
fcis-25108	69	27	.	.	PUNCT
fcis-25108	70	1	different	different	ADJ
fcis-25108	70	2	from	from	ADP
fcis-25108	70	3	the	the	DET
fcis-25108	70	4	seq2seq	seq2seq	PROPN
fcis-25108	70	5	model	model	NOUN
fcis-25108	70	6	,	,	PUNCT
fcis-25108	70	7	transformer	transformer	NOUN
fcis-25108	70	8	relies	rely	VERB
fcis-25108	70	9	entirely	entirely	ADV
fcis-25108	70	10	on	on	ADP
fcis-25108	70	11	the	the	DET
fcis-25108	70	12	self	self	NOUN
fcis-25108	70	13	-	-	PUNCT
fcis-25108	70	14	attention	attention	NOUN
fcis-25108	70	15	mechanism	mechanism	NOUN
fcis-25108	70	16	,	,	PUNCT
fcis-25108	70	17	which	which	PRON
fcis-25108	70	18	allows	allow	VERB
fcis-25108	70	19	the	the	DET
fcis-25108	70	20	model	model	NOUN
fcis-25108	70	21	to	to	PART
fcis-25108	70	22	process	process	VERB
fcis-25108	70	23	input	input	NOUN
fcis-25108	70	24	data	datum	NOUN
fcis-25108	70	25	in	in	ADP
fcis-25108	70	26	parallel	parallel	ADJ
fcis-25108	70	27	and	and	CCONJ
fcis-25108	70	28	better	well	ADJ
fcis-25108	70	29	capture	capture	VERB
fcis-25108	70	30	long	long	ADJ
fcis-25108	70	31	-	-	PUNCT
fcis-25108	70	32	distance	distance	NOUN
fcis-25108	70	33	dependencies	dependency	NOUN
fcis-25108	70	34	in	in	ADP
fcis-25108	70	35	the	the	DET
fcis-25108	70	36	sequence	sequence	NOUN
fcis-25108	70	37	.	.	PUNCT
fcis-25108	71	1	this	this	PRON
fcis-25108	71	2	brings	bring	VERB
fcis-25108	71	3	higher	high	ADJ
fcis-25108	71	4	efficiency	efficiency	NOUN
fcis-25108	71	5	and	and	CCONJ
fcis-25108	71	6	better	well	ADJ
fcis-25108	71	7	performance	performance	NOUN
fcis-25108	71	8	to	to	ADP
fcis-25108	71	9	the	the	DET
fcis-25108	71	10	model	model	NOUN
fcis-25108	71	11	and	and	CCONJ
fcis-25108	71	12	becomes	become	VERB
fcis-25108	71	13	the	the	DET
fcis-25108	71	14	core	core	NOUN
fcis-25108	71	15	of	of	ADP
fcis-25108	71	16	many	many	ADJ
fcis-25108	71	17	leading	lead	VERB
fcis-25108	71	18	translation	translation	NOUN
fcis-25108	71	19	systems	system	NOUN
fcis-25108	71	20	today	today	NOUN
fcis-25108	71	21	.	.	PUNCT
fcis-25108	72	1	3.2	3.2	NUM
fcis-25108	72	2	.	.	PUNCT
fcis-25108	72	3	sentiment	sentiment	NOUN
fcis-25108	72	4	analysis	analysis	NOUN
fcis-25108	72	5	sentiment	sentiment	NOUN
fcis-25108	72	6	analysis	analysis	NOUN
fcis-25108	72	7	,	,	PUNCT
fcis-25108	72	8	also	also	ADV
fcis-25108	72	9	known	know	VERB
fcis-25108	72	10	as	as	ADP
fcis-25108	72	11	tendency	tendency	NOUN
fcis-25108	72	12	analysis	analysis	NOUN
fcis-25108	72	13	,	,	PUNCT
fcis-25108	72	14	aims	aim	VERB
fcis-25108	72	15	to	to	PART
fcis-25108	72	16	analyze	analyze	VERB
fcis-25108	72	17	and	and	CCONJ
fcis-25108	72	18	process	process	VERB
fcis-25108	72	19	the	the	DET
fcis-25108	72	20	text	text	NOUN
fcis-25108	72	21	with	with	ADP
fcis-25108	72	22	subjective	subjective	ADJ
fcis-25108	72	23	emotion	emotion	NOUN
fcis-25108	72	24	[	[	X
fcis-25108	72	25	5	5	NUM
fcis-25108	72	26	]	]	PUNCT
fcis-25108	72	27	.	.	PUNCT
fcis-25108	73	1	different	different	ADJ
fcis-25108	73	2	from	from	ADP
fcis-25108	73	3	the	the	DET
fcis-25108	73	4	traditional	traditional	ADJ
fcis-25108	73	5	sentiment	sentiment	NOUN
fcis-25108	73	6	analysis	analysis	NOUN
fcis-25108	73	7	methods	method	NOUN
fcis-25108	73	8	that	that	PRON
fcis-25108	73	9	have	have	VERB
fcis-25108	73	10	the	the	DET
fcis-25108	73	11	problem	problem	NOUN
fcis-25108	73	12	of	of	ADP
fcis-25108	73	13	semantic	semantic	ADJ
fcis-25108	73	14	loss	loss	NOUN
fcis-25108	73	15	,	,	PUNCT
fcis-25108	73	16	over	over	ADP
fcis-25108	73	17	-	-	PUNCT
fcis-25108	73	18	reliance	reliance	NOUN
fcis-25108	73	19	on	on	ADP
fcis-25108	73	20	prior	prior	ADJ
fcis-25108	73	21	background	background	NOUN
fcis-25108	73	22	knowledge	knowledge	NOUN
fcis-25108	73	23	,	,	PUNCT
fcis-25108	73	24	and	and	CCONJ
fcis-25108	73	25	more	more	ADJ
fcis-25108	73	26	requirement	requirement	NOUN
fcis-25108	73	27	on	on	ADP
fcis-25108	73	28	manual	manual	ADJ
fcis-25108	73	29	intervention	intervention	NOUN
fcis-25108	73	30	,	,	PUNCT
fcis-25108	73	31	dl	dl	PROPN
fcis-25108	73	32	provides	provide	VERB
fcis-25108	73	33	a	a	DET
fcis-25108	73	34	method	method	NOUN
fcis-25108	73	35	to	to	PART
fcis-25108	73	36	learn	learn	VERB
fcis-25108	73	37	and	and	CCONJ
fcis-25108	73	38	extract	extract	VERB
fcis-25108	73	39	features	feature	VERB
fcis-25108	73	40	directly	directly	ADV
fcis-25108	73	41	from	from	ADP
fcis-25108	73	42	data	data	PROPN
fcis-25108	73	43	.	.	PUNCT
fcis-25108	74	1	cnn	cnn	PROPN
fcis-25108	74	2	's	's	PART
fcis-25108	74	3	multiple	multiple	ADJ
fcis-25108	74	4	convolutional	convolutional	ADJ
fcis-25108	74	5	layers	layer	NOUN
fcis-25108	74	6	are	be	AUX
fcis-25108	74	7	able	able	ADJ
fcis-25108	74	8	to	to	PART
