id	sid	tid	token	lemma	pos
fcis-4145	1	1	frontiers	frontier	NOUN
fcis-4145	1	2	in	in	ADP
fcis-4145	1	3	computing	computing	NOUN
fcis-4145	1	4	and	and	CCONJ
fcis-4145	1	5	intelligent	intelligent	ADJ
fcis-4145	1	6	systems	system	NOUN
fcis-4145	1	7	issn	issn	VERB
fcis-4145	1	8	:	:	PUNCT
fcis-4145	1	9	2832	2832	NUM
fcis-4145	1	10	-	-	SYM
fcis-4145	1	11	6024	6024	NUM
fcis-4145	1	12	|	|	NOUN
fcis-4145	1	13	vol	vol	NOUN
fcis-4145	1	14	.	.	PROPN
fcis-4145	2	1	2	2	NUM
fcis-4145	2	2	,	,	PUNCT
fcis-4145	2	3	no	no	INTJ
fcis-4145	2	4	.	.	NOUN
fcis-4145	2	5	2	2	NUM
fcis-4145	2	6	,	,	PUNCT
fcis-4145	2	7	2022	2022	NUM
fcis-4145	2	8	66	66	NUM
fcis-4145	2	9	research	research	NOUN
fcis-4145	2	10	on	on	ADP
fcis-4145	2	11	lovelorn	lovelorn	ADJ
fcis-4145	2	12	emotion	emotion	NOUN
fcis-4145	2	13	recognition	recognition	NOUN
fcis-4145	2	14	based	base	VERB
fcis-4145	2	15	on	on	ADP
fcis-4145	2	16	ernie	ernie	PROPN
fcis-4145	2	17	tiny	tiny	PROPN
fcis-4145	2	18	yuxin	yuxin	PROPN
fcis-4145	2	19	huang	huang	PROPN
fcis-4145	2	20	school	school	PROPN
fcis-4145	2	21	of	of	ADP
fcis-4145	2	22	computing	computing	NOUN
fcis-4145	2	23	and	and	CCONJ
fcis-4145	2	24	data	datum	NOUN
fcis-4145	2	25	science	science	NOUN
fcis-4145	2	26	,	,	PUNCT
fcis-4145	2	27	xiamen	xiamen	PROPN
fcis-4145	2	28	university	university	PROPN
fcis-4145	2	29	malaysia	malaysia	PROPN
fcis-4145	2	30	,	,	PUNCT
fcis-4145	2	31	selangor	selangor	PROPN
fcis-4145	2	32	darul	darul	PROPN
fcis-4145	2	33	ehsan	ehsan	PROPN
fcis-4145	2	34	,	,	PUNCT
fcis-4145	2	35	sepang	sepang	PROPN
fcis-4145	2	36	43900	43900	NUM
fcis-4145	2	37	,	,	PUNCT
fcis-4145	2	38	malaysia	malaysia	PROPN
fcis-4145	2	39	swe1909470@xmu.edu.my	swe1909470@xmu.edu.my	NOUN
fcis-4145	2	40	abstract	abstract	NOUN
fcis-4145	2	41	:	:	PUNCT
fcis-4145	2	42	topics	topic	NOUN
fcis-4145	2	43	related	relate	VERB
fcis-4145	2	44	to	to	AUX
fcis-4145	2	45	sentiment	sentiment	VERB
fcis-4145	2	46	classification	classification	NOUN
fcis-4145	2	47	and	and	CCONJ
fcis-4145	2	48	emotion	emotion	NOUN
fcis-4145	2	49	recognition	recognition	NOUN
fcis-4145	2	50	are	be	AUX
fcis-4145	2	51	an	an	DET
fcis-4145	2	52	important	important	ADJ
fcis-4145	2	53	part	part	NOUN
fcis-4145	2	54	of	of	ADP
fcis-4145	2	55	the	the	DET
fcis-4145	2	56	natural	natural	ADJ
fcis-4145	2	57	language	language	NOUN
fcis-4145	2	58	processing	processing	NOUN
fcis-4145	2	59	research	research	NOUN
fcis-4145	2	60	field	field	NOUN
fcis-4145	2	61	and	and	CCONJ
fcis-4145	2	62	can	can	AUX
fcis-4145	2	63	be	be	AUX
fcis-4145	2	64	used	use	VERB
fcis-4145	2	65	to	to	PART
fcis-4145	2	66	analyze	analyze	VERB
fcis-4145	2	67	users	user	NOUN
fcis-4145	2	68	'	'	PART
fcis-4145	2	69	sentiment	sentiment	NOUN
fcis-4145	2	70	tendencies	tendency	NOUN
fcis-4145	2	71	towards	towards	ADP
fcis-4145	2	72	brands	brand	NOUN
fcis-4145	2	73	,	,	PUNCT
fcis-4145	2	74	understand	understand	VERB
fcis-4145	2	75	the	the	DET
fcis-4145	2	76	public	public	NOUN
fcis-4145	2	77	's	's	PART
fcis-4145	2	78	attitudes	attitude	NOUN
fcis-4145	2	79	and	and	CCONJ
fcis-4145	2	80	opinions	opinion	NOUN
fcis-4145	2	81	on	on	ADP
fcis-4145	2	82	public	public	ADJ
fcis-4145	2	83	opinion	opinion	NOUN
fcis-4145	2	84	events	event	NOUN
fcis-4145	2	85	,	,	PUNCT
fcis-4145	2	86	and	and	CCONJ
fcis-4145	2	87	detect	detect	VERB
fcis-4145	2	88	users	user	NOUN
fcis-4145	2	89	'	'	PART
fcis-4145	2	90	mental	mental	ADJ
fcis-4145	2	91	health	health	NOUN
fcis-4145	2	92	,	,	PUNCT
fcis-4145	2	93	among	among	ADP
fcis-4145	2	94	others	other	NOUN
fcis-4145	2	95	.	.	PUNCT
fcis-4145	3	1	past	past	ADP
fcis-4145	3	2	research	research	NOUN
fcis-4145	3	3	has	have	AUX
fcis-4145	3	4	usually	usually	ADV
fcis-4145	3	5	been	be	AUX
fcis-4145	3	6	based	base	VERB
fcis-4145	3	7	on	on	ADP
fcis-4145	3	8	positive	positive	ADJ
fcis-4145	3	9	and	and	CCONJ
fcis-4145	3	10	negative	negative	ADJ
fcis-4145	3	11	emotions	emotion	NOUN
fcis-4145	3	12	or	or	CCONJ
fcis-4145	3	13	multi	multi	ADJ
fcis-4145	3	14	-	-	ADJ
fcis-4145	3	15	categorized	categorize	VERB
fcis-4145	3	16	emotions	emotion	NOUN
fcis-4145	3	17	such	such	ADJ
fcis-4145	3	18	as	as	ADP
fcis-4145	3	19	happiness	happiness	NOUN
fcis-4145	3	20	,	,	PUNCT
fcis-4145	3	21	anger	anger	NOUN
fcis-4145	3	22	and	and	CCONJ
fcis-4145	3	23	sadness	sadness	NOUN
fcis-4145	3	24	,	,	PUNCT
fcis-4145	3	25	while	while	SCONJ
fcis-4145	3	26	there	there	PRON
fcis-4145	3	27	has	have	AUX
fcis-4145	3	28	been	be	AUX
fcis-4145	3	29	little	little	ADJ
fcis-4145	3	30	research	research	NOUN
fcis-4145	3	31	on	on	ADP
fcis-4145	3	32	the	the	DET
fcis-4145	3	33	recognition	recognition	NOUN
fcis-4145	3	34	of	of	ADP
fcis-4145	3	35	the	the	DET
fcis-4145	3	36	specific	specific	ADJ
fcis-4145	3	37	emotion	emotion	NOUN
fcis-4145	3	38	of	of	ADP
fcis-4145	3	39	lovelorn	lovelorn	ADJ
fcis-4145	3	40	.	.	PUNCT
fcis-4145	4	1	this	this	DET
fcis-4145	4	2	study	study	NOUN
fcis-4145	4	3	aims	aim	VERB
fcis-4145	4	4	to	to	PART
fcis-4145	4	5	identify	identify	VERB
fcis-4145	4	6	the	the	DET
fcis-4145	4	7	lovelorn	lovelorn	ADJ
fcis-4145	4	8	emotion	emotion	NOUN
fcis-4145	4	9	in	in	ADP
fcis-4145	4	10	text	text	NOUN
fcis-4145	4	11	,	,	PUNCT
fcis-4145	4	12	using	use	VERB
fcis-4145	4	13	deep	deep	ADJ
fcis-4145	4	14	learning	learning	NOUN
fcis-4145	4	15	pretrained	pretraine	VERB
fcis-4145	4	16	model	model	NOUN
fcis-4145	4	17	ernir	ernir	NOUN
fcis-4145	4	18	tiny	tiny	ADJ
fcis-4145	4	19	to	to	PART
fcis-4145	4	20	train	train	VERB
fcis-4145	4	21	a	a	DET
fcis-4145	4	22	dataset	dataset	NOUN
fcis-4145	4	23	consisting	consist	VERB
fcis-4145	4	24	of	of	ADP
fcis-4145	4	25	5008	5008	NUM
fcis-4145	4	26	pieces	piece	NOUN
fcis-4145	4	27	of	of	ADP
fcis-4145	4	28	chinese	chinese	ADJ
fcis-4145	4	29	lovelorn	lovelorn	ADJ
fcis-4145	4	30	emotion	emotion	NOUN
fcis-4145	4	31	text	text	NOUN
fcis-4145	4	32	crawled	crawl	VERB
fcis-4145	4	33	from	from	ADP
fcis-4145	4	34	social	social	ADJ
fcis-4145	4	35	media	medium	NOUN
fcis-4145	4	36	platform	platform	NOUN
fcis-4145	4	37	weibo	weibo	NOUN
fcis-4145	4	38	and	and	CCONJ
fcis-4145	4	39	4998	4998	NUM
fcis-4145	4	40	pieces	piece	NOUN
fcis-4145	4	41	of	of	ADP
fcis-4145	4	42	ordinary	ordinary	ADJ
fcis-4145	4	43	text	text	NOUN
fcis-4145	4	44	extracted	extract	VERB
fcis-4145	4	45	from	from	ADP
fcis-4145	4	46	existing	exist	VERB
fcis-4145	4	47	available	available	ADJ
fcis-4145	4	48	dataset	dataset	NOUN
fcis-4145	4	49	.	.	PUNCT
fcis-4145	5	1	and	and	CCONJ
fcis-4145	5	2	finally	finally	ADV
fcis-4145	5	3	,	,	PUNCT
fcis-4145	5	4	it	it	PRON
fcis-4145	5	5	was	be	AUX
fcis-4145	5	6	proved	prove	VERB
fcis-4145	5	7	that	that	SCONJ
fcis-4145	5	8	ernie	ernie	NOUN
fcis-4145	5	9	tiny	tiny	ADJ
fcis-4145	5	10	performs	perform	VERB
fcis-4145	5	11	well	well	ADV
fcis-4145	5	12	in	in	ADP
fcis-4145	5	13	classifying	classify	VERB
fcis-4145	5	14	whether	whether	SCONJ
fcis-4145	5	15	a	a	DET
fcis-4145	5	16	text	text	NOUN
fcis-4145	5	17	contains	contain	VERB
fcis-4145	5	18	lovelorn	lovelorn	ADJ
fcis-4145	5	19	emotion	emotion	NOUN
fcis-4145	5	20	or	or	CCONJ
fcis-4145	5	21	not	not	PART
fcis-4145	5	22	,	,	PUNCT
fcis-4145	5	23	with	with	ADP
fcis-4145	5	24	f1	f1	ADJ
fcis-4145	5	25	score	score	NOUN
fcis-4145	5	26	of	of	ADP
fcis-4145	5	27	0.941929	0.941929	NOUN
fcis-4145	5	28	,	,	PUNCT
fcis-4145	5	29	precision	precision	NOUN
fcis-4145	5	30	score	score	NOUN
fcis-4145	5	31	of	of	ADP
fcis-4145	5	32	0.942300	0.942300	NUM
fcis-4145	5	33	and	and	CCONJ
fcis-4145	5	34	recall	recall	NOUN
fcis-4145	5	35	score	score	NOUN
fcis-4145	5	36	of	of	ADP
fcis-4145	5	37	0.941928	0.941928	NUM
fcis-4145	5	38	obtained	obtain	VERB
fcis-4145	5	39	on	on	ADP
fcis-4145	5	40	the	the	DET
fcis-4145	5	41	test	test	NOUN
fcis-4145	5	42	set	set	NOUN
fcis-4145	5	43	.	.	PUNCT
fcis-4145	6	1	keywords	keyword	NOUN
fcis-4145	6	2	:	:	PUNCT
fcis-4145	6	3	ernie	ernie	PROPN
fcis-4145	6	4	tiny	tiny	ADJ
fcis-4145	6	5	;	;	PUNCT
fcis-4145	6	6	lovelorn	lovelorn	ADJ
fcis-4145	6	7	emotion	emotion	NOUN
fcis-4145	6	8	;	;	PUNCT
fcis-4145	6	9	emotion	emotion	NOUN
fcis-4145	6	10	recognition	recognition	NOUN
fcis-4145	6	11	.	.	PUNCT
fcis-4145	7	1	1	1	X
fcis-4145	7	2	.	.	X
fcis-4145	7	3	introduction	introduction	NOUN
fcis-4145	7	4	sentiment	sentiment	NOUN
fcis-4145	7	5	analysis	analysis	NOUN
fcis-4145	7	6	is	be	AUX
fcis-4145	7	7	an	an	DET
fcis-4145	7	8	attractive	attractive	ADJ
fcis-4145	7	9	and	and	CCONJ
fcis-4145	7	10	valuable	valuable	ADJ
fcis-4145	7	11	task	task	NOUN
fcis-4145	7	12	in	in	ADP
fcis-4145	7	13	the	the	DET
fcis-4145	7	14	discipline	discipline	NOUN
fcis-4145	7	15	of	of	ADP
fcis-4145	7	16	natural	natural	ADJ
fcis-4145	7	17	language	language	NOUN
fcis-4145	7	18	processing	processing	NOUN
fcis-4145	7	19	,	,	PUNCT
fcis-4145	7	20	which	which	PRON
fcis-4145	7	21	is	be	AUX
fcis-4145	7	22	the	the	DET
fcis-4145	7	23	process	process	NOUN
fcis-4145	7	24	of	of	ADP
fcis-4145	7	25	analyzing	analyzing	NOUN
fcis-4145	7	26	,	,	PUNCT
fcis-4145	7	27	processing	processing	NOUN
fcis-4145	7	28	,	,	PUNCT
fcis-4145	7	29	generalizing	generalize	VERB
fcis-4145	7	30	and	and	CCONJ
fcis-4145	7	31	reasoning	reason	VERB
fcis-4145	7	32	about	about	ADP
fcis-4145	7	33	text	text	NOUN
fcis-4145	7	34	with	with	ADP
fcis-4145	7	35	sentiment	sentiment	NOUN
fcis-4145	7	36	,	,	PUNCT
fcis-4145	7	37	focusing	focus	VERB
fcis-4145	7	38	on	on	ADP
fcis-4145	7	39	people	people	NOUN
fcis-4145	7	40	's	's	PART
fcis-4145	7	41	emotional	emotional	ADJ
fcis-4145	7	42	attitudes	attitude	NOUN
fcis-4145	7	43	toward	toward	ADP
fcis-4145	7	44	various	various	ADJ
fcis-4145	7	45	entities	entity	NOUN
fcis-4145	7	46	,	,	PUNCT
fcis-4145	7	47	including	include	VERB
fcis-4145	7	48	products	product	NOUN
fcis-4145	7	49	,	,	PUNCT
fcis-4145	7	50	organizations	organization	NOUN
fcis-4145	7	51	,	,	PUNCT
fcis-4145	7	52	events	event	NOUN
fcis-4145	7	53	,	,	PUNCT
fcis-4145	7	54	etc	etc	X
fcis-4145	7	55	.	.	X
fcis-4145	7	56	according	accord	VERB
fcis-4145	7	57	to	to	ADP
fcis-4145	7	58	the	the	DET
fcis-4145	7	59	granularity	granularity	NOUN
fcis-4145	7	60	of	of	ADP
fcis-4145	7	61	the	the	DET
fcis-4145	7	62	analyzed	analyze	VERB
fcis-4145	7	63	object	object	NOUN
fcis-4145	7	64	,	,	PUNCT
fcis-4145	7	65	it	it	PRON
fcis-4145	7	66	can	can	AUX
fcis-4145	7	67	be	be	AUX
fcis-4145	7	68	divided	divide	VERB
fcis-4145	7	69	into	into	ADP
fcis-4145	7	70	document	document	NOUN
fcis-4145	7	71	,	,	PUNCT
fcis-4145	7	72	sentence	sentence	NOUN
fcis-4145	7	73	and	and	CCONJ
fcis-4145	7	74	aspect	aspect	NOUN
fcis-4145	7	75	level	level	NOUN
fcis-4145	7	76	.	.	PUNCT
fcis-4145	8	1	sentiment	sentiment	NOUN
fcis-4145	8	2	analysis	analysis	NOUN
fcis-4145	8	3	plays	play	VERB
fcis-4145	8	4	an	an	DET
fcis-4145	8	5	important	important	ADJ
fcis-4145	8	6	role	role	NOUN
fcis-4145	8	7	in	in	ADP
fcis-4145	8	8	many	many	ADJ
fcis-4145	8	9	fields	field	NOUN
fcis-4145	8	10	,	,	PUNCT
fcis-4145	8	11	for	for	ADP
fcis-4145	8	12	example	example	NOUN
fcis-4145	8	13	,	,	PUNCT
fcis-4145	8	14	movie	movie	NOUN
fcis-4145	8	15	reviews	review	NOUN
fcis-4145	8	16	can	can	AUX
fcis-4145	8	17	be	be	AUX
fcis-4145	8	18	used	use	VERB
fcis-4145	8	19	to	to	PART
fcis-4145	8	20	analyze	analyze	VERB
fcis-4145	8	21	whether	whether	SCONJ
fcis-4145	8	22	users	user	NOUN
fcis-4145	8	23	rate	rate	VERB
fcis-4145	8	24	movies	movie	NOUN
fcis-4145	8	25	positively	positively	ADV
fcis-4145	8	26	or	or	CCONJ
fcis-4145	8	27	negatively	negatively	ADV
fcis-4145	8	28	;	;	PUNCT
fcis-4145	8	29	sentiment	sentiment	NOUN
fcis-4145	8	30	tendency	tendency	NOUN
fcis-4145	8	31	of	of	ADP
fcis-4145	8	32	users	user	NOUN
fcis-4145	8	33	towards	towards	ADP
fcis-4145	8	34	attributes	attribute	NOUN
fcis-4145	8	35	such	such	ADJ
fcis-4145	8	36	as	as	ADP
fcis-4145	8	37	price	price	NOUN
fcis-4145	8	38	and	and	CCONJ
fcis-4145	8	39	usefulness	usefulness	NOUN
fcis-4145	8	40	of	of	ADP
fcis-4145	8	41	products	product	NOUN
fcis-4145	8	42	can	can	AUX
fcis-4145	8	43	be	be	AUX
fcis-4145	8	44	mined	mine	VERB
fcis-4145	8	45	in	in	ADP
fcis-4145	8	46	product	product	NOUN
fcis-4145	8	47	review	review	NOUN
fcis-4145	8	48	texts	text	NOUN
fcis-4145	8	49	.	.	PUNCT
fcis-4145	9	1	in	in	ADP
fcis-4145	9	2	addition	addition	NOUN
fcis-4145	9	3	,	,	PUNCT
fcis-4145	9	4	sentiment	sentiment	NOUN
fcis-4145	9	5	analysis	analysis	NOUN
fcis-4145	9	6	is	be	AUX
fcis-4145	9	7	also	also	ADV
fcis-4145	9	8	important	important	ADJ
fcis-4145	9	9	in	in	ADP
fcis-4145	9	10	the	the	DET
fcis-4145	9	11	field	field	NOUN
fcis-4145	9	12	of	of	ADP
fcis-4145	9	13	mental	mental	ADJ
fcis-4145	9	14	health	health	NOUN
fcis-4145	9	15	,	,	PUNCT
fcis-4145	9	16	which	which	PRON
fcis-4145	9	17	can	can	AUX
fcis-4145	9	18	understand	understand	VERB
fcis-4145	9	19	the	the	DET
fcis-4145	9	20	psychological	psychological	ADJ
fcis-4145	9	21	needs	need	NOUN
fcis-4145	9	22	of	of	ADP
fcis-4145	9	23	users	user	NOUN
fcis-4145	9	24	and	and	CCONJ
fcis-4145	9	25	provide	provide	VERB
fcis-4145	9	26	emotional	emotional	ADJ
fcis-4145	9	27	support	support	NOUN
fcis-4145	9	28	and	and	CCONJ
fcis-4145	9	29	treatment	treatment	NOUN
fcis-4145	9	30	.	.	PUNCT
fcis-4145	10	1	there	there	PRON
fcis-4145	10	2	are	be	VERB
fcis-4145	10	3	a	a	DET
fcis-4145	10	4	large	large	ADJ
fcis-4145	10	5	number	number	NOUN
fcis-4145	10	6	of	of	ADP
fcis-4145	10	7	emotionally	emotionally	ADV
fcis-4145	10	8	charged	charge	VERB
fcis-4145	10	9	texts	text	NOUN
fcis-4145	10	10	on	on	ADP
fcis-4145	10	11	social	social	ADJ
fcis-4145	10	12	media	medium	NOUN
fcis-4145	10	13	platforms	platform	NOUN
fcis-4145	10	14	,	,	PUNCT
fcis-4145	10	15	including	include	VERB
fcis-4145	10	16	the	the	DET
