id	sid	tid	token	lemma	pos
fcis-30522	1	1	frontiers	frontier	NOUN
fcis-30522	1	2	in	in	ADP
fcis-30522	1	3	computing	computing	NOUN
fcis-30522	1	4	and	and	CCONJ
fcis-30522	1	5	intelligent	intelligent	ADJ
fcis-30522	1	6	systems	system	NOUN
fcis-30522	1	7	issn	issn	VERB
fcis-30522	1	8	:	:	PUNCT
fcis-30522	1	9	2832	2832	NUM
fcis-30522	1	10	-	-	SYM
fcis-30522	1	11	6024	6024	NUM
fcis-30522	1	12	|	|	NOUN
fcis-30522	1	13	vol	vol	NOUN
fcis-30522	1	14	.	.	PROPN
fcis-30522	2	1	12	12	NUM
fcis-30522	2	2	,	,	PUNCT
fcis-30522	2	3	no	no	INTJ
fcis-30522	2	4	.	.	NOUN
fcis-30522	2	5	1	1	NUM
fcis-30522	2	6	,	,	PUNCT
fcis-30522	2	7	2025	2025	NUM
fcis-30522	2	8	70	70	NUM
fcis-30522	2	9	study	study	NOUN
fcis-30522	2	10	on	on	ADP
fcis-30522	2	11	small	small	ADJ
fcis-30522	2	12	sample	sample	NOUN
fcis-30522	2	13	text	text	NOUN
fcis-30522	2	14	classification	classification	NOUN
fcis-30522	2	15	based	base	VERB
fcis-30522	2	16	on	on	ADP
fcis-30522	2	17	multi‐level	multi‐level	NOUN
fcis-30522	2	18	self‐attention	self‐attention	NOUN
fcis-30522	2	19	and	and	CCONJ
fcis-30522	2	20	multi‐feature	multi‐feature	VERB
fcis-30522	2	21	residual	residual	ADJ
fcis-30522	2	22	fusion	fusion	NOUN
fcis-30522	2	23	under	under	ADP
fcis-30522	2	24	data	data	NOUN
fcis-30522	2	25	enhancement	enhancement	NOUN
fcis-30522	2	26	linlin	linlin	PROPN
fcis-30522	2	27	deng	deng	PROPN
fcis-30522	3	1	*	*	PROPN
fcis-30522	3	2	,	,	PUNCT
fcis-30522	3	3	xina	xina	PROPN
fcis-30522	3	4	lu	lu	PROPN
fcis-30522	3	5	school	school	NOUN
fcis-30522	3	6	of	of	ADP
fcis-30522	3	7	economics	economic	NOUN
fcis-30522	3	8	,	,	PUNCT
fcis-30522	3	9	north	north	NOUN
fcis-30522	3	10	minzu	minzu	PROPN
fcis-30522	3	11	university	university	PROPN
fcis-30522	3	12	,	,	PUNCT
fcis-30522	3	13	yinchuan	yinchuan	PROPN
fcis-30522	3	14	,	,	PUNCT
fcis-30522	3	15	ningxia	ningxia	PROPN
fcis-30522	3	16	,	,	PUNCT
fcis-30522	3	17	china	china	PROPN
fcis-30522	3	18	*	*	PUNCT
fcis-30522	3	19	corresponding	correspond	VERB
fcis-30522	3	20	author	author	NOUN
fcis-30522	3	21	:	:	PUNCT
fcis-30522	3	22	linlin	linlin	PROPN
fcis-30522	3	23	deng	deng	PROPN
fcis-30522	3	24	(	(	PUNCT
fcis-30522	3	25	email	email	NOUN
fcis-30522	3	26	:	:	PUNCT
fcis-30522	3	27	m19862109528@163.com	m19862109528@163.com	X
fcis-30522	3	28	)	)	PUNCT
fcis-30522	3	29	abstract	abstract	NOUN
fcis-30522	3	30	:	:	PUNCT
fcis-30522	3	31	in	in	ADP
fcis-30522	3	32	commercial	commercial	ADJ
fcis-30522	3	33	applications	application	NOUN
fcis-30522	3	34	,	,	PUNCT
fcis-30522	3	35	while	while	SCONJ
fcis-30522	3	36	traditional	traditional	ADJ
fcis-30522	3	37	models	model	NOUN
fcis-30522	3	38	can	can	AUX
fcis-30522	3	39	achieve	achieve	VERB
fcis-30522	3	40	comparable	comparable	ADJ
fcis-30522	3	41	performance	performance	NOUN
fcis-30522	3	42	to	to	PART
fcis-30522	3	43	mainstream	mainstream	VERB
fcis-30522	3	44	large	large	ADJ
fcis-30522	3	45	language	language	NOUN
fcis-30522	3	46	models	model	NOUN
fcis-30522	3	47	,	,	PUNCT
fcis-30522	3	48	they	they	PRON
fcis-30522	3	49	generally	generally	ADV
fcis-30522	3	50	necessitate	necessitate	ADJ
fcis-30522	3	51	extensive	extensive	ADJ
fcis-30522	3	52	training	training	NOUN
fcis-30522	3	53	data	datum	NOUN
fcis-30522	3	54	.	.	PUNCT
fcis-30522	4	1	this	this	DET
fcis-30522	4	2	requirement	requirement	NOUN
fcis-30522	4	3	presents	present	VERB
fcis-30522	4	4	a	a	DET
fcis-30522	4	5	significant	significant	ADJ
fcis-30522	4	6	challenge	challenge	NOUN
fcis-30522	4	7	when	when	SCONJ
fcis-30522	4	8	processing	processing	NOUN
fcis-30522	4	9	complex	complex	ADJ
fcis-30522	4	10	,	,	PUNCT
fcis-30522	4	11	lengthy	lengthy	ADJ
fcis-30522	4	12	chinese	chinese	ADJ
fcis-30522	4	13	texts	text	NOUN
fcis-30522	4	14	and	and	CCONJ
fcis-30522	4	15	multi	multi	ADJ
fcis-30522	4	16	-	-	ADJ
fcis-30522	4	17	label	label	ADJ
fcis-30522	4	18	classification	classification	NOUN
fcis-30522	4	19	tasks	task	NOUN
fcis-30522	4	20	with	with	ADP
fcis-30522	4	21	limited	limited	ADJ
fcis-30522	4	22	data	datum	NOUN
fcis-30522	4	23	availability	availability	NOUN
fcis-30522	4	24	.	.	PUNCT
fcis-30522	5	1	furthermore	furthermore	ADV
fcis-30522	5	2	,	,	PUNCT
fcis-30522	5	3	conventional	conventional	ADJ
fcis-30522	5	4	data	datum	NOUN
fcis-30522	5	5	augmentation	augmentation	NOUN
fcis-30522	5	6	techniques	technique	NOUN
fcis-30522	5	7	frequently	frequently	ADV
fcis-30522	5	8	disrupt	disrupt	VERB
fcis-30522	5	9	the	the	DET
fcis-30522	5	10	original	original	ADJ
fcis-30522	5	11	word	word	NOUN
fcis-30522	5	12	order	order	NOUN
fcis-30522	5	13	,	,	PUNCT
fcis-30522	5	14	thereby	thereby	ADV
fcis-30522	5	15	diminishing	diminish	VERB
fcis-30522	5	16	their	their	PRON
fcis-30522	5	17	efficacy	efficacy	NOUN
fcis-30522	5	18	for	for	ADP
fcis-30522	5	19	pretrained	pretraine	VERB
fcis-30522	5	20	language	language	NOUN
fcis-30522	5	21	model	model	NOUN
fcis-30522	5	22	applications	application	NOUN
fcis-30522	5	23	.	.	PUNCT
fcis-30522	6	1	to	to	PART
fcis-30522	6	2	overcome	overcome	VERB
fcis-30522	6	3	these	these	DET
fcis-30522	6	4	limitations	limitation	NOUN
fcis-30522	6	5	,	,	PUNCT
fcis-30522	6	6	we	we	PRON
fcis-30522	6	7	introduce	introduce	VERB
fcis-30522	6	8	the	the	DET
fcis-30522	6	9	macbert	macbert	PROPN
fcis-30522	6	10	-	-	PUNCT
fcis-30522	6	11	cnn	cnn	PROPN
fcis-30522	6	12	-	-	PUNCT
fcis-30522	6	13	bilstm	bilstm	NOUN
fcis-30522	6	14	model	model	NOUN
fcis-30522	6	15	,	,	PUNCT
fcis-30522	6	16	which	which	PRON
fcis-30522	6	17	incorporates	incorporate	VERB
fcis-30522	6	18	a	a	DET
fcis-30522	6	19	multi	multi	ADJ
fcis-30522	6	20	-	-	ADJ
fcis-30522	6	21	level	level	ADJ
fcis-30522	6	22	self	self	NOUN
fcis-30522	6	23	-	-	PUNCT
fcis-30522	6	24	attention	attention	NOUN
fcis-30522	6	25	mechanism	mechanism	NOUN
fcis-30522	6	26	to	to	PART
fcis-30522	6	27	dynamically	dynamically	VERB
fcis-30522	6	28	weight	weight	VERB
fcis-30522	6	29	the	the	DET
fcis-30522	6	30	integrated	integrate	VERB
fcis-30522	6	31	features	feature	NOUN
fcis-30522	6	32	extracted	extract	VERB
fcis-30522	6	33	from	from	ADP
fcis-30522	6	34	macbert	macbert	PROPN
fcis-30522	6	35	,	,	PUNCT
fcis-30522	6	36	cnn	cnn	PROPN
fcis-30522	6	37	,	,	PUNCT
fcis-30522	6	38	and	and	CCONJ
fcis-30522	6	39	bilstm	bilstm	NOUN
fcis-30522	6	40	components	component	NOUN
fcis-30522	6	41	.	.	PUNCT
fcis-30522	7	1	our	our	PRON
fcis-30522	7	2	methodology	methodology	NOUN
fcis-30522	7	3	preserves	preserve	VERB
fcis-30522	7	4	the	the	DET
fcis-30522	7	5	integrity	integrity	NOUN
fcis-30522	7	6	of	of	ADP
fcis-30522	7	7	original	original	ADJ
fcis-30522	7	8	features	feature	NOUN
fcis-30522	7	9	during	during	ADP
fcis-30522	7	10	the	the	DET
fcis-30522	7	11	final	final	ADJ
fcis-30522	7	12	fusion	fusion	NOUN
fcis-30522	7	13	phase	phase	NOUN
fcis-30522	7	14	by	by	ADP
fcis-30522	7	15	combining	combine	VERB
fcis-30522	7	16	weighted	weight	VERB
fcis-30522	7	17	features	feature	NOUN
fcis-30522	7	18	with	with	ADP
fcis-30522	7	19	original	original	ADJ
fcis-30522	7	20	features	feature	NOUN
fcis-30522	7	21	through	through	ADP
fcis-30522	7	22	residual	residual	ADJ
fcis-30522	7	23	connections	connection	NOUN
fcis-30522	7	24	,	,	PUNCT
fcis-30522	7	25	thus	thus	ADV
fcis-30522	7	26	generating	generate	VERB
fcis-30522	7	27	a	a	DET
fcis-30522	7	28	comprehensive	comprehensive	ADJ
fcis-30522	7	29	final	final	ADJ
fcis-30522	7	30	representation	representation	NOUN
fcis-30522	7	31	.	.	PUNCT
fcis-30522	8	1	this	this	DET
fcis-30522	8	2	approach	approach	NOUN
fcis-30522	8	3	culminates	culminate	VERB
fcis-30522	8	4	in	in	ADP
fcis-30522	8	5	our	our	PRON
fcis-30522	8	6	macbert	macbert	NOUN
fcis-30522	8	7	-	-	PUNCT
fcis-30522	8	8	bilstm	bilstm	NOUN
fcis-30522	8	9	-	-	PUNCT
fcis-30522	8	10	cnn	cnn	NOUN
fcis-30522	8	11	-	-	PUNCT
fcis-30522	8	12	resattnet	resattnet	ADJ
fcis-30522	8	13	model	model	NOUN
fcis-30522	8	14	(	(	PUNCT
fcis-30522	8	15	mbcresattnet	mbcresattnet	NOUN
fcis-30522	8	16	)	)	PUNCT
fcis-30522	8	17	,	,	PUNCT
fcis-30522	8	18	specifically	specifically	ADV
fcis-30522	8	19	designed	design	VERB
fcis-30522	8	20	for	for	ADP
fcis-30522	8	21	multi	multi	ADJ
fcis-30522	8	22	-	-	ADJ
fcis-30522	8	23	label	label	ADJ
fcis-30522	8	24	classification	classification	NOUN
fcis-30522	8	25	of	of	ADP
fcis-30522	8	26	small	small	ADJ
fcis-30522	8	27	-	-	PUNCT
fcis-30522	8	28	sample	sample	NOUN
fcis-30522	8	29	chinese	chinese	ADJ
fcis-30522	8	30	literature	literature	PROPN
fcis-30522	8	31	abstracts	abstract	NOUN
fcis-30522	8	32	.	.	PUNCT
fcis-30522	9	1	we	we	PRON
fcis-30522	9	2	conducted	conduct	VERB
fcis-30522	9	3	extensive	extensive	ADJ
fcis-30522	9	4	evaluations	evaluation	NOUN
fcis-30522	9	5	of	of	ADP
fcis-30522	9	6	our	our	PRON
fcis-30522	9	7	model	model	NOUN
fcis-30522	9	8	across	across	ADP
fcis-30522	9	9	three	three	NUM
fcis-30522	9	10	datasets	dataset	NOUN
fcis-30522	9	11	:	:	PUNCT
fcis-30522	9	12	aeda	aeda	NOUN
fcis-30522	9	13	-	-	PUNCT
fcis-30522	9	14	augmented	augmented	ADJ
fcis-30522	9	15	,	,	PUNCT
fcis-30522	9	16	eda	eda	PROPN
fcis-30522	9	17	-	-	PUNCT
fcis-30522	9	18	augmented	augment	VERB
fcis-30522	9	19	,	,	PUNCT
fcis-30522	9	20	and	and	CCONJ
fcis-30522	9	21	original	original	ADJ
fcis-30522	9	22	samples	sample	NOUN
fcis-30522	9	23	,	,	PUNCT
fcis-30522	9	24	benchmarking	benchmarke	VERB
fcis-30522	9	25	against	against	ADP
fcis-30522	9	26	six	six	NUM
fcis-30522	9	27	alternative	alternative	ADJ
fcis-30522	9	28	models	model	NOUN
fcis-30522	9	29	.	.	PUNCT
fcis-30522	10	1	the	the	DET
fcis-30522	10	2	empirical	empirical	ADJ
fcis-30522	10	3	results	result	NOUN
fcis-30522	10	4	demonstrate	demonstrate	VERB
fcis-30522	10	5	that	that	SCONJ
fcis-30522	10	6	incorporating	incorporate	VERB
fcis-30522	10	7	pre	pre	ADJ
fcis-30522	10	8	-	-	ADJ
fcis-30522	10	9	trained	train	VERB
fcis-30522	10	10	language	language	NOUN
fcis-30522	10	11	models	model	NOUN
fcis-30522	10	12	substantially	substantially	ADV
fcis-30522	10	13	enhances	enhance	VERB
fcis-30522	10	14	classification	classification	NOUN
fcis-30522	10	15	performance	performance	NOUN
fcis-30522	10	16	.	.	PUNCT
fcis-30522	11	1	moreover	moreover	ADV
fcis-30522	11	2	,	,	PUNCT
fcis-30522	11	3	the	the	DET
fcis-30522	11	4	multi	multi	ADJ
fcis-30522	11	5	-	-	ADJ
fcis-30522	11	6	level	level	ADJ
fcis-30522	11	7	self	self	NOUN
fcis-30522	11	8	-	-	PUNCT
fcis-30522	11	9	attention	attention	NOUN
fcis-30522	11	10	mechanism	mechanism	NOUN
fcis-30522	11	11	combined	combine	VERB
fcis-30522	11	12	with	with	ADP
fcis-30522	11	13	residual	residual	ADJ
fcis-30522	11	14	feature	feature	NOUN
fcis-30522	11	15	fusion	fusion	NOUN
fcis-30522	11	16	effectively	effectively	ADV
fcis-30522	11	17	captures	capture	VERB
fcis-30522	11	18	global	global	ADJ
fcis-30522	11	19	textual	textual	ADJ
fcis-30522	11	20	patterns	pattern	NOUN
fcis-30522	11	21	,	,	PUNCT
fcis-30522	11	22	resulting	result	VERB