fcis-25108	74	9	extract	extract	VERB
fcis-25108	74	10	more	more	ADJ
fcis-25108	74	11	complex	complex	ADJ
fcis-25108	74	12	text	text	NOUN
fcis-25108	74	13	features	feature	VERB
fcis-25108	74	14	layer	layer	NOUN
fcis-25108	74	15	by	by	ADP
fcis-25108	74	16	layer	layer	NOUN
fcis-25108	74	17	to	to	PART
fcis-25108	74	18	form	form	VERB
fcis-25108	74	19	a	a	DET
fcis-25108	74	20	hierarchical	hierarchical	ADJ
fcis-25108	74	21	text	text	NOUN
fcis-25108	74	22	representation	representation	NOUN
fcis-25108	74	23	,	,	PUNCT
fcis-25108	74	24	which	which	PRON
fcis-25108	74	25	is	be	AUX
fcis-25108	74	26	achieved	achieve	VERB
fcis-25108	74	27	by	by	ADP
fcis-25108	74	28	capturing	capture	VERB
fcis-25108	74	29	the	the	DET
fcis-25108	74	30	n	n	CCONJ
fcis-25108	74	31	-	-	PUNCT
fcis-25108	74	32	gram	gram	NOUN
fcis-25108	74	33	features	feature	NOUN
fcis-25108	74	34	in	in	ADP
fcis-25108	74	35	the	the	DET
fcis-25108	74	36	text	text	NOUN
fcis-25108	74	37	through	through	ADP
fcis-25108	74	38	a	a	DET
fcis-25108	74	39	sliding	slide	VERB
fcis-25108	74	40	window	window	NOUN
fcis-25108	74	41	and	and	CCONJ
fcis-25108	74	42	identifying	identify	VERB
fcis-25108	74	43	keywords	keyword	NOUN
fcis-25108	74	44	or	or	CCONJ
fcis-25108	74	45	phrases	phrase	NOUN
fcis-25108	74	46	related	relate	VERB
fcis-25108	74	47	to	to	ADP
fcis-25108	74	48	specific	specific	ADJ
fcis-25108	74	49	emotions	emotion	NOUN
fcis-25108	74	50	.	.	PUNCT
fcis-25108	75	1	collobert	collobert	VERB
fcis-25108	75	2	et	et	PROPN
fcis-25108	75	3	al	al	PROPN
fcis-25108	75	4	.	.	PROPN
fcis-25108	75	5	proposed	propose	VERB
fcis-25108	75	6	a	a	DET
fcis-25108	75	7	general	general	ADJ
fcis-25108	75	8	framework	framework	NOUN
fcis-25108	75	9	based	base	VERB
fcis-25108	75	10	on	on	ADP
fcis-25108	75	11	cnn	cnn	PROPN
fcis-25108	75	12	to	to	PART
fcis-25108	75	13	solve	solve	VERB
fcis-25108	75	14	a	a	DET
fcis-25108	75	15	large	large	ADJ
fcis-25108	75	16	number	number	NOUN
fcis-25108	75	17	of	of	ADP
fcis-25108	75	18	nlp	nlp	ADJ
fcis-25108	75	19	tasks	task	NOUN
fcis-25108	75	20	in	in	ADP
fcis-25108	75	21	2011[6	2011[6	NUM
fcis-25108	75	22	]	]	PUNCT
fcis-25108	75	23	.	.	PUNCT
fcis-25108	76	1	then	then	ADV
fcis-25108	76	2	in	in	ADP
fcis-25108	76	3	2014	2014	NUM
fcis-25108	76	4	,	,	PUNCT
fcis-25108	76	5	kalchbrenner	kalchbrenner	NOUN
fcis-25108	76	6	et	et	PROPN
fcis-25108	76	7	al	al	PROPN
fcis-25108	76	8	.	.	PROPN
fcis-25108	76	9	successfully	successfully	ADV
fcis-25108	76	10	used	use	VERB
fcis-25108	76	11	cnn	cnn	PROPN
fcis-25108	76	12	to	to	PART
fcis-25108	76	13	extract	extract	VERB
fcis-25108	76	14	significant	significant	ADJ
fcis-25108	76	15	n	n	CCONJ
fcis-25108	76	16	-	-	PUNCT
fcis-25108	76	17	gram	gram	NOUN
fcis-25108	76	18	features	feature	NOUN
fcis-25108	76	19	from	from	ADP
fcis-25108	76	20	input	input	NOUN
fcis-25108	76	21	sentences	sentence	NOUN
fcis-25108	76	22	[	[	X
fcis-25108	76	23	7	7	NUM
fcis-25108	76	24	]	]	PUNCT
fcis-25108	76	25	.	.	PUNCT
fcis-25108	77	1	rnn	rnn	PROPN
fcis-25108	77	2	and	and	CCONJ
fcis-25108	77	3	its	its	PRON
fcis-25108	77	4	extension	extension	NOUN
fcis-25108	77	5	,	,	PUNCT
fcis-25108	77	6	such	such	ADJ
fcis-25108	77	7	as	as	ADP
fcis-25108	77	8	58	58	NUM
fcis-25108	77	9	lstm	lstm	NOUN
fcis-25108	77	10	and	and	CCONJ
fcis-25108	77	11	gru	gru	PROPN
fcis-25108	77	12	,	,	PUNCT
fcis-25108	77	13	are	be	AUX
fcis-25108	77	14	designed	design	VERB
fcis-25108	77	15	to	to	PART
fcis-25108	77	16	solve	solve	VERB
fcis-25108	77	17	the	the	DET
fcis-25108	77	18	problem	problem	NOUN
fcis-25108	77	19	of	of	ADP
fcis-25108	77	20	longdistance	longdistance	NOUN
fcis-25108	77	21	dependency	dependency	NOUN
fcis-25108	77	22	in	in	ADP
fcis-25108	77	23	text	text	NOUN
fcis-25108	77	24	.	.	PUNCT
fcis-25108	78	1	the	the	DET
fcis-25108	78	2	emotion	emotion	NOUN
fcis-25108	78	3	in	in	ADP
fcis-25108	78	4	the	the	DET
fcis-25108	78	5	text	text	NOUN
fcis-25108	78	6	often	often	ADV
fcis-25108	78	7	depends	depend	VERB
fcis-25108	78	8	not	not	PART
fcis-25108	78	9	only	only	ADV
fcis-25108	78	10	on	on	ADP
fcis-25108	78	11	the	the	DET
fcis-25108	78	12	local	local	ADJ
fcis-25108	78	13	vocabulary	vocabulary	ADJ
fcis-25108	78	14	selection	selection	NOUN
fcis-25108	78	15	,	,	PUNCT
fcis-25108	78	16	but	but	CCONJ
fcis-25108	78	17	also	also	ADV
fcis-25108	78	18	on	on	ADP
fcis-25108	78	19	the	the	DET
fcis-25108	78	20	context	context	NOUN
fcis-25108	78	21	,	,	PUNCT
fcis-25108	78	22	the	the	DET
fcis-25108	78	23	information	information	NOUN
fcis-25108	78	24	in	in	ADP
fcis-25108	78	25	the	the	DET
fcis-25108	78	26	front	front	ADJ
fcis-25108	78	27	and	and	CCONJ
fcis-25108	78	28	back	back	ADJ