fcis-4145	10	17	lovelorn	lovelorn	ADJ
fcis-4145	10	18	emotion	emotion	NOUN
fcis-4145	10	19	,	,	PUNCT
fcis-4145	10	20	but	but	CCONJ
fcis-4145	10	21	there	there	PRON
fcis-4145	10	22	has	have	AUX
fcis-4145	10	23	been	be	AUX
fcis-4145	10	24	little	little	ADJ
fcis-4145	10	25	research	research	NOUN
fcis-4145	10	26	on	on	ADP
fcis-4145	10	27	this	this	DET
fcis-4145	10	28	specific	specific	ADJ
fcis-4145	10	29	emotion	emotion	NOUN
fcis-4145	10	30	.	.	PUNCT
fcis-4145	11	1	this	this	DET
fcis-4145	11	2	study	study	NOUN
fcis-4145	11	3	focuses	focus	VERB
fcis-4145	11	4	on	on	ADP
fcis-4145	11	5	detecting	detect	VERB
fcis-4145	11	6	the	the	DET
fcis-4145	11	7	emotion	emotion	NOUN
fcis-4145	11	8	of	of	ADP
fcis-4145	11	9	love	love	NOUN
fcis-4145	11	10	loss	loss	NOUN
fcis-4145	11	11	in	in	ADP
fcis-4145	11	12	texts	text	NOUN
fcis-4145	11	13	,	,	PUNCT
fcis-4145	11	14	which	which	PRON
fcis-4145	11	15	can	can	AUX
fcis-4145	11	16	identify	identify	VERB
fcis-4145	11	17	whether	whether	SCONJ
fcis-4145	11	18	lovelorn	lovelorn	ADJ
fcis-4145	11	19	emotion	emotion	NOUN
fcis-4145	11	20	is	be	AUX
fcis-4145	11	21	contained	contain	VERB
fcis-4145	11	22	from	from	ADP
fcis-4145	11	23	any	any	DET
fcis-4145	11	24	text	text	NOUN
fcis-4145	11	25	,	,	PUNCT
fcis-4145	11	26	helping	help	VERB
fcis-4145	11	27	heterosexual	heterosexual	ADJ
fcis-4145	11	28	dating	date	VERB
fcis-4145	11	29	platforms	platform	NOUN
fcis-4145	11	30	to	to	PART
fcis-4145	11	31	identify	identify	VERB
fcis-4145	11	32	potential	potential	ADJ
fcis-4145	11	33	target	target	NOUN
fcis-4145	11	34	users	user	NOUN
fcis-4145	11	35	of	of	ADP
fcis-4145	11	36	love	love	NOUN
fcis-4145	11	37	loss	loss	NOUN
fcis-4145	11	38	and	and	CCONJ
fcis-4145	11	39	advertise	advertise	VERB
fcis-4145	11	40	their	their	PRON
fcis-4145	11	41	dating	date	VERB
fcis-4145	11	42	software	software	NOUN
fcis-4145	11	43	accordingly	accordingly	ADV
fcis-4145	11	44	.	.	PUNCT
fcis-4145	12	1	while	while	SCONJ
fcis-4145	12	2	social	social	ADJ
fcis-4145	12	3	media	medium	NOUN
fcis-4145	12	4	software	software	NOUN
fcis-4145	12	5	can	can	AUX
fcis-4145	12	6	also	also	ADV
fcis-4145	12	7	cooperate	cooperate	VERB
fcis-4145	12	8	with	with	ADP
fcis-4145	12	9	platforms	platform	NOUN
fcis-4145	12	10	that	that	PRON
fcis-4145	12	11	provide	provide	VERB
fcis-4145	12	12	psychological	psychological	ADJ
fcis-4145	12	13	therapy	therapy	NOUN
fcis-4145	12	14	,	,	PUNCT
fcis-4145	12	15	to	to	PART
fcis-4145	12	16	implement	implement	VERB
fcis-4145	12	17	emotional	emotional	ADJ
fcis-4145	12	18	care	care	NOUN
fcis-4145	12	19	for	for	ADP
fcis-4145	12	20	these	these	DET
fcis-4145	12	21	users	user	NOUN
fcis-4145	12	22	and	and	CCONJ
fcis-4145	12	23	push	push	VERB
fcis-4145	12	24	them	they	PRON
fcis-4145	12	25	information	information	NOUN
fcis-4145	12	26	that	that	PRON
fcis-4145	12	27	helps	help	VERB
fcis-4145	12	28	to	to	PART
fcis-4145	12	29	get	get	VERB
fcis-4145	12	30	out	out	ADP
fcis-4145	12	31	of	of	ADP
fcis-4145	12	32	negative	negative	ADJ
fcis-4145	12	33	emotions	emotion	NOUN
fcis-4145	12	34	,	,	PUNCT
fcis-4145	12	35	so	so	SCONJ
fcis-4145	12	36	that	that	SCONJ
fcis-4145	12	37	the	the	DET
fcis-4145	12	38	target	target	NOUN
fcis-4145	12	39	group	group	NOUN
fcis-4145	12	40	can	can	AUX
fcis-4145	12	41	face	face	VERB
fcis-4145	12	42	lovelorn	lovelorn	ADJ
fcis-4145	12	43	emotions	emotion	NOUN
fcis-4145	12	44	in	in	ADP
fcis-4145	12	45	a	a	DET
fcis-4145	12	46	healthier	healthy	ADJ
fcis-4145	12	47	state	state	NOUN
fcis-4145	12	48	of	of	ADP
fcis-4145	12	49	mind	mind	NOUN
fcis-4145	12	50	,	,	PUNCT
fcis-4145	12	51	showing	show	VERB
fcis-4145	12	52	the	the	DET
fcis-4145	12	53	humanistic	humanistic	ADJ
fcis-4145	12	54	care	care	NOUN
fcis-4145	12	55	of	of	ADP
fcis-4145	12	56	the	the	DET
fcis-4145	12	57	organization	organization	NOUN
fcis-4145	12	58	.	.	PUNCT
fcis-4145	13	1	2	2	X
fcis-4145	13	2	.	.	X
fcis-4145	13	3	literature	literature	NOUN
fcis-4145	13	4	review	review	NOUN
fcis-4145	13	5	detecting	detect	VERB
fcis-4145	13	6	emotions	emotion	NOUN
fcis-4145	13	7	with	with	ADP
fcis-4145	13	8	natural	natural	ADJ
fcis-4145	13	9	language	language	NOUN
fcis-4145	13	10	processing	processing	NOUN
fcis-4145	13	11	techniques	technique	NOUN
fcis-4145	13	12	is	be	AUX
fcis-4145	13	13	one	one	NUM
fcis-4145	13	14	of	of	ADP
fcis-4145	13	15	the	the	DET
fcis-4145	13	16	applications	application	NOUN
fcis-4145	13	17	of	of	ADP
fcis-4145	13	18	artificial	artificial	ADJ
fcis-4145	13	19	intelligence	intelligence	NOUN
fcis-4145	13	20	in	in	ADP
fcis-4145	13	21	mental	mental	ADJ
fcis-4145	13	22	health	health	NOUN
fcis-4145	13	23	.	.	PUNCT
fcis-4145	14	1	to	to	PART
fcis-4145	14	2	address	address	VERB
fcis-4145	14	3	the	the	DET
fcis-4145	14	4	problem	problem	NOUN
fcis-4145	14	5	that	that	SCONJ
fcis-4145	14	6	traditional	traditional	ADJ
fcis-4145	14	7	machine	machine	NOUN
fcis-4145	14	8	learning	learning	NOUN
fcis-4145	14	9	models	model	NOUN
fcis-4145	14	10	fail	fail	VERB
fcis-4145	14	11	to	to	PART
fcis-4145	14	12	effectively	effectively	ADV
fcis-4145	14	13	focus	focus	VERB
fcis-4145	14	14	on	on	ADP
fcis-4145	14	15	some	some	DET
fcis-4145	14	16	features	feature	NOUN
fcis-4145	14	17	in	in	ADP
fcis-4145	14	18	sentiment	sentiment	NOUN
fcis-4145	14	19	analysis	analysis	NOUN
fcis-4145	14	20	tasks	task	NOUN
fcis-4145	14	21	,	,	PUNCT
fcis-4145	14	22	the	the	DET
fcis-4145	14	23	deep	deep	ADJ
fcis-4145	14	24	learning	learning	NOUN
fcis-4145	14	25	model	model	NOUN
fcis-4145	14	26	lstm	lstm	PROPN
fcis-4145	14	27	was	be	AUX
fcis-4145	14	28	used	use	VERB
fcis-4145	14	29	by	by	ADP
fcis-4145	14	30	chen	chen	PROPN
fcis-4145	14	31	et	et	PROPN
fcis-4145	14	32	al	al	PROPN
fcis-4145	14	33	.	.	PUNCT
fcis-4145	14	34	to	to	PART
fcis-4145	14	35	extract	extract	VERB
fcis-4145	14	36	and	and	CCONJ
fcis-4145	14	37	classify	classify	VERB
fcis-4145	14	38	users	user	NOUN
fcis-4145	14	39	'	'	PART
fcis-4145	14	40	emotional	emotional	ADJ
fcis-4145	14	41	features	feature	NOUN
fcis-4145	14	42	and	and	CCONJ
fcis-4145	14	43	applied	apply	VERB
fcis-4145	14	44	to	to	ADP
fcis-4145	14	45	the	the	DET
fcis-4145	14	46	scenario	scenario	NOUN
fcis-4145	14	47	of	of	ADP
fcis-4145	14	48	postpartum	postpartum	NOUN
fcis-4145	14	49	depression	depression	NOUN
fcis-4145	14	50	screening	screen	VERB
fcis-4145	14	51	.	.	PUNCT
fcis-4145	15	1	the	the	DET
fcis-4145	15	2	role	role	NOUN
fcis-4145	15	3	of	of	ADP
fcis-4145	15	4	emoticons	emoticon	NOUN
fcis-4145	15	5	was	be	AUX
fcis-4145	15	6	also	also	ADV
fcis-4145	15	7	considered	consider	VERB
fcis-4145	15	8	by	by	ADP
fcis-4145	15	9	extracting	extract	VERB
fcis-4145	15	10	features	feature	NOUN
fcis-4145	15	11	and	and	CCONJ
fcis-4145	15	12	modeling	modeling	NOUN
fcis-4145	15	13	,	,	PUNCT
fcis-4145	15	14	and	and	CCONJ
fcis-4145	15	15	good	good	ADJ
fcis-4145	15	16	results	result	NOUN
fcis-4145	15	17	were	be	AUX
fcis-4145	15	18	achieved	achieve	VERB
fcis-4145	15	19	in	in	ADP
fcis-4145	15	20	general	general	ADJ
fcis-4145	15	21	agreement	agreement	NOUN
fcis-4145	15	22	with	with	ADP
fcis-4145	15	23	the	the	DET
fcis-4145	15	24	edinburgh	edinburgh	PROPN
fcis-4145	15	25	postpartum	postpartum	PROPN
fcis-4145	15	26	depression	depression	NOUN
fcis-4145	15	27	scale	scale	NOUN
fcis-4145	15	28	.	.	PUNCT
fcis-4145	16	1	the	the	DET
fcis-4145	16	2	study	study	NOUN
fcis-4145	16	3	significantly	significantly	ADV
fcis-4145	16	4	reduces	reduce	VERB
fcis-4145	16	5	the	the	DET
fcis-4145	16	6	length	length	NOUN
fcis-4145	16	7	of	of	ADP
fcis-4145	16	8	sentiment	sentiment	NOUN
fcis-4145	16	9	screening	screen	VERB
fcis-4145	16	10	and	and	CCONJ
fcis-4145	16	11	served	serve	VERB
fcis-4145	16	12	as	as	ADP
fcis-4145	16	13	a	a	DET
fcis-4145	16	14	good	good	ADJ
fcis-4145	16	15	reference	reference	NOUN
fcis-4145	16	16	for	for	ADP
fcis-4145	16	17	document	document	NOUN
fcis-4145	16	18	-	-	PUNCT
fcis-4145	16	19	level	level	NOUN
fcis-4145	16	20	emotion	emotion	NOUN
fcis-4145	16	21	classification	classification	NOUN
fcis-4145	16	22	tasks	task	NOUN
fcis-4145	16	23	in	in	ADP
fcis-4145	16	24	specific	specific	ADJ
fcis-4145	16	25	domains	domain	NOUN
fcis-4145	16	26	.	.	PUNCT
fcis-4145	17	1	liu	liu	PROPN
fcis-4145	17	2	,	,	PUNCT
fcis-4145	17	3	shi	shi	PROPN
fcis-4145	17	4	and	and	CCONJ
fcis-4145	17	5	jiang	jiang	PROPN
fcis-4145	17	6	used	use	VERB
fcis-4145	17	7	the	the	DET
fcis-4145	17	8	(	(	PUNCT
fcis-4145	17	9	bsc	bsc	PROPN
fcis-4145	17	10	+	+	CCONJ
fcis-4145	17	11	rfs)-fs	rfs)-fs	NOUN
fcis-4145	17	12	,	,	PUNCT
fcis-4145	17	13	wec	wec	NOUN
fcis-4145	17	14	-	-	PUNCT
fcis-4145	17	15	fs	fs	PROPN
fcis-4145	17	16	model	model	NOUN
fcis-4145	17	17	proposed	propose	VERB
fcis-4145	17	18	by	by	ADP
fcis-4145	17	19	the	the	DET
fcis-4145	17	20	optimal	optimal	ADJ
fcis-4145	17	21	classification	classification	NOUN
fcis-4145	17	22	model	model	NOUN
fcis-4145	17	23	fusing	fuse	VERB
fcis-4145	17	24	single	single	ADJ
fcis-4145	17	25	and	and	CCONJ
fcis-4145	17	26	multidimensional	multidimensional	ADJ
fcis-4145	17	27	features	feature	NOUN
fcis-4145	17	28	to	to	PART
fcis-4145	17	29	identify	identify	VERB
fcis-4145	17	30	posts	post	NOUN
fcis-4145	17	31	with	with	ADP
fcis-4145	17	32	suicidal	suicidal	ADJ
fcis-4145	17	33	ideation	ideation	NOUN
fcis-4145	17	34	(	(	PUNCT
fcis-4145	17	35	including	include	VERB
fcis-4145	17	36	the	the	DET
fcis-4145	17	37	stress	stress	NOUN
fcis-4145	17	38	of	of	ADP
fcis-4145	17	39	love	love	NOUN
fcis-4145	17	40	loss	loss	NOUN
fcis-4145	17	41	)	)	PUNCT
fcis-4145	17	42	in	in	ADP
fcis-4145	17	43	social	social	ADJ
fcis-4145	17	44	media	medium	NOUN
fcis-4145	17	45	,	,	PUNCT
fcis-4145	17	46	and	and	CCONJ
fcis-4145	17	47	the	the	DET
fcis-4145	17	48	results	result	NOUN
fcis-4145	17	49	demonstrated	demonstrate	VERB
fcis-4145	17	50	that	that	SCONJ
fcis-4145	17	51	an	an	DET
fcis-4145	17	52	appropriate	appropriate	ADJ
fcis-4145	17	53	combination	combination	NOUN
fcis-4145	17	54	strategy	strategy	NOUN
fcis-4145	17	55	and	and	CCONJ
fcis-4145	17	56	the	the	DET
fcis-4145	17	57	selection	selection	NOUN
fcis-4145	17	58	of	of	ADP
fcis-4145	17	59	a	a	DET
fcis-4145	17	60	targeted	target	VERB
fcis-4145	17	61	classification	classification	NOUN
fcis-4145	17	62	model	model	NOUN
fcis-4145	17	63	contributed	contribute	VERB
fcis-4145	17	64	to	to	ADP
fcis-4145	17	65	the	the	DET
fcis-4145	17	66	improvement	improvement	NOUN
fcis-4145	17	67	of	of	ADP
fcis-4145	17	68	suicidal	suicidal	ADJ
fcis-4145	17	69	ideation	ideation	NOUN
fcis-4145	17	70	detection	detection	NOUN
fcis-4145	17	71	rate	rate	NOUN
fcis-4145	17	72	.	.	PUNCT
fcis-4145	18	1	traditional	traditional	ADJ
fcis-4145	18	2	sentiment	sentiment	NOUN
fcis-4145	18	3	analysis	analysis	NOUN
fcis-4145	18	4	techniques	technique	NOUN
fcis-4145	18	5	include	include	VERB
fcis-4145	18	6	supervised	supervised	ADJ
fcis-4145	18	7	methods	method	NOUN
fcis-4145	18	8	,	,	PUNCT
fcis-4145	18	9	such	such	ADJ
fcis-4145	18	10	as	as	ADP
fcis-4145	18	11	support	support	NOUN
fcis-4145	18	12	vector	vector	NOUN
fcis-4145	18	13	machines	machine	NOUN
fcis-4145	18	14	,	,	PUNCT
fcis-4145	18	15	maximum	maximum	PROPN
fcis-4145	18	16	entropy	entropy	NOUN
fcis-4145	18	17	,	,	PUNCT
fcis-4145	18	18	plain	plain	ADJ
fcis-4145	18	19	bayes	bayes	NOUN
fcis-4145	18	20	and	and	CCONJ
fcis-4145	18	21	other	other	ADJ
fcis-4145	18	22	supervised	supervised	ADJ
fcis-4145	18	23	machine	machine	NOUN
fcis-4145	18	24	learning	learning	NOUN
fcis-4145	18	25	methods	method	NOUN
fcis-4145	18	26	,	,	PUNCT
fcis-4145	18	27	and	and	CCONJ
fcis-4145	18	28	unsupervised	unsupervised	ADJ
fcis-4145	18	29	methods	method	NOUN
fcis-4145	18	30	,	,	PUNCT
fcis-4145	18	31	including	include	VERB
fcis-4145	18	32	methods	method	NOUN
fcis-4145	18	33	based	base	VERB
fcis-4145	18	34	on	on	ADP
fcis-4145	18	35	sentiment	sentiment	NOUN
fcis-4145	18	36	lexicon	lexicon	NOUN
fcis-4145	18	37	,	,	PUNCT
fcis-4145	18	38	syntactic	syntactic	ADJ
fcis-4145	18	39	analysis	analysis	NOUN
fcis-4145	18	40	and	and	CCONJ
fcis-4145	18	41	syntactic	syntactic	ADJ
fcis-4145	18	42	patterns	pattern	NOUN
fcis-4145	18	43	.	.	PUNCT
fcis-4145	19	1	the	the	DET
fcis-4145	19	2	lexicon	lexicon	NOUN
fcis-4145	19	3	-	-	PUNCT
fcis-4145	19	4	based	base	VERB
fcis-4145	19	5	methods	method	NOUN
fcis-4145	19	6	mainly	mainly	ADV
fcis-4145	19	7	use	use	VERB
fcis-4145	19	8	a	a	DET
fcis-4145	19	9	series	series	NOUN
fcis-4145	19	10	of	of	ADP
fcis-4145	19	11	sentiment	sentiment	NOUN
fcis-4145	19	12	lexicons	lexicon	NOUN
fcis-4145	19	13	and	and	CCONJ
fcis-4145	19	14	rules	rule	NOUN
fcis-4145	19	15	developed	develop	VERB
fcis-4145	19	16	for	for	ADP
fcis-4145	19	17	paragraph	paragraph	NOUN
fcis-4145	19	18	disassembly	disassembly	NOUN
fcis-4145	19	19	and	and	CCONJ
fcis-4145	19	20	syntactic	syntactic	ADJ
fcis-4145	19	21	analysis	analysis	NOUN
fcis-4145	19	22	of	of	ADP
fcis-4145	19	23	texts	text	NOUN
fcis-4145	19	24	to	to	PART
fcis-4145	19	25	calculate	calculate	VERB
fcis-4145	19	26	sentiment	sentiment	NOUN
fcis-4145	19	27	values	value	NOUN
fcis-4145	19	28	,	,	PUNCT
fcis-4145	19	29	and	and	CCONJ
fcis-4145	19	30	finally	finally	ADV
fcis-4145	19	31	use	use	VERB
fcis-4145	19	32	the	the	DET
fcis-4145	19	33	sentiment	sentiment	NOUN
fcis-4145	19	34	values	value	NOUN
fcis-4145	19	35	as	as	ADP
fcis-4145	19	36	a	a	DET
fcis-4145	19	37	basis	basis	NOUN
fcis-4145	19	38	for	for	ADP
fcis-4145	19	39	judging	judge	VERB
fcis-4145	19	40	the	the	DET
fcis-4145	19	41	sentiment	sentiment	NOUN
fcis-4145	19	42	tendency	tendency	NOUN
fcis-4145	19	43	of	of	ADP
fcis-4145	19	44	texts	text	NOUN
fcis-4145	19	45	.	.	PUNCT
fcis-4145	20	1	most	most	ADJ