fcis-30522	11	23	in	in	ADP
fcis-30522	11	24	significant	significant	ADJ
fcis-30522	11	25	performance	performance	NOUN
fcis-30522	11	26	improvements	improvement	NOUN
fcis-30522	11	27	.	.	PUNCT
fcis-30522	12	1	in	in	ADP
fcis-30522	12	2	the	the	DET
fcis-30522	12	3	context	context	NOUN
fcis-30522	12	4	of	of	ADP
fcis-30522	12	5	pre	pre	ADJ
fcis-30522	12	6	-	-	ADJ
fcis-30522	12	7	trained	train	VERB
fcis-30522	12	8	language	language	NOUN
fcis-30522	12	9	models	model	NOUN
fcis-30522	12	10	,	,	PUNCT
fcis-30522	12	11	aeda	aeda	NOUN
fcis-30522	12	12	demonstrates	demonstrate	VERB
fcis-30522	12	13	superior	superior	ADJ
fcis-30522	12	14	efficacy	efficacy	NOUN
fcis-30522	12	15	compared	compare	VERB
fcis-30522	12	16	to	to	ADP
fcis-30522	12	17	eda	eda	PROPN
fcis-30522	12	18	in	in	ADP
fcis-30522	12	19	maintaining	maintain	VERB
fcis-30522	12	20	original	original	ADJ
fcis-30522	12	21	semantic	semantic	ADJ
fcis-30522	12	22	integrity	integrity	NOUN
fcis-30522	12	23	.	.	PUNCT
fcis-30522	13	1	additionally	additionally	ADV
fcis-30522	13	2	,	,	PUNCT
fcis-30522	13	3	the	the	DET
fcis-30522	13	4	residual	residual	ADJ
fcis-30522	13	5	feature	feature	NOUN
fcis-30522	13	6	fusion	fusion	NOUN
fcis-30522	13	7	methodology	methodology	NOUN
fcis-30522	13	8	preserves	preserve	VERB
fcis-30522	13	9	critical	critical	ADJ
fcis-30522	13	10	original	original	ADJ
fcis-30522	13	11	information	information	NOUN
fcis-30522	13	12	while	while	SCONJ
fcis-30522	13	13	markedly	markedly	ADV
fcis-30522	13	14	improving	improve	VERB
fcis-30522	13	15	model	model	NOUN
fcis-30522	13	16	performance	performance	NOUN
fcis-30522	13	17	.	.	PUNCT
fcis-30522	14	1	with	with	ADP
fcis-30522	14	2	the	the	DET
fcis-30522	14	3	implementation	implementation	NOUN
fcis-30522	14	4	of	of	ADP
fcis-30522	14	5	aeda	aeda	ADJ
fcis-30522	14	6	augmentation	augmentation	NOUN
fcis-30522	14	7	,	,	PUNCT
fcis-30522	14	8	all	all	DET
fcis-30522	14	9	evaluated	evaluate	VERB
fcis-30522	14	10	models	model	NOUN
fcis-30522	14	11	exhibited	exhibit	VERB
fcis-30522	14	12	performance	performance	NOUN
fcis-30522	14	13	gains	gain	NOUN
fcis-30522	14	14	exceeding	exceed	VERB
fcis-30522	14	15	10	10	NUM
fcis-30522	14	16	%	%	NOUN
fcis-30522	14	17	,	,	PUNCT
fcis-30522	14	18	with	with	ADP
fcis-30522	14	19	our	our	PRON
fcis-30522	14	20	mbcresattnet	mbcresattnet	NOUN
fcis-30522	14	21	model	model	NOUN
fcis-30522	14	22	attaining	attain	VERB
fcis-30522	14	23	96.17	96.17	NUM
fcis-30522	14	24	%	%	NOUN
fcis-30522	14	25	accuracy	accuracy	NOUN
fcis-30522	14	26	—	—	PUNCT
fcis-30522	14	27	representing	represent	VERB
fcis-30522	14	28	a	a	DET
fcis-30522	14	29	substantial	substantial	ADJ
fcis-30522	14	30	13.41	13.41	NUM
fcis-30522	14	31	%	%	NOUN
fcis-30522	14	32	improvement	improvement	NOUN
fcis-30522	14	33	over	over	ADP
fcis-30522	14	34	baseline	baseline	ADJ
fcis-30522	14	35	methods	method	NOUN
fcis-30522	14	36	.	.	PUNCT
fcis-30522	15	1	keywords	keyword	NOUN
fcis-30522	15	2	:	:	PUNCT
fcis-30522	15	3	macbert	macbert	PROPN
fcis-30522	15	4	;	;	PUNCT
fcis-30522	15	5	data	datum	NOUN
fcis-30522	15	6	enhancement	enhancement	NOUN
fcis-30522	15	7	;	;	PUNCT
fcis-30522	15	8	self	self	NOUN
fcis-30522	15	9	-	-	PUNCT
fcis-30522	15	10	attention	attention	NOUN
fcis-30522	15	11	mechanism	mechanism	NOUN
fcis-30522	15	12	;	;	PUNCT
fcis-30522	15	13	text	text	NOUN
fcis-30522	15	14	categorization	categorization	NOUN
fcis-30522	15	15	;	;	PUNCT
fcis-30522	15	16	small	small	ADJ
fcis-30522	15	17	samples	sample	NOUN
fcis-30522	15	18	.	.	PUNCT
fcis-30522	16	1	1	1	X
fcis-30522	16	2	.	.	X
fcis-30522	16	3	introduction	introduction	NOUN
fcis-30522	16	4	currently	currently	ADV
fcis-30522	16	5	,	,	PUNCT
fcis-30522	16	6	multimodal	multimodal	ADJ
fcis-30522	16	7	large	large	ADJ
fcis-30522	16	8	language	language	NOUN
fcis-30522	16	9	models	model	NOUN
fcis-30522	16	10	excel	excel	VERB
fcis-30522	16	11	in	in	ADP
fcis-30522	16	12	text	text	NOUN
fcis-30522	16	13	classification	classification	NOUN
fcis-30522	16	14	,	,	PUNCT
fcis-30522	16	15	but	but	CCONJ
fcis-30522	16	16	using	use	VERB
fcis-30522	16	17	paid	pay	VERB
fcis-30522	16	18	apis	apis	PROPN
fcis-30522	16	19	is	be	AUX
fcis-30522	16	20	costly	costly	ADJ
fcis-30522	16	21	.	.	PUNCT
fcis-30522	17	1	task	task	NOUN
fcis-30522	17	2	-	-	PUNCT
fcis-30522	17	3	specific	specific	ADJ
fcis-30522	17	4	deep	deep	ADJ
fcis-30522	17	5	learning	learning	NOUN
fcis-30522	17	6	models	model	NOUN
fcis-30522	17	7	are	be	AUX
fcis-30522	17	8	cheaper	cheap	ADJ
fcis-30522	17	9	but	but	CCONJ
fcis-30522	17	10	depend	depend	VERB
fcis-30522	17	11	heavily	heavily	ADV
fcis-30522	17	12	on	on	ADP
fcis-30522	17	13	training	training	NOUN
fcis-30522	17	14	data	datum	NOUN
fcis-30522	17	15	,	,	PUNCT
fcis-30522	17	16	especially	especially	ADV
fcis-30522	17	17	for	for	ADP
fcis-30522	17	18	complex	complex	ADJ
fcis-30522	17	19	chinese	chinese	ADJ
fcis-30522	17	20	texts	text	NOUN
fcis-30522	17	21	.	.	PUNCT
fcis-30522	18	1	for	for	ADP
fcis-30522	18	2	high	high	ADJ
fcis-30522	18	3	performance	performance	NOUN
fcis-30522	18	4	with	with	ADP
fcis-30522	18	5	small	small	ADJ
fcis-30522	18	6	samples	sample	NOUN
fcis-30522	18	7	,	,	PUNCT
fcis-30522	18	8	models	model	NOUN
fcis-30522	18	9	need	need	VERB
fcis-30522	18	10	strong	strong	ADJ
fcis-30522	18	11	local	local	ADJ
fcis-30522	18	12	feature	feature	NOUN
fcis-30522	18	13	extraction	extraction	NOUN
fcis-30522	18	14	,	,	PUNCT
fcis-30522	18	15	contextual	contextual	ADJ
fcis-30522	18	16	relationship	relationship	NOUN
fcis-30522	18	17	building	building	NOUN
fcis-30522	18	18	,	,	PUNCT
fcis-30522	18	19	and	and	CCONJ
fcis-30522	18	20	information	information	NOUN
fcis-30522	18	21	retention	retention	NOUN
fcis-30522	18	22	,	,	PUNCT
fcis-30522	18	23	plus	plus	CCONJ
fcis-30522	18	24	appropriate	appropriate	ADJ
fcis-30522	18	25	data	datum	NOUN
fcis-30522	18	26	augmentation	augmentation	NOUN
fcis-30522	18	27	without	without	ADP
fcis-30522	18	28	altering	alter	VERB
fcis-30522	18	29	original	original	ADJ
fcis-30522	18	30	semantics	semantic	NOUN
fcis-30522	18	31	.	.	PUNCT
fcis-30522	19	1	deep	deep	ADJ
fcis-30522	19	2	learning	learning	NOUN
fcis-30522	19	3	has	have	VERB
fcis-30522	19	4	advantages	advantage	NOUN
fcis-30522	19	5	over	over	ADP
fcis-30522	19	6	traditional	traditional	ADJ
fcis-30522	19	7	machine	machine	NOUN
fcis-30522	19	8	learning	learning	NOUN
fcis-30522	19	9	in	in	ADP
fcis-30522	19	10	text	text	NOUN
fcis-30522	19	11	analysis	analysis	NOUN
fcis-30522	19	12	,	,	PUNCT
fcis-30522	19	13	reducing	reduce	VERB
fcis-30522	19	14	manual	manual	ADJ
fcis-30522	19	15	feature	feature	NOUN
fcis-30522	19	16	engineering	engineering	NOUN
fcis-30522	19	17	through	through	ADP
fcis-30522	19	18	pre	pre	ADJ
fcis-30522	19	19	-	-	ADJ
fcis-30522	19	20	trained	train	VERB
fcis-30522	19	21	language	language	NOUN
fcis-30522	19	22	models	model	NOUN
fcis-30522	19	23	.	.	PUNCT
fcis-30522	20	1	while	while	SCONJ
fcis-30522	20	2	traditional	traditional	ADJ
fcis-30522	20	3	methods	method	NOUN
fcis-30522	20	4	struggle	struggle	VERB
fcis-30522	20	5	with	with	ADP
fcis-30522	20	6	unstructured	unstructured	ADJ
fcis-30522	20	7	text	text	NOUN
fcis-30522	20	8	,	,	PUNCT
fcis-30522	20	9	deep	deep	ADJ
fcis-30522	20	10	learning	learning	NOUN
fcis-30522	20	11	excels	excel	NOUN
fcis-30522	20	12	with	with	ADP
fcis-30522	20	13	complex	complex	ADJ
fcis-30522	20	14	semantics	semantic	NOUN
fcis-30522	20	15	.	.	PUNCT
fcis-30522	21	1	yoon	yoon	PROPN
fcis-30522	21	2	kim	kim	PROPN
fcis-30522	21	3	's	's	PART
fcis-30522	21	4	work[1	work[1	PROPN
fcis-30522	21	5	]	]	X
fcis-30522	21	6	effectively	effectively	ADV
fcis-30522	21	7	applied	apply	VERB
fcis-30522	21	8	cnns	cnn	NOUN
fcis-30522	21	9	to	to	PART
fcis-30522	21	10	text	text	VERB
fcis-30522	21	11	classification	classification	NOUN
fcis-30522	21	12	,	,	PUNCT
fcis-30522	21	13	influencing	influence	VERB
fcis-30522	21	14	subsequent	subsequent	ADJ
fcis-30522	21	15	research	research	NOUN
fcis-30522	21	16	.	.	PUNCT
fcis-30522	22	1	now	now	ADV
fcis-30522	22	2	deep	deep	ADJ
fcis-30522	22	3	learning	learning	NOUN
fcis-30522	22	4	is	be	AUX
fcis-30522	22	5	widely	widely	ADV
fcis-30522	22	6	used	use	VERB
fcis-30522	22	7	in	in	ADP
fcis-30522	22	8	nlp	nlp	ADJ
fcis-30522	22	9	tasks	task	NOUN
fcis-30522	22	10	:	:	PUNCT
fcis-30522	23	1	[	[	X
fcis-30522	23	2	2	2	NUM
fcis-30522	23	3	]	]	PUNCT
fcis-30522	23	4	applied	apply	VERB
fcis-30522	23	5	textcnn	textcnn	NOUN
fcis-30522	23	6	to	to	ADP
fcis-30522	23	7	adverse	adverse	ADJ
fcis-30522	23	8	drug	drug	NOUN
fcis-30522	23	9	event	event	NOUN
fcis-30522	23	10	detection	detection	NOUN
fcis-30522	23	11	;	;	PUNCT
fcis-30522	23	12	[	[	X
fcis-30522	23	13	3	3	X
fcis-30522	23	14	]	]	PUNCT
fcis-30522	23	15	used	use	VERB
fcis-30522	23	16	improved	improved	ADJ
fcis-30522	23	17	multi	multi	ADJ
fcis-30522	23	18	-	-	ADJ
fcis-30522	23	19	channel	channel	ADJ
fcis-30522	23	20	textcnn	textcnn	NOUN
fcis-30522	23	21	for	for	ADP
fcis-30522	23	22	short	short	ADJ
fcis-30522	23	23	texts	text	NOUN
fcis-30522	23	24	;	;	PUNCT
fcis-30522	23	25	[	[	X
fcis-30522	23	26	4	4	X
fcis-30522	23	27	]	]	PUNCT
fcis-30522	23	28	achieved	achieve	VERB
fcis-30522	23	29	excellence	excellence	NOUN
fcis-30522	23	30	in	in	ADP
fcis-30522	23	31	multi	multi	ADJ
fcis-30522	23	32	-	-	ADJ
fcis-30522	23	33	label	label	ADJ
fcis-30522	23	34	classification	classification	NOUN
fcis-30522	23	35	using	use	VERB
fcis-30522	23	36	cnn	cnn	PROPN
fcis-30522	23	37	;	;	PUNCT
fcis-30522	23	38	[	[	X
fcis-30522	23	39	5	5	NUM
fcis-30522	23	40	]	]	PUNCT
fcis-30522	23	41	proposed	propose	VERB
fcis-30522	23	42	an	an	DET
fcis-30522	23	43	effective	effective	ADJ
fcis-30522	23	44	cnn	cnn	ADJ
fcis-30522	23	45	-	-	PUNCT
fcis-30522	23	46	bilstm	bilstm	NOUN
fcis-30522	23	47	model	model	NOUN
fcis-30522	23	48	with	with	ADP
fcis-30522	23	49	attention	attention	NOUN
fcis-30522	23	50	for	for	ADP
fcis-30522	23	51	depression	depression	NOUN
fcis-30522	23	52	detection	detection	NOUN
fcis-30522	23	53	.	.	PUNCT
fcis-30522	24	1	most	most	ADJ
fcis-30522	24	2	research	research	NOUN
fcis-30522	24	3	focuses	focus	VERB
fcis-30522	24	4	on	on	ADP
fcis-30522	24	5	downstream	downstream	ADJ
fcis-30522	24	6	models	model	NOUN
fcis-30522	24	7	rather	rather	ADV
fcis-30522	24	8	than	than	ADP
fcis-30522	24	9	upstream	upstream	ADJ
fcis-30522	24	10	text	text	NOUN
fcis-30522	24	11	processing	processing	NOUN
fcis-30522	24	12	.	.	PUNCT
fcis-30522	25	1	google	google	PROPN
fcis-30522	25	2	's	's	PART
fcis-30522	25	3	bert[6	bert[6	NOUN
fcis-30522	25	4	]	]	PUNCT
fcis-30522	25	5	uses	use	VERB
fcis-30522	25	6	masked	mask	VERB
fcis-30522	25	7	language	language	NOUN
fcis-30522	25	8	modeling	modeling	NOUN
fcis-30522	25	9	,	,	PUNCT
fcis-30522	25	10	transformers	transformer	NOUN
fcis-30522	25	11	,	,	PUNCT