fcis-25108	78	29	sentences	sentence	NOUN
fcis-25108	78	30	.	.	PUNCT
fcis-25108	79	1	in	in	ADP
fcis-25108	79	2	2011	2011	NUM
fcis-25108	79	3	,	,	PUNCT
fcis-25108	79	4	mikolov	mikolov	PROPN
fcis-25108	79	5	et	et	PROPN
fcis-25108	79	6	al	al	PROPN
fcis-25108	79	7	.	.	PROPN
fcis-25108	79	8	successfully	successfully	ADV
fcis-25108	79	9	used	use	VERB
fcis-25108	79	10	rnn	rnn	NOUN
fcis-25108	79	11	for	for	ADP
fcis-25108	79	12	language	language	NOUN
fcis-25108	79	13	modeling	modeling	NOUN
fcis-25108	79	14	[	[	X
fcis-25108	79	15	8	8	NUM
fcis-25108	79	16	]	]	PUNCT
fcis-25108	79	17	.	.	PUNCT
fcis-25108	80	1	sutskever	sutskever	PROPN
fcis-25108	80	2	et	et	PROPN
fcis-25108	80	3	al	al	PROPN
fcis-25108	80	4	.	.	PROPN
fcis-25108	80	5	proposed	propose	VERB
fcis-25108	80	6	a	a	DET
fcis-25108	80	7	general	general	ADJ
fcis-25108	80	8	deep	deep	ADJ
fcis-25108	80	9	lstm	lstm	ADJ
fcis-25108	80	10	encoder	encoder	NOUN
fcis-25108	80	11	-	-	PUNCT
fcis-25108	80	12	decoder	decoder	NOUN
fcis-25108	80	13	framework	framework	NOUN
fcis-25108	80	14	that	that	PRON
fcis-25108	80	15	enables	enable	VERB
fcis-25108	80	16	mapping	mapping	NOUN
fcis-25108	80	17	between	between	ADP
fcis-25108	80	18	sequences	sequence	NOUN
fcis-25108	80	19	in	in	ADP
fcis-25108	80	20	2014[9	2014[9	NUM
fcis-25108	80	21	]	]	PUNCT
fcis-25108	80	22	.	.	PUNCT
fcis-25108	81	1	3.3	3.3	NUM
fcis-25108	81	2	.	.	PUNCT
fcis-25108	82	1	question	question	NOUN
fcis-25108	82	2	answering	answering	NOUN
fcis-25108	82	3	system	system	NOUN
fcis-25108	82	4	the	the	DET
fcis-25108	82	5	question	question	NOUN
fcis-25108	82	6	answering	answer	VERB
fcis-25108	82	7	system	system	NOUN
fcis-25108	82	8	is	be	AUX
fcis-25108	82	9	designed	design	VERB
fcis-25108	82	10	to	to	PART
fcis-25108	82	11	provide	provide	VERB
fcis-25108	82	12	users	user	NOUN
fcis-25108	82	13	with	with	ADP
fcis-25108	82	14	accurate	accurate	ADJ
fcis-25108	82	15	answers	answer	NOUN
fcis-25108	82	16	.	.	PUNCT
fcis-25108	83	1	driven	drive	VERB
fcis-25108	83	2	by	by	ADP
fcis-25108	83	3	dl	dl	PROPN
fcis-25108	83	4	technology	technology	PROPN
fcis-25108	83	5	,	,	PUNCT
fcis-25108	83	6	the	the	DET
fcis-25108	83	7	ability	ability	NOUN
fcis-25108	83	8	of	of	ADP
fcis-25108	83	9	question	question	NOUN
fcis-25108	83	10	answering	answering	NOUN
fcis-25108	83	11	system	system	NOUN
fcis-25108	83	12	has	have	AUX
fcis-25108	83	13	been	be	AUX
fcis-25108	83	14	significantly	significantly	ADV
fcis-25108	83	15	enhanced	enhance	VERB
fcis-25108	83	16	,	,	PUNCT
fcis-25108	83	17	especially	especially	ADV
fcis-25108	83	18	when	when	SCONJ
fcis-25108	83	19	dealing	deal	VERB
fcis-25108	83	20	with	with	ADP
fcis-25108	83	21	large	large	ADJ
fcis-25108	83	22	and	and	CCONJ
fcis-25108	83	23	complex	complex	ADJ
fcis-25108	83	24	text	text	NOUN
fcis-25108	83	25	databases	database	NOUN
fcis-25108	83	26	.	.	PUNCT
fcis-25108	84	1	traditional	traditional	ADJ
fcis-25108	84	2	question	question	NOUN
fcis-25108	84	3	answering	answering	NOUN
fcis-25108	84	4	methods	method	NOUN
fcis-25108	84	5	usually	usually	ADV
fcis-25108	84	6	rely	rely	VERB
fcis-25108	84	7	on	on	ADP
fcis-25108	84	8	manually	manually	ADV
fcis-25108	84	9	formulated	formulate	VERB
fcis-25108	84	10	rules	rule	NOUN
fcis-25108	84	11	or	or	CCONJ
fcis-25108	84	12	shallow	shallow	ADJ
fcis-25108	84	13	text	text	NOUN
fcis-25108	84	14	matching	matching	NOUN
fcis-25108	84	15	techniques	technique	NOUN
fcis-25108	84	16	,	,	PUNCT
fcis-25108	84	17	but	but	CCONJ
fcis-25108	84	18	these	these	DET
fcis-25108	84	19	methods	method	NOUN
fcis-25108	84	20	lack	lack	VERB
fcis-25108	84	21	the	the	DET
fcis-25108	84	22	ability	ability	NOUN
fcis-25108	84	23	to	to	PART
fcis-25108	84	24	deal	deal	VERB
fcis-25108	84	25	with	with	ADP
fcis-25108	84	26	diverse	diverse	ADJ
fcis-25108	84	27	and	and	CCONJ
fcis-25108	84	28	complex	complex	ADJ
fcis-25108	84	29	problems	problem	NOUN
fcis-25108	84	30	.	.	PUNCT
fcis-25108	85	1	in	in	ADP
fcis-25108	85	2	contrast	contrast	NOUN
fcis-25108	85	3	,	,	PUNCT
fcis-25108	85	4	dl	dl	PROPN
fcis-25108	85	5	provides	provide	VERB
fcis-25108	85	6	a	a	DET
fcis-25108	85	7	way	way	NOUN
fcis-25108	85	8	to	to	PART
fcis-25108	85	9	learn	learn	VERB
fcis-25108	85	10	directly	directly	ADV
fcis-25108	85	11	from	from	ADP
fcis-25108	85	12	data	datum	NOUN
fcis-25108	85	13	,	,	PUNCT
fcis-25108	85	14	and	and	CCONJ
fcis-25108	85	15	to	to	PART
fcis-25108	85	16	understand	understand	VERB
fcis-25108	85	17	text	text	NOUN
fcis-25108	85	18	content	content	NOUN
fcis-25108	85	19	and	and	CCONJ
fcis-25108	85	20	contextual	contextual	ADJ
fcis-25108	85	21	relationships	relationship	NOUN
fcis-25108	85	22	more	more	ADV