fcis-4145	20	2	of	of	ADP
fcis-4145	20	3	the	the	DET
fcis-4145	20	4	machine	machine	NOUN
fcis-4145	20	5	learning	learning	NOUN
fcis-4145	20	6	-	-	PUNCT
fcis-4145	20	7	based	base	VERB
fcis-4145	20	8	methods	method	NOUN
fcis-4145	20	9	treat	treat	VERB
fcis-4145	20	10	this	this	DET
fcis-4145	20	11	problem	problem	NOUN
fcis-4145	20	12	as	as	ADP
fcis-4145	20	13	a	a	DET
fcis-4145	20	14	classification	classification	NOUN
fcis-4145	20	15	problem	problem	NOUN
fcis-4145	20	16	,	,	PUNCT
fcis-4145	20	17	manually	manually	ADV
fcis-4145	20	18	labeling	label	VERB
fcis-4145	20	19	the	the	DET
fcis-4145	20	20	training	training	NOUN
fcis-4145	20	21	text	text	NOUN
fcis-4145	20	22	,	,	PUNCT
fcis-4145	20	23	structuring	structure	VERB
fcis-4145	20	24	the	the	DET
fcis-4145	20	25	text	text	NOUN
fcis-4145	20	26	content	content	NOUN
fcis-4145	20	27	,	,	PUNCT
fcis-4145	20	28	and	and	CCONJ
fcis-4145	20	29	inputting	inputte	VERB
fcis-4145	20	30	it	it	PRON
fcis-4145	20	31	to	to	ADP
fcis-4145	20	32	the	the	DET
fcis-4145	20	33	algorithm	algorithm	NOUN
fcis-4145	20	34	for	for	ADP
fcis-4145	20	35	training	training	NOUN
fcis-4145	20	36	and	and	CCONJ
fcis-4145	20	37	learning	learning	NOUN
fcis-4145	20	38	,	,	PUNCT
fcis-4145	20	39	and	and	CCONJ
fcis-4145	20	40	classifying	classify	VERB
fcis-4145	20	41	the	the	DET
fcis-4145	20	42	target	target	NOUN
fcis-4145	20	43	sentiment	sentiment	NOUN
fcis-4145	20	44	into	into	ADP
fcis-4145	20	45	positive	positive	ADJ
fcis-4145	20	46	and	and	CCONJ
fcis-4145	20	47	negative	negative	ADJ
fcis-4145	20	48	categories	category	NOUN
fcis-4145	20	49	to	to	PART
fcis-4145	20	50	judge	judge	VERB
fcis-4145	20	51	the	the	DET
fcis-4145	20	52	sentiment	sentiment	NOUN
fcis-4145	20	53	polarity	polarity	NOUN
fcis-4145	20	54	.	.	PUNCT
fcis-4145	21	1	these	these	DET
fcis-4145	21	2	early	early	ADJ
fcis-4145	21	3	methods	method	NOUN
fcis-4145	21	4	are	be	AUX
fcis-4145	21	5	relatively	relatively	ADV
fcis-4145	21	6	simple	simple	ADJ
fcis-4145	21	7	and	and	CCONJ
fcis-4145	21	8	have	have	VERB
fcis-4145	21	9	good	good	ADJ
fcis-4145	21	10	results	result	NOUN
fcis-4145	21	11	for	for	ADP
fcis-4145	21	12	some	some	DET
fcis-4145	21	13	related	relate	VERB
fcis-4145	21	14	tasks	task	NOUN
fcis-4145	21	15	.	.	PUNCT
fcis-4145	22	1	however	however	ADV
fcis-4145	22	2	,	,	PUNCT
fcis-4145	22	3	sentiment	sentiment	NOUN
fcis-4145	22	4	lexicon	lexicon	NOUN
fcis-4145	22	5	-	-	PUNCT
fcis-4145	22	6	based	base	VERB
fcis-4145	22	7	approaches	approach	NOUN
fcis-4145	22	8	lack	lack	VERB
fcis-4145	22	9	semantic	semantic	ADJ
fcis-4145	22	10	associations	association	NOUN
fcis-4145	22	11	between	between	ADP
fcis-4145	22	12	contexts	context	NOUN
fcis-4145	22	13	,	,	PUNCT
fcis-4145	22	14	and	and	CCONJ
fcis-4145	22	15	the	the	DET
fcis-4145	22	16	training	training	NOUN
fcis-4145	22	17	of	of	ADP
fcis-4145	22	18	machine	machine	NOUN
fcis-4145	22	19	learning	learning	NOUN
fcis-4145	22	20	models	model	NOUN
fcis-4145	22	21	relies	rely	VERB
fcis-4145	22	22	on	on	ADP
fcis-4145	22	23	the	the	DET
fcis-4145	22	24	quality	quality	NOUN
fcis-4145	22	25	of	of	ADP
fcis-4145	22	26	annotated	annotate	VERB
fcis-4145	22	27	datasets	dataset	NOUN
fcis-4145	22	28	,	,	PUNCT
fcis-4145	22	29	67	67	NUM
fcis-4145	22	30	which	which	PRON
fcis-4145	22	31	require	require	VERB
fcis-4145	22	32	high	high	ADJ
fcis-4145	22	33	quality	quality	NOUN
fcis-4145	22	34	feature	feature	NOUN
fcis-4145	22	35	construction	construction	NOUN
fcis-4145	22	36	and	and	CCONJ
fcis-4145	22	37	selection	selection	NOUN
fcis-4145	22	38	,	,	PUNCT
fcis-4145	22	39	and	and	CCONJ
fcis-4145	22	40	therefore	therefore	ADV
fcis-4145	22	41	have	have	VERB
fcis-4145	22	42	limitations	limitation	NOUN
fcis-4145	22	43	.	.	PUNCT
fcis-4145	23	1	with	with	ADP
fcis-4145	23	2	the	the	DET
fcis-4145	23	3	popularity	popularity	NOUN
fcis-4145	23	4	of	of	ADP
fcis-4145	23	5	deep	deep	ADJ
fcis-4145	23	6	learning	learning	NOUN
fcis-4145	23	7	,	,	PUNCT
fcis-4145	23	8	the	the	DET
fcis-4145	23	9	application	application	NOUN
fcis-4145	23	10	of	of	ADP
fcis-4145	23	11	deep	deep	ADJ
fcis-4145	23	12	learning	learning	NOUN
fcis-4145	23	13	architectures	architecture	NOUN
fcis-4145	23	14	in	in	ADP
fcis-4145	23	15	sentiment	sentiment	NOUN
fcis-4145	23	16	analysis	analysis	NOUN
fcis-4145	23	17	tasks	task	NOUN
fcis-4145	23	18	has	have	AUX
fcis-4145	23	19	become	become	VERB
fcis-4145	23	20	a	a	DET
fcis-4145	23	21	very	very	ADV
fcis-4145	23	22	popular	popular	ADJ
fcis-4145	23	23	research	research	NOUN
fcis-4145	23	24	topic	topic	NOUN
fcis-4145	23	25	in	in	ADP
fcis-4145	23	26	recent	recent	ADJ
fcis-4145	23	27	years	year	NOUN
fcis-4145	23	28	,	,	PUNCT
fcis-4145	23	29	using	use	VERB
fcis-4145	23	30	multilayer	multilayer	ADJ
fcis-4145	23	31	neural	neural	ADJ
fcis-4145	23	32	networks	network	NOUN
fcis-4145	23	33	to	to	PART
fcis-4145	23	34	learn	learn	VERB
fcis-4145	23	35	features	feature	NOUN
fcis-4145	23	36	and	and	CCONJ
fcis-4145	23	37	extract	extract	VERB
fcis-4145	23	38	them	they	PRON
fcis-4145	23	39	automatically	automatically	ADV
fcis-4145	23	40	,	,	PUNCT
fcis-4145	23	41	overcoming	overcome	VERB
fcis-4145	23	42	many	many	ADJ
fcis-4145	23	43	of	of	ADP
fcis-4145	23	44	the	the	DET
fcis-4145	23	45	problems	problem	NOUN
fcis-4145	23	46	of	of	ADP
fcis-4145	23	47	traditional	traditional	ADJ
fcis-4145	23	48	classification	classification	NOUN
fcis-4145	23	49	techniques	technique	NOUN
fcis-4145	23	50	.	.	PUNCT
fcis-4145	24	1	a	a	DET
fcis-4145	24	2	large	large	ADJ
fcis-4145	24	3	number	number	NOUN
fcis-4145	24	4	of	of	ADP
fcis-4145	24	5	sentiment	sentiment	NOUN
fcis-4145	24	6	analysis	analysis	NOUN
fcis-4145	24	7	studies	study	NOUN
fcis-4145	24	8	based	base	VERB
fcis-4145	24	9	on	on	ADP
fcis-4145	24	10	deep	deep	ADJ
fcis-4145	24	11	learning	learning	NOUN
fcis-4145	24	12	(	(	PUNCT
fcis-4145	24	13	e.g.	e.g.	ADV
fcis-4145	24	14	,	,	PUNCT
fcis-4145	24	15	convolutional	convolutional	ADJ
fcis-4145	24	16	neural	neural	ADJ
fcis-4145	24	17	networks	network	NOUN
fcis-4145	24	18	cnn	cnn	PROPN
fcis-4145	24	19	,	,	PUNCT
fcis-4145	24	20	recurrent	recurrent	ADJ
fcis-4145	24	21	neural	neural	ADJ
fcis-4145	24	22	networks	network	NOUN
fcis-4145	24	23	rnn	rnn	VERB
fcis-4145	24	24	,	,	PUNCT
fcis-4145	24	25	long	long	ADJ
fcis-4145	24	26	-	-	PUNCT
fcis-4145	24	27	short	short	ADJ
fcis-4145	24	28	term	term	NOUN
fcis-4145	24	29	memory	memory	NOUN
fcis-4145	24	30	networks	network	NOUN
fcis-4145	24	31	lstm	lstm	PROPN
fcis-4145	24	32	,	,	PUNCT
fcis-4145	24	33	etc	etc	X
fcis-4145	24	34	.	.	X
fcis-4145	24	35	)	)	PUNCT
fcis-4145	24	36	have	have	AUX
fcis-4145	24	37	emerged	emerge	VERB
fcis-4145	24	38	and	and	CCONJ
fcis-4145	24	39	obtained	obtain	VERB
fcis-4145	24	40	state	state	NOUN
fcis-4145	24	41	-	-	PUNCT
fcis-4145	24	42	of	of	ADP
fcis-4145	24	43	-	-	PUNCT
fcis-4145	24	44	the	the	DET
fcis-4145	24	45	-	-	PUNCT
fcis-4145	24	46	art	art	NOUN
fcis-4145	24	47	results	result	NOUN
fcis-4145	24	48	in	in	ADP
fcis-4145	24	49	many	many	ADJ
fcis-4145	24	50	tasks	task	NOUN
fcis-4145	24	51	of	of	ADP
fcis-4145	24	52	sentiment	sentiment	NOUN
fcis-4145	24	53	classification	classification	NOUN
fcis-4145	24	54	.	.	PUNCT
fcis-4145	25	1	kim	kim	PROPN
fcis-4145	25	2	discussed	discuss	VERB
fcis-4145	25	3	that	that	SCONJ
fcis-4145	25	4	fine	fine	ADJ
fcis-4145	25	5	-	-	PUNCT
fcis-4145	25	6	tuning	tune	VERB
fcis-4145	25	7	task	task	NOUN
fcis-4145	25	8	-	-	PUNCT
fcis-4145	25	9	specific	specific	ADJ
fcis-4145	25	10	word	word	NOUN
fcis-4145	25	11	vectors	vector	NOUN
fcis-4145	25	12	in	in	ADP
fcis-4145	25	13	a	a	DET
fcis-4145	25	14	cnn	cnn	PROPN
fcis-4145	25	15	model	model	NOUN
fcis-4145	25	16	can	can	AUX
fcis-4145	25	17	improve	improve	VERB
fcis-4145	25	18	performance	performance	NOUN
fcis-4145	25	19	.	.	PUNCT
fcis-4145	26	1	with	with	ADP
fcis-4145	26	2	the	the	DET
fcis-4145	26	3	performance	performance	NOUN
fcis-4145	26	4	of	of	ADP
fcis-4145	26	5	pre	pre	ADJ
fcis-4145	26	6	-	-	ADJ
fcis-4145	26	7	trained	train	VERB
fcis-4145	26	8	word	word	NOUN
fcis-4145	26	9	vectors	vector	NOUN
fcis-4145	26	10	on	on	ADP
fcis-4145	26	11	sentence	sentence	NOUN
fcis-4145	26	12	classification	classification	NOUN
fcis-4145	26	13	tasks	task	NOUN
fcis-4145	26	14	,	,	PUNCT
fcis-4145	26	15	it	it	PRON
fcis-4145	26	16	is	be	AUX
fcis-4145	26	17	demonstrated	demonstrate	VERB
fcis-4145	26	18	that	that	SCONJ
fcis-4145	26	19	simple	simple	ADJ
fcis-4145	26	20	single	single	ADJ
fcis-4145	26	21	-	-	PUNCT
fcis-4145	26	22	layer	layer	NOUN
fcis-4145	26	23	cnn	cnn	NOUN
fcis-4145	26	24	models	model	NOUN
fcis-4145	26	25	with	with	ADP
fcis-4145	26	26	less	less	ADJ
fcis-4145	26	27	hyperparameter	hyperparameter	NOUN
fcis-4145	26	28	tuning	tuning	NOUN
fcis-4145	26	29	and	and	CCONJ
fcis-4145	26	30	static	static	ADJ
fcis-4145	26	31	word	word	NOUN
fcis-4145	26	32	vectors	vector	NOUN
fcis-4145	26	33	can	can	AUX
fcis-4145	26	34	achieve	achieve	VERB
fcis-4145	26	35	better	well	ADJ
fcis-4145	26	36	performance	performance	NOUN
fcis-4145	26	37	on	on	ADP
fcis-4145	26	38	multiple	multiple	ADJ
fcis-4145	26	39	benchmarks	benchmark	NOUN
fcis-4145	26	40	,	,	PUNCT
fcis-4145	26	41	optimizing	optimize	VERB
fcis-4145	26	42	tasks	task	NOUN
fcis-4145	26	43	including	include	VERB
fcis-4145	26	44	sentiment	sentiment	NOUN
fcis-4145	26	45	analysis	analysis	NOUN
fcis-4145	26	46	and	and	CCONJ
fcis-4145	26	47	problem	problem	NOUN
fcis-4145	26	48	classification	classification	NOUN
fcis-4145	26	49	.	.	PUNCT
fcis-4145	27	1	tang	tang	PROPN
fcis-4145	27	2	,	,	PUNCT
fcis-4145	27	3	qin	qin	INTJ
fcis-4145	27	4	,	,	PUNCT
fcis-4145	27	5	and	and	CCONJ
fcis-4145	27	6	liu	liu	PROPN
fcis-4145	27	7	used	use	VERB
fcis-4145	27	8	the	the	DET
fcis-4145	27	9	idea	idea	NOUN
fcis-4145	27	10	of	of	ADP
fcis-4145	27	11	memory	memory	NOUN
fcis-4145	27	12	network	network	NOUN
fcis-4145	27	13	for	for	ADP
fcis-4145	27	14	aspect	aspect	NOUN
fcis-4145	27	15	-	-	PUNCT
fcis-4145	27	16	level	level	NOUN
fcis-4145	27	17	sentiment	sentiment	NOUN
fcis-4145	27	18	analysis	analysis	NOUN
fcis-4145	27	19	,	,	PUNCT
fcis-4145	27	20	constructed	construct	VERB
fcis-4145	27	21	the	the	DET
fcis-4145	27	22	memory	memory	NOUN
fcis-4145	27	23	by	by	ADP
fcis-4145	27	24	context	context	NOUN
fcis-4145	27	25	information	information	NOUN
fcis-4145	27	26	,	,	PUNCT
fcis-4145	27	27	introduced	introduce	VERB
fcis-4145	27	28	attention	attention	NOUN
fcis-4145	27	29	to	to	PART
fcis-4145	27	30	capture	capture	VERB
fcis-4145	27	31	the	the	DET
fcis-4145	27	32	important	important	ADJ
fcis-4145	27	33	information	information	NOUN
fcis-4145	27	34	in	in	ADP
fcis-4145	27	35	the	the	DET
fcis-4145	27	36	sentiment	sentiment	NOUN
fcis-4145	27	37	tendency	tendency	NOUN
fcis-4145	27	38	of	of	ADP
fcis-4145	27	39	different	different	ADJ
fcis-4145	27	40	aspects	aspect	NOUN
fcis-4145	27	41	,	,	PUNCT
fcis-4145	27	42	and	and	CCONJ
fcis-4145	27	43	achieve	achieve	VERB
fcis-4145	27	44	great	great	ADJ
fcis-4145	27	45	results	result	NOUN
fcis-4145	27	46	on	on	ADP
fcis-4145	27	47	the	the	DET
fcis-4145	27	48	experimental	experimental	ADJ
fcis-4145	27	49	dataset	dataset	NOUN
fcis-4145	27	50	,	,	PUNCT
fcis-4145	27	51	which	which	PRON
fcis-4145	27	52	was	be	AUX
fcis-4145	27	53	simpler	simple	ADJ
fcis-4145	27	54	and	and	CCONJ
fcis-4145	27	55	faster	fast	ADV
fcis-4145	27	56	to	to	PART
fcis-4145	27	57	compute	compute	VERB
fcis-4145	27	58	than	than	ADP
fcis-4145	27	59	the	the	DET
fcis-4145	27	60	neural	neural	ADJ
fcis-4145	27	61	network	network	NOUN
fcis-4145	27	62	models	model	NOUN
fcis-4145	27	63	such	such	ADJ
fcis-4145	27	64	as	as	ADP
fcis-4145	27	65	rnn	rnn	NOUN
fcis-4145	27	66	and	and	CCONJ
fcis-4145	27	67	lstm	lstm	NOUN
fcis-4145	27	68	.	.	PUNCT
fcis-4145	28	1	in	in	ADP
fcis-4145	28	2	addition	addition	NOUN
fcis-4145	28	3	,	,	PUNCT
fcis-4145	28	4	the	the	DET
fcis-4145	28	5	research	research	NOUN
fcis-4145	28	6	demonstrated	demonstrate	VERB
fcis-4145	28	7	that	that	SCONJ
fcis-4145	28	8	combining	combine	VERB
fcis-4145	28	9	content	content	NOUN
fcis-4145	28	10	information	information	NOUN
fcis-4145	28	11	and	and	CCONJ
fcis-4145	28	12	location	location	NOUN
fcis-4145	28	13	information	information	NOUN
fcis-4145	28	14	to	to	PART
fcis-4145	28	15	learn	learn	VERB
fcis-4145	28	16	context	context	NOUN
fcis-4145	28	17	weight	weight	NOUN
fcis-4145	28	18	was	be	AUX
fcis-4145	28	19	a	a	DET
fcis-4145	28	20	more	more	ADV
fcis-4145	28	21	suitable	suitable	ADJ
fcis-4145	28	22	method	method	NOUN
fcis-4145	28	23	for	for	ADP
fcis-4145	28	24	aspect	aspect	NOUN
fcis-4145	28	25	-	-	PUNCT
fcis-4145	28	26	level	level	NOUN
fcis-4145	28	27	sentiment	sentiment	NOUN
fcis-4145	28	28	analysis	analysis	NOUN
fcis-4145	28	29	,	,	PUNCT
fcis-4145	28	30	and	and	CCONJ
fcis-4145	28	31	the	the	DET
fcis-4145	28	32	multi	multi	ADJ
fcis-4145	28	33	-	-	ADJ
fcis-4145	28	34	layer	layer	ADJ
fcis-4145	28	35	computational	computational	ADJ
fcis-4145	28	36	unit	unit	NOUN
fcis-4145	28	37	can	can	AUX
fcis-4145	28	38	learn	learn	VERB
fcis-4145	28	39	more	more	ADJ
fcis-4145	28	40	abstractive	abstractive	ADJ
fcis-4145	28	41	information	information	NOUN
fcis-4145	28	42	,	,	PUNCT
fcis-4145	28	43	which	which	PRON
fcis-4145	28	44	can	can	AUX
fcis-4145	28	45	improve	improve	VERB
fcis-4145	28	46	the	the	DET
fcis-4145	28	47	model	model	NOUN
fcis-4145	28	48	performance	performance	NOUN
fcis-4145	28	49	.	.	PUNCT
fcis-4145	29	1	an	an	DET
fcis-4145	29	2	attention	attention	NOUN
fcis-4145	29	3	-	-	PUNCT
fcis-4145	29	4	based	base	VERB
fcis-4145	29	5	bi	bi	ADJ
fcis-4145	29	6	-	-	ADJ
fcis-4145	29	7	directional	directional	ADJ