fcis-30522	25	12	and	and	CCONJ
fcis-30522	25	13	self	self	NOUN
fcis-30522	25	14	-	-	PUNCT
fcis-30522	25	15	attention	attention	NOUN
fcis-30522	25	16	to	to	PART
fcis-30522	25	17	capture	capture	VERB
fcis-30522	25	18	sentence	sentence	NOUN
fcis-30522	25	19	structure	structure	NOUN
fcis-30522	25	20	and	and	CCONJ
fcis-30522	25	21	context	context	NOUN
fcis-30522	25	22	,	,	PUNCT
fcis-30522	25	23	solving	solve	VERB
fcis-30522	25	24	word2vec	word2vec	PROPN
fcis-30522	25	25	's	's	PART
fcis-30522	25	26	polysemy	polysemy	NOUN
fcis-30522	25	27	limitations	limitation	NOUN
fcis-30522	25	28	.	.	PUNCT
fcis-30522	26	1	[	[	X
fcis-30522	26	2	7	7	X
fcis-30522	26	3	]	]	PUNCT
fcis-30522	26	4	and	and	CCONJ
fcis-30522	26	5	[	[	X
fcis-30522	26	6	8	8	NUM
fcis-30522	26	7	]	]	PUNCT
fcis-30522	26	8	showed	show	VERB
fcis-30522	26	9	bert	bert	PROPN
fcis-30522	26	10	improves	improve	VERB
fcis-30522	26	11	classification	classification	NOUN
fcis-30522	26	12	performance	performance	NOUN
fcis-30522	26	13	.	.	PUNCT
fcis-30522	27	1	[	[	X
fcis-30522	27	2	9	9	NUM
fcis-30522	27	3	]	]	PUNCT
fcis-30522	27	4	proposed	propose	VERB
fcis-30522	27	5	ff	ff	NOUN
fcis-30522	27	6	-	-	PUNCT
fcis-30522	27	7	bert	bert	NOUN
fcis-30522	27	8	for	for	ADP
fcis-30522	27	9	flash	flash	NOUN
fcis-30522	27	10	flood	flood	NOUN
fcis-30522	27	11	classification	classification	NOUN
fcis-30522	27	12	.	.	PUNCT
fcis-30522	28	1	[	[	X
fcis-30522	28	2	10	10	NUM
fcis-30522	28	3	]	]	PUNCT
fcis-30522	28	4	demonstrated	demonstrate	VERB
fcis-30522	28	5	bert+bilstm	bert+bilstm	PROPN
fcis-30522	28	6	+	+	NOUN
fcis-30522	28	7	attention	attention	NOUN
fcis-30522	28	8	outperforms	outperform	NOUN
fcis-30522	28	9	word2vec	word2vec	X
fcis-30522	28	10	in	in	ADP
fcis-30522	28	11	chinese	chinese	ADJ
fcis-30522	28	12	medicine	medicine	NOUN
fcis-30522	28	13	text	text	NOUN
fcis-30522	28	14	classification	classification	NOUN
fcis-30522	28	15	.	.	PUNCT
fcis-30522	29	1	bert	bert	PROPN
fcis-30522	29	2	quickly	quickly	ADV
fcis-30522	29	3	became	become	VERB
fcis-30522	29	4	fundamental	fundamental	ADJ
fcis-30522	29	5	in	in	ADP
fcis-30522	29	6	nlp	nlp	NOUN
fcis-30522	29	7	,	,	PUNCT
fcis-30522	29	8	inspiring	inspire	VERB
fcis-30522	29	9	improved	improved	ADJ
fcis-30522	29	10	models	model	NOUN
fcis-30522	29	11	like	like	ADP
fcis-30522	29	12	roberta[11	roberta[11	PROPN
fcis-30522	29	13	]	]	PUNCT
fcis-30522	29	14	,	,	PUNCT
fcis-30522	29	15	albert[12	albert[12	PROPN
fcis-30522	29	16	]	]	PUNCT
fcis-30522	29	17	,	,	PUNCT
fcis-30522	29	18	distilbert	distilbert	VERB
fcis-30522	29	19	[	[	X
fcis-30522	29	20	13	13	NUM
fcis-30522	29	21	]	]	PUNCT
fcis-30522	29	22	,	,	PUNCT
fcis-30522	29	23	and	and	CCONJ
fcis-30522	29	24	chinesebert[14	chinesebert[14	PROPN
fcis-30522	29	25	]	]	PUNCT
fcis-30522	29	26	which	which	PRON
fcis-30522	29	27	incorporates	incorporate	VERB
fcis-30522	29	28	grapheme	grapheme	ADJ
fcis-30522	29	29	and	and	CCONJ
fcis-30522	29	30	pinyin	pinyin	NOUN
fcis-30522	29	31	information	information	NOUN
fcis-30522	29	32	.	.	PUNCT
fcis-30522	30	1	macbert[15	macbert[15	PROPN
fcis-30522	30	2	]	]	PUNCT
fcis-30522	30	3	introduced	introduce	VERB
fcis-30522	30	4	correction	correction	NOUN
fcis-30522	30	5	-	-	PUNCT
fcis-30522	30	6	type	type	NOUN
fcis-30522	30	7	masked	mask	VERB
fcis-30522	30	8	language	language	NOUN
fcis-30522	30	9	modeling	modeling	NOUN
fcis-30522	30	10	,	,	PUNCT
fcis-30522	30	11	effectively	effectively	ADV
fcis-30522	30	12	addressing	address	VERB
fcis-30522	30	13	pre	pre	ADJ
fcis-30522	30	14	-	-	ADJ
fcis-30522	30	15	training	training	ADJ
fcis-30522	30	16	and	and	CCONJ
fcis-30522	30	17	downstream	downstream	ADJ
fcis-30522	30	18	task	task	NOUN
fcis-30522	30	19	inconsistencies	inconsistency	NOUN
fcis-30522	30	20	,	,	PUNCT
fcis-30522	30	21	outperforming	outperform	VERB
fcis-30522	30	22	original	original	ADJ
fcis-30522	30	23	bert	bert	NOUN
fcis-30522	30	24	in	in	ADP
fcis-30522	30	25	chinese	chinese	ADJ
fcis-30522	30	26	nlp	nlp	NOUN
fcis-30522	30	27	tasks	task	NOUN
fcis-30522	30	28	.	.	PUNCT
fcis-30522	31	1	this	this	DET
fcis-30522	31	2	paper	paper	NOUN
fcis-30522	31	3	uses	use	VERB
fcis-30522	31	4	macbert	macbert	NOUN
fcis-30522	31	5	for	for	ADP
fcis-30522	31	6	upstream	upstream	ADJ
fcis-30522	31	7	feature	feature	NOUN
fcis-30522	31	8	extraction	extraction	NOUN
fcis-30522	31	9	and	and	CCONJ
fcis-30522	31	10	textcnn	textcnn	VERB
fcis-30522	31	11	for	for	ADP
fcis-30522	31	12	downstream	downstream	ADJ
fcis-30522	31	13	local	local	ADJ
fcis-30522	31	14	feature	feature	NOUN
fcis-30522	31	15	capture	capture	NOUN
fcis-30522	31	16	.	.	PUNCT
fcis-30522	32	1	since	since	SCONJ
fcis-30522	32	2	cnn	cnn	PROPN
fcis-30522	32	3	may	may	AUX
fcis-30522	32	4	miss	miss	VERB
fcis-30522	32	5	global	global	ADJ
fcis-30522	32	6	features[16	features[16	PROPN
fcis-30522	32	7	]	]	PUNCT
fcis-30522	32	8	,	,	PUNCT
fcis-30522	32	9	bilstm	bilstm	NOUN
fcis-30522	32	10	is	be	AUX
fcis-30522	32	11	added	add	VERB
fcis-30522	32	12	for	for	ADP
fcis-30522	32	13	contextual	contextual	ADJ
fcis-30522	32	14	connections	connection	NOUN
fcis-30522	32	15	as	as	SCONJ
fcis-30522	32	16	suggested	suggest	VERB
fcis-30522	32	17	by	by	ADP
fcis-30522	32	18	[	[	X
fcis-30522	32	19	17	17	NUM
fcis-30522	32	20	]	]	PUNCT
fcis-30522	32	21	.	.	PUNCT
fcis-30522	33	1	[	[	X
fcis-30522	33	2	18	18	NUM
fcis-30522	33	3	]	]	PUNCT
fcis-30522	33	4	noted	note	VERB
fcis-30522	33	5	cnn	cnn	PROPN
fcis-30522	33	6	-	-	PUNCT
fcis-30522	33	7	bilstm	bilstm	NOUN
fcis-30522	33	8	combinations	combination	NOUN
fcis-30522	33	9	compensate	compensate	NOUN
fcis-30522	33	10	for	for	ADP
fcis-30522	33	11	each	each	DET
fcis-30522	33	12	other	other	ADJ
fcis-30522	33	13	's	's	PART
fcis-30522	33	14	weaknesses	weakness	NOUN
fcis-30522	33	15	in	in	ADP
fcis-30522	33	16	feature	feature	NOUN
fcis-30522	33	17	extraction	extraction	NOUN
fcis-30522	33	18	.	.	PUNCT
fcis-30522	34	1	the	the	DET
fcis-30522	34	2	self	self	NOUN
fcis-30522	34	3	-	-	PUNCT
fcis-30522	34	4	attention	attention	NOUN
fcis-30522	34	5	mechanism	mechanism	NOUN
fcis-30522	34	6	highlights	highlight	VERB
fcis-30522	34	7	key	key	ADJ
fcis-30522	34	8	words	word	NOUN
fcis-30522	34	9	and	and	CCONJ
fcis-30522	34	10	captures	capture	VERB
fcis-30522	34	11	global	global	ADJ
fcis-30522	34	12	features	feature	NOUN
fcis-30522	34	13	.	.	PUNCT
fcis-30522	35	1	[	[	X
fcis-30522	35	2	19	19	NUM
fcis-30522	35	3	]	]	PUNCT
fcis-30522	35	4	achieved	achieve	VERB
fcis-30522	35	5	accurate	accurate	ADJ
fcis-30522	35	6	polarity	polarity	NOUN
fcis-30522	35	7	prediction	prediction	NOUN
fcis-30522	35	8	using	use	VERB
fcis-30522	35	9	bert	bert	NOUN
fcis-30522	35	10	-	-	PUNCT
fcis-30522	35	11	bilstm	bilstm	NOUN
fcis-30522	35	12	with	with	ADP
fcis-30522	35	13	improved	improved	ADJ
fcis-30522	35	14	self	self	NOUN
fcis-30522	35	15	-	-	PUNCT
fcis-30522	35	16	attention	attention	NOUN
fcis-30522	35	17	.	.	PUNCT
fcis-30522	36	1	[	[	X
fcis-30522	36	2	20	20	NUM
fcis-30522	36	3	]	]	PUNCT
fcis-30522	36	4	used	use	VERB
fcis-30522	36	5	multi	multi	ADJ
fcis-30522	36	6	-	-	ADJ
fcis-30522	36	7	head	head	ADJ
fcis-30522	36	8	self	self	NOUN
fcis-30522	36	9	-	-	PUNCT
fcis-30522	36	10	attention	attention	NOUN
fcis-30522	36	11	for	for	ADP
fcis-30522	36	12	complex	complex	ADJ
fcis-30522	36	13	contextual	contextual	ADJ
fcis-30522	36	14	relationships	relationship	NOUN
fcis-30522	36	15	.	.	PUNCT
fcis-30522	37	1	[	[	X
fcis-30522	37	2	21	21	NUM
fcis-30522	37	3	]	]	PUNCT
fcis-30522	37	4	improved	improve	VERB
fcis-30522	37	5	short	short	ADJ
fcis-30522	37	6	text	text	NOUN
fcis-30522	37	7	classification	classification	NOUN
fcis-30522	37	8	performance	performance	NOUN
fcis-30522	37	9	.	.	PUNCT
fcis-30522	38	1	[	[	X
fcis-30522	38	2	22	22	NUM
fcis-30522	38	3	]	]	PUNCT
fcis-30522	38	4	constructed	construct	VERB
fcis-30522	38	5	a	a	DET
fcis-30522	38	6	quantum	quantum	ADJ
fcis-30522	38	7	self	self	NOUN
fcis-30522	38	8	-	-	PUNCT
fcis-30522	38	9	attention	attention	NOUN
fcis-30522	38	10	neural	neural	ADJ
fcis-30522	38	11	network	network	NOUN
fcis-30522	38	12	with	with	ADP
fcis-30522	38	13	excellent	excellent	ADJ
fcis-30522	38	14	text	text	NOUN
fcis-30522	38	15	classification	classification	NOUN
fcis-30522	38	16	performance	performance	NOUN
fcis-30522	38	17	.	.	PUNCT
fcis-30522	39	1	this	this	DET
fcis-30522	39	2	paper	paper	NOUN
fcis-30522	39	3	implements	implement	VERB
fcis-30522	39	4	self	self	NOUN
fcis-30522	39	5	-	-	PUNCT
fcis-30522	39	6	attention	attention	NOUN
fcis-30522	39	7	mechanisms	mechanism	NOUN
fcis-30522	39	8	at	at	ADP
fcis-30522	39	9	multiple	multiple	ADJ
fcis-30522	39	10	levels	level	NOUN
fcis-30522	39	11	.	.	PUNCT
fcis-30522	40	1	[	[	X
fcis-30522	40	2	23	23	NUM
fcis-30522	40	3	]	]	PUNCT
fcis-30522	40	4	improved	improve	VERB
fcis-30522	40	5	classification	classification	NOUN
fcis-30522	40	6	by	by	ADP
fcis-30522	40	7	fusing	fuse	VERB
fcis-30522	40	8	features	feature	NOUN
fcis-30522	40	9	from	from	ADP
fcis-30522	40	10	cnn	cnn	PROPN
fcis-30522	40	11	and	and	CCONJ
fcis-30522	40	12	lstm	lstm	NOUN
fcis-30522	40	13	models	model	NOUN
fcis-30522	40	14	.	.	PUNCT
fcis-30522	41	1	[	[	X
fcis-30522	41	2	24	24	NUM
fcis-30522	41	3	]	]	PUNCT
fcis-30522	41	4	enhanced	enhanced	ADJ
fcis-30522	41	5	performance	performance	NOUN
fcis-30522	41	6	using	use	VERB
fcis-30522	41	7	feature	feature	NOUN
fcis-30522	41	8	selection	selection	NOUN
fcis-30522	41	9	fusion	fusion	NOUN
fcis-30522	41	10	.	.	PUNCT
fcis-30522	42	1	with	with	ADP
fcis-30522	42	2	limited	limited	ADJ
fcis-30522	42	3	small	small	ADJ
fcis-30522	42	4	-	-	PUNCT
fcis-30522	42	5	sample	sample	NOUN
fcis-30522	42	6	information	information	NOUN
fcis-30522	42	7	,	,	PUNCT
fcis-30522	42	8	this	this	DET
fcis-30522	42	9	71	71	NUM
fcis-30522	42	10	paper	paper	NOUN
fcis-30522	42	11	retains	retain	VERB
fcis-30522	42	12	both	both	DET
fcis-30522	42	13	original	original	ADJ
fcis-30522	42	14	and	and	CCONJ
fcis-30522	42	15	attention	attention	NOUN
fcis-30522	42	16	-	-	PUNCT
fcis-30522	42	17	weighted	weight	VERB
fcis-30522	42	18	features	feature	NOUN
fcis-30522	42	19	through	through	ADP
fcis-30522	42	20	residual	residual	ADJ
fcis-30522	42	21	connections	connection	NOUN
fcis-30522	42	22	.	.	PUNCT
fcis-30522	43	1	for	for	ADP
fcis-30522	43	2	small	small	ADJ
fcis-30522	43	3	-	-	PUNCT
fcis-30522	43	4	sample	sample	NOUN
fcis-30522	43	5	data	datum	NOUN
fcis-30522	43	6	augmentation	augmentation	NOUN
fcis-30522	43	7	,	,	PUNCT
fcis-30522	43	8	[	[	X
fcis-30522	43	9	25	25	NUM
fcis-30522	43	10	]	]	PUNCT
fcis-30522	43	11	proposed	propose	VERB
fcis-30522	43	12	symbolic	symbolic	ADJ
fcis-30522	43	13	augmentation	augmentation	NOUN
fcis-30522	43	14	like	like	ADP
fcis-30522	43	15	word	word	NOUN
fcis-30522	43	16	replacement	replacement	NOUN
fcis-30522	43	17	,	,	PUNCT
fcis-30522	43	18	which	which	PRON