fcis-25108	85	23	deeply	deeply	ADV
fcis-25108	85	24	.	.	PUNCT
fcis-25108	86	1	bert	bert	PROPN
fcis-25108	86	2	,	,	PUNCT
fcis-25108	86	3	a	a	DET
fcis-25108	86	4	bidirectional	bidirectional	ADJ
fcis-25108	86	5	deep	deep	ADJ
fcis-25108	86	6	language	language	NOUN
fcis-25108	86	7	model	model	NOUN
fcis-25108	86	8	based	base	VERB
fcis-25108	86	9	on	on	ADP
fcis-25108	86	10	transformer	transformer	NOUN
fcis-25108	86	11	,	,	PUNCT
fcis-25108	86	12	is	be	AUX
fcis-25108	86	13	proposed	propose	VERB
fcis-25108	86	14	by	by	ADP
fcis-25108	86	15	devlin	devlin	PROPN
fcis-25108	86	16	et	et	PROPN
fcis-25108	86	17	al	al	PROPN
fcis-25108	86	18	.	.	PUNCT
fcis-25108	87	1	in	in	ADP
fcis-25108	87	2	2018[10	2018[10	NUM
fcis-25108	87	3	]	]	PUNCT
fcis-25108	87	4	.	.	PUNCT
fcis-25108	88	1	as	as	SCONJ
fcis-25108	88	2	shown	show	VERB
fcis-25108	88	3	in	in	ADP
fcis-25108	88	4	figure	figure	NOUN
fcis-25108	88	5	2	2	NUM
fcis-25108	88	6	,	,	PUNCT
fcis-25108	88	7	the	the	DET
fcis-25108	88	8	model	model	NOUN
fcis-25108	88	9	is	be	AUX
fcis-25108	88	10	composed	compose	VERB
fcis-25108	88	11	of	of	ADP
fcis-25108	88	12	multi	multi	ADJ
fcis-25108	88	13	-	-	ADJ
fcis-25108	88	14	layer	layer	ADJ
fcis-25108	88	15	bidirectional	bidirectional	ADJ
fcis-25108	88	16	transformer	transformer	NOUN
fcis-25108	88	17	decoder	decoder	NOUN
fcis-25108	88	18	.	.	PUNCT
fcis-25108	89	1	it	it	PRON
fcis-25108	89	2	learns	learn	VERB
fcis-25108	89	3	the	the	DET
fcis-25108	89	4	internal	internal	ADJ
fcis-25108	89	5	structure	structure	NOUN
fcis-25108	89	6	of	of	ADP
fcis-25108	89	7	language	language	NOUN
fcis-25108	89	8	by	by	ADP
fcis-25108	89	9	pre	pre	VERB
fcis-25108	89	10	-	-	NOUN
fcis-25108	89	11	training	train	VERB
fcis-25108	89	12	a	a	DET
fcis-25108	89	13	large	large	ADJ
fcis-25108	89	14	number	number	NOUN
fcis-25108	89	15	of	of	ADP
fcis-25108	89	16	unlabeled	unlabeled	ADJ
fcis-25108	89	17	texts	text	NOUN
fcis-25108	89	18	and	and	CCONJ
fcis-25108	89	19	semantics	semantic	NOUN
fcis-25108	89	20	,	,	PUNCT
fcis-25108	89	21	and	and	CCONJ
fcis-25108	89	22	then	then	ADV
fcis-25108	89	23	optimize	optimize	VERB
fcis-25108	89	24	the	the	DET
fcis-25108	89	25	results	result	NOUN
fcis-25108	89	26	through	through	ADP
fcis-25108	89	27	label	label	NOUN
fcis-25108	89	28	data	datum	NOUN
fcis-25108	89	29	for	for	ADP
fcis-25108	89	30	question	question	NOUN
fcis-25108	89	31	answering	answering	NOUN
fcis-25108	89	32	tasks	task	NOUN
fcis-25108	89	33	.	.	PUNCT
fcis-25108	90	1	in	in	ADP
fcis-25108	90	2	this	this	DET
fcis-25108	90	3	case	case	NOUN
fcis-25108	90	4	,	,	PUNCT
fcis-25108	90	5	bert	bert	PROPN
fcis-25108	90	6	can	can	AUX
fcis-25108	90	7	capture	capture	VERB
fcis-25108	90	8	bidirectional	bidirectional	ADJ
fcis-25108	90	9	context	context	NOUN
fcis-25108	90	10	semantics	semantic	NOUN
fcis-25108	90	11	.	.	PUNCT
fcis-25108	91	1	another	another	DET
fcis-25108	91	2	model	model	NOUN
fcis-25108	91	3	named	name	VERB
fcis-25108	91	4	text	text	NOUN
fcis-25108	91	5	-	-	PUNCT
fcis-25108	91	6	to	to	ADP
fcis-25108	91	7	-	-	PUNCT
fcis-25108	91	8	text	text	NOUN
fcis-25108	91	9	transfer	transfer	NOUN
fcis-25108	91	10	transformer(t5	transformer(t5	NOUN
fcis-25108	91	11	)	)	PUNCT
fcis-25108	91	12	adopts	adopt	VERB
fcis-25108	91	13	a	a	DET
fcis-25108	91	14	new	new	ADJ
fcis-25108	91	15	methods[11	methods[11	NOUN
fcis-25108	91	16	]	]	PUNCT
fcis-25108	91	17	.	.	PUNCT
fcis-25108	92	1	it	it	PRON
fcis-25108	92	2	treats	treat	VERB
fcis-25108	92	3	all	all	DET
fcis-25108	92	4	nlp	nlp	NOUN
fcis-25108	92	5	tasks	task	NOUN
fcis-25108	92	6	as	as	ADP
fcis-25108	92	7	text	text	NOUN
fcis-25108	92	8	-	-	PUNCT
fcis-25108	92	9	to	to	ADP
fcis-25108	92	10	-	-	PUNCT
fcis-25108	92	11	text	text	NOUN
fcis-25108	92	12	conversion	conversion	NOUN
fcis-25108	92	13	problems	problem	NOUN
fcis-25108	92	14	.	.	PUNCT
fcis-25108	93	1	t5	t5	PROPN
fcis-25108	93	2	emphasizes	emphasize	VERB
fcis-25108	93	3	the	the	DET
fcis-25108	93	4	combination	combination	NOUN
fcis-25108	93	5	of	of	ADP
fcis-25108	93	6	taskindependent	taskindependent	NOUN
fcis-25108	93	7	pre	pre	NOUN
fcis-25108	93	8	-	-	NOUN
fcis-25108	93	9	training	training	NOUN
fcis-25108	93	10	and	and	CCONJ
fcis-25108	93	11	task	task	NOUN
fcis-25108	93	12	-	-	PUNCT
fcis-25108	93	13	specific	specific	ADJ
fcis-25108	93	14	fine	fine	ADJ
fcis-25108	93	15	-	-	PUNCT
fcis-25108	93	16	tuning	tuning	NOUN
fcis-25108	93	17	,	,	PUNCT
fcis-25108	93	18	which	which	PRON
fcis-25108	93	19	enables	enable	VERB
fcis-25108	93	20	the	the	DET
fcis-25108	93	21	model	model	NOUN
fcis-25108	93	22	to	to	PART
fcis-25108	93	23	show	show	VERB
fcis-25108	93	24	a	a	DET