fcis-4145	29	8	cnn	cnn	PROPN
fcis-4145	29	9	-	-	PUNCT
fcis-4145	29	10	rnn	rnn	NOUN
fcis-4145	29	11	depth	depth	NOUN
fcis-4145	29	12	model	model	NOUN
fcis-4145	29	13	(	(	PUNCT
fcis-4145	29	14	abcdm	abcdm	NOUN
fcis-4145	29	15	)	)	PUNCT
fcis-4145	29	16	was	be	AUX
fcis-4145	29	17	proposed	propose	VERB
fcis-4145	29	18	by	by	ADP
fcis-4145	29	19	basiri	basiri	PROPN
fcis-4145	29	20	et	et	PROPN
fcis-4145	29	21	al	al	PROPN
fcis-4145	29	22	.	.	PROPN
fcis-4145	30	1	for	for	ADP
fcis-4145	30	2	sentiment	sentiment	NOUN
fcis-4145	30	3	analysis	analysis	NOUN
fcis-4145	30	4	.	.	PUNCT
fcis-4145	31	1	by	by	ADP
fcis-4145	31	2	using	use	VERB
fcis-4145	31	3	two	two	NUM
fcis-4145	31	4	independent	independent	ADJ
fcis-4145	31	5	bidirectional	bidirectional	ADJ
fcis-4145	31	6	lstm	lstm	NOUN
fcis-4145	31	7	and	and	CCONJ
fcis-4145	31	8	gru	gru	NOUN
fcis-4145	31	9	layers	layer	NOUN
fcis-4145	31	10	,	,	PUNCT
fcis-4145	31	11	temporal	temporal	ADJ
fcis-4145	31	12	information	information	NOUN
fcis-4145	31	13	in	in	ADP
fcis-4145	31	14	both	both	DET
fcis-4145	31	15	directions	direction	NOUN
fcis-4145	31	16	was	be	AUX
fcis-4145	31	17	considered	consider	VERB
fcis-4145	31	18	to	to	PART
fcis-4145	31	19	extract	extract	VERB
fcis-4145	31	20	context	context	NOUN
fcis-4145	31	21	.	.	PUNCT
fcis-4145	32	1	also	also	ADV
fcis-4145	32	2	,	,	PUNCT
fcis-4145	32	3	the	the	DET
fcis-4145	32	4	attention	attention	NOUN
fcis-4145	32	5	mechanism	mechanism	NOUN
fcis-4145	32	6	was	be	AUX
fcis-4145	32	7	applied	apply	VERB
fcis-4145	32	8	in	in	ADP
fcis-4145	32	9	the	the	DET
fcis-4145	32	10	output	output	NOUN
fcis-4145	32	11	of	of	ADP
fcis-4145	32	12	the	the	DET
fcis-4145	32	13	model	model	NOUN
fcis-4145	32	14	's	's	PART
fcis-4145	32	15	bidirectional	bidirectional	ADJ
fcis-4145	32	16	layers	layer	NOUN
fcis-4145	32	17	to	to	PART
fcis-4145	32	18	emphasize	emphasize	VERB
fcis-4145	32	19	contributions	contribution	NOUN
fcis-4145	32	20	of	of	ADP
fcis-4145	32	21	different	different	ADJ
fcis-4145	32	22	words	word	NOUN
fcis-4145	32	23	,	,	PUNCT
fcis-4145	32	24	achieving	achieve	VERB
fcis-4145	32	25	sota	sota	NOUN
fcis-4145	32	26	results	result	NOUN
fcis-4145	32	27	on	on	ADP
fcis-4145	32	28	the	the	DET
fcis-4145	32	29	twitter	twitter	NOUN
fcis-4145	32	30	dataset	dataset	VERB
fcis-4145	32	31	.	.	PUNCT
fcis-4145	33	1	the	the	DET
fcis-4145	33	2	bert+attention	bert+attention	PROPN
fcis-4145	33	3	model	model	NOUN
fcis-4145	33	4	proposed	propose	VERB
fcis-4145	33	5	by	by	ADP
fcis-4145	33	6	wang	wang	PROPN
fcis-4145	33	7	and	and	CCONJ
fcis-4145	33	8	tong	tong	PROPN
fcis-4145	33	9	improved	improve	VERB
fcis-4145	33	10	the	the	DET
fcis-4145	33	11	bert	bert	ADJ
fcis-4145	33	12	pre	pre	ADJ
fcis-4145	33	13	-	-	ADJ
fcis-4145	33	14	training	training	ADJ
fcis-4145	33	15	model	model	NOUN
fcis-4145	33	16	by	by	ADP
fcis-4145	33	17	introducing	introduce	VERB
fcis-4145	33	18	an	an	DET
fcis-4145	33	19	attention	attention	NOUN
fcis-4145	33	20	mechanism	mechanism	NOUN
fcis-4145	33	21	and	and	CCONJ
fcis-4145	33	22	weighting	weight	VERB
fcis-4145	33	23	key	key	ADJ
fcis-4145	33	24	features	feature	NOUN
fcis-4145	33	25	to	to	PART
fcis-4145	33	26	analyze	analyze	VERB
fcis-4145	33	27	the	the	DET
fcis-4145	33	28	emotions	emotion	NOUN
fcis-4145	33	29	expressed	express	VERB
fcis-4145	33	30	on	on	ADP
fcis-4145	33	31	microblogs	microblog	NOUN
fcis-4145	33	32	during	during	ADP
fcis-4145	33	33	the	the	DET
fcis-4145	33	34	epidemic	epidemic	NOUN
fcis-4145	33	35	,	,	PUNCT
fcis-4145	33	36	and	and	CCONJ
fcis-4145	33	37	obtained	obtain	VERB
fcis-4145	33	38	higher	high	ADJ
fcis-4145	33	39	accuracy	accuracy	NOUN
fcis-4145	33	40	than	than	ADP
fcis-4145	33	41	the	the	DET
fcis-4145	33	42	textcnn	textcnn	NOUN
fcis-4145	33	43	,	,	PUNCT
fcis-4145	33	44	bilstm	bilstm	NOUN
fcis-4145	33	45	and	and	CCONJ
fcis-4145	33	46	the	the	DET
fcis-4145	33	47	bilstm+attention	bilstm+attention	PROPN
fcis-4145	33	48	model	model	NOUN
fcis-4145	33	49	.	.	PUNCT
fcis-4145	34	1	3	3	X
fcis-4145	34	2	.	.	X
fcis-4145	34	3	research	research	NOUN
fcis-4145	34	4	methodology	methodology	NOUN
fcis-4145	34	5	3.1	3.1	NUM
fcis-4145	34	6	.	.	PUNCT
fcis-4145	35	1	ernie	ernie	PROPN
fcis-4145	35	2	tiny	tiny	PROPN
fcis-4145	35	3	model	model	PROPN
fcis-4145	35	4	ernie	ernie	PROPN
fcis-4145	35	5	tiny	tiny	ADJ
fcis-4145	35	6	,	,	PUNCT
fcis-4145	35	7	proposed	propose	VERB
fcis-4145	35	8	by	by	ADP
fcis-4145	35	9	baidu	baidu	PROPN
fcis-4145	35	10	,	,	PUNCT
fcis-4145	35	11	is	be	AUX
fcis-4145	35	12	obtained	obtain	VERB
fcis-4145	35	13	by	by	ADP
fcis-4145	35	14	compressing	compress	VERB
fcis-4145	35	15	ernie	ernie	PROPN
fcis-4145	35	16	2.0	2.0	NUM
fcis-4145	35	17	base	base	PROPN
fcis-4145	35	18	model	model	NOUN
fcis-4145	35	19	through	through	ADP
fcis-4145	35	20	the	the	DET
fcis-4145	35	21	method	method	NOUN
fcis-4145	35	22	of	of	ADP
fcis-4145	35	23	model	model	NOUN
fcis-4145	35	24	structure	structure	NOUN
fcis-4145	35	25	compression	compression	NOUN
fcis-4145	35	26	and	and	CCONJ
fcis-4145	35	27	model	model	NOUN
fcis-4145	35	28	distillation	distillation	NOUN
fcis-4145	35	29	,	,	PUNCT
fcis-4145	35	30	using	use	VERB
fcis-4145	35	31	3layer	3layer	NUM
fcis-4145	35	32	transformer	transformer	NOUN
fcis-4145	35	33	structure	structure	NOUN
fcis-4145	35	34	,	,	PUNCT
fcis-4145	35	35	which	which	PRON
fcis-4145	35	36	significantly	significantly	ADV
fcis-4145	35	37	improves	improve	VERB
fcis-4145	35	38	the	the	DET
fcis-4145	35	39	prediction	prediction	NOUN
fcis-4145	35	40	speed	speed	NOUN
fcis-4145	35	41	by	by	ADP
fcis-4145	35	42	4	4	NUM
fcis-4145	35	43	times	time	NOUN
fcis-4145	35	44	and	and	CCONJ
fcis-4145	35	45	can	can	AUX
fcis-4145	35	46	quickly	quickly	ADV
fcis-4145	35	47	land	land	VERB
fcis-4145	35	48	the	the	DET
fcis-4145	35	49	project	project	NOUN
fcis-4145	35	50	.	.	PUNCT
fcis-4145	36	1	and	and	CCONJ
fcis-4145	36	2	ernie	ernie	PROPN
fcis-4145	36	3	2.0	2.0	NUM
fcis-4145	36	4	consists	consist	VERB
fcis-4145	36	5	of	of	ADP
fcis-4145	36	6	two	two	NUM
fcis-4145	36	7	parts	part	NOUN
fcis-4145	36	8	,	,	PUNCT
fcis-4145	36	9	transformer	transformer	ADJ
fcis-4145	36	10	encoder	encoder	NOUN
fcis-4145	36	11	and	and	CCONJ
fcis-4145	36	12	task	task	NOUN
fcis-4145	36	13	embedding	embed	VERB
fcis-4145	36	14	,	,	PUNCT
fcis-4145	36	15	which	which	PRON
fcis-4145	36	16	improves	improve	VERB
fcis-4145	36	17	the	the	DET
fcis-4145	36	18	masking	masking	NOUN
fcis-4145	36	19	strategy	strategy	NOUN
fcis-4145	36	20	and	and	CCONJ
fcis-4145	36	21	corpus	corpus	NOUN
fcis-4145	36	22	on	on	ADP
fcis-4145	36	23	the	the	DET
fcis-4145	36	24	basis	basis	NOUN
fcis-4145	36	25	of	of	ADP
fcis-4145	36	26	bert	bert	NOUN
fcis-4145	36	27	,	,	PUNCT
fcis-4145	36	28	making	make	VERB
fcis-4145	36	29	it	it	PRON
fcis-4145	36	30	more	more	ADV
fcis-4145	36	31	suitable	suitable	ADJ
fcis-4145	36	32	for	for	ADP
fcis-4145	36	33	chinese	chinese	ADJ
fcis-4145	36	34	sentiment	sentiment	NOUN
fcis-4145	36	35	classification	classification	NOUN
fcis-4145	36	36	tasks	task	NOUN
fcis-4145	36	37	.	.	PUNCT
fcis-4145	37	1	(	(	PUNCT
fcis-4145	37	2	1	1	X
fcis-4145	37	3	)	)	PUNCT
fcis-4145	37	4	transformer	transformer	NOUN
fcis-4145	37	5	encoder	encoder	NOUN
fcis-4145	37	6	consistent	consistent	ADJ
fcis-4145	37	7	with	with	ADP
fcis-4145	37	8	several	several	ADJ
fcis-4145	37	9	pre	pre	ADJ
fcis-4145	37	10	-	-	ADJ
fcis-4145	37	11	trained	train	VERB
fcis-4145	37	12	models	model	NOUN
fcis-4145	37	13	such	such	ADJ
fcis-4145	37	14	as	as	ADP
fcis-4145	37	15	bert	bert	PROPN
fcis-4145	37	16	,	,	PUNCT
fcis-4145	37	17	the	the	DET
fcis-4145	37	18	ernie	ernie	PROPN
fcis-4145	37	19	2.0	2.0	NUM
fcis-4145	37	20	model	model	NOUN
fcis-4145	37	21	also	also	ADV
fcis-4145	37	22	uses	use	VERB
fcis-4145	37	23	a	a	DET
fcis-4145	37	24	12	12	NUM
fcis-4145	37	25	-	-	PUNCT
fcis-4145	37	26	layer	layer	NOUN
fcis-4145	37	27	transformer	transformer	NOUN
fcis-4145	37	28	as	as	ADP
fcis-4145	37	29	encoder	encoder	NOUN
fcis-4145	37	30	.	.	PUNCT
fcis-4145	38	1	unlike	unlike	ADP
fcis-4145	38	2	bert	bert	PROPN
fcis-4145	38	3	,	,	PUNCT
fcis-4145	38	4	the	the	DET
fcis-4145	38	5	transformer	transformer	NOUN
fcis-4145	38	6	encoding	encoding	NOUN
fcis-4145	38	7	layer	layer	NOUN
fcis-4145	38	8	of	of	ADP
fcis-4145	38	9	bert	bert	PROPN
fcis-4145	38	10	is	be	AUX
fcis-4145	38	11	used	use	VERB
fcis-4145	38	12	in	in	ADP
fcis-4145	38	13	the	the	DET
fcis-4145	38	14	first	first	ADJ
fcis-4145	38	15	6	6	NUM
fcis-4145	38	16	layers	layer	NOUN
fcis-4145	38	17	,	,	PUNCT
fcis-4145	38	18	but	but	CCONJ
fcis-4145	38	19	in	in	ADP
fcis-4145	38	20	the	the	DET
fcis-4145	38	21	7th	7th	ADJ
fcis-4145	38	22	custom	custom	NOUN
fcis-4145	38	23	knowledge	knowledge	NOUN
fcis-4145	38	24	fusion	fusion	NOUN
fcis-4145	38	25	layer	layer	NOUN
fcis-4145	38	26	bertlayermix	bertlayermix	NOUN
fcis-4145	38	27	,	,	PUNCT
fcis-4145	38	28	the	the	DET
fcis-4145	38	29	aligned	aligned	ADJ
fcis-4145	38	30	entity	entity	NOUN
fcis-4145	38	31	vector	vector	NOUN
fcis-4145	38	32	and	and	CCONJ
fcis-4145	38	33	the	the	DET
fcis-4145	38	34	denominator	denominator	NOUN
fcis-4145	38	35	vector	vector	NOUN
fcis-4145	38	36	are	be	AUX
fcis-4145	38	37	summed	sum	VERB
fcis-4145	38	38	for	for	ADP
fcis-4145	38	39	the	the	DET
fcis-4145	38	40	first	first	ADJ
fcis-4145	38	41	time	time	NOUN
fcis-4145	38	42	and	and	CCONJ
fcis-4145	38	43	transmitted	transmit	VERB
fcis-4145	38	44	to	to	ADP
fcis-4145	38	45	the	the	DET
fcis-4145	38	46	knowledge	knowledge	NOUN
fcis-4145	38	47	encoding	encoding	NOUN
fcis-4145	38	48	module	module	NOUN
fcis-4145	38	49	and	and	CCONJ
fcis-4145	38	50	the	the	DET
fcis-4145	38	51	text	text	NOUN
fcis-4145	38	52	encoding	encoding	NOUN
fcis-4145	38	53	module	module	NOUN
fcis-4145	38	54	,	,	PUNCT
fcis-4145	38	55	respectively	respectively	ADV
fcis-4145	38	56	,	,	PUNCT
fcis-4145	38	57	and	and	CCONJ
fcis-4145	38	58	in	in	ADP
fcis-4145	38	59	the	the	DET
fcis-4145	38	60	remaining	remain	VERB
fcis-4145	38	61	5	5	NUM
fcis-4145	38	62	custom	custom	NOUN
fcis-4145	38	63	knowledge	knowledge	NOUN
fcis-4145	38	64	encoding	encode	VERB
fcis-4145	38	65	layers	layer	NOUN
fcis-4145	38	66	bertlayer	bertlayer	PROPN
fcis-4145	38	67	,	,	PUNCT
fcis-4145	38	68	the	the	DET
fcis-4145	38	69	entity	entity	NOUN
fcis-4145	38	70	sequence	sequence	NOUN
fcis-4145	38	71	and	and	CCONJ
fcis-4145	38	72	the	the	DET
fcis-4145	38	73	text	text	NOUN
fcis-4145	38	74	sequence	sequence	NOUN
fcis-4145	38	75	,	,	PUNCT
fcis-4145	38	76	which	which	PRON
fcis-4145	38	77	have	have	AUX
fcis-4145	38	78	been	be	AUX
fcis-4145	38	79	fused	fuse	VERB
fcis-4145	38	80	with	with	ADP
fcis-4145	38	81	both	both	DET
fcis-4145	38	82	information	information	NOUN
fcis-4145	38	83	,	,	PUNCT
fcis-4145	38	84	are	be	AUX
fcis-4145	38	85	encoded	encode	VERB
fcis-4145	38	86	using	use	VERB
fcis-4145	38	87	the	the	DET
fcis-4145	38	88	self	self	NOUN
fcis-4145	38	89	-	-	PUNCT
fcis-4145	38	90	attention	attention	NOUN
fcis-4145	38	91	mechanism	mechanism	NOUN
fcis-4145	38	92	,	,	PUNCT
fcis-4145	38	93	respectively	respectively	ADV
fcis-4145	38	94	.	.	PUNCT
fcis-4145	39	1	transformer	transformer	NOUN
fcis-4145	39	2	captures	capture	VERB
fcis-4145	39	3	the	the	DET
fcis-4145	39	4	contextual	contextual	ADJ
fcis-4145	39	5	information	information	NOUN
fcis-4145	39	6	of	of	ADP
fcis-4145	39	7	each	each	PRON
fcis-4145	39	8	token	token	VERB
fcis-4145	39	9	in	in	ADP
fcis-4145	39	10	a	a	DET
fcis-4145	39	11	text	text	NOUN
fcis-4145	39	12	sequence	sequence	NOUN
fcis-4145	39	13	by	by	ADP
fcis-4145	39	14	self	self	NOUN
fcis-4145	39	15	-	-	PUNCT
fcis-4145	39	16	attention	attention	NOUN
fcis-4145	39	17	and	and	CCONJ
fcis-4145	39	18	generates	generate	VERB
fcis-4145	39	19	contextual	contextual	ADJ
fcis-4145	39	20	representation	representation	NOUN
fcis-4145	39	21	embeddings	embedding	NOUN
fcis-4145	39	22	.	.	PUNCT
fcis-4145	40	1	for	for	ADP
fcis-4145	40	2	a	a	DET
fcis-4145	40	3	given	give	VERB
fcis-4145	40	4	sequence	sequence	NOUN
fcis-4145	40	5	,	,	PUNCT
fcis-4145	40	6	the	the	DET
fcis-4145	40	7	starting	starting	NOUN
fcis-4145	40	8	position	position	NOUN
fcis-4145	40	9	is	be	AUX
fcis-4145	40	10	a	a	DET
fcis-4145	40	11	predefined	predefine	VERB
fcis-4145	40	12	separator	separator	NOUN
fcis-4145	41	1	[	[	X
fcis-4145	41	2	cls	cls	NOUN
fcis-4145	41	3	]	]	X
fcis-4145	41	4	;	;	PUNCT
fcis-4145	41	5	for	for	ADP
fcis-4145	41	6	tasks	task	NOUN
fcis-4145	41	7	where	where	SCONJ
fcis-4145	41	8	the	the	DET
fcis-4145	41	9	input	input	NOUN
fcis-4145	41	10	is	be	AUX
fcis-4145	41	11	multi	multi	ADJ
fcis-4145	41	12	-	-	ADJ
fcis-4145	41	13	segment	segment	ADJ
fcis-4145	41	14	,	,	PUNCT
fcis-4145	41	15	the	the	DET
fcis-4145	41	16	different	different	ADJ
fcis-4145	41	17	segments	segment	NOUN
fcis-4145	41	18	are	be	AUX
fcis-4145	41	19	separated	separate	VERB
fcis-4145	41	20	by	by	ADP
fcis-4145	41	21	a	a	DET
fcis-4145	41	22	predefined	predefine	VERB
fcis-4145	41	23	[	[	X
fcis-4145	41	24	sep	sep	NOUN
fcis-4145	41	25	]	]	X
fcis-4145	41	26	.	.	PUNCT
fcis-4145	42	1	(	(	PUNCT
fcis-4145	42	2	2	2	X
fcis-4145	42	3	)	)	PUNCT
fcis-4145	42	4	task	task	NOUN
fcis-4145	42	5	embedding	embed	VERB
fcis-4145	42	6	the	the	DET
fcis-4145	42	7	task	task	NOUN
fcis-4145	42	8	embedding	embed	VERB
fcis-4145	42	9	in	in	ADP
fcis-4145	42	10	the	the	DET
fcis-4145	42	11	model	model	NOUN
fcis-4145	42	12	is	be	AUX
fcis-4145	42	13	used	use	VERB
fcis-4145	42	14	to	to	PART
fcis-4145	42	15	apply	apply	VERB
fcis-4145	42	16	different	different	ADJ
fcis-4145	42	17	tasks	task	NOUN
fcis-4145	42	18	,	,	PUNCT
fcis-4145	42	19	and	and	CCONJ