fcis-30522	43	19	risks	risk	VERB
fcis-30522	43	20	changing	change	VERB
fcis-30522	43	21	original	original	ADJ
fcis-30522	43	22	semantics	semantic	NOUN
fcis-30522	43	23	.	.	PUNCT
fcis-30522	44	1	[	[	X
fcis-30522	44	2	26	26	NUM
fcis-30522	44	3	]	]	PUNCT
fcis-30522	44	4	introduced	introduce	VERB
fcis-30522	44	5	easy	easy	ADJ
fcis-30522	44	6	data	datum	NOUN
fcis-30522	44	7	augmentation	augmentation	NOUN
fcis-30522	44	8	(	(	PUNCT
fcis-30522	44	9	eda	eda	PROPN
fcis-30522	44	10	)	)	PUNCT
fcis-30522	44	11	with	with	ADP
fcis-30522	44	12	four	four	NUM
fcis-30522	44	13	operations	operation	NOUN
fcis-30522	44	14	:	:	PUNCT
fcis-30522	44	15	synonym	synonym	NOUN
fcis-30522	44	16	replacement	replacement	NOUN
fcis-30522	44	17	,	,	PUNCT
fcis-30522	44	18	random	random	ADJ
fcis-30522	44	19	insertion	insertion	NOUN
fcis-30522	44	20	,	,	PUNCT
fcis-30522	44	21	word	word	NOUN
fcis-30522	44	22	swapping	swapping	NOUN
fcis-30522	44	23	,	,	PUNCT
fcis-30522	44	24	and	and	CCONJ
fcis-30522	44	25	deletion	deletion	NOUN
fcis-30522	44	26	.	.	PUNCT
fcis-30522	45	1	[	[	X
fcis-30522	45	2	27	27	NUM
fcis-30522	45	3	]	]	PUNCT
fcis-30522	45	4	achieved	achieve	VERB
fcis-30522	45	5	improvements	improvement	NOUN
fcis-30522	45	6	using	use	VERB
fcis-30522	45	7	eda	eda	PROPN
fcis-30522	45	8	for	for	ADP
fcis-30522	45	9	medical	medical	ADJ
fcis-30522	45	10	data	datum	NOUN
fcis-30522	45	11	.	.	PUNCT
fcis-30522	46	1	however	however	ADV
fcis-30522	46	2	,	,	PUNCT
fcis-30522	46	3	with	with	ADP
fcis-30522	46	4	pre	pre	ADJ
fcis-30522	46	5	-	-	ADJ
fcis-30522	46	6	trained	train	VERB
fcis-30522	46	7	models	model	NOUN
fcis-30522	46	8	like	like	ADP
fcis-30522	46	9	bert	bert	PROPN
fcis-30522	46	10	,	,	PUNCT
fcis-30522	46	11	eda	eda	PROPN
fcis-30522	46	12	may	may	AUX
fcis-30522	46	13	change	change	VERB
fcis-30522	46	14	word	word	NOUN
fcis-30522	46	15	order	order	NOUN
fcis-30522	46	16	or	or	CCONJ
fcis-30522	46	17	cause	cause	VERB
fcis-30522	46	18	information	information	NOUN
fcis-30522	46	19	loss	loss	NOUN
fcis-30522	46	20	,	,	PUNCT
fcis-30522	46	21	potentially	potentially	ADV
fcis-30522	46	22	decreasing	decrease	VERB
fcis-30522	46	23	model	model	NOUN
fcis-30522	46	24	performance	performance	NOUN
fcis-30522	46	25	.	.	PUNCT
fcis-30522	47	1	the	the	DET
fcis-30522	47	2	aeda	aeda	ADJ
fcis-30522	47	3	method	method	NOUN
fcis-30522	47	4	[	[	X
fcis-30522	47	5	28	28	NUM
fcis-30522	47	6	]	]	X
fcis-30522	47	7	randomly	randomly	ADJ
fcis-30522	47	8	inserts	insert	NOUN
fcis-30522	47	9	punctuation	punctuation	NOUN
fcis-30522	47	10	,	,	PUNCT
fcis-30522	47	11	maintaining	maintain	VERB
fcis-30522	47	12	word	word	NOUN
fcis-30522	47	13	order	order	NOUN
fcis-30522	47	14	while	while	SCONJ
fcis-30522	47	15	improving	improve	VERB
fcis-30522	47	16	generalization	generalization	NOUN
fcis-30522	47	17	.	.	PUNCT
fcis-30522	48	1	this	this	DET
fcis-30522	48	2	paper	paper	NOUN
fcis-30522	48	3	explores	explore	NOUN
fcis-30522	48	4	aeda	aeda	NOUN
fcis-30522	48	5	's	's	PART
fcis-30522	48	6	combination	combination	NOUN
fcis-30522	48	7	with	with	ADP
fcis-30522	48	8	chinese	chinese	ADJ
fcis-30522	48	9	pre	pre	ADJ
fcis-30522	48	10	-	-	ADJ
fcis-30522	48	11	trained	train	VERB
fcis-30522	48	12	language	language	NOUN
fcis-30522	48	13	models	model	NOUN
fcis-30522	49	1	.	.	PUNCT
fcis-30522	50	1	the	the	DET
fcis-30522	50	2	resulting	result	VERB
fcis-30522	50	3	mbcresattnet	mbcresattnet	NOUN
fcis-30522	50	4	model	model	NOUN
fcis-30522	50	5	with	with	ADP
fcis-30522	50	6	aeda	aeda	ADJ
fcis-30522	50	7	data	datum	NOUN
fcis-30522	50	8	augmentation	augmentation	NOUN
fcis-30522	50	9	performs	perform	VERB
fcis-30522	50	10	excellently	excellently	ADV
fcis-30522	50	11	in	in	ADP
fcis-30522	50	12	multi	multi	ADJ
fcis-30522	50	13	-	-	ADJ
fcis-30522	50	14	label	label	ADJ
fcis-30522	50	15	classification	classification	NOUN
fcis-30522	50	16	of	of	ADP
fcis-30522	50	17	small	small	ADJ
fcis-30522	50	18	-	-	PUNCT
fcis-30522	50	19	sample	sample	NOUN
fcis-30522	50	20	chinese	chinese	ADJ
fcis-30522	50	21	literature	literature	PROPN
fcis-30522	50	22	abstracts	abstract	NOUN
fcis-30522	50	23	,	,	PUNCT
fcis-30522	50	24	providing	provide	VERB
fcis-30522	50	25	valuable	valuable	ADJ
fcis-30522	50	26	reference	reference	NOUN
fcis-30522	50	27	for	for	ADP
fcis-30522	50	28	combining	combine	VERB
fcis-30522	50	29	data	datum	NOUN
fcis-30522	50	30	augmentation	augmentation	NOUN
fcis-30522	50	31	with	with	ADP
fcis-30522	50	32	pre	pre	ADJ
fcis-30522	50	33	-	-	ADJ
fcis-30522	50	34	trained	train	VERB
fcis-30522	50	35	language	language	NOUN
fcis-30522	50	36	models	model	NOUN
fcis-30522	50	37	in	in	ADP
fcis-30522	50	38	smallsample	smallsample	ADJ
fcis-30522	50	39	multi	multi	ADJ
fcis-30522	50	40	-	-	ADJ
fcis-30522	50	41	label	label	ADJ
fcis-30522	50	42	classification	classification	NOUN
fcis-30522	50	43	.	.	PUNCT
fcis-30522	51	1	2	2	X
fcis-30522	51	2	.	.	X
fcis-30522	51	3	materials	material	NOUN
fcis-30522	51	4	and	and	CCONJ
fcis-30522	51	5	methods	method	NOUN
fcis-30522	51	6	2.1	2.1	NUM
fcis-30522	51	7	.	.	PUNCT
fcis-30522	52	1	mbcresattnet	mbcresattnet	NOUN
fcis-30522	52	2	model	model	NOUN
fcis-30522	52	3	to	to	PART
fcis-30522	52	4	reduce	reduce	VERB
fcis-30522	52	5	the	the	DET
fcis-30522	52	6	risk	risk	NOUN
fcis-30522	52	7	of	of	ADP
fcis-30522	52	8	overfitting	overfitte	VERB
fcis-30522	52	9	and	and	CCONJ
fcis-30522	52	10	improve	improve	VERB
fcis-30522	52	11	the	the	DET
fcis-30522	52	12	model	model	NOUN
fcis-30522	52	13	's	's	PART
fcis-30522	52	14	generalization	generalization	NOUN
fcis-30522	52	15	ability	ability	NOUN
fcis-30522	52	16	,	,	PUNCT
fcis-30522	52	17	the	the	DET
fcis-30522	52	18	dropout	dropout	NOUN
fcis-30522	52	19	algorithm	algorithm	NOUN
fcis-30522	52	20	is	be	AUX
fcis-30522	52	21	introduced	introduce	VERB
fcis-30522	52	22	based	base	VERB
fcis-30522	52	23	on	on	ADP
fcis-30522	52	24	the	the	DET
fcis-30522	52	25	above	above	ADJ
fcis-30522	52	26	model	model	NOUN
fcis-30522	52	27	.	.	PUNCT
fcis-30522	53	1	the	the	DET
fcis-30522	53	2	risk	risk	NOUN
fcis-30522	53	3	of	of	ADP
fcis-30522	53	4	overfitting	overfitting	NOUN
fcis-30522	53	5	is	be	AUX
fcis-30522	53	6	reduced	reduce	VERB
fcis-30522	53	7	by	by	ADP
fcis-30522	53	8	discarding	discard	VERB
fcis-30522	53	9	a	a	DET
fcis-30522	53	10	portion	portion	NOUN
fcis-30522	53	11	of	of	ADP
fcis-30522	53	12	the	the	DET
fcis-30522	53	13	neurons	neuron	NOUN
fcis-30522	53	14	.	.	PUNCT
fcis-30522	54	1	at	at	ADP
fcis-30522	54	2	the	the	DET
fcis-30522	54	3	same	same	ADJ
fcis-30522	54	4	time	time	NOUN
fcis-30522	54	5	,	,	PUNCT
fcis-30522	54	6	the	the	DET
fcis-30522	54	7	model	model	NOUN
fcis-30522	54	8	's	's	PART
fcis-30522	54	9	capability	capability	NOUN
fcis-30522	54	10	is	be	AUX
fcis-30522	54	11	enhanced	enhance	VERB
fcis-30522	54	12	because	because	SCONJ
fcis-30522	54	13	a	a	DET
fcis-30522	54	14	multi	multi	ADJ
fcis-30522	54	15	-	-	ADJ
fcis-30522	54	16	level	level	ADJ
fcis-30522	54	17	selfattention	selfattention	NOUN
fcis-30522	54	18	mechanism	mechanism	NOUN
fcis-30522	54	19	is	be	AUX
fcis-30522	54	20	introduced	introduce	VERB
fcis-30522	54	21	into	into	ADP
fcis-30522	54	22	the	the	DET
fcis-30522	54	23	model	model	NOUN
fcis-30522	54	24	to	to	PART
fcis-30522	54	25	ensure	ensure	VERB
fcis-30522	54	26	that	that	SCONJ
fcis-30522	54	27	information	information	NOUN
fcis-30522	54	28	is	be	AUX
fcis-30522	54	29	not	not	PART
fcis-30522	54	30	lost	lose	VERB
fcis-30522	54	31	during	during	ADP
fcis-30522	54	32	processing	processing	NOUN
fcis-30522	54	33	.	.	PUNCT
fcis-30522	55	1	in	in	ADP
fcis-30522	55	2	the	the	DET
fcis-30522	55	3	final	final	ADJ
fcis-30522	55	4	feature	feature	NOUN
fcis-30522	55	5	fusion	fusion	NOUN
fcis-30522	55	6	stage	stage	NOUN
fcis-30522	55	7	,	,	PUNCT
fcis-30522	55	8	the	the	DET
fcis-30522	55	9	comprehensive	comprehensive	ADJ
fcis-30522	55	10	features	feature	NOUN
fcis-30522	55	11	obtained	obtain	VERB
fcis-30522	55	12	by	by	ADP
fcis-30522	55	13	concatenating	concatenate	VERB
fcis-30522	55	14	bilstm_out	bilstm_out	PRON
fcis-30522	55	15	and	and	CCONJ
fcis-30522	55	16	textcnn_out	textcnn_out	NUM
fcis-30522	55	17	are	be	AUX
fcis-30522	55	18	first	first	ADV
fcis-30522	55	19	saved	save	VERB
fcis-30522	55	20	.	.	PUNCT
fcis-30522	56	1	after	after	SCONJ
fcis-30522	56	2	these	these	DET
fcis-30522	56	3	comprehensive	comprehensive	ADJ
fcis-30522	56	4	features	feature	NOUN
fcis-30522	56	5	are	be	AUX
fcis-30522	56	6	processed	process	VERB
fcis-30522	56	7	by	by	ADP
fcis-30522	56	8	the	the	DET
fcis-30522	56	9	final	final	ADJ
fcis-30522	56	10	self	self	NOUN
fcis-30522	56	11	-	-	PUNCT
fcis-30522	56	12	attention	attention	NOUN
fcis-30522	56	13	mechanism	mechanism	NOUN
fcis-30522	56	14	,	,	PUNCT
fcis-30522	56	15	a	a	DET
fcis-30522	56	16	weighted	weight	VERB
fcis-30522	56	17	feature	feature	NOUN
fcis-30522	56	18	representation	representation	NOUN
fcis-30522	56	19	is	be	AUX
fcis-30522	56	20	obtained	obtain	VERB
fcis-30522	56	21	.	.	PUNCT
fcis-30522	57	1	this	this	PRON
fcis-30522	57	2	weighted	weight	VERB
fcis-30522	57	3	feature	feature	NOUN
fcis-30522	57	4	and	and	CCONJ
fcis-30522	57	5	the	the	DET
fcis-30522	57	6	previously	previously	ADV
fcis-30522	57	7	retained	retain	VERB
fcis-30522	57	8	original	original	ADJ
fcis-30522	57	9	comprehensive	comprehensive	ADJ
fcis-30522	57	10	features	feature	NOUN
fcis-30522	57	11	are	be	AUX
fcis-30522	57	12	fused	fuse	VERB
fcis-30522	57	13	into	into	ADP
fcis-30522	57	14	the	the	DET
fcis-30522	57	15	final	final	ADJ
fcis-30522	57	16	feature	feature	NOUN
fcis-30522	57	17	representation	representation	NOUN
fcis-30522	57	18	by	by	ADP
fcis-30522	57	19	means	mean	NOUN
fcis-30522	57	20	of	of	ADP
fcis-30522	57	21	a	a	DET
fcis-30522	57	22	residual	residual	ADJ
fcis-30522	57	23	connection	connection	NOUN
fcis-30522	57	24	,	,	PUNCT
fcis-30522	57	25	and	and	CCONJ
fcis-30522	57	26	then	then	ADV
fcis-30522	57	27	passed	pass	VERB
fcis-30522	57	28	through	through	ADP
fcis-30522	57	29	a	a	DET
fcis-30522	57	30	fully	fully	ADV
fcis-30522	57	31	connected	connect	VERB
fcis-30522	57	32	layer	layer	NOUN
fcis-30522	57	33	to	to	PART
fcis-30522	57	34	complete	complete	VERB
fcis-30522	57	35	the	the	DET
fcis-30522	57	36	final	final	ADJ
fcis-30522	57	37	classification	classification	NOUN
fcis-30522	57	38	task	task	NOUN
fcis-30522	57	39	.	.	PUNCT
fcis-30522	58	1	this	this	DET
fcis-30522	58	2	method	method	NOUN
fcis-30522	58	3	ensures	ensure	VERB
fcis-30522	58	4	that	that	SCONJ
fcis-30522	58	5	the	the	DET
fcis-30522	58	6	original	original	ADJ
fcis-30522	58	7	feature	feature	NOUN
fcis-30522	58	8	information	information	NOUN
fcis-30522	58	9	is	be	AUX
fcis-30522	58	10	not	not	PART
fcis-30522	58	11	lost	lose	VERB
fcis-30522	58	12	and	and	CCONJ
fcis-30522	58	13	can	can	AUX
fcis-30522	58	14	avoid	avoid	VERB