fcis-25108	93	25	high	high	ADJ
fcis-25108	93	26	degree	degree	NOUN
fcis-25108	93	27	of	of	ADP
fcis-25108	93	28	flexibility	flexibility	NOUN
fcis-25108	93	29	and	and	CCONJ
fcis-25108	93	30	accuracy	accuracy	NOUN
fcis-25108	93	31	in	in	ADP
fcis-25108	93	32	question	question	NOUN
fcis-25108	93	33	and	and	CCONJ
fcis-25108	93	34	answer	answer	VERB
fcis-25108	93	35	scenarios	scenario	NOUN
fcis-25108	93	36	.	.	PUNCT
fcis-25108	94	1	fig	fig	NOUN
fcis-25108	94	2	2	2	NUM
fcis-25108	94	3	.	.	PUNCT
fcis-25108	95	1	bert	bert	PROPN
fcis-25108	95	2	model	model	PROPN
fcis-25108	95	3	4	4	NUM
fcis-25108	95	4	.	.	PUNCT
fcis-25108	95	5	conclusion	conclusion	NOUN
fcis-25108	95	6	this	this	DET
fcis-25108	95	7	paper	paper	NOUN
fcis-25108	95	8	mainly	mainly	ADV
fcis-25108	95	9	introduces	introduce	VERB
fcis-25108	95	10	cnn	cnn	PROPN
fcis-25108	95	11	and	and	CCONJ
fcis-25108	95	12	rnn	rnn	VERB
fcis-25108	95	13	in	in	ADP
fcis-25108	95	14	dl	dl	PROPN
fcis-25108	95	15	,	,	PUNCT
fcis-25108	95	16	and	and	CCONJ
fcis-25108	95	17	expounds	expound	VERB
fcis-25108	95	18	the	the	DET
fcis-25108	95	19	research	research	NOUN
fcis-25108	95	20	progress	progress	NOUN
fcis-25108	95	21	of	of	ADP
fcis-25108	95	22	each	each	DET
fcis-25108	95	23	task	task	NOUN
fcis-25108	95	24	in	in	ADP
fcis-25108	95	25	the	the	DET
fcis-25108	95	26	field	field	NOUN
fcis-25108	95	27	of	of	ADP
fcis-25108	95	28	nlp	nlp	NOUN
fcis-25108	95	29	.	.	PUNCT
fcis-25108	96	1	although	although	SCONJ
fcis-25108	96	2	dl	dl	PROPN
fcis-25108	96	3	has	have	AUX
fcis-25108	96	4	achieved	achieve	VERB
fcis-25108	96	5	great	great	ADJ
fcis-25108	96	6	success	success	NOUN
fcis-25108	96	7	in	in	ADP
fcis-25108	96	8	various	various	ADJ
fcis-25108	96	9	tasks	task	NOUN
fcis-25108	96	10	of	of	ADP
fcis-25108	96	11	nlp	nlp	NOUN
fcis-25108	96	12	,	,	PUNCT
fcis-25108	96	13	there	there	PRON
fcis-25108	96	14	are	be	VERB
fcis-25108	96	15	still	still	ADV
fcis-25108	96	16	difficulties	difficulty	NOUN
fcis-25108	96	17	that	that	PRON
fcis-25108	96	18	need	need	VERB
fcis-25108	96	19	to	to	PART
fcis-25108	96	20	be	be	AUX
fcis-25108	96	21	overcome	overcome	VERB
fcis-25108	96	22	.	.	PUNCT
fcis-25108	97	1	the	the	PRON
fcis-25108	97	2	larger	large	ADJ
fcis-25108	97	3	the	the	DET
fcis-25108	97	4	deep	deep	ADJ
fcis-25108	97	5	neural	neural	ADJ
fcis-25108	97	6	network	network	NOUN
fcis-25108	97	7	model	model	NOUN
fcis-25108	97	8	is	be	AUX
fcis-25108	97	9	,	,	PUNCT
fcis-25108	97	10	the	the	PRON
fcis-25108	97	11	longer	long	ADJ
fcis-25108	97	12	the	the	DET
fcis-25108	97	13	training	training	NOUN
fcis-25108	97	14	time	time	NOUN
fcis-25108	97	15	needs	need	VERB
fcis-25108	97	16	.	.	PUNCT
fcis-25108	98	1	how	how	SCONJ
fcis-25108	98	2	to	to	PART
fcis-25108	98	3	reduce	reduce	VERB
fcis-25108	98	4	the	the	DET
fcis-25108	98	5	model	model	NOUN
fcis-25108	98	6	volume	volume	NOUN
fcis-25108	98	7	while	while	SCONJ
fcis-25108	98	8	maintaining	maintain	VERB
fcis-25108	98	9	the	the	DET
fcis-25108	98	10	performance	performance	NOUN
fcis-25108	98	11	is	be	AUX
fcis-25108	98	12	a	a	DET
fcis-25108	98	13	direction	direction	NOUN
fcis-25108	98	14	for	for	ADP
fcis-25108	98	15	future	future	ADJ
fcis-25108	98	16	research	research	NOUN
fcis-25108	98	17	.	.	PUNCT
fcis-25108	99	1	looking	look	VERB
fcis-25108	99	2	ahead	ahead	ADV
fcis-25108	99	3	,	,	PUNCT
fcis-25108	99	4	dl	dl	PROPN
fcis-25108	99	5	may	may	AUX
fcis-25108	99	6	also	also	ADV
fcis-25108	99	7	be	be	AUX
fcis-25108	99	8	combined	combine	VERB
fcis-25108	99	9	with	with	ADP
fcis-25108	99	10	other	other	ADJ
fcis-25108	99	11	cuttingedge	cuttingedge	NOUN
fcis-25108	99	12	technologies	technology	NOUN
fcis-25108	99	13	such	such	ADJ
fcis-25108	99	14	as	as	ADP
fcis-25108	99	15	quantum	quantum	NOUN
fcis-25108	99	16	computing	computing	NOUN
fcis-25108	99	17	and	and	CCONJ
fcis-25108	99	18	edge	edge	NOUN
fcis-25108	99	19	computing	compute	VERB
fcis-25108	99	20	to	to	PART
fcis-25108	99	21	achieve	achieve	VERB
fcis-25108	99	22	higher	high	ADJ
fcis-25108	99	23	computational	computational	ADJ
fcis-25108	99	24	efficiency	efficiency	NOUN
fcis-25108	99	25	and	and	CCONJ
fcis-25108	99	26	wider	wide	ADJ
fcis-25108	99	27	application	application	NOUN
fcis-25108	99	28	scenarios	scenario	NOUN
fcis-25108	99	29	.	.	PUNCT
fcis-25108	100	1	the	the	DET
fcis-25108	100	2	application	application	NOUN
fcis-25108	100	3	of	of	ADP
fcis-25108	100	4	dl	dl	PROPN
fcis-25108	100	5	in	in	ADP
fcis-25108	100	6	nlp	nlp	NOUN
fcis-25108	100	7	will	will	AUX
fcis-25108	100	8	be	be	AUX
fcis-25108	100	9	more	more	ADV
fcis-25108	100	10	diverse	diverse	ADJ
fcis-25108	100	11	and	and	CCONJ
fcis-25108	100	12	extensive	extensive	ADJ