fcis-4145	42	20	each	each	DET
fcis-4145	42	21	task	task	NOUN
fcis-4145	42	22	i	i	PROPN
fcis-4145	42	23	d	d	PROPN
fcis-4145	42	24	has	have	VERB
fcis-4145	42	25	its	its	PRON
fcis-4145	42	26	own	own	ADJ
fcis-4145	42	27	specific	specific	ADJ
fcis-4145	42	28	task	task	NOUN
fcis-4145	42	29	embedding	embed	VERB
fcis-4145	42	30	,	,	PUNCT
fcis-4145	42	31	which	which	PRON
fcis-4145	42	32	is	be	AUX
fcis-4145	42	33	input	input	NOUN
fcis-4145	42	34	to	to	ADP
fcis-4145	42	35	the	the	DET
fcis-4145	42	36	model	model	NOUN
fcis-4145	42	37	along	along	ADP
fcis-4145	42	38	with	with	ADP
fcis-4145	42	39	token	token	ADJ
fcis-4145	42	40	embedding	embed	VERB
fcis-4145	42	41	,	,	PUNCT
fcis-4145	42	42	position	position	NOUN
fcis-4145	42	43	embedding	embed	VERB
fcis-4145	42	44	and	and	CCONJ
fcis-4145	42	45	sentence	sentence	NOUN
fcis-4145	42	46	embedding	embed	VERB
fcis-4145	42	47	.	.	PUNCT
fcis-4145	43	1	in	in	ADP
fcis-4145	43	2	the	the	DET
fcis-4145	43	3	finetuning	finetuning	NOUN
fcis-4145	43	4	phase	phase	NOUN
fcis-4145	43	5	,	,	PUNCT
fcis-4145	43	6	task	task	NOUN
fcis-4145	43	7	ids	id	NOUN
fcis-4145	43	8	can	can	AUX
fcis-4145	43	9	be	be	AUX
fcis-4145	43	10	selected	select	VERB
fcis-4145	43	11	to	to	PART
fcis-4145	43	12	initialize	initialize	VERB
fcis-4145	43	13	the	the	DET
fcis-4145	43	14	model	model	NOUN
fcis-4145	43	15	.	.	PUNCT
fcis-4145	44	1	and	and	CCONJ
fcis-4145	44	2	the	the	DET
fcis-4145	44	3	sentence	sentence	NOUN
fcis-4145	44	4	loss	loss	NOUN
fcis-4145	44	5	and	and	CCONJ
fcis-4145	44	6	token	token	VERB
fcis-4145	44	7	loss	loss	NOUN
fcis-4145	44	8	are	be	AUX
fcis-4145	44	9	used	use	VERB
fcis-4145	44	10	as	as	ADP
fcis-4145	44	11	the	the	DET
fcis-4145	44	12	loss	loss	NOUN
fcis-4145	44	13	function	function	NOUN
fcis-4145	44	14	.	.	PUNCT
fcis-4145	45	1	the	the	DET
fcis-4145	45	2	structure	structure	NOUN
fcis-4145	45	3	of	of	ADP
fcis-4145	45	4	ernie	ernie	PROPN
fcis-4145	45	5	2.0	2.0	NUM
fcis-4145	45	6	model	model	NOUN
fcis-4145	45	7	is	be	AUX
fcis-4145	45	8	shown	show	VERB
fcis-4145	45	9	in	in	ADP
fcis-4145	45	10	figure	figure	NOUN
fcis-4145	45	11	1	1	NUM
fcis-4145	45	12	.	.	PUNCT
fcis-4145	45	13	figure	figure	NOUN
fcis-4145	45	14	1	1	NUM
fcis-4145	45	15	.	.	PUNCT
fcis-4145	46	1	ernie	ernie	PROPN
fcis-4145	46	2	2.0	2.0	NUM
fcis-4145	46	3	model	model	NOUN
fcis-4145	46	4	structure	structure	NOUN
fcis-4145	46	5	(	(	PUNCT
fcis-4145	46	6	3	3	X
fcis-4145	46	7	)	)	PUNCT
fcis-4145	46	8	model	model	NOUN
fcis-4145	46	9	distillation	distillation	NOUN
fcis-4145	46	10	ernie	ernie	PROPN
fcis-4145	46	11	tiny	tiny	ADJ
fcis-4145	46	12	compresses	compress	VERB
fcis-4145	46	13	the	the	DET
fcis-4145	46	14	pre	pre	ADJ
fcis-4145	46	15	-	-	ADJ
fcis-4145	46	16	trained	train	VERB
fcis-4145	46	17	model	model	NOUN
fcis-4145	46	18	through	through	ADP
fcis-4145	46	19	a	a	DET
fcis-4145	46	20	progressive	progressive	ADJ
fcis-4145	46	21	distillation	distillation	NOUN
fcis-4145	46	22	framework	framework	NOUN
fcis-4145	46	23	using	use	VERB
fcis-4145	46	24	four	four	NUM
fcis-4145	46	25	stages	stage	NOUN
fcis-4145	46	26	of	of	ADP
fcis-4145	46	27	distillation	distillation	NOUN
fcis-4145	46	28	,	,	PUNCT
fcis-4145	46	29	including	include	VERB
fcis-4145	46	30	general	general	ADJ
fcis-4145	46	31	-	-	PUNCT
fcis-4145	46	32	enhanced	enhance	VERB
fcis-4145	46	33	distillation	distillation	NOUN
fcis-4145	46	34	,	,	PUNCT
fcis-4145	46	35	where	where	SCONJ
fcis-4145	46	36	the	the	DET
fcis-4145	46	37	student	student	NOUN
fcis-4145	46	38	model	model	NOUN
fcis-4145	46	39	receives	receive	VERB
fcis-4145	46	40	augmented	augment	VERB
fcis-4145	46	41	knowledge	knowledge	NOUN
fcis-4145	46	42	from	from	ADP
fcis-4145	46	43	finetuned	finetune	VERB
fcis-4145	46	44	teachers	teacher	NOUN
fcis-4145	46	45	and	and	CCONJ
fcis-4145	46	46	generalization	generalization	NOUN
fcis-4145	46	47	capability	capability	NOUN
fcis-4145	46	48	is	be	AUX
fcis-4145	46	49	improved	improve	VERB
fcis-4145	46	50	.	.	PUNCT
fcis-4145	47	1	task	task	NOUN
fcis-4145	47	2	-	-	PUNCT
fcis-4145	47	3	adaptive	adaptive	ADJ
fcis-4145	47	4	distillation	distillation	NOUN
fcis-4145	47	5	smooths	smooth	VERB
fcis-4145	47	6	the	the	DET
fcis-4145	47	7	transition	transition	NOUN
fcis-4145	47	8	through	through	ADP
fcis-4145	47	9	designed	design	VERB
fcis-4145	47	10	learning	learning	NOUN
fcis-4145	47	11	objectives	objective	NOUN
fcis-4145	47	12	.	.	PUNCT
fcis-4145	48	1	figure	figure	NOUN
fcis-4145	48	2	2	2	NUM
fcis-4145	48	3	shows	show	VERB
fcis-4145	48	4	the	the	DET
fcis-4145	48	5	model	model	NOUN
fcis-4145	48	6	distillation	distillation	NOUN
fcis-4145	48	7	work	work	NOUN
fcis-4145	48	8	flow	flow	NOUN
fcis-4145	48	9	of	of	ADP
fcis-4145	48	10	ernie	ernie	PROPN
fcis-4145	48	11	tiny	tiny	PROPN
fcis-4145	48	12	.	.	PUNCT
fcis-4145	49	1	figure	figure	NOUN
fcis-4145	49	2	2	2	NUM
fcis-4145	49	3	.	.	PUNCT
fcis-4145	49	4	workflow	workflow	NOUN
fcis-4145	49	5	of	of	ADP
fcis-4145	49	6	ernie	ernie	PROPN
fcis-4145	49	7	tiny	tiny	ADJ
fcis-4145	49	8	based	base	VERB
fcis-4145	49	9	on	on	ADP
fcis-4145	49	10	previous	previous	ADJ
fcis-4145	49	11	studies	study	NOUN
fcis-4145	49	12	,	,	PUNCT
fcis-4145	49	13	it	it	PRON
fcis-4145	49	14	can	can	AUX
fcis-4145	49	15	be	be	AUX
fcis-4145	49	16	seen	see	VERB
fcis-4145	49	17	that	that	SCONJ
fcis-4145	49	18	deep	deep	ADJ
fcis-4145	49	19	learning	learning	NOUN
fcis-4145	49	20	models	model	NOUN
fcis-4145	49	21	,	,	PUNCT
fcis-4145	49	22	especially	especially	ADV
fcis-4145	49	23	pre	pre	ADJ
fcis-4145	49	24	-	-	ADJ
fcis-4145	49	25	trained	train	VERB
fcis-4145	49	26	models	model	NOUN
fcis-4145	49	27	like	like	ADP
fcis-4145	49	28	bert	bert	PROPN
fcis-4145	49	29	,	,	PUNCT
fcis-4145	49	30	have	have	VERB
fcis-4145	49	31	superior	superior	ADJ
fcis-4145	49	32	results	result	NOUN
fcis-4145	49	33	for	for	ADP
fcis-4145	49	34	sentiment	sentiment	NOUN
fcis-4145	49	35	classification	classification	NOUN
fcis-4145	49	36	tasks	task	NOUN
fcis-4145	49	37	.	.	PUNCT
fcis-4145	50	1	at	at	ADP
fcis-4145	50	2	the	the	DET
fcis-4145	50	3	same	same	ADJ
fcis-4145	50	4	time	time	NOUN
fcis-4145	50	5	,	,	PUNCT
fcis-4145	50	6	there	there	PRON
fcis-4145	50	7	are	be	VERB
fcis-4145	50	8	fewer	few	ADJ
fcis-4145	50	9	studies	study	NOUN
fcis-4145	50	10	on	on	ADP
fcis-4145	50	11	the	the	DET
fcis-4145	50	12	psychological	psychological	ADJ
fcis-4145	50	13	emotion	emotion	NOUN
fcis-4145	50	14	analysis	analysis	NOUN
fcis-4145	50	15	for	for	ADP
fcis-4145	50	16	the	the	DET
fcis-4145	50	17	emotion	emotion	NOUN
fcis-4145	50	18	of	of	ADP
fcis-4145	50	19	lost	lost	ADJ
fcis-4145	50	20	love	love	NOUN
fcis-4145	50	21	,	,	PUNCT
fcis-4145	50	22	so	so	SCONJ
fcis-4145	50	23	this	this	DET
fcis-4145	50	24	paper	paper	NOUN
fcis-4145	50	25	focuses	focus	VERB
fcis-4145	50	26	on	on	ADP
fcis-4145	50	27	68	68	NUM
fcis-4145	50	28	the	the	DET
fcis-4145	50	29	study	study	NOUN
fcis-4145	50	30	of	of	ADP
fcis-4145	50	31	the	the	DET
fcis-4145	50	32	recognition	recognition	NOUN
fcis-4145	50	33	of	of	ADP
fcis-4145	50	34	the	the	DET
fcis-4145	50	35	lovelorn	lovelorn	ADJ
fcis-4145	50	36	emotion	emotion	NOUN
fcis-4145	50	37	using	use	VERB
fcis-4145	50	38	deep	deep	ADJ
fcis-4145	50	39	learning	learning	NOUN
fcis-4145	50	40	model	model	NOUN
fcis-4145	50	41	.	.	PUNCT
fcis-4145	51	1	3.2	3.2	NUM
fcis-4145	51	2	.	.	PUNCT
fcis-4145	51	3	experiment	experiment	NOUN
fcis-4145	51	4	3.2.1	3.2.1	NUM
fcis-4145	51	5	.	.	PUNCT
fcis-4145	52	1	data	datum	NOUN
fcis-4145	52	2	preparation	preparation	NOUN
fcis-4145	52	3	and	and	CCONJ
fcis-4145	52	4	processing	process	VERB
fcis-4145	52	5	the	the	DET
fcis-4145	52	6	first	first	ADJ
fcis-4145	52	7	step	step	NOUN
fcis-4145	52	8	was	be	AUX
fcis-4145	52	9	to	to	PART
fcis-4145	52	10	obtain	obtain	VERB
fcis-4145	52	11	the	the	DET
fcis-4145	52	12	data	datum	NOUN
fcis-4145	52	13	.	.	PUNCT
fcis-4145	53	1	since	since	SCONJ
fcis-4145	53	2	there	there	PRON
fcis-4145	53	3	was	be	VERB
fcis-4145	53	4	no	no	DET
fcis-4145	53	5	readily	readily	ADV
fcis-4145	53	6	available	available	ADJ
fcis-4145	53	7	textual	textual	ADJ
fcis-4145	53	8	dataset	dataset	NOUN
fcis-4145	53	9	for	for	ADP
fcis-4145	53	10	love	love	NOUN
fcis-4145	53	11	loss	loss	NOUN
fcis-4145	53	12	,	,	PUNCT
fcis-4145	53	13	a	a	DET
fcis-4145	53	14	web	web	NOUN
fcis-4145	53	15	crawler	crawler	NOUN
fcis-4145	53	16	program	program	NOUN
fcis-4145	53	17	was	be	AUX
fcis-4145	53	18	written	write	VERB
fcis-4145	53	19	to	to	PART
fcis-4145	53	20	crawl	crawl	VERB
fcis-4145	53	21	the	the	DET
fcis-4145	53	22	comments	comment	NOUN
fcis-4145	53	23	of	of	ADP
fcis-4145	53	24	articles	article	NOUN
fcis-4145	53	25	about	about	ADP
fcis-4145	53	26	"	"	PUNCT
fcis-4145	53	27	love	love	NOUN
fcis-4145	53	28	loss	loss	NOUN
fcis-4145	53	29	"	"	PUNCT
fcis-4145	53	30	,	,	PUNCT
fcis-4145	53	31	"	"	PUNCT
fcis-4145	53	32	regret	regret	NOUN
fcis-4145	53	33	of	of	ADP
fcis-4145	53	34	youth	youth	NOUN
fcis-4145	53	35	"	"	PUNCT
fcis-4145	53	36	and	and	CCONJ
fcis-4145	53	37	other	other	ADJ
fcis-4145	53	38	related	related	ADJ
fcis-4145	53	39	words	word	NOUN
fcis-4145	53	40	on	on	ADP
fcis-4145	53	41	chinese	chinese	ADJ
fcis-4145	53	42	social	social	ADJ
fcis-4145	53	43	media	medium	NOUN
fcis-4145	53	44	platforms	platform	NOUN
fcis-4145	53	45	weibo	weibo	VERB
fcis-4145	53	46	.	.	PUNCT
fcis-4145	54	1	then	then	ADV
fcis-4145	54	2	data	datum	NOUN
fcis-4145	54	3	cleaning	cleaning	NOUN
fcis-4145	54	4	was	be	AUX
fcis-4145	54	5	performed	perform	VERB
fcis-4145	54	6	,	,	PUNCT
fcis-4145	54	7	using	use	VERB
fcis-4145	54	8	python	python	NOUN
fcis-4145	54	9	to	to	PART
fcis-4145	54	10	remove	remove	VERB
fcis-4145	54	11	the	the	DET
fcis-4145	54	12	empty	empty	ADJ
fcis-4145	54	13	and	and	CCONJ
fcis-4145	54	14	duplicate	duplicate	ADJ
fcis-4145	54	15	lines	line	NOUN
fcis-4145	54	16	in	in	ADP
fcis-4145	54	17	the	the	DET
fcis-4145	54	18	comments	comment	NOUN
fcis-4145	54	19	,	,	PUNCT
fcis-4145	54	20	and	and	CCONJ
fcis-4145	54	21	then	then	ADV
fcis-4145	54	22	manually	manually	ADV
fcis-4145	54	23	filter	filter	VERB
fcis-4145	54	24	the	the	DET
fcis-4145	54	25	comments	comment	NOUN
fcis-4145	54	26	that	that	PRON
fcis-4145	54	27	are	be	AUX
fcis-4145	54	28	not	not	PART
fcis-4145	54	29	related	relate	VERB
fcis-4145	54	30	to	to	ADP
fcis-4145	54	31	the	the	DET
fcis-4145	54	32	emotion	emotion	NOUN
fcis-4145	54	33	of	of	ADP
fcis-4145	54	34	lost	lost	ADJ
fcis-4145	54	35	love	love	NOUN
fcis-4145	54	36	,	,	PUNCT
fcis-4145	54	37	and	and	CCONJ
fcis-4145	54	38	finally	finally	ADV
fcis-4145	54	39	got	get	VERB
fcis-4145	54	40	5008	5008	NUM
fcis-4145	54	41	pieces	piece	NOUN
fcis-4145	54	42	of	of	ADP
fcis-4145	54	43	text	text	NOUN
fcis-4145	54	44	with	with	ADP
fcis-4145	54	45	the	the	DET
fcis-4145	54	46	emotion	emotion	NOUN
fcis-4145	54	47	of	of	ADP
fcis-4145	54	48	lovelorn	lovelorn	ADJ
fcis-4145	54	49	,	,	PUNCT
fcis-4145	54	50	and	and	CCONJ
fcis-4145	54	51	set	set	VERB
fcis-4145	54	52	the	the	DET
fcis-4145	54	53	label	label	NOUN
fcis-4145	54	54	to	to	ADP
fcis-4145	54	55	1	1	NUM
fcis-4145	54	56	,	,	PUNCT
fcis-4145	54	57	which	which	PRON
fcis-4145	54	58	means	mean	VERB
fcis-4145	54	59	it	it	PRON
fcis-4145	54	60	contains	contain	VERB
fcis-4145	54	61	the	the	DET
fcis-4145	54	62	emotion	emotion	NOUN
fcis-4145	54	63	of	of	ADP
fcis-4145	54	64	lost	lost	ADJ
fcis-4145	54	65	love	love	NOUN
fcis-4145	54	66	.	.	PUNCT
fcis-4145	55	1	the	the	DET
fcis-4145	55	2	next	next	ADJ
fcis-4145	55	3	step	step	NOUN
fcis-4145	55	4	was	be	AUX
fcis-4145	55	5	merging	merge	VERB
fcis-4145	55	6	the	the	DET
fcis-4145	55	7	data	datum	NOUN
fcis-4145	55	8	.	.	PUNCT
fcis-4145	56	1	the	the	DET
fcis-4145	56	2	comments	comment	NOUN
fcis-4145	56	3	from	from	ADP
fcis-4145	56	4	the	the	DET
fcis-4145	56	5	existing	exist	VERB
fcis-4145	56	6	general	general	ADJ
fcis-4145	56	7	weibo	weibo	NOUN
fcis-4145	56	8	sentiment	sentiment	NOUN
fcis-4145	56	9	seven	seven	NUM
fcis-4145	56	10	-	-	PUNCT
fcis-4145	56	11	category	category	NOUN
fcis-4145	56	12	comment	comment	NOUN
fcis-4145	56	13	dataset	dataset	VERB
fcis-4145	56	14	ocemotion	ocemotion	NOUN
fcis-4145	56	15	[	[	X
fcis-4145	56	16	10	10	NUM
fcis-4145	56	17	]	]	PUNCT
fcis-4145	56	18	,	,	PUNCT
fcis-4145	56	19	which	which	PRON
fcis-4145	56	20	is	be	AUX
fcis-4145	56	21	not	not	PART
fcis-4145	56	22	specific	specific	ADJ
fcis-4145	56	23	to	to	ADP
fcis-4145	56	24	any	any	DET
fcis-4145	56	25	topic	topic	NOUN
fcis-4145	56	26	,	,	PUNCT
fcis-4145	56	27	was	be	AUX
fcis-4145	56	28	extracted	extract	VERB
fcis-4145	56	29	and	and	CCONJ
fcis-4145	56	30	merged	merge	VERB
fcis-4145	56	31	with	with	ADP
fcis-4145	56	32	the	the	DET
fcis-4145	56	33	lost	lose	VERB
fcis-4145	56	34	love	love	NOUN
fcis-4145	56	35	sentiment	sentiment	NOUN
fcis-4145	56	36	text	text	NOUN
fcis-4145	56	37	into	into	ADP
fcis-4145	56	38	a	a	DET
fcis-4145	56	39	csv	csv	NOUN
fcis-4145	56	40	file	file	NOUN
fcis-4145	56	41	.	.	PUNCT
fcis-4145	57	1	since	since	SCONJ
fcis-4145	57	2	ordinary	ordinary	ADJ
fcis-4145	57	3	comments	comment	NOUN
fcis-4145	57	4	may	may	AUX
fcis-4145	57	5	also	also	ADV
fcis-4145	57	6	contain	contain	VERB
fcis-4145	57	7	comments	comment	NOUN
fcis-4145	57	8	related	relate	VERB
fcis-4145	57	9	to	to	PART
fcis-4145	57	10	love	love	VERB
fcis-4145	57	11	loss	loss	NOUN
fcis-4145	57	12	,	,	PUNCT
fcis-4145	57	13	4998	4998	NUM
fcis-4145	57	14	comments	comment	NOUN
fcis-4145	57	15	under	under	ADP
fcis-4145	57	16	the	the	DET
fcis-4145	57	17	"	"	PUNCT
fcis-4145	57	18	happiness	happiness	NOUN
fcis-4145	57	19	"	"	PUNCT
fcis-4145	57	20	sentiment	sentiment	NOUN
fcis-4145	57	21	tag	tag	NOUN
fcis-4145	57	22	were	be	AUX