fcis-30522	58	15	the	the	DET
fcis-30522	58	16	loss	loss	NOUN
fcis-30522	58	17	of	of	ADP
fcis-30522	58	18	feature	feature	NOUN
fcis-30522	58	19	information	information	NOUN
fcis-30522	58	20	between	between	ADP
fcis-30522	58	21	layers	layer	NOUN
fcis-30522	58	22	.	.	PUNCT
fcis-30522	59	1	the	the	DET
fcis-30522	59	2	final	final	ADJ
fcis-30522	59	3	model	model	NOUN
fcis-30522	59	4	structure	structure	NOUN
fcis-30522	59	5	is	be	AUX
fcis-30522	59	6	shown	show	VERB
fcis-30522	59	7	in	in	ADP
fcis-30522	59	8	(	(	PUNCT
fcis-30522	59	9	fig	fig	NOUN
fcis-30522	59	10	1	1	NUM
fcis-30522	59	11	)	)	PUNCT
fcis-30522	59	12	fig	fig	NOUN
fcis-30522	59	13	1	1	NUM
fcis-30522	59	14	.	.	PUNCT
fcis-30522	60	1	mbcresattnet	mbcresattnet	NOUN
fcis-30522	60	2	model	model	NOUN
fcis-30522	60	3	structure	structure	NOUN
fcis-30522	60	4	diagram	diagram	PROPN
fcis-30522	60	5	2.2	2.2	NUM
fcis-30522	60	6	.	.	PUNCT
fcis-30522	61	1	data	datum	NOUN
fcis-30522	61	2	enhancement	enhancement	NOUN
fcis-30522	61	3	technology	technology	NOUN
fcis-30522	61	4	2.2.1	2.2.1	NUM
fcis-30522	61	5	.	.	PUNCT
fcis-30522	62	1	eda	eda	PROPN
fcis-30522	62	2	data	data	PROPN
fcis-30522	62	3	enhancement	enhancement	NOUN
fcis-30522	62	4	eda	eda	PROPN
fcis-30522	62	5	(	(	PUNCT
fcis-30522	62	6	easy	easy	ADJ
fcis-30522	62	7	data	datum	NOUN
fcis-30522	62	8	augmentation	augmentation	NOUN
fcis-30522	62	9	)	)	PUNCT
fcis-30522	62	10	,	,	PUNCT
fcis-30522	62	11	proposed	propose	VERB
fcis-30522	62	12	by	by	ADP
fcis-30522	62	13	jason	jason	PROPN
fcis-30522	62	14	wei	wei	PROPN
fcis-30522	62	15	et	et	PROPN
fcis-30522	62	16	al	al	PROPN
fcis-30522	62	17	.	.	PROPN
fcis-30522	62	18	,	,	PUNCT
fcis-30522	62	19	expands	expand	VERB
fcis-30522	62	20	datasets	dataset	NOUN
fcis-30522	62	21	and	and	CCONJ
fcis-30522	62	22	improves	improve	VERB
fcis-30522	62	23	model	model	NOUN
fcis-30522	62	24	generalization	generalization	NOUN
fcis-30522	62	25	when	when	SCONJ
fcis-30522	62	26	data	datum	NOUN
fcis-30522	62	27	is	be	AUX
fcis-30522	62	28	scarce	scarce	ADJ
fcis-30522	62	29	.	.	PUNCT
fcis-30522	63	1	by	by	ADP
fcis-30522	63	2	manipulating	manipulate	VERB
fcis-30522	63	3	text	text	NOUN
fcis-30522	63	4	data	datum	NOUN
fcis-30522	63	5	through	through	ADP
fcis-30522	63	6	four	four	NUM
fcis-30522	63	7	basic	basic	ADJ
fcis-30522	63	8	operations	operation	NOUN
fcis-30522	63	9	—	—	PUNCT
fcis-30522	63	10	synonym	synonym	NOUN
fcis-30522	63	11	replacement	replacement	NOUN
fcis-30522	63	12	,	,	PUNCT
fcis-30522	63	13	random	random	ADJ
fcis-30522	63	14	insertion	insertion	NOUN
fcis-30522	63	15	,	,	PUNCT
fcis-30522	63	16	random	random	ADJ
fcis-30522	63	17	swapping	swapping	NOUN
fcis-30522	63	18	,	,	PUNCT
fcis-30522	63	19	and	and	CCONJ
fcis-30522	63	20	random	random	ADJ
fcis-30522	63	21	deletion	deletion	NOUN
fcis-30522	63	22	—	—	PUNCT
fcis-30522	63	23	eda	eda	PROPN
fcis-30522	63	24	creates	create	VERB
fcis-30522	63	25	diverse	diverse	ADJ
fcis-30522	63	26	training	training	NOUN
fcis-30522	63	27	samples	sample	NOUN
fcis-30522	63	28	that	that	PRON
fcis-30522	63	29	help	help	VERB
fcis-30522	63	30	models	model	NOUN
fcis-30522	63	31	learn	learn	VERB
fcis-30522	63	32	varied	varied	ADJ
fcis-30522	63	33	expressions	expression	NOUN
fcis-30522	63	34	and	and	CCONJ
fcis-30522	63	35	improve	improve	VERB
fcis-30522	63	36	robustness	robustness	NOUN
fcis-30522	63	37	.	.	PUNCT
fcis-30522	64	1	this	this	DET
fcis-30522	64	2	technique	technique	NOUN
fcis-30522	64	3	is	be	AUX
fcis-30522	64	4	ideal	ideal	ADJ
fcis-30522	64	5	for	for	ADP
fcis-30522	64	6	limited	limited	ADJ
fcis-30522	64	7	-	-	PUNCT
fcis-30522	64	8	data	datum	NOUN
fcis-30522	64	9	tasks	task	NOUN
fcis-30522	64	10	like	like	ADP
fcis-30522	64	11	text	text	NOUN
fcis-30522	64	12	classification	classification	NOUN
fcis-30522	64	13	and	and	CCONJ
fcis-30522	64	14	sentiment	sentiment	NOUN
fcis-30522	64	15	analysis	analysis	NOUN
fcis-30522	64	16	,	,	PUNCT
fcis-30522	64	17	where	where	SCONJ
fcis-30522	64	18	it	it	PRON
fcis-30522	64	19	reduces	reduce	VERB
fcis-30522	64	20	overfitting	overfitte	VERB
fcis-30522	64	21	and	and	CCONJ
fcis-30522	64	22	enhances	enhance	VERB
fcis-30522	64	23	generalization	generalization	NOUN
fcis-30522	64	24	.	.	PUNCT
fcis-30522	65	1	its	its	PRON
fcis-30522	65	2	advantage	advantage	NOUN
fcis-30522	65	3	lies	lie	VERB
fcis-30522	65	4	in	in	ADP
fcis-30522	65	5	improving	improve	VERB
fcis-30522	65	6	linguistic	linguistic	ADJ
fcis-30522	65	7	understanding	understanding	NOUN
fcis-30522	65	8	without	without	ADP
fcis-30522	65	9	complex	complex	ADJ
fcis-30522	65	10	external	external	ADJ
fcis-30522	65	11	resources	resource	NOUN
fcis-30522	65	12	.	.	PUNCT
fcis-30522	66	1	however	however	ADV
fcis-30522	66	2	,	,	PUNCT
fcis-30522	66	3	operations	operation	NOUN
fcis-30522	66	4	like	like	ADP
fcis-30522	66	5	synonym	synonym	NOUN
fcis-30522	66	6	replacement	replacement	NOUN
fcis-30522	66	7	may	may	AUX
fcis-30522	66	8	alter	alter	VERB
fcis-30522	66	9	original	original	ADJ
fcis-30522	66	10	semantics	semantic	NOUN
fcis-30522	66	11	,	,	PUNCT
fcis-30522	66	12	potentially	potentially	ADV
fcis-30522	66	13	harming	harm	VERB
fcis-30522	66	14	performance	performance	NOUN
fcis-30522	66	15	in	in	ADP
fcis-30522	66	16	semantically	semantically	ADV
fcis-30522	66	17	sensitive	sensitive	ADJ
fcis-30522	66	18	tasks	task	NOUN
fcis-30522	66	19	.	.	PUNCT
fcis-30522	67	1	2.2.2	2.2.2	X
fcis-30522	67	2	.	.	PUNCT
fcis-30522	67	3	aeda	aeda	ADJ
fcis-30522	67	4	data	datum	NOUN
fcis-30522	67	5	enhancement	enhancement	NOUN
fcis-30522	67	6	aeda	aeda	NOUN
fcis-30522	67	7	is	be	AUX
fcis-30522	67	8	a	a	DET
fcis-30522	67	9	simpler	simple	ADJ
fcis-30522	67	10	text	text	NOUN
fcis-30522	67	11	data	datum	NOUN
fcis-30522	67	12	augmentation	augmentation	NOUN
fcis-30522	67	13	method	method	NOUN
fcis-30522	67	14	compared	compare	VERB
fcis-30522	67	15	to	to	ADP
fcis-30522	67	16	eda	eda	PROPN
fcis-30522	67	17	.	.	PUNCT
fcis-30522	68	1	the	the	DET
fcis-30522	68	2	paper	paper	NOUN
fcis-30522	68	3	introducing	introduce	VERB
fcis-30522	68	4	aeda	aeda	NOUN
fcis-30522	68	5	was	be	AUX
fcis-30522	68	6	published	publish	VERB
fcis-30522	68	7	by	by	ADP
fcis-30522	68	8	the	the	DET
fcis-30522	68	9	imp	imp	NOUN
fcis-30522	68	10	lab	lab	NOUN
fcis-30522	68	11	of	of	ADP
fcis-30522	68	12	the	the	DET
fcis-30522	68	13	university	university	NOUN
fcis-30522	68	14	of	of	ADP
fcis-30522	68	15	parma	parma	PROPN
fcis-30522	68	16	,	,	PUNCT
fcis-30522	68	17	italy	italy	PROPN
fcis-30522	68	18	,	,	PUNCT
fcis-30522	68	19	at	at	ADP
fcis-30522	68	20	the	the	DET
fcis-30522	68	21	emnlp	emnlp	ADJ
fcis-30522	68	22	2021	2021	NUM
fcis-30522	68	23	conference	conference	NOUN
fcis-30522	68	24	.	.	PUNCT
fcis-30522	69	1	aeda	aeda	PROPN
fcis-30522	69	2	was	be	AUX
fcis-30522	69	3	inspired	inspire	VERB
fcis-30522	69	4	by	by	ADP
fcis-30522	69	5	eda	eda	PROPN
fcis-30522	69	6	,	,	PUNCT
fcis-30522	69	7	which	which	PRON
fcis-30522	69	8	proposes	propose	VERB
fcis-30522	69	9	four	four	NUM
fcis-30522	69	10	simple	simple	ADJ
fcis-30522	69	11	data	datum	NOUN
fcis-30522	69	12	augmentation	augmentation	NOUN
fcis-30522	69	13	operations	operation	NOUN
fcis-30522	69	14	:	:	PUNCT
fcis-30522	69	15	synonym	synonym	NOUN
fcis-30522	69	16	replacement	replacement	NOUN
fcis-30522	69	17	(	(	PUNCT
fcis-30522	69	18	replacing	replace	VERB
fcis-30522	69	19	words	word	NOUN
fcis-30522	69	20	in	in	ADP
fcis-30522	69	21	a	a	DET
fcis-30522	69	22	sentence	sentence	NOUN
fcis-30522	69	23	with	with	ADP
fcis-30522	69	24	synonyms	synonym	NOUN
fcis-30522	69	25	from	from	ADP
fcis-30522	69	26	a	a	DET
fcis-30522	69	27	synonym	synonym	NOUN
fcis-30522	69	28	list	list	NOUN
fcis-30522	69	29	)	)	PUNCT
fcis-30522	69	30	,	,	PUNCT
fcis-30522	69	31	random	random	ADJ
fcis-30522	69	32	swap	swap	NOUN
fcis-30522	69	33	(	(	PUNCT
fcis-30522	69	34	randomly	randomly	ADV
fcis-30522	69	35	swapping	swap	VERB
fcis-30522	69	36	two	two	NUM
fcis-30522	69	37	words	word	NOUN
fcis-30522	69	38	in	in	ADP
fcis-30522	69	39	a	a	DET
fcis-30522	69	40	sentence	sentence	NOUN
fcis-30522	69	41	to	to	PART
fcis-30522	69	42	change	change	VERB
fcis-30522	69	43	the	the	DET
fcis-30522	69	44	order	order	NOUN
fcis-30522	69	45	of	of	ADP
fcis-30522	69	46	words	word	NOUN
fcis-30522	69	47	)	)	PUNCT
fcis-30522	69	48	,	,	PUNCT
fcis-30522	69	49	random	random	ADJ
fcis-30522	69	50	insertion	insertion	NOUN
fcis-30522	69	51	(	(	PUNCT
fcis-30522	69	52	randomly	randomly	ADV
fcis-30522	69	53	inserting	insert	VERB
fcis-30522	69	54	a	a	DET
fcis-30522	69	55	synonym	synonym	NOUN
fcis-30522	69	56	for	for	ADP
fcis-30522	69	57	a	a	DET
fcis-30522	69	58	word	word	NOUN
fcis-30522	69	59	in	in	ADP
fcis-30522	69	60	the	the	DET
fcis-30522	69	61	sentence	sentence	NOUN
fcis-30522	69	62	)	)	PUNCT
fcis-30522	69	63	,	,	PUNCT
fcis-30522	69	64	and	and	CCONJ
fcis-30522	69	65	random	random	ADJ
fcis-30522	69	66	deletion	deletion	NOUN
fcis-30522	69	67	(	(	PUNCT
fcis-30522	69	68	randomly	randomly	ADV
fcis-30522	69	69	deleting	delete	VERB
fcis-30522	69	70	words	word	NOUN
fcis-30522	69	71	in	in	ADP
fcis-30522	69	72	the	the	DET
fcis-30522	69	73	sentence	sentence	NOUN
fcis-30522	69	74	)	)	PUNCT
fcis-30522	69	75	.	.	PUNCT
fcis-30522	70	1	however	however	ADV
fcis-30522	70	2	,	,	PUNCT
fcis-30522	70	3	eda	eda	PROPN
fcis-30522	70	4	sometimes	sometimes	ADV
fcis-30522	70	5	yields	yield	VERB
fcis-30522	70	6	poor	poor	ADJ
fcis-30522	70	7	results	result	NOUN
fcis-30522	70	8	when	when	SCONJ
fcis-30522	70	9	using	use	VERB
fcis-30522	70	10	a	a	DET
fcis-30522	70	11	pre	pre	ADJ
fcis-30522	70	12	-	-	ADJ
fcis-30522	70	13	trained	train	VERB
fcis-30522	70	14	language	language	NOUN
fcis-30522	70	15	model	model	NOUN
fcis-30522	70	16	72	72	NUM
fcis-30522	70	17	and	and	CCONJ
fcis-30522	70	18	may	may	AUX
fcis-30522	70	19	even	even	ADV
fcis-30522	70	20	worsen	worsen	VERB
fcis-30522	70	21	the	the	DET
fcis-30522	70	22	outcomes	outcome	NOUN
fcis-30522	70	23	.	.	PUNCT
fcis-30522	71	1	this	this	PRON
fcis-30522	71	2	may	may	AUX
fcis-30522	71	3	be	be	AUX
fcis-30522	71	4	because	because	SCONJ
fcis-30522	71	5	operations	operation	NOUN
fcis-30522	71	6	like	like	ADP
fcis-30522	71	7	synonym	synonym	NOUN
fcis-30522	71	8	replacement	replacement	NOUN
fcis-30522	71	9	,	,	PUNCT
fcis-30522	71	10	random	random	ADJ
fcis-30522	71	11	swap	swap	NOUN
fcis-30522	71	12	,	,	PUNCT
fcis-30522	71	13	random	random	ADJ
fcis-30522	71	14	insertion	insertion	NOUN
fcis-30522	71	15	,	,	PUNCT
fcis-30522	71	16	and	and	CCONJ
fcis-30522	71	17	random	random	ADJ
fcis-30522	71	18	deletion	deletion	NOUN
fcis-30522	71	19	alter	alter	VERB
fcis-30522	71	20	the	the	DET
fcis-30522	71	21	sequence	sequence	NOUN
fcis-30522	71	22	information	information	NOUN
fcis-30522	71	23	of	of	ADP
fcis-30522	71	24	the	the	DET
fcis-30522	71	25	original	original	ADJ