fcis-25108	100	13	.	.	PUNCT
fcis-25108	101	1	acknowledgments	acknowledgment	NOUN
fcis-25108	101	2	the	the	DET
fcis-25108	101	3	authors	author	NOUN
fcis-25108	101	4	gratefully	gratefully	ADV
fcis-25108	101	5	acknowledge	acknowledge	VERB
fcis-25108	101	6	the	the	DET
fcis-25108	101	7	financial	financial	ADJ
fcis-25108	101	8	support	support	NOUN
fcis-25108	101	9	from	from	ADP
fcis-25108	101	10	liaoning	liaoning	PROPN
fcis-25108	101	11	provincial	provincial	ADJ
fcis-25108	101	12	department	department	PROPN
fcis-25108	101	13	of	of	ADP
fcis-25108	101	14	education	education	PROPN
fcis-25108	101	15	scientific	scientific	ADJ
fcis-25108	101	16	research	research	NOUN
fcis-25108	101	17	funding	funding	NOUN
fcis-25108	101	18	project	project	NOUN
fcis-25108	101	19	funds	fund	NOUN
fcis-25108	101	20	(	(	PUNCT
fcis-25108	101	21	lfw202004	lfw202004	NOUN
fcis-25108	101	22	)	)	PUNCT
fcis-25108	101	23	and	and	CCONJ
fcis-25108	101	24	postgraduate	postgraduate	VERB
fcis-25108	101	25	research	research	NOUN
fcis-25108	101	26	project	project	NOUN
fcis-25108	101	27	of	of	ADP
fcis-25108	101	28	shenyang	shenyang	PROPN
fcis-25108	101	29	normal	normal	PROPN
fcis-25108	101	30	university	university	PROPN
fcis-25108	101	31	(	(	PUNCT
fcis-25108	101	32	synuxj2024060	synuxj2024060	PROPN
fcis-25108	101	33	)	)	PUNCT
fcis-25108	101	34	.	.	PUNCT
fcis-25108	102	1	references	reference	NOUN
fcis-25108	103	1	[	[	X
fcis-25108	103	2	1	1	NUM
fcis-25108	103	3	]	]	X
fcis-25108	103	4	hinton	hinton	PROPN
fcis-25108	103	5	g	g	PROPN
fcis-25108	103	6	e	e	PROPN
fcis-25108	103	7	,	,	PUNCT
fcis-25108	103	8	salakhutdinov	salakhutdinov	PROPN
fcis-25108	103	9	r	r	NOUN
fcis-25108	103	10	r.	r.	NOUN
fcis-25108	103	11	reducing	reduce	VERB
fcis-25108	103	12	the	the	DET
fcis-25108	103	13	dimensionality	dimensionality	NOUN
fcis-25108	103	14	of	of	ADP
fcis-25108	103	15	data	datum	NOUN
fcis-25108	103	16	with	with	ADP
fcis-25108	103	17	neural	neural	ADJ
fcis-25108	103	18	networks[j	networks[j	PROPN
fcis-25108	103	19	]	]	X
fcis-25108	103	20	.	.	PUNCT
fcis-25108	104	1	science	science	NOUN
fcis-25108	104	2	,	,	PUNCT
fcis-25108	104	3	2006	2006	NUM
fcis-25108	104	4	,	,	PUNCT
fcis-25108	104	5	313(5786	313(5786	NUM
fcis-25108	104	6	):	):	PUNCT
fcis-25108	104	7	504507	504507	NUM
fcis-25108	104	8	.	.	PUNCT
fcis-25108	105	1	[	[	X
fcis-25108	105	2	2	2	X
fcis-25108	105	3	]	]	X
fcis-25108	105	4	chen	chen	PROPN
fcis-25108	105	5	w.	w.	PROPN
fcis-25108	105	6	research	research	PROPN
fcis-25108	105	7	on	on	ADP
fcis-25108	105	8	artificial	artificial	ADJ
fcis-25108	105	9	intelligence	intelligence	NOUN
fcis-25108	105	10	in	in	ADP
fcis-25108	105	11	natural	natural	ADJ
fcis-25108	105	12	language	language	NOUN
fcis-25108	105	13	processing[j	processing[j	NOUN
fcis-25108	105	14	]	]	PUNCT
fcis-25108	105	15	.	.	PUNCT
fcis-25108	106	1	information	information	NOUN
fcis-25108	106	2	recording	recording	NOUN
fcis-25108	106	3	materials	material	NOUN
fcis-25108	106	4	,	,	PUNCT
fcis-25108	106	5	2023	2023	NUM
fcis-25108	106	6	,	,	PUNCT
fcis-25108	106	7	24	24	NUM
fcis-25108	106	8	(	(	PUNCT
fcis-25108	106	9	10	10	NUM
fcis-25108	106	10	)	)	PUNCT
fcis-25108	106	11	:	:	PUNCT
fcis-25108	107	1	92	92	NUM
fcis-25108	107	2	-	-	SYM
fcis-25108	107	3	94	94	NUM
fcis-25108	107	4	.	.	PUNCT
fcis-25108	108	1	[	[	X
fcis-25108	108	2	3	3	X
fcis-25108	108	3	]	]	X
fcis-25108	108	4	wei	wei	PROPN
fcis-25108	108	5	j	j	PROPN
fcis-25108	108	6	,	,	PUNCT
fcis-25108	108	7	liu	liu	PROPN
fcis-25108	108	8	a	a	PROPN
fcis-25108	108	9	,	,	PUNCT
fcis-25108	108	10	tang	tang	PROPN
fcis-25108	108	11	j.	j.	PROPN
fcis-25108	108	12	research	research	PROPN
fcis-25108	108	13	on	on	ADP
fcis-25108	108	14	alarm	alarm	NOUN
fcis-25108	108	15	model	model	NOUN
fcis-25108	108	16	of	of	ADP
fcis-25108	108	17	digital	digital	ADJ
fcis-25108	108	18	tv	tv	NOUN
fcis-25108	108	19	monitoring	monitoring	NOUN
fcis-25108	108	20	platform	platform	NOUN
fcis-25108	108	21	based	base	VERB
fcis-25108	108	22	on	on	ADP
fcis-25108	108	23	deep	deep	ADJ
fcis-25108	108	24	learning	learn	VERB
fcis-25108	108	25	neural	neural	ADJ
fcis-25108	108	26	network	network	NOUN
fcis-25108	108	27	technology[j	technology[j	PROPN
fcis-25108	108	28	]	]	PUNCT
fcis-25108	108	29	.	.	PUNCT
fcis-25108	109	1	cable	cable	NOUN
fcis-25108	109	2	tv	tv	NOUN
fcis-25108	109	3	technology	technology	NOUN
fcis-25108	109	4	,	,	PUNCT
fcis-25108	109	5	2017	2017	NUM
fcis-25108	109	6	,	,	PUNCT
fcis-25108	109	7	(	(	PUNCT
fcis-25108	109	8	07	07	NUM
fcis-25108	109	9	)	)	PUNCT
fcis-25108	109	10	:	:	PUNCT
fcis-25108	110	1	78	78	NUM
fcis-25108	110	2	-	-	SYM
fcis-25108	110	3	82	82	NUM
fcis-25108	110	4	.	.	PUNCT