fcis-4145	57	23	selected	select	VERB
fcis-4145	57	24	as	as	ADP
fcis-4145	57	25	nonlove	nonlove	NOUN
fcis-4145	57	26	-	-	PUNCT
fcis-4145	57	27	loss	loss	NOUN
fcis-4145	57	28	sentiment	sentiment	NOUN
fcis-4145	57	29	data	datum	NOUN
fcis-4145	57	30	,	,	PUNCT
fcis-4145	57	31	and	and	CCONJ
fcis-4145	57	32	the	the	DET
fcis-4145	57	33	tag	tag	NOUN
fcis-4145	57	34	was	be	AUX
fcis-4145	57	35	set	set	VERB
fcis-4145	57	36	to	to	ADP
fcis-4145	57	37	0	0	NUM
fcis-4145	57	38	,	,	PUNCT
fcis-4145	57	39	resulting	result	VERB
fcis-4145	57	40	in	in	ADP
fcis-4145	57	41	a	a	DET
fcis-4145	57	42	dichotomous	dichotomous	ADJ
fcis-4145	57	43	(	(	PUNCT
fcis-4145	57	44	lovelorn	lovelorn	ADJ
fcis-4145	57	45	and	and	CCONJ
fcis-4145	57	46	normal	normal	ADJ
fcis-4145	57	47	)	)	PUNCT
fcis-4145	57	48	weibo	weibo	NOUN
fcis-4145	57	49	sentiment	sentiment	NOUN
fcis-4145	57	50	dataset	dataset	VERB
fcis-4145	57	51	.	.	PUNCT
fcis-4145	58	1	the	the	DET
fcis-4145	58	2	sample	sample	NOUN
fcis-4145	58	3	comments	comment	NOUN
fcis-4145	58	4	were	be	AUX
fcis-4145	58	5	translated	translate	VERB
fcis-4145	58	6	into	into	ADP
fcis-4145	58	7	english	english	PROPN
fcis-4145	58	8	,	,	PUNCT
fcis-4145	58	9	as	as	SCONJ
fcis-4145	58	10	shown	show	VERB
fcis-4145	58	11	in	in	ADP
fcis-4145	58	12	table	table	NOUN
fcis-4145	58	13	1	1	NUM
fcis-4145	58	14	.	.	PUNCT
fcis-4145	59	1	finally	finally	ADV
fcis-4145	59	2	,	,	PUNCT
fcis-4145	59	3	the	the	DET
fcis-4145	59	4	dataset	dataset	NOUN
fcis-4145	59	5	was	be	AUX
fcis-4145	59	6	divided	divide	VERB
fcis-4145	59	7	.	.	PUNCT
fcis-4145	60	1	the	the	DET
fcis-4145	60	2	training	training	NOUN
fcis-4145	60	3	set	set	NOUN
fcis-4145	60	4	,	,	PUNCT
fcis-4145	60	5	validation	validation	NOUN
fcis-4145	60	6	set	set	NOUN
fcis-4145	60	7	and	and	CCONJ
fcis-4145	60	8	test	test	NOUN
fcis-4145	60	9	set	set	VERB
fcis-4145	60	10	were	be	AUX
fcis-4145	60	11	divided	divide	VERB
fcis-4145	60	12	according	accord	VERB
fcis-4145	60	13	to	to	ADP
fcis-4145	60	14	8:1:1	8:1:1	PROPN
fcis-4145	60	15	and	and	CCONJ
fcis-4145	60	16	the	the	DET
fcis-4145	60	17	data	datum	NOUN
fcis-4145	60	18	were	be	AUX
fcis-4145	60	19	randomly	randomly	ADV
fcis-4145	60	20	shuffled	shuffle	VERB
fcis-4145	60	21	,	,	PUNCT
fcis-4145	60	22	making	make	VERB
fcis-4145	60	23	the	the	DET
fcis-4145	60	24	data	datum	NOUN
fcis-4145	60	25	of	of	ADP
fcis-4145	60	26	different	different	ADJ
fcis-4145	60	27	categories	category	NOUN
fcis-4145	60	28	as	as	ADP
fcis-4145	60	29	uniformly	uniformly	ADV
fcis-4145	60	30	distributed	distribute	VERB
fcis-4145	60	31	as	as	ADP
fcis-4145	60	32	possible	possible	ADJ
fcis-4145	60	33	.	.	PUNCT
fcis-4145	61	1	table	table	NOUN
fcis-4145	61	2	1	1	NUM
fcis-4145	61	3	.	.	PUNCT
fcis-4145	61	4	sample	sample	NOUN
fcis-4145	61	5	review	review	NOUN
fcis-4145	61	6	text	text	NOUN
fcis-4145	61	7	text	text	NOUN
fcis-4145	61	8	label	label	NOUN
fcis-4145	61	9	(	(	PUNCT
fcis-4145	61	10	1	1	NUM
fcis-4145	61	11	:	:	PUNCT
fcis-4145	61	12	lovelorn	lovelorn	ADJ
fcis-4145	61	13	;	;	PUNCT
fcis-4145	61	14	0	0	NUM
fcis-4145	61	15	:	:	PUNCT
fcis-4145	61	16	normal	normal	ADJ
fcis-4145	61	17	)	)	PUNCT
fcis-4145	61	18	no	no	ADV
fcis-4145	61	19	more	more	ADV
fcis-4145	61	20	,	,	PUNCT
fcis-4145	61	21	he	he	PRON
fcis-4145	61	22	said	say	VERB
fcis-4145	61	23	he	he	PRON
fcis-4145	61	24	does	do	AUX
fcis-4145	61	25	n't	not	PART
fcis-4145	61	26	love	love	VERB
fcis-4145	61	27	me	i	PRON
fcis-4145	61	28	anymore	anymore	ADV
fcis-4145	61	29	.	.	PUNCT
fcis-4145	62	1	1	1	NUM
fcis-4145	62	2	i	i	PRON
fcis-4145	62	3	said	say	VERB
fcis-4145	62	4	the	the	DET
fcis-4145	62	5	breakup	breakup	NOUN
fcis-4145	62	6	,	,	PUNCT
fcis-4145	62	7	but	but	CCONJ
fcis-4145	62	8	why	why	SCONJ
fcis-4145	62	9	do	do	AUX
fcis-4145	62	10	i	i	PRON
fcis-4145	62	11	miss	miss	VERB
fcis-4145	62	12	him	he	PRON
fcis-4145	62	13	so	so	ADV
fcis-4145	62	14	much	much	ADV
fcis-4145	62	15	,	,	PUNCT
fcis-4145	62	16	and	and	CCONJ
fcis-4145	62	17	also	also	ADV
fcis-4145	62	18	sad	sad	ADJ
fcis-4145	62	19	,	,	PUNCT
fcis-4145	62	20	i	i	PRON
fcis-4145	62	21	seem	seem	VERB
fcis-4145	62	22	to	to	PART
fcis-4145	62	23	have	have	AUX
fcis-4145	62	24	lost	lose	VERB
fcis-4145	62	25	him	he	PRON
fcis-4145	62	26	.	.	PUNCT
fcis-4145	63	1	1	1	NUM
fcis-4145	63	2	eight	eight	NUM
fcis-4145	63	3	years	year	NOUN
fcis-4145	63	4	,	,	PUNCT
fcis-4145	63	5	how	how	SCONJ
fcis-4145	63	6	can	can	AUX
fcis-4145	63	7	i	i	PRON
fcis-4145	63	8	give	give	VERB
fcis-4145	63	9	up	up	ADP
fcis-4145	63	10	.	.	PUNCT
fcis-4145	64	1	you	you	PRON
fcis-4145	64	2	are	be	AUX
fcis-4145	64	3	so	so	ADV
fcis-4145	64	4	good	good	ADJ
fcis-4145	64	5	,	,	PUNCT
fcis-4145	64	6	so	so	ADV
fcis-4145	64	7	good	good	ADJ
fcis-4145	64	8	that	that	SCONJ
fcis-4145	64	9	i	i	PRON
fcis-4145	64	10	want	want	VERB
fcis-4145	64	11	to	to	PART
fcis-4145	64	12	spend	spend	VERB
fcis-4145	64	13	my	my	PRON
fcis-4145	64	14	life	life	NOUN
fcis-4145	64	15	with	with	ADP
fcis-4145	64	16	you	you	PRON
fcis-4145	64	17	.	.	PUNCT
fcis-4145	65	1	but	but	CCONJ
fcis-4145	65	2	i	i	PRON
fcis-4145	65	3	was	be	AUX
fcis-4145	65	4	disqualified	disqualify	VERB
fcis-4145	65	5	.	.	PUNCT
fcis-4145	66	1	i	i	PRON
fcis-4145	66	2	'm	be	AUX
fcis-4145	66	3	lost	lose	VERB
fcis-4145	66	4	,	,	PUNCT
fcis-4145	66	5	i	i	PRON
fcis-4145	66	6	do	do	AUX
fcis-4145	66	7	n't	not	PART
fcis-4145	66	8	want	want	VERB
fcis-4145	66	9	to	to	PART
fcis-4145	66	10	go	go	VERB
fcis-4145	66	11	the	the	DET
fcis-4145	66	12	rest	rest	NOUN
fcis-4145	66	13	of	of	ADP
fcis-4145	66	14	the	the	DET
fcis-4145	66	15	way	way	NOUN
fcis-4145	66	16	.	.	PUNCT
fcis-4145	67	1	1	1	NUM
fcis-4145	67	2	today	today	NOUN
fcis-4145	67	3	go	go	VERB
fcis-4145	67	4	out	out	ADP
fcis-4145	67	5	to	to	PART
fcis-4145	67	6	play	play	VERB
fcis-4145	67	7	so	so	ADV
fcis-4145	67	8	cool	cool	ADJ
fcis-4145	67	9	ah	ah	INTJ
fcis-4145	67	10	,	,	PUNCT
fcis-4145	67	11	both	both	CCONJ
fcis-4145	67	12	wind	wind	NOUN
fcis-4145	67	13	and	and	CCONJ
fcis-4145	67	14	sun	sun	NOUN
fcis-4145	67	15	and	and	CCONJ
fcis-4145	67	16	sun	sun	NOUN
fcis-4145	67	17	uv	uv	NOUN
fcis-4145	67	18	weak	weak	ADJ
fcis-4145	67	19	.	.	PUNCT
fcis-4145	68	1	0	0	NUM
fcis-4145	68	2	happiness	happiness	NOUN
fcis-4145	68	3	is	be	AUX
fcis-4145	68	4	not	not	PART
fcis-4145	68	5	about	about	ADP
fcis-4145	68	6	having	have	VERB
fcis-4145	68	7	more	more	ADJ
fcis-4145	68	8	,	,	PUNCT
fcis-4145	68	9	but	but	CCONJ
fcis-4145	68	10	about	about	ADP
fcis-4145	68	11	thinking	think	VERB
fcis-4145	68	12	less	less	ADJ
fcis-4145	68	13	.	.	PUNCT
fcis-4145	68	14	0	0	PUNCT
fcis-4145	69	1	wake	wake	VERB
fcis-4145	69	2	up	up	ADP
fcis-4145	69	3	and	and	CCONJ
fcis-4145	69	4	find	find	VERB
fcis-4145	69	5	that	that	PRON
fcis-4145	69	6	can	can	AUX
fcis-4145	69	7	still	still	ADV
fcis-4145	69	8	stay	stay	VERB
fcis-4145	69	9	in	in	ADP
fcis-4145	69	10	bed	bed	NOUN
fcis-4145	69	11	!	!	PUNCT
fcis-4145	69	12	0	0	PUNCT
fcis-4145	70	1	3.2.2	3.2.2	X
fcis-4145	70	2	.	.	PUNCT
fcis-4145	71	1	build	build	VERB
fcis-4145	71	2	and	and	CCONJ
fcis-4145	71	3	train	train	NOUN
fcis-4145	71	4	model	model	NOUN
fcis-4145	71	5	after	after	ADP
fcis-4145	71	6	processing	process	VERB
fcis-4145	71	7	the	the	DET
fcis-4145	71	8	data	datum	NOUN
fcis-4145	71	9	into	into	ADP
fcis-4145	71	10	a	a	DET
fcis-4145	71	11	format	format	NOUN
fcis-4145	71	12	acceptable	acceptable	ADJ
fcis-4145	71	13	to	to	ADP
fcis-4145	71	14	the	the	DET
fcis-4145	71	15	model	model	NOUN
fcis-4145	71	16	,	,	PUNCT
fcis-4145	71	17	using	use	VERB
fcis-4145	71	18	paddlehub	paddlehub	NOUN
fcis-4145	71	19	,	,	PUNCT
fcis-4145	71	20	a	a	DET
fcis-4145	71	21	pre	pre	ADJ
fcis-4145	71	22	-	-	ADJ
fcis-4145	71	23	training	training	ADJ
fcis-4145	71	24	model	model	NOUN
fcis-4145	71	25	management	management	NOUN
fcis-4145	71	26	and	and	CCONJ
fcis-4145	71	27	migration	migration	NOUN
fcis-4145	71	28	learning	learn	VERB
fcis-4145	71	29	tool	tool	NOUN
fcis-4145	71	30	,	,	PUNCT
fcis-4145	71	31	the	the	DET
fcis-4145	71	32	pre	pre	ADJ
fcis-4145	71	33	-	-	ADJ
fcis-4145	71	34	trained	train	VERB
fcis-4145	71	35	model	model	NOUN
fcis-4145	71	36	ernie	ernie	PROPN
fcis-4145	71	37	tiny	tiny	ADJ
fcis-4145	71	38	was	be	AUX
fcis-4145	71	39	loaded	load	VERB
fcis-4145	71	40	and	and	CCONJ
fcis-4145	71	41	the	the	DET
fcis-4145	71	42	text	text	NOUN
fcis-4145	71	43	was	be	AUX
fcis-4145	71	44	processed	process	VERB
fcis-4145	71	45	using	use	VERB
fcis-4145	71	46	the	the	DET
fcis-4145	71	47	model	model	NOUN
fcis-4145	71	48	's	's	PART
fcis-4145	71	49	built	build	VERB
fcis-4145	71	50	-	-	PUNCT
fcis-4145	71	51	in	in	ADP
fcis-4145	71	52	tokenizer	tokenizer	NOUN
fcis-4145	71	53	.	.	PUNCT
fcis-4145	72	1	next	next	ADV
fcis-4145	72	2	,	,	PUNCT
fcis-4145	72	3	the	the	DET
fcis-4145	72	4	adamw	adamw	NOUN
fcis-4145	72	5	optimizer	optimizer	NOUN
fcis-4145	72	6	was	be	AUX
fcis-4145	72	7	selected	select	VERB
fcis-4145	72	8	as	as	ADP
fcis-4145	72	9	the	the	DET
fcis-4145	72	10	optimization	optimization	NOUN
fcis-4145	72	11	strategy	strategy	NOUN
fcis-4145	72	12	for	for	ADP
fcis-4145	72	13	pretrained	pretraine	VERB
fcis-4145	72	14	model	model	PROPN
fcis-4145	72	15	ernie	ernie	PROPN
fcis-4145	72	16	tiny	tiny	ADJ
fcis-4145	72	17	,	,	PUNCT
fcis-4145	72	18	setting	set	VERB
fcis-4145	72	19	the	the	DET
fcis-4145	72	20	learning	learning	NOUN
fcis-4145	72	21	rate	rate	NOUN
fcis-4145	72	22	and	and	CCONJ
fcis-4145	72	23	using	use	VERB
fcis-4145	72	24	the	the	DET
fcis-4145	72	25	model	model	NOUN
fcis-4145	72	26	default	default	NOUN
fcis-4145	72	27	parameters	parameter	NOUN
fcis-4145	72	28	.	.	PUNCT
fcis-4145	73	1	then	then	ADV
fcis-4145	73	2	run	run	VERB
fcis-4145	73	3	the	the	DET
fcis-4145	73	4	configuration	configuration	NOUN
fcis-4145	73	5	to	to	PART
fcis-4145	73	6	control	control	VERB
fcis-4145	73	7	the	the	DET
fcis-4145	73	8	finetuning	finetuning	NOUN
fcis-4145	73	9	of	of	ADP
fcis-4145	73	10	the	the	DET
fcis-4145	73	11	training	training	NOUN
fcis-4145	73	12	task	task	NOUN
fcis-4145	73	13	,	,	PUNCT
fcis-4145	73	14	specify	specify	VERB
fcis-4145	73	15	the	the	DET
fcis-4145	73	16	training	training	NOUN
fcis-4145	73	17	and	and	CCONJ
fcis-4145	73	18	validation	validation	NOUN
fcis-4145	73	19	sets	set	NOUN
fcis-4145	73	20	of	of	ADP
fcis-4145	73	21	the	the	DET
fcis-4145	73	22	model	model	NOUN
fcis-4145	73	23	,	,	PUNCT
fcis-4145	73	24	and	and	CCONJ
fcis-4145	73	25	configure	configure	VERB
fcis-4145	73	26	the	the	DET
fcis-4145	73	27	training	training	NOUN
fcis-4145	73	28	parameters	parameter	NOUN
fcis-4145	73	29	such	such	ADJ
fcis-4145	73	30	as	as	ADP
fcis-4145	73	31	the	the	DET
fcis-4145	73	32	number	number	NOUN
fcis-4145	73	33	of	of	ADP
fcis-4145	73	34	training	training	NOUN
fcis-4145	73	35	epochs	epoch	NOUN
fcis-4145	73	36	,	,	PUNCT
fcis-4145	73	37	batch	batch	NOUN
fcis-4145	73	38	size	size	NOUN
fcis-4145	73	39	,	,	PUNCT
fcis-4145	73	40	etc	etc	X
fcis-4145	73	41	.	.	X
fcis-4145	74	1	4	4	X
fcis-4145	74	2	.	.	NOUN
fcis-4145	74	3	result	result	NOUN
fcis-4145	74	4	discussion	discussion	NOUN
fcis-4145	74	5	in	in	ADP
fcis-4145	74	6	this	this	DET
fcis-4145	74	7	experiment	experiment	NOUN
fcis-4145	75	1	,	,	PUNCT
fcis-4145	75	2	adamw	adamw	PROPN
fcis-4145	75	3	was	be	AUX
fcis-4145	75	4	chosen	choose	VERB
fcis-4145	75	5	as	as	ADP
fcis-4145	75	6	the	the	DET
fcis-4145	75	7	optimizer	optimizer	NOUN
fcis-4145	75	8	of	of	ADP
fcis-4145	75	9	the	the	DET
fcis-4145	75	10	pre	pre	ADJ
fcis-4145	75	11	-	-	ADJ
fcis-4145	75	12	trained	train	VERB
fcis-4145	75	13	model	model	NOUN
fcis-4145	75	14	to	to	PART
fcis-4145	75	15	solve	solve	VERB
fcis-4145	75	16	the	the	DET
fcis-4145	75	17	problem	problem	NOUN
fcis-4145	75	18	of	of	ADP
fcis-4145	75	19	l2	l2	NOUN
fcis-4145	75	20	regularization	regularization	NOUN
fcis-4145	75	21	failure	failure	NOUN
fcis-4145	75	22	in	in	ADP
fcis-4145	75	23	the	the	DET
fcis-4145	75	24	adam	adam	PROPN
fcis-4145	75	25	optimizer	optimizer	NOUN
fcis-4145	75	26	[	[	X
fcis-4145	75	27	11	11	NUM
fcis-4145	75	28	]	]	PUNCT
fcis-4145	75	29	.	.	PUNCT
fcis-4145	76	1	finally	finally	ADV
fcis-4145	76	2	,	,	PUNCT
fcis-4145	76	3	the	the	DET
fcis-4145	76	4	test	test	NOUN
fcis-4145	76	5	set	set	NOUN
fcis-4145	76	6	was	be	AUX
fcis-4145	76	7	evaluated	evaluate	VERB
fcis-4145	76	8	with	with	ADP
fcis-4145	76	9	macro	macro	ADJ
fcis-4145	76	10	f1	f1	NOUN
fcis-4145	76	11	-	-	PUNCT
fcis-4145	76	12	score	score	NOUN
fcis-4145	76	13	,	,	PUNCT
fcis-4145	76	14	precision	precision	NOUN
fcis-4145	76	15	(	(	PUNCT
fcis-4145	76	16	p	p	NOUN
fcis-4145	76	17	)	)	PUNCT
fcis-4145	76	18	,	,	PUNCT
fcis-4145	76	19	and	and	CCONJ
fcis-4145	76	20	recall	recall	NOUN
fcis-4145	76	21	(	(	PUNCT
fcis-4145	76	22	r	r	NOUN
fcis-4145	76	23	)	)	PUNCT
fcis-4145	76	24	metrics	metric	NOUN
fcis-4145	76	25	which	which	PRON
fcis-4145	76	26	are	be	AUX
fcis-4145	76	27	the	the	DET
fcis-4145	76	28	common	common	ADJ
fcis-4145	76	29	indexes	index	NOUN
fcis-4145	76	30	used	use	VERB
fcis-4145	76	31	in	in	ADP
fcis-4145	76	32	classification	classification	NOUN
fcis-4145	76	33	tasks	task	NOUN
fcis-4145	76	34	.	.	PUNCT
fcis-4145	77	1	f1	f1	NOUN
fcis-4145	77	2	value	value	NOUN
fcis-4145	77	3	is	be	AUX
fcis-4145	77	4	the	the	DET
fcis-4145	77	5	harmonized	harmonized	ADJ
fcis-4145	77	6	average	average	NOUN
fcis-4145	77	7	of	of	ADP
fcis-4145	77	8	p	p	NOUN
fcis-4145	77	9	and	and	CCONJ
fcis-4145	77	10	r	r	NOUN
fcis-4145	77	11	values	value	NOUN
fcis-4145	77	12	,	,	PUNCT
fcis-4145	77	13	and	and	CCONJ
fcis-4145	77	14	macro	macro	ADJ
fcis-4145	77	15	f1	f1	PROPN
fcis-4145	77	16	score	score	NOUN
fcis-4145	77	17	directly	directly	ADV
fcis-4145	77	18	calculate	calculate	VERB