fcis-30522	71	26	text	text	NOUN
fcis-30522	71	27	.	.	PUNCT
fcis-30522	72	1	in	in	ADP
fcis-30522	72	2	contrast	contrast	NOUN
fcis-30522	72	3	,	,	PUNCT
fcis-30522	72	4	the	the	DET
fcis-30522	72	5	aeda	aeda	ADJ
fcis-30522	72	6	method	method	NOUN
fcis-30522	72	7	only	only	ADV
fcis-30522	72	8	inserts	insert	VERB
fcis-30522	72	9	punctuation	punctuation	NOUN
fcis-30522	72	10	marks	mark	NOUN
fcis-30522	72	11	,	,	PUNCT
fcis-30522	72	12	resulting	result	VERB
fcis-30522	72	13	in	in	ADP
fcis-30522	72	14	minimal	minimal	ADJ
fcis-30522	72	15	modification	modification	NOUN
fcis-30522	72	16	to	to	ADP
fcis-30522	72	17	the	the	DET
fcis-30522	72	18	sequence	sequence	NOUN
fcis-30522	72	19	information	information	NOUN
fcis-30522	72	20	of	of	ADP
fcis-30522	72	21	the	the	DET
fcis-30522	72	22	original	original	ADJ
fcis-30522	72	23	data	datum	NOUN
fcis-30522	72	24	.	.	PUNCT
fcis-30522	73	1	modifying	modify	VERB
fcis-30522	73	2	words	word	NOUN
fcis-30522	73	3	may	may	AUX
fcis-30522	73	4	lead	lead	VERB
fcis-30522	73	5	to	to	ADP
fcis-30522	73	6	semantic	semantic	ADJ
fcis-30522	73	7	changes	change	NOUN
fcis-30522	73	8	and	and	CCONJ
fcis-30522	73	9	thus	thus	ADV
fcis-30522	73	10	negative	negative	ADJ
fcis-30522	73	11	effects	effect	NOUN
fcis-30522	73	12	.	.	PUNCT
fcis-30522	74	1	aeda	aeda	ADJ
fcis-30522	74	2	,	,	PUNCT
fcis-30522	74	3	on	on	ADP
fcis-30522	74	4	the	the	DET
fcis-30522	74	5	other	other	ADJ
fcis-30522	74	6	hand	hand	NOUN
fcis-30522	74	7	,	,	PUNCT
fcis-30522	74	8	only	only	ADV
fcis-30522	74	9	adds	add	VERB
fcis-30522	74	10	punctuation	punctuation	NOUN
fcis-30522	74	11	marks	mark	NOUN
fcis-30522	74	12	.	.	PUNCT
fcis-30522	75	1	although	although	SCONJ
fcis-30522	75	2	it	it	PRON
fcis-30522	75	3	introduces	introduce	VERB
fcis-30522	75	4	noise	noise	NOUN
fcis-30522	75	5	,	,	PUNCT
fcis-30522	75	6	it	it	PRON
fcis-30522	75	7	does	do	AUX
fcis-30522	75	8	not	not	PART
fcis-30522	75	9	change	change	VERB
fcis-30522	75	10	the	the	DET
fcis-30522	75	11	order	order	NOUN
fcis-30522	75	12	of	of	ADP
fcis-30522	75	13	the	the	DET
fcis-30522	75	14	original	original	ADJ
fcis-30522	75	15	text	text	NOUN
fcis-30522	75	16	,	,	PUNCT
fcis-30522	75	17	and	and	CCONJ
fcis-30522	75	18	it	it	PRON
fcis-30522	75	19	does	do	AUX
fcis-30522	75	20	not	not	PART
fcis-30522	75	21	insert	insert	VERB
fcis-30522	75	22	too	too	ADV
fcis-30522	75	23	many	many	ADJ
fcis-30522	75	24	punctuation	punctuation	NOUN
fcis-30522	75	25	marks	mark	NOUN
fcis-30522	75	26	,	,	PUNCT
fcis-30522	75	27	so	so	CCONJ
fcis-30522	75	28	the	the	DET
fcis-30522	75	29	added	add	VERB
fcis-30522	75	30	noise	noise	NOUN
fcis-30522	75	31	is	be	AUX
fcis-30522	75	32	minimal	minimal	ADJ
fcis-30522	75	33	.	.	PUNCT
fcis-30522	76	1	the	the	DET
fcis-30522	76	2	specific	specific	ADJ
fcis-30522	76	3	method	method	NOUN
fcis-30522	76	4	of	of	ADP
fcis-30522	76	5	aeda	aeda	ADJ
fcis-30522	76	6	data	datum	NOUN
fcis-30522	76	7	enhancement	enhancement	NOUN
fcis-30522	76	8	is	be	AUX
fcis-30522	76	9	as	as	SCONJ
fcis-30522	76	10	follows	follow	VERB
fcis-30522	76	11	:	:	PUNCT
fcis-30522	76	12	in	in	ADP
fcis-30522	76	13	sentence	sentence	NOUN
fcis-30522	76	14	length	length	NOUN
fcis-30522	76	15	1	1	NUM
fcis-30522	76	16	,	,	PUNCT
fcis-30522	76	17	a	a	DET
fcis-30522	76	18	number	number	NOUN
fcis-30522	76	19	is	be	AUX
fcis-30522	76	20	randomly	randomly	ADV
fcis-30522	76	21	selected	select	VERB
fcis-30522	76	22	as	as	ADP
fcis-30522	76	23	the	the	DET
fcis-30522	76	24	number	number	NOUN
fcis-30522	76	25	of	of	ADP
fcis-30522	76	26	punctuation	punctuation	NOUN
fcis-30522	76	27	marks	mark	NOUN
fcis-30522	76	28	to	to	PART
fcis-30522	76	29	be	be	AUX
fcis-30522	76	30	inserted	insert	VERB
fcis-30522	76	31	,	,	PUNCT
fcis-30522	76	32	and	and	CCONJ
fcis-30522	76	33	the	the	DET
fcis-30522	76	34	positions	position	NOUN
fcis-30522	76	35	of	of	ADP
fcis-30522	76	36	the	the	DET
fcis-30522	76	37	punctuation	punctuation	NOUN
fcis-30522	76	38	marks	mark	NOUN
fcis-30522	76	39	to	to	PART
fcis-30522	76	40	be	be	AUX
fcis-30522	76	41	inserted	insert	VERB
fcis-30522	76	42	in	in	ADP
fcis-30522	76	43	the	the	DET
fcis-30522	76	44	sentence	sentence	NOUN
fcis-30522	76	45	are	be	AUX
fcis-30522	76	46	randomly	randomly	ADV
fcis-30522	76	47	inserted	insert	VERB
fcis-30522	76	48	.	.	PUNCT
fcis-30522	77	1	there	there	PRON
fcis-30522	77	2	are	be	VERB
fcis-30522	77	3	mainly	mainly	ADV
fcis-30522	77	4	six	six	NUM
fcis-30522	77	5	types	type	NOUN
fcis-30522	77	6	of	of	ADP
fcis-30522	77	7	punctuation	punctuation	NOUN
fcis-30522	77	8	marks	mark	NOUN
fcis-30522	77	9	to	to	PART
fcis-30522	77	10	be	be	AUX
fcis-30522	77	11	inserted	insert	VERB
fcis-30522	77	12	:	:	PUNCT
fcis-30522	77	13	[	[	X
fcis-30522	77	14	“	"	PUNCT
fcis-30522	77	15	.	.	PUNCT
fcis-30522	77	16	”	"	PUNCT
fcis-30522	77	17	,	,	PUNCT
fcis-30522	77	18	“	"	PUNCT
fcis-30522	77	19	;	;	PUNCT
fcis-30522	77	20	”	"	PUNCT
fcis-30522	77	21	,	,	PUNCT
fcis-30522	77	22	“	"	PUNCT
fcis-30522	77	23	?	?	PUNCT
fcis-30522	77	24	”	"	PUNCT
fcis-30522	77	25	,	,	PUNCT
fcis-30522	77	26	“	"	PUNCT
fcis-30522	77	27	:	:	PUNCT
fcis-30522	77	28	”	"	PUNCT
fcis-30522	77	29	,	,	PUNCT
fcis-30522	77	30	“	"	PUNCT
fcis-30522	77	31	!	!	PUNCT
fcis-30522	77	32	”	"	PUNCT
fcis-30522	77	33	,	,	PUNCT
fcis-30522	77	34	“	"	PUNCT
fcis-30522	77	35	,	,	PUNCT
fcis-30522	77	36	”	"	PUNCT
fcis-30522	77	37	]	]	X
fcis-30522	77	38	2.3	2.3	NUM
fcis-30522	77	39	.	.	PUNCT
fcis-30522	78	1	experimental	experimental	ADJ
fcis-30522	78	2	data	datum	NOUN
fcis-30522	78	3	the	the	DET
fcis-30522	78	4	data	datum	NOUN
fcis-30522	78	5	used	use	VERB
fcis-30522	78	6	in	in	ADP
fcis-30522	78	7	this	this	DET
fcis-30522	78	8	paper	paper	NOUN
fcis-30522	78	9	comprises	comprise	VERB
fcis-30522	78	10	the	the	DET
fcis-30522	78	11	abstract	abstract	ADJ
fcis-30522	78	12	texts	text	NOUN
fcis-30522	78	13	of	of	ADP
fcis-30522	78	14	collected	collect	VERB
fcis-30522	78	15	literature	literature	NOUN
fcis-30522	78	16	.	.	PUNCT
fcis-30522	79	1	the	the	DET
fcis-30522	79	2	sample	sample	NOUN
fcis-30522	79	3	content	content	NOUN
fcis-30522	79	4	has	have	AUX
fcis-30522	79	5	been	be	AUX
fcis-30522	79	6	manually	manually	ADV
fcis-30522	79	7	cleaned	clean	VERB
fcis-30522	79	8	,	,	PUNCT
fcis-30522	79	9	resulting	result	VERB
fcis-30522	79	10	in	in	ADP
fcis-30522	79	11	2,532	2,532	NUM
fcis-30522	79	12	samples	sample	NOUN
fcis-30522	79	13	.	.	PUNCT
fcis-30522	80	1	there	there	PRON
fcis-30522	80	2	are	be	VERB
fcis-30522	80	3	ten	ten	NUM
fcis-30522	80	4	categories	category	NOUN
fcis-30522	80	5	in	in	ADP
fcis-30522	80	6	total	total	ADJ
fcis-30522	80	7	:	:	PUNCT
fcis-30522	80	8	law	law	NOUN
fcis-30522	80	9	,	,	PUNCT
fcis-30522	80	10	engineering	engineering	NOUN
fcis-30522	80	11	,	,	PUNCT
fcis-30522	80	12	education	education	NOUN
fcis-30522	80	13	,	,	PUNCT
fcis-30522	80	14	management	management	NOUN
fcis-30522	80	15	,	,	PUNCT
fcis-30522	80	16	economics	economic	NOUN
fcis-30522	80	17	,	,	PUNCT
fcis-30522	80	18	science	science	NOUN
fcis-30522	80	19	,	,	PUNCT
fcis-30522	80	20	history	history	NOUN
fcis-30522	80	21	,	,	PUNCT
fcis-30522	80	22	agriculture	agriculture	NOUN
fcis-30522	80	23	,	,	PUNCT
fcis-30522	80	24	literature	literature	NOUN
fcis-30522	80	25	,	,	PUNCT
fcis-30522	80	26	and	and	CCONJ
fcis-30522	80	27	medicine	medicine	NOUN
fcis-30522	80	28	.	.	PUNCT
fcis-30522	81	1	table	table	NOUN
fcis-30522	81	2	1	1	NUM
fcis-30522	81	3	presents	present	VERB
fcis-30522	81	4	the	the	DET
fcis-30522	81	5	category	category	NOUN
fcis-30522	81	6	and	and	CCONJ
fcis-30522	81	7	label	label	NOUN
fcis-30522	81	8	information	information	NOUN
fcis-30522	81	9	of	of	ADP
fcis-30522	81	10	the	the	DET
fcis-30522	81	11	original	original	ADJ
fcis-30522	81	12	samples	sample	NOUN
fcis-30522	81	13	.	.	PUNCT
fcis-30522	82	1	table	table	NOUN
fcis-30522	82	2	2	2	NUM
fcis-30522	82	3	provides	provide	VERB
fcis-30522	82	4	an	an	DET
fcis-30522	82	5	example	example	NOUN
fcis-30522	82	6	of	of	ADP
fcis-30522	82	7	a	a	DET
fcis-30522	82	8	sample	sample	NOUN
fcis-30522	82	9	in	in	ADP
fcis-30522	82	10	the	the	DET
fcis-30522	82	11	eda	eda	PROPN
fcis-30522	82	12	data	data	PROPN
fcis-30522	82	13	augmentation	augmentation	NOUN
fcis-30522	82	14	section	section	NOUN
fcis-30522	82	15	,	,	PUNCT
fcis-30522	82	16	and	and	CCONJ
fcis-30522	82	17	table	table	NOUN
fcis-30522	82	18	3	3	NUM
fcis-30522	82	19	presents	present	VERB
fcis-30522	82	20	an	an	DET
fcis-30522	82	21	example	example	NOUN
fcis-30522	82	22	of	of	ADP
fcis-30522	82	23	a	a	DET
fcis-30522	82	24	sample	sample	NOUN
fcis-30522	82	25	in	in	ADP
fcis-30522	82	26	the	the	DET
fcis-30522	82	27	aeda	aeda	ADJ
fcis-30522	82	28	data	data	NOUN
fcis-30522	82	29	augmentation	augmentation	NOUN
fcis-30522	82	30	section	section	NOUN
fcis-30522	82	31	.	.	PUNCT
fcis-30522	83	1	table	table	NOUN
fcis-30522	83	2	1	1	NUM
fcis-30522	83	3	.	.	PUNCT
fcis-30522	83	4	sample	sample	NOUN
fcis-30522	83	5	label	label	NOUN
fcis-30522	83	6	names	name	NOUN
fcis-30522	83	7	and	and	CCONJ
fcis-30522	83	8	distribution	distribution	NOUN
fcis-30522	83	9	category	category	NOUN
fcis-30522	83	10	name	name	NOUN
fcis-30522	83	11	sample	sample	NOUN
fcis-30522	83	12	size	size	NOUN
fcis-30522	83	13	label	label	NOUN
fcis-30522	83	14	law	law	NOUN
fcis-30522	83	15	230	230	NUM
fcis-30522	83	16	0	0	NUM
fcis-30522	83	17	engineering	engineer	VERB
fcis-30522	83	18	267	267	NUM
fcis-30522	83	19	1	1	NUM
fcis-30522	83	20	management	management	NOUN
fcis-30522	83	21	248	248	NUM
fcis-30522	83	22	2	2	NUM
fcis-30522	83	23	pedagogical	pedagogical	ADJ
fcis-30522	83	24	226	226	NUM
fcis-30522	83	25	3	3	NUM
fcis-30522	83	26	economics	economic	NOUN
fcis-30522	83	27	245	245	NUM
fcis-30522	83	28	4	4	NUM
fcis-30522	83	29	science	science	NOUN
fcis-30522	83	30	261	261	NUM
fcis-30522	83	31	5	5	NUM
fcis-30522	83	32	history	history	NOUN
fcis-30522	83	33	295	295	NUM
fcis-30522	83	34	6	6	NUM
fcis-30522	83	35	agronomy	agronomy	NOUN
fcis-30522	83	36	266	266	NUM
fcis-30522	83	37	7	7	NUM
fcis-30522	83	38	literary	literary	ADJ
fcis-30522	83	39	267	267	NUM
fcis-30522	83	40	8	8	NUM
fcis-30522	83	41	medical	medical	ADJ
fcis-30522	83	42	science	science	NOUN
fcis-30522	83	43	227	227	NUM
fcis-30522	83	44	9	9	NUM
fcis-30522	83	45	2.4	2.4	NUM
fcis-30522	83	46	.	.	PUNCT
fcis-30522	84	1	experimental	experimental	ADJ
fcis-30522	84	2	environment	environment	NOUN
fcis-30522	84	3	and	and	CCONJ
fcis-30522	84	4	model	model	NOUN
fcis-30522	84	5	parameters	parameter	NOUN
fcis-30522	84	6	the	the	DET
fcis-30522	84	7	environmental	environmental	ADJ
fcis-30522	84	8	configuration	configuration	NOUN
fcis-30522	84	9	is	be	AUX
fcis-30522	84	10	shown	show	VERB
fcis-30522	84	11	in	in	ADP
fcis-30522	84	12	table	table	NOUN
fcis-30522	84	13	2	2	NUM
fcis-30522	84	14	.	.	PUNCT
fcis-30522	84	15	table	table	NOUN