fcis-25108	111	1	[	[	X
fcis-25108	111	2	4	4	NUM
fcis-25108	111	3	]	]	X
fcis-25108	111	4	lecun	lecun	PROPN
fcis-25108	111	5	y	y	PROPN
fcis-25108	111	6	,	,	PUNCT
fcis-25108	111	7	bottou	bottou	PROPN
fcis-25108	111	8	l	l	PROPN
fcis-25108	111	9	,	,	PUNCT
fcis-25108	111	10	bengio	bengio	PROPN
fcis-25108	111	11	y	y	PROPN
fcis-25108	111	12	,	,	PUNCT
fcis-25108	111	13	et	et	PROPN
fcis-25108	111	14	al	al	PROPN
fcis-25108	111	15	.	.	PUNCT
fcis-25108	111	16	gradient	gradient	NOUN
fcis-25108	111	17	-	-	PUNCT
fcis-25108	111	18	based	base	VERB
fcis-25108	111	19	learning	learning	NOUN
fcis-25108	111	20	applied	apply	VERB
fcis-25108	111	21	to	to	ADP
fcis-25108	111	22	document	document	NOUN
fcis-25108	111	23	recognition[j	recognition[j	NOUN
fcis-25108	111	24	]	]	PUNCT
fcis-25108	111	25	.	.	PUNCT
fcis-25108	112	1	proceedings	proceeding	NOUN
fcis-25108	112	2	of	of	ADP
fcis-25108	112	3	the	the	DET
fcis-25108	112	4	ieee	ieee	NOUN
fcis-25108	112	5	,	,	PUNCT
fcis-25108	112	6	1998	1998	NUM
fcis-25108	112	7	,	,	PUNCT
fcis-25108	112	8	86(11	86(11	NUM
fcis-25108	112	9	):	):	PUNCT
fcis-25108	112	10	2278	2278	NUM
fcis-25108	112	11	-	-	SYM
fcis-25108	112	12	2324	2324	NUM
fcis-25108	112	13	.	.	PUNCT
fcis-25108	113	1	[	[	X
fcis-25108	113	2	5	5	X
fcis-25108	113	3	]	]	PUNCT
fcis-25108	113	4	yan	yan	PROPN
fcis-25108	114	1	f	f	PROPN
fcis-25108	114	2	,	,	PUNCT
fcis-25108	114	3	wang	wang	PROPN
fcis-25108	114	4	j.	j.	PROPN
fcis-25108	114	5	research	research	PROPN
fcis-25108	114	6	on	on	ADP
fcis-25108	114	7	sentiment	sentiment	NOUN
fcis-25108	114	8	analysis	analysis	NOUN
fcis-25108	114	9	of	of	ADP
fcis-25108	114	10	micro	micro	NOUN
fcis-25108	114	11	-	-	NOUN
fcis-25108	114	12	blog	blog	NOUN
fcis-25108	114	13	based	base	VERB
fcis-25108	114	14	on	on	ADP
fcis-25108	114	15	attention	attention	NOUN
fcis-25108	114	16	-	-	PUNCT
fcis-25108	114	17	bilstm[j	bilstm[j	NOUN
fcis-25108	114	18	]	]	PUNCT
fcis-25108	114	19	.	.	PUNCT
fcis-25108	115	1	frontiers	frontier	NOUN
fcis-25108	115	2	in	in	ADP
fcis-25108	115	3	computing	computing	NOUN
fcis-25108	115	4	and	and	CCONJ
fcis-25108	115	5	intelligent	intelligent	ADJ
fcis-25108	115	6	systems	system	NOUN
fcis-25108	115	7	,	,	PUNCT
fcis-25108	115	8	2024	2024	NUM
fcis-25108	115	9	,	,	PUNCT
fcis-25108	115	10	7(3	7(3	NUM
fcis-25108	115	11	):	):	PUNCT
fcis-25108	115	12	49	49	NUM
fcis-25108	115	13	-	-	SYM
fcis-25108	115	14	51	51	NUM
fcis-25108	115	15	.	.	PUNCT
fcis-25108	116	1	[	[	X
fcis-25108	116	2	6	6	NUM
fcis-25108	116	3	]	]	PUNCT
fcis-25108	116	4	collobert	collobert	PROPN
fcis-25108	116	5	r	r	PROPN
fcis-25108	116	6	,	,	PUNCT
fcis-25108	116	7	weston	weston	PROPN
fcis-25108	116	8	j	j	PROPN
fcis-25108	116	9	,	,	PUNCT
fcis-25108	116	10	bottou	bottou	PROPN
fcis-25108	116	11	l	l	PROPN
fcis-25108	116	12	,	,	PUNCT
fcis-25108	116	13	et	et	PROPN
fcis-25108	116	14	al	al	PROPN
fcis-25108	116	15	.	.	PROPN
fcis-25108	116	16	natural	natural	ADJ
fcis-25108	116	17	language	language	NOUN
fcis-25108	116	18	processing	processing	NOUN
fcis-25108	116	19	(	(	PUNCT
fcis-25108	116	20	almost	almost	ADV
fcis-25108	116	21	)	)	PUNCT
fcis-25108	116	22	from	from	ADP
fcis-25108	116	23	scratch[j	scratch[j	PROPN
fcis-25108	116	24	]	]	PUNCT
fcis-25108	116	25	.	.	PUNCT
fcis-25108	117	1	journal	journal	PROPN
fcis-25108	117	2	of	of	ADP
fcis-25108	117	3	machine	machine	NOUN
fcis-25108	117	4	learning	learn	VERB
fcis-25108	117	5	research	research	NOUN
fcis-25108	117	6	,	,	PUNCT
fcis-25108	117	7	2011	2011	NUM
fcis-25108	117	8	,	,	PUNCT
fcis-25108	117	9	12	12	NUM
fcis-25108	117	10	:	:	PUNCT
fcis-25108	117	11	2493−	2493−	NUM
fcis-25108	117	12	2537	2537	NUM
fcis-25108	117	13	.	.	PUNCT
fcis-25108	118	1	[	[	X
fcis-25108	118	2	7	7	X
fcis-25108	118	3	]	]	SYM
fcis-25108	118	4	kalchbrenner	kalchbrenner	X
fcis-25108	118	5	n	n	CCONJ
fcis-25108	118	6	,	,	PUNCT
fcis-25108	118	7	grefenstette	grefenstette	NOUN
fcis-25108	118	8	e	e	NOUN
fcis-25108	118	9	,	,	PUNCT
fcis-25108	118	10	blunsom	blunsom	NOUN
fcis-25108	118	11	p.	p.	NOUN
fcis-25108	118	12	a	a	DET
fcis-25108	118	13	convolutional	convolutional	ADJ
fcis-25108	118	14	neural	neural	ADJ
fcis-25108	118	15	network	network	NOUN
fcis-25108	118	16	for	for	ADP
fcis-25108	118	17	modelling	model	VERB
fcis-25108	118	18	sentences[j	sentences[j	PROPN
fcis-25108	118	19	]	]	PUNCT
fcis-25108	118	20	.	.	PUNCT
fcis-25108	119	1	arxiv:1404.2188v1	arxiv:1404.2188v1	PROPN
fcis-25108	119	2	,	,	PUNCT
fcis-25108	119	3	2014	2014	NUM
fcis-25108	119	4	.	.	PUNCT
fcis-25108	120	1	[	[	X
fcis-25108	120	2	8	8	NUM
fcis-25108	120	3	]	]	X
fcis-25108	120	4	mikolov	mikolov	PROPN
fcis-25108	120	5	t	t	PROPN