fcis-4145	77	19	the	the	DET
fcis-4145	77	20	average	average	NOUN
fcis-4145	77	21	of	of	ADP
fcis-4145	77	22	f1	f1	ADJ
fcis-4145	77	23	values	value	NOUN
fcis-4145	77	24	for	for	ADP
fcis-4145	77	25	each	each	DET
fcis-4145	77	26	category	category	NOUN
fcis-4145	77	27	.	.	PUNCT
fcis-4145	78	1	the	the	DET
fcis-4145	78	2	precision	precision	NOUN
fcis-4145	78	3	is	be	AUX
fcis-4145	78	4	the	the	DET
fcis-4145	78	5	percentage	percentage	NOUN
fcis-4145	78	6	of	of	ADP
fcis-4145	78	7	samples	sample	NOUN
fcis-4145	78	8	with	with	ADP
fcis-4145	78	9	positive	positive	ADJ
fcis-4145	78	10	true	true	ADJ
fcis-4145	78	11	values	value	NOUN
fcis-4145	78	12	among	among	ADP
fcis-4145	78	13	all	all	DET
fcis-4145	78	14	samples	sample	NOUN
fcis-4145	78	15	with	with	ADP
fcis-4145	78	16	positive	positive	ADJ
fcis-4145	78	17	predicted	predict	VERB
fcis-4145	78	18	values	value	NOUN
fcis-4145	78	19	,	,	PUNCT
fcis-4145	78	20	and	and	CCONJ
fcis-4145	78	21	the	the	DET
fcis-4145	78	22	recall	recall	NOUN
fcis-4145	78	23	is	be	AUX
fcis-4145	78	24	the	the	DET
fcis-4145	78	25	percentage	percentage	NOUN
fcis-4145	78	26	of	of	ADP
fcis-4145	78	27	samples	sample	NOUN
fcis-4145	78	28	with	with	ADP
fcis-4145	78	29	positive	positive	ADJ
fcis-4145	78	30	predicted	predict	VERB
fcis-4145	78	31	values	value	NOUN
fcis-4145	78	32	among	among	ADP
fcis-4145	78	33	all	all	DET
fcis-4145	78	34	samples	sample	NOUN
fcis-4145	78	35	with	with	ADP
fcis-4145	78	36	positive	positive	ADJ
fcis-4145	78	37	true	true	ADJ
fcis-4145	78	38	values	value	NOUN
fcis-4145	78	39	.	.	PUNCT
fcis-4145	79	1	the	the	DET
fcis-4145	79	2	final	final	ADJ
fcis-4145	79	3	test	test	NOUN
fcis-4145	79	4	score	score	NOUN
fcis-4145	79	5	results	result	NOUN
fcis-4145	79	6	were	be	AUX
fcis-4145	79	7	obtained	obtain	VERB
fcis-4145	79	8	as	as	SCONJ
fcis-4145	79	9	shown	show	VERB
fcis-4145	79	10	in	in	ADP
fcis-4145	79	11	table	table	NOUN
fcis-4145	79	12	2	2	NUM
fcis-4145	79	13	,	,	PUNCT
fcis-4145	79	14	with	with	ADP
fcis-4145	79	15	f1	f1	PROPN
fcis-4145	79	16	score	score	NOUN
fcis-4145	79	17	,	,	PUNCT
fcis-4145	79	18	precision	precision	NOUN
fcis-4145	79	19	score	score	NOUN
fcis-4145	79	20	and	and	CCONJ
fcis-4145	79	21	recall	recall	VERB
fcis-4145	79	22	score	score	NOUN
fcis-4145	79	23	of	of	ADP
fcis-4145	79	24	0.941929	0.941929	NUM
fcis-4145	79	25	,	,	PUNCT
fcis-4145	79	26	0.942300	0.942300	NUM
fcis-4145	79	27	and	and	CCONJ
fcis-4145	79	28	0.941928	0.941928	NUM
fcis-4145	79	29	respectively	respectively	ADV
fcis-4145	79	30	.	.	PUNCT
fcis-4145	80	1	it	it	PRON
fcis-4145	80	2	can	can	AUX
fcis-4145	80	3	be	be	AUX
fcis-4145	80	4	seen	see	VERB
fcis-4145	80	5	that	that	SCONJ
fcis-4145	80	6	ernie	ernie	PROPN
fcis-4145	80	7	tiny	tiny	ADJ
fcis-4145	80	8	model	model	NOUN
fcis-4145	80	9	can	can	AUX
fcis-4145	80	10	accurately	accurately	ADV
fcis-4145	80	11	identify	identify	VERB
fcis-4145	80	12	and	and	CCONJ
fcis-4145	80	13	classify	classify	VERB
fcis-4145	80	14	the	the	DET
fcis-4145	80	15	emotions	emotion	NOUN
fcis-4145	80	16	of	of	ADP
fcis-4145	80	17	lovelorn	lovelorn	ADJ
fcis-4145	80	18	,	,	PUNCT
fcis-4145	80	19	and	and	CCONJ
fcis-4145	80	20	get	get	VERB
fcis-4145	80	21	satisfying	satisfy	VERB
fcis-4145	80	22	results	result	NOUN
fcis-4145	80	23	.	.	PUNCT
fcis-4145	81	1	table	table	NOUN
fcis-4145	81	2	2	2	NUM
fcis-4145	81	3	.	.	PUNCT
fcis-4145	81	4	evaluation	evaluation	NOUN
fcis-4145	81	5	index	index	NOUN
fcis-4145	81	6	result	result	VERB
fcis-4145	81	7	f1	f1	ADJ
fcis-4145	81	8	-	-	PUNCT
fcis-4145	81	9	score	score	NOUN
fcis-4145	81	10	precision	precision	NOUN
fcis-4145	81	11	recall	recall	VERB
fcis-4145	81	12	0.941929	0.941929	NUM
fcis-4145	81	13	0.942300	0.942300	NUM
fcis-4145	81	14	0.941928	0.941928	NUM
fcis-4145	81	15	5	5	NUM
fcis-4145	81	16	.	.	PUNCT
fcis-4145	81	17	conclusion	conclusion	NOUN
fcis-4145	81	18	in	in	ADP
fcis-4145	81	19	this	this	DET
fcis-4145	81	20	study	study	NOUN
fcis-4145	81	21	,	,	PUNCT
fcis-4145	81	22	the	the	DET
fcis-4145	81	23	basic	basic	ADJ
fcis-4145	81	24	architecture	architecture	NOUN
fcis-4145	81	25	and	and	CCONJ
fcis-4145	81	26	principles	principle	NOUN
fcis-4145	81	27	of	of	ADP
fcis-4145	81	28	the	the	DET
fcis-4145	81	29	deep	deep	ADJ
fcis-4145	81	30	learning	learning	NOUN
fcis-4145	81	31	model	model	PROPN
fcis-4145	81	32	ernie	ernie	PROPN
fcis-4145	81	33	tiny	tiny	ADJ
fcis-4145	81	34	were	be	AUX
fcis-4145	81	35	introduced	introduce	VERB
fcis-4145	81	36	.	.	PUNCT
fcis-4145	82	1	three	three	NUM
fcis-4145	82	2	approaches	approach	NOUN
fcis-4145	82	3	to	to	PART
fcis-4145	82	4	sentiment	sentiment	NOUN
fcis-4145	82	5	classification	classification	NOUN
fcis-4145	82	6	(	(	PUNCT
fcis-4145	82	7	sentiment	sentiment	NOUN
fcis-4145	82	8	lexiconbased	lexiconbase	VERB
fcis-4145	82	9	approach	approach	NOUN
fcis-4145	82	10	,	,	PUNCT
fcis-4145	82	11	machine	machine	NOUN
fcis-4145	82	12	learning	learning	NOUN
fcis-4145	82	13	based	base	VERB
fcis-4145	82	14	approach	approach	NOUN
fcis-4145	82	15	and	and	CCONJ
fcis-4145	82	16	deep	deep	ADJ
fcis-4145	82	17	learning	learning	NOUN
fcis-4145	82	18	approach	approach	NOUN
fcis-4145	82	19	)	)	PUNCT
fcis-4145	82	20	were	be	AUX
fcis-4145	82	21	explored	explore	VERB
fcis-4145	82	22	.	.	PUNCT
fcis-4145	83	1	in	in	ADP
fcis-4145	83	2	order	order	NOUN
fcis-4145	83	3	to	to	PART
fcis-4145	83	4	address	address	VERB
fcis-4145	83	5	the	the	DET
fcis-4145	83	6	problem	problem	NOUN
fcis-4145	83	7	that	that	SCONJ
fcis-4145	83	8	few	few	ADJ
fcis-4145	83	9	existing	exist	VERB
fcis-4145	83	10	sentiment	sentiment	NOUN
fcis-4145	83	11	analysis	analysis	NOUN
fcis-4145	83	12	studies	study	NOUN
fcis-4145	83	13	detect	detect	VERB
fcis-4145	83	14	love	love	NOUN
fcis-4145	83	15	loss	loss	NOUN
fcis-4145	83	16	emotion	emotion	NOUN
fcis-4145	83	17	texts	text	NOUN
fcis-4145	83	18	,	,	PUNCT
fcis-4145	83	19	the	the	DET
fcis-4145	83	20	author	author	NOUN
fcis-4145	83	21	crawled	crawl	VERB
fcis-4145	83	22	love	love	NOUN
fcis-4145	83	23	loss	loss	NOUN
fcis-4145	83	24	emotion	emotion	NOUN
fcis-4145	83	25	texts	text	NOUN
fcis-4145	83	26	on	on	ADP
fcis-4145	83	27	the	the	DET
fcis-4145	83	28	social	social	ADJ
fcis-4145	83	29	media	medium	NOUN
fcis-4145	83	30	platform	platform	NOUN
fcis-4145	83	31	weibo	weibo	NOUN
fcis-4145	83	32	,	,	PUNCT
fcis-4145	83	33	combined	combine	VERB
fcis-4145	83	34	with	with	ADP
fcis-4145	83	35	normal	normal	ADJ
fcis-4145	83	36	sentiment	sentiment	NOUN
fcis-4145	83	37	reviews	review	NOUN
fcis-4145	83	38	as	as	ADP
fcis-4145	83	39	dataset	dataset	NOUN
fcis-4145	83	40	,	,	PUNCT
fcis-4145	83	41	and	and	CCONJ
fcis-4145	83	42	the	the	DET
fcis-4145	83	43	ernie	ernie	NOUN
fcis-4145	83	44	tiny	tiny	ADJ
fcis-4145	83	45	model	model	NOUN
fcis-4145	83	46	was	be	AUX
fcis-4145	83	47	trained	train	VERB
fcis-4145	83	48	to	to	PART
fcis-4145	83	49	recognize	recognize	VERB
fcis-4145	83	50	lovelorn	lovelorn	ADJ
fcis-4145	83	51	emotion	emotion	NOUN
fcis-4145	83	52	texts	text	NOUN
fcis-4145	83	53	,	,	PUNCT
fcis-4145	83	54	which	which	PRON
fcis-4145	83	55	can	can	AUX
fcis-4145	83	56	effectively	effectively	ADV
fcis-4145	83	57	classify	classify	VERB
fcis-4145	83	58	texts	text	NOUN
fcis-4145	83	59	containing	contain	VERB
fcis-4145	83	60	love	love	NOUN
fcis-4145	83	61	-	-	PUNCT
fcis-4145	83	62	loss	loss	NOUN
fcis-4145	83	63	emotion	emotion	NOUN
fcis-4145	83	64	from	from	ADP
fcis-4145	83	65	ordinary	ordinary	ADJ
fcis-4145	83	66	non	non	ADJ
fcis-4145	83	67	-	-	ADJ
fcis-4145	83	68	love	love	ADJ
fcis-4145	83	69	-	-	PUNCT
fcis-4145	83	70	loss	loss	NOUN
fcis-4145	83	71	texts	text	NOUN
fcis-4145	83	72	,	,	PUNCT
fcis-4145	83	73	and	and	CCONJ
fcis-4145	83	74	finally	finally	ADV
fcis-4145	83	75	achieve	achieve	VERB
fcis-4145	83	76	satisfactory	satisfactory	ADJ
fcis-4145	83	77	results	result	NOUN
fcis-4145	83	78	.	.	PUNCT
fcis-4145	84	1	on	on	ADP
fcis-4145	84	2	the	the	DET
fcis-4145	84	3	one	one	NUM
fcis-4145	84	4	hand	hand	NOUN
fcis-4145	84	5	,	,	PUNCT
fcis-4145	84	6	this	this	DET
fcis-4145	84	7	study	study	NOUN
fcis-4145	84	8	contributes	contribute	VERB
fcis-4145	84	9	to	to	ADP
fcis-4145	84	10	the	the	DET
fcis-4145	84	11	work	work	NOUN
fcis-4145	84	12	on	on	ADP
fcis-4145	84	13	emotion	emotion	NOUN
fcis-4145	84	14	perception	perception	NOUN
fcis-4145	84	15	in	in	ADP
fcis-4145	84	16	digital	digital	ADJ
fcis-4145	84	17	mental	mental	ADJ
fcis-4145	84	18	health	health	NOUN
fcis-4145	84	19	treatment	treatment	NOUN
fcis-4145	84	20	solutions	solution	NOUN
fcis-4145	84	21	,	,	PUNCT
fcis-4145	84	22	reflecting	reflect	VERB
fcis-4145	84	23	the	the	DET
fcis-4145	84	24	important	important	ADJ
fcis-4145	84	25	value	value	NOUN
fcis-4145	84	26	of	of	ADP
fcis-4145	84	27	sentiment	sentiment	NOUN
fcis-4145	84	28	analysis	analysis	NOUN
fcis-4145	84	29	in	in	ADP
fcis-4145	84	30	this	this	DET
fcis-4145	84	31	area	area	NOUN
fcis-4145	84	32	and	and	CCONJ
fcis-4145	84	33	having	have	VERB
fcis-4145	84	34	valuable	valuable	ADJ
fcis-4145	84	35	social	social	ADJ
fcis-4145	84	36	significance	significance	NOUN
fcis-4145	84	37	.	.	PUNCT
fcis-4145	85	1	on	on	ADP
fcis-4145	85	2	the	the	DET
fcis-4145	85	3	other	other	ADJ
fcis-4145	85	4	hand	hand	NOUN
fcis-4145	85	5	,	,	PUNCT
fcis-4145	85	6	since	since	SCONJ
fcis-4145	85	7	punctuation	punctuation	NOUN
fcis-4145	85	8	marks	mark	NOUN
fcis-4145	85	9	may	may	AUX
fcis-4145	85	10	also	also	ADV
fcis-4145	85	11	contain	contain	VERB
fcis-4145	85	12	some	some	DET
fcis-4145	85	13	enhanced	enhanced	ADJ
fcis-4145	85	14	emotions	emotion	NOUN
fcis-4145	85	15	,	,	PUNCT
fcis-4145	85	16	and	and	CCONJ
fcis-4145	85	17	the	the	DET
fcis-4145	85	18	number	number	NOUN
fcis-4145	85	19	of	of	ADP
fcis-4145	85	20	training	training	NOUN
fcis-4145	85	21	texts	text	NOUN
fcis-4145	85	22	is	be	AUX
fcis-4145	85	23	not	not	PART
fcis-4145	85	24	very	very	ADV
fcis-4145	85	25	large	large	ADJ
fcis-4145	85	26	,	,	PUNCT
fcis-4145	85	27	future	future	ADJ
fcis-4145	85	28	work	work	NOUN
fcis-4145	85	29	is	be	AUX
fcis-4145	85	30	to	to	PART
fcis-4145	85	31	consider	consider	VERB
fcis-4145	85	32	the	the	DET
fcis-4145	85	33	role	role	NOUN
fcis-4145	85	34	of	of	ADP
fcis-4145	85	35	punctuation	punctuation	NOUN
fcis-4145	85	36	marks	mark	NOUN
fcis-4145	85	37	into	into	ADP
fcis-4145	85	38	the	the	DET
fcis-4145	85	39	problem	problem	NOUN
fcis-4145	85	40	of	of	ADP
fcis-4145	85	41	sentiment	sentiment	NOUN
fcis-4145	85	42	analysis	analysis	NOUN
fcis-4145	85	43	with	with	ADP
fcis-4145	85	44	more	more	ADJ
fcis-4145	85	45	training	training	NOUN
fcis-4145	85	46	data	datum	NOUN
fcis-4145	85	47	,	,	PUNCT
fcis-4145	85	48	further	far	ADV
fcis-4145	85	49	improve	improve	VERB
fcis-4145	85	50	the	the	DET
fcis-4145	85	51	accuracy	accuracy	NOUN
fcis-4145	85	52	of	of	ADP
fcis-4145	85	53	sentiment	sentiment	NOUN
fcis-4145	85	54	classification	classification	NOUN
fcis-4145	85	55	.	.	PUNCT
fcis-4145	86	1	references	reference	NOUN
fcis-4145	86	2	[	[	X
fcis-4145	86	3	1	1	NUM
fcis-4145	86	4	]	]	X
fcis-4145	86	5	chen	chen	PROPN
fcis-4145	86	6	,	,	PUNCT
fcis-4145	86	7	y.	y.	PROPN
fcis-4145	86	8	,	,	PUNCT
fcis-4145	86	9	zhou	zhou	PROPN
fcis-4145	86	10	,	,	PUNCT
fcis-4145	86	11	b.	b.	PROPN
fcis-4145	86	12	,	,	PUNCT
fcis-4145	86	13	zhang	zhang	PROPN
fcis-4145	86	14	,	,	PUNCT
fcis-4145	86	15	w.	w.	PROPN
fcis-4145	86	16	,	,	PUNCT
fcis-4145	86	17	gong	gong	PROPN
fcis-4145	86	18	,	,	PUNCT
fcis-4145	86	19	w.	w.	PROPN
fcis-4145	86	20	,	,	PUNCT
fcis-4145	86	21	&	&	CCONJ
fcis-4145	86	22	sun	sun	PROPN
fcis-4145	86	23	,	,	PUNCT
fcis-4145	86	24	g.	g.	PROPN
fcis-4145	86	25	(	(	PUNCT
fcis-4145	86	26	2018	2018	NUM
fcis-4145	86	27	)	)	PUNCT
fcis-4145	86	28	.	.	PUNCT
fcis-4145	87	1	sentiment	sentiment	NOUN
fcis-4145	87	2	analysis	analysis	NOUN
fcis-4145	87	3	based	base	VERB
fcis-4145	87	4	on	on	ADP
fcis-4145	87	5	deep	deep	ADJ
fcis-4145	87	6	learning	learning	NOUN
fcis-4145	87	7	and	and	CCONJ
fcis-4145	87	8	its	its	PRON
fcis-4145	87	9	application	application	NOUN
fcis-4145	87	10	in	in	ADP
fcis-4145	87	11	screening	screen	VERB
fcis-4145	87	12	for	for	ADP
fcis-4145	87	13	perinatal	perinatal	ADJ
fcis-4145	87	14	depression	depression	NOUN
fcis-4145	87	15	.	.	PUNCT
fcis-4145	88	1	2018	2018	NUM
fcis-4145	88	2	ieee	ieee	PROPN
fcis-4145	88	3	third	third	ADJ
fcis-4145	88	4	international	international	ADJ
fcis-4145	88	5	conference	conference	NOUN
fcis-4145	88	6	on	on	ADP
fcis-4145	88	7	data	datum	NOUN
fcis-4145	88	8	science	science	NOUN
fcis-4145	88	9	in	in	ADP
fcis-4145	88	10	cyberspace	cyberspace	NOUN
fcis-4145	88	11	(	(	PUNCT
fcis-4145	88	12	dsc	dsc	NOUN
fcis-4145	88	13	)	)	PUNCT
fcis-4145	88	14	.	.	PUNCT
fcis-4145	89	1	https://doi.org/10.1109/dsc.2018.00073	https://doi.org/10.1109/dsc.2018.00073	NOUN
fcis-4145	90	1	[	[	X
fcis-4145	90	2	2	2	NUM
fcis-4145	90	3	]	]	X
fcis-4145	90	4	liu	liu	PROPN
fcis-4145	90	5	,	,	PUNCT
fcis-4145	90	6	j.	j.	PROPN
fcis-4145	90	7	,	,	PUNCT
fcis-4145	90	8	shi	shi	PROPN
fcis-4145	90	9	,	,	PUNCT
fcis-4145	90	10	m.	m.	NOUN
fcis-4145	90	11	,	,	PUNCT
fcis-4145	90	12	&	&	CCONJ
fcis-4145	90	13	jiang	jiang	PROPN
fcis-4145	90	14	,	,	PUNCT
fcis-4145	90	15	h.	h.	PROPN
fcis-4145	90	16	(	(	PUNCT
fcis-4145	90	17	2022	2022	NUM
fcis-4145	90	18	)	)	PUNCT
fcis-4145	90	19	.	.	PUNCT
fcis-4145	91	1	detecting	detect	VERB
fcis-4145	91	2	suicidal	suicidal	ADJ
fcis-4145	91	3	ideation	ideation	NOUN
fcis-4145	91	4	in	in	ADP
fcis-4145	91	5	social	social	ADJ
fcis-4145	91	6	media	medium	NOUN
fcis-4145	91	7	:	:	PUNCT
fcis-4145	91	8	an	an	DET
fcis-4145	91	9	ensemble	ensemble	ADJ
fcis-4145	91	10	method	method	NOUN
fcis-4145	91	11	based	base	VERB