fcis-30522	84	16	2	2	NUM
fcis-30522	84	17	.	.	PUNCT
fcis-30522	84	18	experimental	experimental	ADJ
fcis-30522	84	19	environment	environment	NOUN
fcis-30522	84	20	configuration	configuration	NOUN
fcis-30522	84	21	related	relate	VERB
fcis-30522	84	22	environment	environment	NOUN
fcis-30522	84	23	specific	specific	ADJ
fcis-30522	84	24	version	version	NOUN
fcis-30522	84	25	operating	operating	NOUN
fcis-30522	84	26	system	system	NOUN
fcis-30522	84	27	windows10	windows10	PROPN
fcis-30522	84	28	gpu	gpu	PROPN
fcis-30522	84	29	geforce	geforce	NOUN
fcis-30522	84	30	rtx4090(24	rtx4090(24	VERB
fcis-30522	84	31	gb	gb	PRON
fcis-30522	84	32	)	)	PUNCT
fcis-30522	84	33	cpu	cpu	NOUN
fcis-30522	84	34	16vcpu	16vcpu	NUM
fcis-30522	84	35	intel(r	intel(r	NOUN
fcis-30522	84	36	)	)	PUNCT
fcis-30522	84	37	xeon(r	xeon(r	PROPN
fcis-30522	84	38	)	)	PUNCT
fcis-30522	84	39	platinum	platinum	NOUN
fcis-30522	84	40	8352v	8352v	NUM
fcis-30522	84	41	cpu@2.10ghz	cpu@2.10ghz	PROPN
fcis-30522	84	42	development	development	NOUN
fcis-30522	84	43	language	language	NOUN
fcis-30522	84	44	python3.10.8	python3.10.8	NOUN
fcis-30522	84	45	development	development	NOUN
fcis-30522	84	46	framework	framework	NOUN
fcis-30522	84	47	pytorch2.1.0	pytorch2.1.0	NOUN
fcis-30522	84	48	development	development	NOUN
fcis-30522	84	49	tool	tool	NOUN
fcis-30522	84	50	pycharm	pycharm	VERB
fcis-30522	84	51	professional	professional	PROPN
fcis-30522	84	52	edition	edition	NOUN
fcis-30522	84	53	the	the	DET
fcis-30522	84	54	relevant	relevant	ADJ
fcis-30522	84	55	parameters	parameter	NOUN
fcis-30522	84	56	of	of	ADP
fcis-30522	84	57	the	the	DET
fcis-30522	84	58	deep	deep	ADJ
fcis-30522	84	59	learning	learning	NOUN
fcis-30522	84	60	model	model	NOUN
fcis-30522	84	61	are	be	AUX
fcis-30522	84	62	shown	show	VERB
fcis-30522	84	63	in	in	ADP
fcis-30522	84	64	table	table	NOUN
fcis-30522	84	65	3	3	NUM
fcis-30522	84	66	.	.	PUNCT
fcis-30522	84	67	table	table	NOUN
fcis-30522	84	68	3	3	NUM
fcis-30522	84	69	.	.	PUNCT
fcis-30522	84	70	related	relate	VERB
fcis-30522	84	71	parameters	parameter	NOUN
fcis-30522	84	72	parameter	parameter	NOUN
fcis-30522	84	73	name	name	NOUN
fcis-30522	84	74	specific	specific	ADJ
fcis-30522	84	75	meaning	meaning	NOUN
fcis-30522	84	76	value	value	NOUN
fcis-30522	84	77	epoch	epoch	NOUN
fcis-30522	84	78	number	number	NOUN
fcis-30522	84	79	of	of	ADP
fcis-30522	84	80	training	training	NOUN
fcis-30522	84	81	sessions	session	NOUN
fcis-30522	84	82	35	35	NUM
fcis-30522	84	83	learning_rate	learning_rate	ADJ
fcis-30522	84	84	learning	learning	NOUN
fcis-30522	84	85	rate	rate	NOUN
fcis-30522	85	1	1e-5	1e-5	NOUN
fcis-30522	85	2	train_batch_size	train_batch_size	NOUN
fcis-30522	85	3	number	number	NOUN
fcis-30522	85	4	of	of	ADP
fcis-30522	85	5	samples	sample	NOUN
fcis-30522	85	6	included	include	VERB
fcis-30522	85	7	in	in	ADP
fcis-30522	85	8	each	each	DET
fcis-30522	85	9	batch	batch	NOUN
fcis-30522	85	10	during	during	ADP
fcis-30522	85	11	training	train	VERB
fcis-30522	85	12	28	28	NUM
fcis-30522	85	13	test_batch_size	test_batch_size	NOUN
fcis-30522	85	14	number	number	NOUN
fcis-30522	85	15	of	of	ADP
fcis-30522	85	16	samples	sample	NOUN
fcis-30522	85	17	per	per	ADP
fcis-30522	85	18	batch	batch	NOUN
fcis-30522	85	19	included	include	VERB
fcis-30522	85	20	in	in	ADP
fcis-30522	85	21	the	the	DET
fcis-30522	85	22	model	model	NOUN
fcis-30522	85	23	evaluation	evaluation	NOUN
fcis-30522	85	24	or	or	CCONJ
fcis-30522	85	25	testing	test	VERB
fcis-30522	85	26	16	16	NUM
fcis-30522	85	27	filter_size	filter_size	NOUN
fcis-30522	85	28	convolutional	convolutional	ADJ
fcis-30522	85	29	kernel	kernel	NOUN
fcis-30522	85	30	size	size	NOUN
fcis-30522	85	31	[	[	X
fcis-30522	85	32	2,3,4	2,3,4	NUM
fcis-30522	85	33	]	]	SYM
fcis-30522	85	34	weight_decay	weight_decay	NOUN
fcis-30522	85	35	weight	weight	NOUN
fcis-30522	85	36	attenuation	attenuation	NOUN
fcis-30522	85	37	0.015	0.015	NUM
fcis-30522	85	38	drop_out	drop_out	NOUN
fcis-30522	85	39	discard	discard	ADJ
fcis-30522	85	40	rate	rate	NOUN
fcis-30522	85	41	0.5	0.5	NUM
fcis-30522	85	42	max_length	max_length	NOUN
fcis-30522	85	43	maximum	maximum	ADJ
fcis-30522	85	44	truncation	truncation	NOUN
fcis-30522	85	45	length	length	NOUN
fcis-30522	85	46	of	of	ADP
fcis-30522	85	47	the	the	DET
fcis-30522	85	48	model	model	NOUN
fcis-30522	85	49	(	(	PUNCT
fcis-30522	85	50	see	see	VERB
fcis-30522	85	51	the	the	DET
fcis-30522	85	52	following	follow	VERB
fcis-30522	85	53	paragraph	paragraph	NOUN
fcis-30522	85	54	for	for	ADP
fcis-30522	85	55	analysis	analysis	NOUN
fcis-30522	85	56	)	)	PUNCT
fcis-30522	85	57	235	235	NUM
fcis-30522	85	58	2.5	2.5	NUM
fcis-30522	85	59	.	.	PUNCT
fcis-30522	86	1	selection	selection	NOUN
fcis-30522	86	2	of	of	ADP
fcis-30522	86	3	evaluation	evaluation	NOUN
fcis-30522	86	4	indicators	indicator	NOUN
fcis-30522	86	5	evaluation	evaluation	NOUN
fcis-30522	86	6	metrics	metric	NOUN
fcis-30522	86	7	the	the	DET
fcis-30522	86	8	accuracy	accuracy	NOUN
fcis-30522	86	9	and	and	CCONJ
fcis-30522	86	10	f1	f1	NOUN
fcis-30522	86	11	value	value	NOUN
fcis-30522	86	12	are	be	AUX
fcis-30522	86	13	used	use	VERB
fcis-30522	86	14	as	as	ADP
fcis-30522	86	15	evaluation	evaluation	NOUN
fcis-30522	86	16	criteria	criterion	NOUN
fcis-30522	86	17	,	,	PUNCT
fcis-30522	86	18	and	and	CCONJ
fcis-30522	86	19	the	the	DET
fcis-30522	86	20	calculation	calculation	NOUN
fcis-30522	86	21	equation	equation	NOUN
fcis-30522	86	22	is	be	AUX
fcis-30522	86	23	:	:	PUNCT
fcis-30522	86	24	(	(	PUNCT
fcis-30522	86	25	)	)	PUNCT
fcis-30522	86	26	2	2	NUM
fcis-30522	86	27	1	1	NUM
fcis-30522	86	28	tp	tp	PART
fcis-30522	86	29	tn	tn	NOUN
fcis-30522	86	30	accuracy	accuracy	NOUN
fcis-30522	86	31	n	n	ADP
fcis-30522	86	32	tp	tp	NOUN
fcis-30522	86	33	precision	precision	NOUN
fcis-30522	86	34	tp	tp	ADP
fcis-30522	86	35	fp	fp	INTJ
fcis-30522	86	36	tp	tp	PART
fcis-30522	86	37	recall	recall	VERB
fcis-30522	86	38	tp	tp	ADP
fcis-30522	86	39	fn	fn	PROPN
fcis-30522	86	40	precision	precision	NOUN
fcis-30522	86	41	recall	recall	NOUN
fcis-30522	86	42	f	f	PROPN
fcis-30522	86	43	precision	precision	NOUN
fcis-30522	86	44	recall	recall	VERB
fcis-30522	86	45			PROPN
fcis-30522	87	1			PROPN
fcis-30522	87	2			PROPN
fcis-30522	87	3			ADV
fcis-30522	87	4			NUM
fcis-30522	87	5			PROPN
fcis-30522	87	6			PROPN
fcis-30522	87	7			PROPN
fcis-30522	87	8			PROPN
fcis-30522	87	9			X
fcis-30522	87	10	(	(	PUNCT
fcis-30522	87	11	1	1	X
fcis-30522	87	12	)	)	PUNCT
fcis-30522	87	13	tp	tp	NOUN
fcis-30522	87	14	(	(	PUNCT
fcis-30522	87	15	true	true	ADJ
fcis-30522	87	16	positive	positive	ADJ
fcis-30522	87	17	)	)	PUNCT
fcis-30522	87	18	is	be	AUX
fcis-30522	87	19	the	the	DET
fcis-30522	87	20	number	number	NOUN
fcis-30522	87	21	of	of	ADP
fcis-30522	87	22	samples	sample	NOUN
fcis-30522	87	23	correctly	correctly	ADV
fcis-30522	87	24	predicted	predict	VERB
fcis-30522	87	25	by	by	ADP
fcis-30522	87	26	the	the	DET
fcis-30522	87	27	classification	classification	NOUN
fcis-30522	87	28	model	model	NOUN
fcis-30522	87	29	as	as	ADP
fcis-30522	87	30	belonging	belong	VERB
fcis-30522	87	31	to	to	ADP
fcis-30522	87	32	a	a	DET
fcis-30522	87	33	specific	specific	ADJ
fcis-30522	87	34	subject	subject	ADJ
fcis-30522	87	35	category	category	NOUN
fcis-30522	87	36	.	.	PUNCT
fcis-30522	88	1	tn	tn	PROPN
fcis-30522	88	2	(	(	PUNCT
fcis-30522	88	3	true	true	ADJ
fcis-30522	88	4	negative	negative	ADJ
fcis-30522	88	5	)	)	PUNCT
fcis-30522	88	6	is	be	AUX
fcis-30522	88	7	the	the	DET
fcis-30522	88	8	number	number	NOUN
fcis-30522	88	9	of	of	ADP
fcis-30522	88	10	samples	sample	NOUN
fcis-30522	88	11	correctly	correctly	ADV
fcis-30522	88	12	predicted	predict	VERB
fcis-30522	88	13	by	by	ADP
fcis-30522	88	14	the	the	DET
fcis-30522	88	15	classification	classification	NOUN
fcis-30522	88	16	model	model	NOUN
fcis-30522	88	17	as	as	ADP
fcis-30522	88	18	not	not	PART
fcis-30522	88	19	belonging	belong	VERB
fcis-30522	88	20	to	to	ADP
fcis-30522	88	21	that	that	DET
fcis-30522	88	22	subject	subject	NOUN
fcis-30522	88	23	category	category	NOUN
fcis-30522	88	24	.	.	PUNCT
fcis-30522	89	1	fp	fp	X
fcis-30522	89	2	(	(	PUNCT
fcis-30522	89	3	false	false	ADJ
fcis-30522	89	4	positive	positive	ADJ
fcis-30522	89	5	)	)	PUNCT
fcis-30522	89	6	is	be	AUX
fcis-30522	89	7	the	the	DET
fcis-30522	89	8	number	number	NOUN
fcis-30522	89	9	of	of	ADP
fcis-30522	89	10	samples	sample	NOUN
fcis-30522	89	11	incorrectly	incorrectly	ADV
fcis-30522	89	12	predicted	predict	VERB
fcis-30522	89	13	by	by	ADP
fcis-30522	89	14	the	the	DET
fcis-30522	89	15	classification	classification	NOUN
fcis-30522	89	16	model	model	NOUN
fcis-30522	89	17	as	as	ADP
fcis-30522	89	18	belonging	belong	VERB
fcis-30522	89	19	to	to	ADP
fcis-30522	89	20	that	that	DET
fcis-30522	89	21	subject	subject	NOUN
fcis-30522	89	22	category	category	NOUN
fcis-30522	89	23	.	.	PUNCT
fcis-30522	90	1	fn	fn	INTJ
fcis-30522	90	2	(	(	PUNCT
fcis-30522	90	3	false	false	ADJ
fcis-30522	90	4	negative	negative	NOUN
fcis-30522	90	5	)	)	PUNCT
fcis-30522	90	6	is	be	AUX
fcis-30522	90	7	the	the	DET
fcis-30522	90	8	number	number	NOUN
fcis-30522	90	9	of	of	ADP
fcis-30522	90	10	samples	sample	NOUN
fcis-30522	90	11	incorrectly	incorrectly	ADV
fcis-30522	90	12	predicted	predict	VERB
fcis-30522	90	13	by	by	ADP
fcis-30522	90	14	the	the	DET
fcis-30522	90	15	classification	classification	NOUN
fcis-30522	90	16	model	model	NOUN
fcis-30522	90	17	as	as	ADP
fcis-30522	90	18	not	not	PART
fcis-30522	90	19	belonging	belong	VERB
fcis-30522	90	20	to	to	ADP
fcis-30522	90	21	that	that	DET
fcis-30522	90	22	subject	subject	NOUN
fcis-30522	90	23	category	category	NOUN
fcis-30522	90	24	.	.	PUNCT
fcis-30522	91	1	3	3	X
fcis-30522	91	2	.	.	NOUN
fcis-30522	91	3	results	result	NOUN
fcis-30522	91	4	and	and	CCONJ
fcis-30522	91	5	discussion	discussion	NOUN
fcis-30522	91	6	3.1	3.1	NUM
fcis-30522	91	7	.	.	PUNCT
fcis-30522	92	1	word	word	NOUN
fcis-30522	92	2	length	length	NOUN
fcis-30522	92	3	analysis	analysis	NOUN
fcis-30522	92	4	the	the	DET
fcis-30522	92	5	maximum	maximum	ADJ
fcis-30522	92	6	input	input	NOUN
fcis-30522	92	7	length	length	NOUN
fcis-30522	92	8	of	of	ADP
fcis-30522	92	9	the	the	DET
fcis-30522	92	10	macbert	macbert	NOUN
fcis-30522	92	11	model	model	NOUN
fcis-30522	92	12	is	be	AUX
fcis-30522	92	13	512	512	NUM
fcis-30522	92	14	tokens	token	NOUN
fcis-30522	92	15	.	.	PUNCT
fcis-30522	93	1	here	here	ADV
fcis-30522	93	2	,	,	PUNCT
fcis-30522	93	3	we	we	PRON
fcis-30522	93	4	set	set	VERB
fcis-30522	93	5	max_length	max_length	NOUN
fcis-30522	93	6	,	,	PUNCT
fcis-30522	93	7	the	the	DET
fcis-30522	93	8	maximum	maximum	ADJ
fcis-30522	93	9	input	input	NOUN
fcis-30522	93	10	length	length	NOUN
fcis-30522	93	11	of	of	ADP
fcis-30522	93	12	the	the	DET
fcis-30522	93	13	text	text	NOUN
fcis-30522	93	14	,	,	PUNCT
fcis-30522	93	15	to	to	ADP