fcis-25108	120	6	,	,	PUNCT
fcis-25108	120	7	kombrink	kombrink	PROPN
fcis-25108	120	8	s	s	PART
fcis-25108	120	9	,	,	PUNCT
fcis-25108	120	10	burget	burget	VERB
fcis-25108	120	11	l	l	NOUN
fcis-25108	120	12	,	,	PUNCT
fcis-25108	120	13	et	et	PROPN
fcis-25108	120	14	al	al	PROPN
fcis-25108	120	15	.	.	PUNCT
fcis-25108	120	16	extensions	extension	NOUN
fcis-25108	120	17	of	of	ADP
fcis-25108	120	18	recurrent	recurrent	ADJ
fcis-25108	120	19	neural	neural	ADJ
fcis-25108	120	20	network	network	NOUN
fcis-25108	120	21	language	language	NOUN
fcis-25108	120	22	model[c]//2011	model[c]//2011	PROPN
fcis-25108	120	23	ieee	ieee	PROPN
fcis-25108	120	24	international	international	ADJ
fcis-25108	120	25	conference	conference	NOUN
fcis-25108	120	26	on	on	ADP
fcis-25108	120	27	acoustics	acoustic	NOUN
fcis-25108	120	28	,	,	PUNCT
fcis-25108	120	29	speech	speech	NOUN
fcis-25108	120	30	and	and	CCONJ
fcis-25108	120	31	signal	signal	NOUN
fcis-25108	120	32	processing	processing	NOUN
fcis-25108	120	33	(	(	PUNCT
fcis-25108	120	34	icassp	icassp	PROPN
fcis-25108	120	35	)	)	PUNCT
fcis-25108	120	36	.	.	PUNCT
fcis-25108	121	1	ieee	ieee	PROPN
fcis-25108	121	2	,	,	PUNCT
fcis-25108	121	3	2011	2011	NUM
fcis-25108	121	4	:	:	PUNCT
fcis-25108	121	5	5528	5528	NUM
fcis-25108	121	6	-	-	SYM
fcis-25108	121	7	5531	5531	NUM
fcis-25108	121	8	.	.	PUNCT
fcis-25108	122	1	[	[	X
fcis-25108	122	2	9	9	NUM
fcis-25108	122	3	]	]	PUNCT
fcis-25108	122	4	sutskever	sutskever	VERB
fcis-25108	122	5	i	i	PRON
fcis-25108	122	6	,	,	PUNCT
fcis-25108	122	7	vinyals	vinyal	NOUN
fcis-25108	122	8	o	o	NOUN
fcis-25108	122	9	,	,	PUNCT
fcis-25108	122	10	le	le	X
fcis-25108	122	11	q	q	PROPN
fcis-25108	122	12	v.	v.	ADP
fcis-25108	122	13	sequence	sequence	NOUN
fcis-25108	122	14	to	to	ADP
fcis-25108	122	15	sequence	sequence	NOUN
fcis-25108	122	16	learning	learn	VERB
fcis-25108	122	17	with	with	ADP
fcis-25108	122	18	neural	neural	ADJ
fcis-25108	122	19	networks[j	networks[j	PROPN
fcis-25108	122	20	]	]	X
fcis-25108	122	21	.	.	PUNCT
fcis-25108	123	1	advances	advance	NOUN
fcis-25108	123	2	in	in	ADP
fcis-25108	123	3	neural	neural	ADJ
fcis-25108	123	4	information	information	NOUN
fcis-25108	123	5	processing	processing	NOUN
fcis-25108	123	6	systems	system	NOUN
fcis-25108	123	7	,	,	PUNCT
fcis-25108	123	8	2014	2014	NUM
fcis-25108	123	9	,	,	PUNCT
fcis-25108	123	10	27	27	NUM
fcis-25108	123	11	.	.	PUNCT
fcis-25108	124	1	[	[	X
fcis-25108	124	2	10	10	NUM
fcis-25108	124	3	]	]	X
fcis-25108	124	4	devlin	devlin	PROPN
fcis-25108	124	5	j.	j.	PROPN
fcis-25108	124	6	bert	bert	PROPN
fcis-25108	124	7	:	:	PUNCT
fcis-25108	124	8	pre	pre	ADJ
fcis-25108	124	9	-	-	NOUN
fcis-25108	124	10	training	training	NOUN
fcis-25108	124	11	of	of	ADP
fcis-25108	124	12	deep	deep	ADJ
fcis-25108	124	13	bidirectional	bidirectional	ADJ
fcis-25108	124	14	transformers	transformer	NOUN
fcis-25108	124	15	for	for	ADP
fcis-25108	124	16	language	language	NOUN
fcis-25108	124	17	understanding[j	understanding[j	NOUN
fcis-25108	124	18	]	]	PUNCT
fcis-25108	124	19	.	.	PUNCT
fcis-25108	125	1	arxiv:1810.04805	arxiv:1810.04805	PROPN
fcis-25108	125	2	,	,	PUNCT
fcis-25108	125	3	2018	2018	NUM
fcis-25108	125	4	.	.	PUNCT
fcis-25108	126	1	[	[	X
fcis-25108	126	2	11	11	NUM
fcis-25108	126	3	]	]	PUNCT
fcis-25108	126	4	raffel	raffel	NOUN
fcis-25108	126	5	c	c	NOUN
fcis-25108	126	6	,	,	PUNCT
fcis-25108	126	7	shazeer	shazeer	NOUN
fcis-25108	126	8	n	n	CCONJ
fcis-25108	126	9	,	,	PUNCT
fcis-25108	126	10	roberts	roberts	PROPN
fcis-25108	126	11	a	a	PROPN
fcis-25108	126	12	,	,	PUNCT
fcis-25108	126	13	et	et	PROPN
fcis-25108	126	14	al	al	PROPN
fcis-25108	126	15	.	.	PUNCT
fcis-25108	126	16	exploring	explore	VERB
fcis-25108	126	17	the	the	DET
fcis-25108	126	18	limits	limit	NOUN
fcis-25108	126	19	of	of	ADP
fcis-25108	126	20	transfer	transfer	NOUN
fcis-25108	126	21	learning	learn	VERB
fcis-25108	126	22	with	with	ADP
fcis-25108	126	23	a	a	DET
fcis-25108	126	24	unified	unified	ADJ
fcis-25108	126	25	text	text	NOUN
fcis-25108	126	26	-	-	PUNCT
fcis-25108	126	27	to	to	ADP
fcis-25108	126	28	-	-	PUNCT
fcis-25108	126	29	text	text	NOUN
fcis-25108	126	30	transformer[j	transformer[j	NOUN
fcis-25108	126	31	]	]	PUNCT
fcis-25108	126	32	.	.	PUNCT
fcis-25108	127	1	journal	journal	PROPN
fcis-25108	127	2	of	of	ADP
fcis-25108	127	3	machine	machine	NOUN
fcis-25108	127	4	learning	learn	VERB
fcis-25108	127	5	research	research	NOUN
fcis-25108	127	6	,	,	PUNCT
fcis-25108	127	7	2020	2020	NUM
fcis-25108	127	8	,	,	PUNCT
fcis-25108	127	9	21(140	21(140	NUM
fcis-25108	127	10	):	):	PUNCT
fcis-25108	127	11	1	1	NUM
fcis-25108	127	12	-	-	SYM
fcis-25108	127	13	67	67	NUM
fcis-25108	127	14	.	.	PUNCT