fcis-4145	91	12	on	on	ADP
fcis-4145	91	13	feature	feature	NOUN
fcis-4145	91	14	fusion	fusion	NOUN
fcis-4145	91	15	.	.	PUNCT
fcis-4145	92	1	international	international	ADJ
fcis-4145	92	2	journal	journal	PROPN
fcis-4145	92	3	of	of	ADP
fcis-4145	92	4	environmental	environmental	ADJ
fcis-4145	92	5	research	research	NOUN
fcis-4145	92	6	and	and	CCONJ
fcis-4145	92	7	public	public	ADJ
fcis-4145	92	8	health	health	NOUN
fcis-4145	92	9	,	,	PUNCT
fcis-4145	92	10	19(13	19(13	NUM
fcis-4145	92	11	)	)	PUNCT
fcis-4145	92	12	,	,	PUNCT
fcis-4145	92	13	8197	8197	NUM
fcis-4145	92	14	.	.	PUNCT
fcis-4145	93	1	doi:10.3390	doi:10.3390	NOUN
fcis-4145	93	2	/	/	PUNCT
fcis-4145	94	1	ijerph19138197	ijerph19138197	PROPN
fcis-4145	95	1	[	[	X
fcis-4145	95	2	3	3	NUM
fcis-4145	95	3	]	]	X
fcis-4145	95	4	zhang	zhang	PROPN
fcis-4145	95	5	,	,	PUNCT
fcis-4145	95	6	l.	l.	PROPN
fcis-4145	95	7	,	,	PUNCT
fcis-4145	95	8	wang	wang	PROPN
fcis-4145	95	9	,	,	PUNCT
fcis-4145	95	10	s.	s.	PROPN
fcis-4145	95	11	,	,	PUNCT
fcis-4145	95	12	&	&	CCONJ
fcis-4145	95	13	liu	liu	PROPN
fcis-4145	95	14	,	,	PUNCT
fcis-4145	95	15	b.	b.	PROPN
fcis-4145	95	16	(	(	PUNCT
fcis-4145	95	17	2018	2018	NUM
fcis-4145	95	18	)	)	PUNCT
fcis-4145	95	19	.	.	PUNCT
fcis-4145	96	1	deep	deep	ADJ
fcis-4145	96	2	learning	learn	VERB
fcis-4145	96	3	for	for	ADP
fcis-4145	96	4	sentiment	sentiment	NOUN
fcis-4145	96	5	analysis	analysis	NOUN
fcis-4145	96	6	:	:	PUNCT
fcis-4145	96	7	a	a	DET
fcis-4145	96	8	survey	survey	NOUN
fcis-4145	96	9	.	.	PUNCT
fcis-4145	97	1	cornell	cornell	PROPN
fcis-4145	97	2	university	university	PROPN
fcis-4145	97	3	arxiv	arxiv	PROPN
fcis-4145	97	4	.	.	PUNCT
fcis-4145	98	1	https://doi.org/10.48550/arxiv.1801.07883	https://doi.org/10.48550/arxiv.1801.07883	PROPN
fcis-4145	99	1	[	[	X
fcis-4145	99	2	4	4	NUM
fcis-4145	99	3	]	]	X
fcis-4145	99	4	kim	kim	PROPN
fcis-4145	99	5	,	,	PUNCT
fcis-4145	99	6	y.	y.	PROPN
fcis-4145	99	7	(	(	PUNCT
fcis-4145	99	8	2014	2014	NUM
fcis-4145	99	9	)	)	PUNCT
fcis-4145	99	10	.	.	PUNCT
fcis-4145	100	1	convolutional	convolutional	ADJ
fcis-4145	100	2	neural	neural	ADJ
fcis-4145	100	3	networks	network	NOUN
fcis-4145	100	4	for	for	ADP
fcis-4145	100	5	sentence	sentence	NOUN
fcis-4145	100	6	classification	classification	NOUN
fcis-4145	100	7	.	.	PUNCT
fcis-4145	101	1	proceedings	proceeding	NOUN
fcis-4145	101	2	of	of	ADP
fcis-4145	101	3	the	the	DET
fcis-4145	101	4	2014	2014	NUM
fcis-4145	101	5	conference	conference	NOUN
fcis-4145	101	6	on	on	ADP
fcis-4145	101	7	69	69	NUM
fcis-4145	101	8	empirical	empirical	ADJ
fcis-4145	101	9	methods	method	NOUN
fcis-4145	101	10	in	in	ADP
fcis-4145	101	11	natural	natural	ADJ
fcis-4145	101	12	language	language	NOUN
fcis-4145	101	13	processing	processing	NOUN
fcis-4145	101	14	(	(	PUNCT
fcis-4145	101	15	emnlp	emnlp	ADJ
fcis-4145	101	16	)	)	PUNCT
fcis-4145	101	17	.	.	PUNCT
fcis-4145	102	1	doi:10.3115	doi:10.3115	PROPN
fcis-4145	102	2	/	/	SYM
fcis-4145	102	3	v1	v1	PROPN
fcis-4145	102	4	/	/	SYM
fcis-4145	102	5	d14	d14	NOUN
fcis-4145	102	6	-	-	PUNCT
fcis-4145	102	7	1181	1181	NUM
fcis-4145	102	8	[	[	X
fcis-4145	102	9	5	5	NUM
fcis-4145	102	10	]	]	SYM
fcis-4145	102	11	tang	tang	PROPN
fcis-4145	102	12	,	,	PUNCT
fcis-4145	102	13	d.	d.	PROPN
fcis-4145	102	14	,	,	PUNCT
fcis-4145	102	15	qin	qin	PROPN
fcis-4145	102	16	,	,	PUNCT
fcis-4145	102	17	b.	b.	PROPN
fcis-4145	102	18	,	,	PUNCT
fcis-4145	102	19	&	&	CCONJ
fcis-4145	102	20	liu	liu	PROPN
fcis-4145	102	21	,	,	PUNCT
fcis-4145	102	22	t.	t.	PROPN
fcis-4145	102	23	(	(	PUNCT
fcis-4145	102	24	2016	2016	NUM
fcis-4145	102	25	)	)	PUNCT
fcis-4145	102	26	.	.	PUNCT
fcis-4145	103	1	aspect	aspect	NOUN
fcis-4145	103	2	level	level	NOUN
fcis-4145	103	3	sentiment	sentiment	NOUN
fcis-4145	103	4	classification	classification	NOUN
fcis-4145	103	5	with	with	ADP
fcis-4145	103	6	deep	deep	ADJ
fcis-4145	103	7	memory	memory	NOUN
fcis-4145	103	8	network	network	NOUN
fcis-4145	103	9	.	.	PUNCT
fcis-4145	104	1	proceedings	proceeding	NOUN
fcis-4145	104	2	of	of	ADP
fcis-4145	104	3	the	the	DET
fcis-4145	104	4	2016	2016	NUM
fcis-4145	104	5	conference	conference	NOUN
fcis-4145	104	6	on	on	ADP
fcis-4145	104	7	empirical	empirical	ADJ
fcis-4145	104	8	methods	method	NOUN
fcis-4145	104	9	in	in	ADP
fcis-4145	104	10	natural	natural	ADJ
fcis-4145	104	11	language	language	NOUN
fcis-4145	104	12	processing	processing	NOUN
fcis-4145	104	13	.	.	PUNCT
fcis-4145	105	1	https://doi.org/10.18653/v1/d16-1021	https://doi.org/10.18653/v1/d16-1021	PROPN
fcis-4145	106	1	[	[	X
fcis-4145	106	2	6	6	NUM
fcis-4145	106	3	]	]	PUNCT
fcis-4145	106	4	basiri	basiri	NOUN
fcis-4145	106	5	,	,	PUNCT
fcis-4145	106	6	m.	m.	PROPN
fcis-4145	106	7	e.	e.	PROPN
fcis-4145	106	8	,	,	PUNCT
fcis-4145	106	9	nemati	nemati	PROPN
fcis-4145	106	10	,	,	PUNCT
fcis-4145	106	11	s.	s.	PROPN
fcis-4145	106	12	,	,	PUNCT
fcis-4145	106	13	abdar	abdar	NOUN
fcis-4145	106	14	,	,	PUNCT
fcis-4145	106	15	m.	m.	NOUN
fcis-4145	106	16	,	,	PUNCT
fcis-4145	106	17	cambria	cambria	PROPN
fcis-4145	106	18	,	,	PUNCT
fcis-4145	106	19	e.	e.	PROPN
fcis-4145	106	20	,	,	PUNCT
fcis-4145	106	21	&	&	CCONJ
fcis-4145	106	22	acharya	acharya	PROPN
fcis-4145	106	23	,	,	PUNCT
fcis-4145	106	24	u.	u.	PROPN
fcis-4145	106	25	r.	r.	PROPN
fcis-4145	106	26	(	(	PUNCT
fcis-4145	106	27	2021	2021	NUM
fcis-4145	106	28	)	)	PUNCT
fcis-4145	106	29	.	.	PUNCT
fcis-4145	107	1	abcdm	abcdm	NOUN
fcis-4145	107	2	:	:	PUNCT
fcis-4145	107	3	an	an	DET
fcis-4145	107	4	attention	attention	NOUN
fcis-4145	107	5	-	-	PUNCT
fcis-4145	107	6	based	base	VERB
fcis-4145	107	7	bidirectional	bidirectional	ADJ
fcis-4145	107	8	cnnrnn	cnnrnn	NOUN
fcis-4145	107	9	deep	deep	ADJ
fcis-4145	107	10	model	model	NOUN
fcis-4145	107	11	for	for	ADP
fcis-4145	107	12	sentiment	sentiment	NOUN
fcis-4145	107	13	analysis	analysis	NOUN
fcis-4145	107	14	.	.	PUNCT
fcis-4145	108	1	future	future	ADJ
fcis-4145	108	2	generation	generation	NOUN
fcis-4145	108	3	computer	computer	NOUN
fcis-4145	108	4	systems	system	NOUN
fcis-4145	108	5	,	,	PUNCT
fcis-4145	108	6	115	115	NUM
fcis-4145	108	7	,	,	PUNCT
fcis-4145	108	8	279	279	NUM
fcis-4145	108	9	-	-	SYM
fcis-4145	108	10	294	294	NUM
fcis-4145	108	11	.	.	PUNCT
fcis-4145	109	1	doi:10.1016	doi:10.1016	PROPN
fcis-4145	109	2	/	/	PUNCT
fcis-4145	110	1	j.future.2020.08.005	j.future.2020.08.005	X
fcis-4145	111	1	[	[	X
fcis-4145	111	2	7	7	X
fcis-4145	111	3	]	]	X
fcis-4145	111	4	wang	wang	PROPN
fcis-4145	111	5	,	,	PUNCT
fcis-4145	111	6	x.	x.	PROPN
fcis-4145	111	7	,	,	PUNCT
fcis-4145	111	8	&	&	CCONJ
fcis-4145	111	9	tong	tong	PROPN
fcis-4145	111	10	,	,	PUNCT
fcis-4145	111	11	y.	y.	PROPN
fcis-4145	111	12	(	(	PUNCT
fcis-4145	111	13	2021	2021	NUM
fcis-4145	111	14	)	)	PUNCT
fcis-4145	111	15	.	.	PUNCT
fcis-4145	112	1	application	application	NOUN
fcis-4145	112	2	of	of	ADP
fcis-4145	112	3	bert+attention	bert+attention	PROPN
fcis-4145	112	4	model	model	NOUN
fcis-4145	112	5	in	in	ADP
fcis-4145	112	6	emotion	emotion	NOUN
fcis-4145	112	7	recognition	recognition	NOUN
fcis-4145	112	8	of	of	ADP
fcis-4145	112	9	metizens	metizen	NOUN
fcis-4145	112	10	during	during	ADP
fcis-4145	112	11	epidemic	epidemic	ADJ
fcis-4145	112	12	period	period	NOUN
fcis-4145	112	13	.	.	PUNCT
fcis-4145	113	1	journal	journal	PROPN
fcis-4145	113	2	of	of	ADP
fcis-4145	113	3	physics	physics	PROPN
fcis-4145	113	4	:	:	PUNCT
fcis-4145	113	5	conference	conference	NOUN
fcis-4145	113	6	series	series	NOUN
fcis-4145	113	7	,	,	PUNCT
fcis-4145	113	8	1982(1	1982(1	NUM
fcis-4145	113	9	)	)	PUNCT
fcis-4145	113	10	,	,	PUNCT
fcis-4145	113	11	012102	012102	NUM
fcis-4145	113	12	.	.	PUNCT
fcis-4145	114	1	doi:10.1088/1742	doi:10.1088/1742	NOUN
fcis-4145	114	2	-	-	PUNCT
fcis-4145	114	3	6596/1982/1/012102	6596/1982/1/012102	NUM
fcis-4145	115	1	[	[	X
fcis-4145	115	2	8	8	NUM
fcis-4145	115	3	]	]	X
fcis-4145	115	4	sun	sun	NOUN
fcis-4145	115	5	,	,	PUNCT
fcis-4145	115	6	y.	y.	PROPN
fcis-4145	115	7	,	,	PUNCT
fcis-4145	115	8	wang	wang	PROPN
fcis-4145	115	9	,	,	PUNCT
fcis-4145	115	10	s.	s.	PROPN
fcis-4145	115	11	,	,	PUNCT
fcis-4145	115	12	li	li	PROPN
fcis-4145	115	13	,	,	PUNCT
fcis-4145	115	14	y.	y.	PROPN
fcis-4145	115	15	,	,	PUNCT
fcis-4145	115	16	feng	feng	PROPN
fcis-4145	115	17	,	,	PUNCT
fcis-4145	115	18	s.	s.	PROPN
fcis-4145	115	19	,	,	PUNCT
fcis-4145	115	20	tian	tian	PROPN
fcis-4145	115	21	,	,	PUNCT
fcis-4145	115	22	h.	h.	PROPN
fcis-4145	115	23	,	,	PUNCT
fcis-4145	115	24	wu	wu	PROPN
fcis-4145	115	25	,	,	PUNCT
fcis-4145	115	26	h.	h.	PROPN
fcis-4145	115	27	,	,	PUNCT
fcis-4145	115	28	&	&	CCONJ
fcis-4145	115	29	wang	wang	PROPN
fcis-4145	115	30	,	,	PUNCT
fcis-4145	115	31	h.	h.	PROPN
fcis-4145	115	32	(	(	PUNCT
fcis-4145	115	33	2020	2020	NUM
fcis-4145	115	34	)	)	PUNCT
fcis-4145	115	35	.	.	PUNCT
fcis-4145	116	1	ernie	ernie	PROPN
fcis-4145	116	2	2.0	2.0	NUM
fcis-4145	116	3	:	:	PUNCT
fcis-4145	116	4	a	a	DET
fcis-4145	116	5	continual	continual	ADJ
fcis-4145	116	6	pre	pre	ADJ
fcis-4145	116	7	-	-	ADJ
fcis-4145	116	8	training	training	ADJ
fcis-4145	116	9	framework	framework	NOUN
fcis-4145	116	10	for	for	ADP
fcis-4145	116	11	language	language	NOUN
fcis-4145	116	12	understanding	understanding	NOUN
fcis-4145	116	13	.	.	PUNCT
fcis-4145	117	1	proceedings	proceeding	NOUN
fcis-4145	117	2	of	of	ADP
fcis-4145	117	3	the	the	DET
fcis-4145	117	4	aaai	aaai	PROPN
fcis-4145	117	5	conference	conference	NOUN
fcis-4145	117	6	on	on	ADP
fcis-4145	117	7	artificial	artificial	ADJ
fcis-4145	117	8	intelligence	intelligence	NOUN
fcis-4145	117	9	,	,	PUNCT
fcis-4145	117	10	34(05	34(05	PROPN
fcis-4145	117	11	)	)	PUNCT
fcis-4145	117	12	,	,	PUNCT
fcis-4145	117	13	8968	8968	NUM
fcis-4145	117	14	-	-	SYM
fcis-4145	117	15	8975	8975	NUM
fcis-4145	117	16	.	.	PUNCT
fcis-4145	118	1	doi:10.1609	doi:10.1609	NOUN
fcis-4145	118	2	/	/	SYM
fcis-4145	118	3	aaai.v34i05.6428	aaai.v34i05.6428	NOUN
fcis-4145	119	1	[	[	X
fcis-4145	119	2	9	9	NUM
fcis-4145	119	3	]	]	SYM
fcis-4145	119	4	su	su	PROPN
fcis-4145	119	5	,	,	PUNCT
fcis-4145	119	6	w.	w.	PROPN
fcis-4145	119	7	,	,	PUNCT
fcis-4145	119	8	chen	chen	PROPN
fcis-4145	119	9	,	,	PUNCT
fcis-4145	119	10	x.	x.	PROPN
fcis-4145	119	11	,	,	PUNCT
fcis-4145	119	12	feng	feng	PROPN
fcis-4145	119	13	,	,	PUNCT
fcis-4145	119	14	s.	s.	PROPN
fcis-4145	119	15	,	,	PUNCT
fcis-4145	119	16	liu	liu	PROPN
fcis-4145	119	17	,	,	PUNCT
fcis-4145	119	18	j.	j.	PROPN
fcis-4145	119	19	,	,	PUNCT
fcis-4145	119	20	liu	liu	PROPN
fcis-4145	119	21	,	,	PUNCT
fcis-4145	119	22	w.	w.	PROPN
fcis-4145	119	23	,	,	PUNCT
fcis-4145	119	24	sun	sun	PROPN
fcis-4145	119	25	,	,	PUNCT
fcis-4145	119	26	y.	y.	PROPN
fcis-4145	119	27	,	,	PUNCT
fcis-4145	119	28	.	.	PUNCT
fcis-4145	119	29	.	.	PUNCT
fcis-4145	119	30	.	.	PUNCT
fcis-4145	120	1	wang	wang	PROPN
fcis-4145	120	2	,	,	PUNCT
fcis-4145	120	3	h.	h.	PROPN
fcis-4145	120	4	(	(	PUNCT
fcis-4145	120	5	2021	2021	NUM
fcis-4145	120	6	)	)	PUNCT
fcis-4145	120	7	.	.	PUNCT
fcis-4145	121	1	ernie	ernie	NOUN
fcis-4145	121	2	-	-	PUNCT
fcis-4145	121	3	tiny	tiny	ADJ
fcis-4145	121	4	:	:	PUNCT
fcis-4145	121	5	a	a	DET
fcis-4145	121	6	progressive	progressive	ADJ
fcis-4145	121	7	distillation	distillation	NOUN
fcis-4145	121	8	framework	framework	NOUN
fcis-4145	121	9	for	for	ADP
fcis-4145	121	10	pretrained	pretraine	VERB
fcis-4145	121	11	transformer	transformer	NOUN
fcis-4145	121	12	compression	compression	NOUN
fcis-4145	121	13	.	.	PUNCT
fcis-4145	122	1	arxiv	arxiv	NOUN
fcis-4145	122	2	:	:	PUNCT
fcis-4145	122	3	computation	computation	NOUN
fcis-4145	122	4	and	and	CCONJ
fcis-4145	122	5	language	language	NOUN
fcis-4145	122	6	.	.	PUNCT
fcis-4145	123	1	[	[	X
fcis-4145	123	2	10	10	NUM
fcis-4145	123	3	]	]	SYM
fcis-4145	123	4	li	li	PROPN
fcis-4145	123	5	,	,	PUNCT
fcis-4145	123	6	m.	m.	NOUN
fcis-4145	123	7	,	,	PUNCT
fcis-4145	123	8	long	long	ADV
fcis-4145	123	9	,	,	PUNCT
fcis-4145	123	10	y.	y.	PROPN
fcis-4145	123	11	,	,	PUNCT
fcis-4145	123	12	lu	lu	PROPN
fcis-4145	123	13	,	,	PUNCT
fcis-4145	123	14	q.	q.	PROPN
fcis-4145	123	15	,	,	PUNCT
fcis-4145	123	16	&	&	CCONJ
fcis-4145	123	17	li	li	PROPN
fcis-4145	123	18	,	,	PUNCT
fcis-4145	123	19	w.	w.	PROPN
fcis-4145	123	20	(	(	PUNCT
fcis-4145	123	21	2016	2016	NUM
fcis-4145	123	22	)	)	PUNCT
fcis-4145	123	23	.	.	PUNCT
fcis-4145	124	1	emotion	emotion	NOUN
fcis-4145	124	2	corpus	corpus	PROPN
fcis-4145	124	3	construction	construction	NOUN
fcis-4145	124	4	based	base	VERB
fcis-4145	124	5	on	on	ADP
fcis-4145	124	6	selection	selection	NOUN
fcis-4145	124	7	from	from	ADP
fcis-4145	124	8	hashtags	hashtag	NOUN
fcis-4145	124	9	.	.	PUNCT
fcis-4145	125	1	language	language	NOUN
fcis-4145	125	2	resources	resource	NOUN
fcis-4145	125	3	and	and	CCONJ
fcis-4145	125	4	evaluation	evaluation	NOUN
fcis-4145	125	5	,	,	PUNCT
fcis-4145	125	6	1845–1849	1845–1849	NUM
fcis-4145	125	7	.	.	PUNCT
fcis-4145	126	1	[	[	X
fcis-4145	126	2	11	11	NUM
fcis-4145	126	3	]	]	X
fcis-4145	126	4	loshchilov	loshchilov	PROPN
fcis-4145	126	5	,	,	PUNCT
fcis-4145	126	6	i.	i.	PROPN
fcis-4145	126	7	,	,	PUNCT
fcis-4145	126	8	&	&	CCONJ
fcis-4145	126	9	hutter	hutter	PROPN
fcis-4145	126	10	,	,	PUNCT
fcis-4145	126	11	f.	f.	PROPN
fcis-4145	126	12	(	(	PUNCT
fcis-4145	126	13	2018	2018	NUM
fcis-4145	126	14	)	)	PUNCT
fcis-4145	126	15	.	.	PUNCT
fcis-4145	127	1	decoupled	decouple	VERB
fcis-4145	127	2	weight	weight	NOUN
fcis-4145	127	3	decay	decay	NOUN
fcis-4145	127	4	regularization	regularization	NOUN
fcis-4145	127	5	.	.	PUNCT
fcis-4145	128	1	international	international	ADJ
fcis-4145	128	2	conference	conference	NOUN
fcis-4145	128	3	on	on	ADP
fcis-4145	128	4	learning	learn	VERB
fcis-4145	128	5	representations	representation	NOUN
fcis-4145	128	6	.	.	PUNCT