fcis-30522	93	16	x	x	X
fcis-30522	93	17	*	*	NOUN
fcis-30522	93	18	,	,	PUNCT
fcis-30522	93	19	which	which	PRON
fcis-30522	93	20	represents	represent	VERB
fcis-30522	93	21	the	the	DET
fcis-30522	93	22	word	word	NOUN
fcis-30522	93	23	length	length	NOUN
fcis-30522	93	24	used	use	VERB
fcis-30522	93	25	by	by	ADP
fcis-30522	93	26	macbert	macbert	NOUN
fcis-30522	93	27	.	.	PUNCT
fcis-30522	94	1	assuming	assume	VERB
fcis-30522	94	2	that	that	SCONJ
fcis-30522	94	3	the	the	DET
fcis-30522	94	4	summary	summary	NOUN
fcis-30522	94	5	text	text	NOUN
fcis-30522	94	6	is	be	AUX
fcis-30522	94	7	set	set	VERB
fcis-30522	94	8	to	to	PART
fcis-30522	94	9	texti	texti	VERB
fcis-30522	94	10	,	,	PUNCT
fcis-30522	94	11	it	it	PRON
fcis-30522	94	12	contains	contain	VERB
fcis-30522	94	13	xi	xi	ADP
fcis-30522	94	14	tokens	token	NOUN
fcis-30522	94	15	.	.	PUNCT
fcis-30522	95	1	if	if	SCONJ
fcis-30522	95	2	each	each	DET
fcis-30522	95	3	token	token	NOUN
fcis-30522	95	4	is	be	AUX
fcis-30522	95	5	represented	represent	VERB
fcis-30522	95	6	as	as	ADP
fcis-30522	95	7	an	an	DET
fcis-30522	95	8	ndimensional	ndimensional	ADJ
fcis-30522	95	9	vector	vector	NOUN
fcis-30522	95	10	,	,	PUNCT
fcis-30522	95	11	the	the	DET
fcis-30522	95	12	text	text	NOUN
fcis-30522	95	13	matrix	matrix	NOUN
fcis-30522	95	14	can	can	AUX
fcis-30522	95	15	be	be	AUX
fcis-30522	95	16	expressed	express	VERB
fcis-30522	95	17	as	as	SCONJ
fcis-30522	95	18	follows	follow	VERB
fcis-30522	95	19	,	,	PUNCT
fcis-30522	95	20	as	as	SCONJ
fcis-30522	95	21	shown	show	VERB
fcis-30522	95	22	in	in	ADP
fcis-30522	95	23	the	the	DET
fcis-30522	95	24	following	follow	VERB
fcis-30522	95	25	matrix	matrix	NOUN
fcis-30522	95	26	:	:	PUNCT
fcis-30522	95	27	1	1	NUM
fcis-30522	95	28	11	11	NUM
fcis-30522	95	29	12	12	NUM
fcis-30522	95	30	1	1	NUM
fcis-30522	95	31	2	2	NUM
fcis-30522	95	32	21	21	NUM
fcis-30522	95	33	22	22	NUM
fcis-30522	95	34	2	2	NUM
fcis-30522	95	35	1	1	NUM
fcis-30522	95	36	2	2	NUM
fcis-30522	95	37	...	...	PUNCT
fcis-30522	95	38	...	...	PUNCT
fcis-30522	96	1	......	......	PUNCT
fcis-30522	96	2	...	...	PUNCT
fcis-30522	96	3	...	...	PUNCT
fcis-30522	96	4	...	...	PUNCT
fcis-30522	96	5	...	...	PUNCT
fcis-30522	97	1	i	i	PRON
fcis-30522	97	2	i	i	PRON
fcis-30522	98	1	i	i	PRON
fcis-30522	98	2	i	i	VERB
fcis-30522	99	1	i	i	VERB
fcis-30522	99	2	n	n	VERB
fcis-30522	99	3	n	n	NOUN
fcis-30522	99	4	i	i	NOUN
fcis-30522	99	5	x	x	PUNCT
fcis-30522	100	1	x	x	PUNCT
fcis-30522	100	2	x	x	PUNCT
fcis-30522	100	3	x	x	SYM
fcis-30522	100	4	n	n	NOUN
fcis-30522	100	5	x	x	SYM
fcis-30522	100	6	n	n	PRON
fcis-30522	100	7	word	word	NOUN
fcis-30522	100	8	a	a	DET
fcis-30522	100	9	a	a	DET
fcis-30522	100	10	a	a	DET
fcis-30522	100	11	word	word	NOUN
fcis-30522	100	12	a	a	DET
fcis-30522	100	13	a	a	DET
fcis-30522	100	14	a	a	DET
fcis-30522	100	15	w	w	NOUN
fcis-30522	100	16	word	word	NOUN
fcis-30522	100	17	a	a	DET
fcis-30522	100	18	a	a	PRON
fcis-30522	100	19	a	a	DET
fcis-30522	100	20			NOUN
fcis-30522	101	1			PROPN
fcis-30522	101	2			PROPN
fcis-30522	101	3			NOUN
fcis-30522	101	4			PROPN
fcis-30522	101	5			NOUN
fcis-30522	101	6			PROPN
fcis-30522	101	7			PROPN
fcis-30522	102	1			PROPN
fcis-30522	103	1			PROPN
fcis-30522	104	1			INTJ
fcis-30522	105	1			PROPN
fcis-30522	106	1			INTJ
fcis-30522	107	1			PROPN
fcis-30522	108	1			INTJ
fcis-30522	109	1			PROPN
fcis-30522	110	1			INTJ
fcis-30522	111	1			PROPN
fcis-30522	112	1			INTJ
fcis-30522	113	1			PROPN
fcis-30522	114	1			INTJ
fcis-30522	115	1			PROPN
fcis-30522	115	2			PROPN
fcis-30522	116	1			ADJ
fcis-30522	116	2			PROPN
fcis-30522	116	3			PROPN
fcis-30522	116	4			NOUN
fcis-30522	116	5	(	(	PUNCT
fcis-30522	116	6	2	2	NUM
fcis-30522	116	7	)	)	PUNCT
fcis-30522	116	8	73	73	NUM
fcis-30522	117	1	if	if	SCONJ
fcis-30522	117	2	xi	xi	X
fcis-30522	117	3	≥	≥	X
fcis-30522	117	4	x	x	X
fcis-30522	117	5	*	*	PUNCT
fcis-30522	117	6	,	,	PUNCT
fcis-30522	117	7	the	the	DET
fcis-30522	117	8	first	first	ADJ
fcis-30522	117	9	x*row	x*row	NUM
fcis-30522	117	10	of	of	ADP
fcis-30522	117	11	the	the	DET
fcis-30522	117	12	summary	summary	NOUN
fcis-30522	117	13	text	text	NOUN
fcis-30522	117	14	matrix	matrix	NOUN
fcis-30522	117	15	is	be	AUX
fcis-30522	117	16	taken	take	VERB
fcis-30522	117	17	as	as	ADP
fcis-30522	117	18	input	input	NOUN
fcis-30522	117	19	;	;	PUNCT
fcis-30522	117	20	the	the	DET
fcis-30522	117	21	result	result	NOUN
fcis-30522	117	22	is	be	AUX
fcis-30522	117	23	as	as	SCONJ
fcis-30522	117	24	follows	follow	VERB
fcis-30522	117	25	：	：	PUNCT
fcis-30522	117	26	1	1	NUM
fcis-30522	117	27	11	11	NUM
fcis-30522	117	28	12	12	NUM
fcis-30522	117	29	1	1	NUM
fcis-30522	117	30	2	2	NUM
fcis-30522	117	31	21	21	NUM
fcis-30522	117	32	22	22	NUM
fcis-30522	117	33	2	2	NUM
fcis-30522	117	34	1	1	NUM
fcis-30522	117	35	2	2	NUM
fcis-30522	117	36	,	,	PUNCT
fcis-30522	117	37	...	...	PUNCT
fcis-30522	117	38	...	...	PUNCT
fcis-30522	117	39	...	...	PUNCT
fcis-30522	117	40	...	...	PUNCT
fcis-30522	117	41	...	...	PUNCT
fcis-30522	117	42	...	...	PUNCT
fcis-30522	117	43	...	...	PUNCT
fcis-30522	117	44	...	...	PUNCT
fcis-30522	118	1	i	i	PRON
fcis-30522	118	2	x	x	SYM
fcis-30522	118	3	n	n	VERB
fcis-30522	118	4	n	n	CCONJ
fcis-30522	118	5	n	n	NOUN
fcis-30522	118	6	x	x	NOUN
fcis-30522	118	7	x	x	PUNCT
fcis-30522	118	8	x	x	PUNCT
fcis-30522	118	9	x	x	SYM
fcis-30522	118	10	n	n	NOUN
fcis-30522	118	11	x	x	SYM
fcis-30522	118	12	x	x	X
fcis-30522	118	13	word	word	NOUN
fcis-30522	118	14	a	a	DET
fcis-30522	118	15	a	a	DET
fcis-30522	118	16	a	a	DET
fcis-30522	118	17	word	word	NOUN
fcis-30522	118	18	a	a	DET
fcis-30522	118	19	a	a	DET
fcis-30522	118	20	a	a	DET
fcis-30522	118	21	word	word	NOUN
fcis-30522	118	22	a	a	DET
fcis-30522	118	23	a	a	DET
fcis-30522	118	24	a	a	DET
fcis-30522	118	25			NOUN
fcis-30522	118	26			PROPN
fcis-30522	118	27			PROPN
fcis-30522	118	28			PROPN
fcis-30522	118	29			PROPN
fcis-30522	118	30			VERB
fcis-30522	118	31			PROPN
fcis-30522	118	32			NUM
fcis-30522	119	1			NOUN
fcis-30522	119	2			PROPN
fcis-30522	119	3			NOUN
fcis-30522	119	4			PROPN
fcis-30522	119	5			PROPN
fcis-30522	120	1			PROPN
fcis-30522	121	1			PROPN
fcis-30522	122	1			INTJ
fcis-30522	123	1			PROPN
fcis-30522	124	1			INTJ
fcis-30522	125	1			PROPN
fcis-30522	126	1			INTJ
fcis-30522	127	1			PROPN
fcis-30522	128	1			INTJ
fcis-30522	129	1			PROPN
fcis-30522	130	1			INTJ
fcis-30522	131	1			PROPN
fcis-30522	132	1			INTJ
fcis-30522	133	1			PROPN
fcis-30522	133	2			PROPN
fcis-30522	134	1			ADJ
fcis-30522	134	2			PROPN
fcis-30522	134	3			PROPN
fcis-30522	134	4			NOUN
fcis-30522	134	5	(	(	PUNCT
fcis-30522	134	6	3	3	NUM
fcis-30522	134	7	)	)	PUNCT
fcis-30522	134	8	if	if	SCONJ
fcis-30522	134	9	xi	xi	X
fcis-30522	134	10	<	<	X
fcis-30522	134	11	x	x	X
fcis-30522	134	12	*	*	PROPN
fcis-30522	134	13	,	,	PUNCT
fcis-30522	134	14	a	a	DET
fcis-30522	134	15	n	n	ADV
fcis-30522	134	16	-	-	PUNCT
fcis-30522	134	17	dimensional	dimensional	ADJ
fcis-30522	134	18	zero	zero	NUM
fcis-30522	134	19	vector	vector	NOUN
fcis-30522	134	20	is	be	AUX
fcis-30522	134	21	added	add	VERB
fcis-30522	134	22	below	below	ADP
fcis-30522	134	23	the	the	DET
fcis-30522	134	24	original	original	ADJ
fcis-30522	134	25	summary	summary	NOUN
fcis-30522	134	26	text	text	NOUN
fcis-30522	134	27	matrix	matrix	NOUN
fcis-30522	134	28	to	to	PART
fcis-30522	134	29	supplement	supplement	VERB
fcis-30522	134	30	the	the	DET
fcis-30522	134	31	number	number	NOUN
fcis-30522	134	32	of	of	ADP
fcis-30522	134	33	rows	row	NOUN
fcis-30522	134	34	to	to	ADP
fcis-30522	134	35	x	x	X
fcis-30522	134	36	*	*	NOUN
fcis-30522	134	37	;	;	PUNCT
fcis-30522	134	38	the	the	DET
fcis-30522	134	39	result	result	NOUN
fcis-30522	134	40	is	be	AUX
fcis-30522	134	41	as	as	SCONJ
fcis-30522	134	42	follows	follow	VERB
fcis-30522	134	43	:	:	PUNCT
fcis-30522	134	44	1	1	NUM
fcis-30522	134	45	11	11	NUM
fcis-30522	134	46	12	12	NUM
fcis-30522	134	47	1	1	NUM
fcis-30522	134	48	2	2	NUM
fcis-30522	134	49	21	21	NUM
fcis-30522	134	50	22	22	NUM
fcis-30522	134	51	2	2	NUM
fcis-30522	134	52	1	1	NUM
fcis-30522	134	53	2	2	NUM
fcis-30522	134	54	...	...	PUNCT
fcis-30522	134	55	...	...	PUNCT
fcis-30522	134	56	...	...	PUNCT
fcis-30522	134	57	...	...	PUNCT
fcis-30522	134	58	...	...	PUNCT
fcis-30522	134	59	...	...	PUNCT
fcis-30522	134	60	...	...	PUNCT
fcis-30522	135	1	...	...	PUNCT
fcis-30522	136	1	0	0	NUM
fcis-30522	136	2	0	0	NUM
fcis-30522	136	3	0	0	NUM
fcis-30522	136	4	...	...	PUNCT
fcis-30522	137	1	0	0	NUM
fcis-30522	137	2	0	0	NUM
fcis-30522	137	3	0	0	NUM
fcis-30522	137	4	0	0	NUM
fcis-30522	137	5	...	...	PUNCT
fcis-30522	137	6	0	0	NUM
fcis-30522	137	7	...	...	PUNCT
fcis-30522	137	8	...	...	PUNCT
fcis-30522	137	9	...	...	PUNCT
fcis-30522	137	10	...	...	PUNCT
fcis-30522	137	11	...	...	PUNCT
fcis-30522	138	1	0	0	NUM
fcis-30522	138	2	0	0	NUM
fcis-30522	138	3	0	0	NUM
fcis-30522	138	4	...	...	PUNCT
fcis-30522	138	5	0	0	PUNCT
fcis-30522	139	1	i	i	PRON
fcis-30522	139	2	i	i	PRON
fcis-30522	140	1	i	i	VERB
fcis-30522	140	2	i	i	VERB
fcis-30522	140	3	n	n	VERB
fcis-30522	140	4	n	n	NOUN
fcis-30522	140	5	x	x	NOUN
fcis-30522	140	6	x	x	PUNCT
fcis-30522	140	7	x	x	SYM
fcis-30522	140	8	x	x	SYM
fcis-30522	140	9	n	n	NOUN
fcis-30522	140	10	i	i	NOUN
fcis-30522	140	11	x	x	SYM
fcis-30522	140	12	n	n	PRON
fcis-30522	140	13	word	word	VERB
fcis-30522	140	14	a	a	DET
fcis-30522	140	15	a	a	DET
fcis-30522	140	16	a	a	DET
fcis-30522	140	17	word	word	NOUN
fcis-30522	140	18	a	a	DET
fcis-30522	140	19	a	a	DET
fcis-30522	140	20	a	a	DET
fcis-30522	140	21	word	word	NOUN
fcis-30522	140	22	a	a	DET
fcis-30522	140	23	a	a	DET
fcis-30522	140	24	a	a	NOUN
fcis-30522	140	25	x	x	SYM
fcis-30522	140	26	x	x	SYM
fcis-30522	140	27			PROPN
fcis-30522	140	28			PROPN
fcis-30522	140	29			NOUN
fcis-30522	140	30			PROPN
fcis-30522	140	31			PROPN
fcis-30522	141	1			NOUN
fcis-30522	141	2			PROPN
fcis-30522	141	3			NOUN
fcis-30522	141	4			PROPN
fcis-30522	142	1			PROPN
fcis-30522	142	2			PROPN
fcis-30522	143	1			PROPN
fcis-30522	144	1			INTJ
fcis-30522	145	1			PROPN
fcis-30522	146	1			INTJ
fcis-30522	147	1			PROPN
fcis-30522	148	1			INTJ
fcis-30522	149	1			PROPN
fcis-30522	150	1			INTJ
fcis-30522	151	1			PROPN
fcis-30522	152	1			INTJ
fcis-30522	153	1			PROPN
fcis-30522	154	1			INTJ
fcis-30522	155	1			PROPN
fcis-30522	156	1			INTJ
fcis-30522	157	1			PROPN
fcis-30522	158	1			INTJ
fcis-30522	159	1			PROPN
fcis-30522	160	1			INTJ
fcis-30522	161	1			PROPN
fcis-30522	162	1			INTJ
fcis-30522	163	1			PROPN
fcis-30522	164	1			INTJ
fcis-30522	165	1			PROPN
fcis-30522	166	1			INTJ
fcis-30522	167	1			PROPN
fcis-30522	168	1			INTJ
fcis-30522	169	1			PROPN
fcis-30522	170	1			INTJ
fcis-30522	171	1			PROPN
fcis-30522	172	1			INTJ
fcis-30522	173	1			PROPN
fcis-30522	174	1			INTJ
fcis-30522	175	1			PROPN
fcis-30522	176	1			INTJ
fcis-30522	177	1			PROPN
fcis-30522	178	1			INTJ
fcis-30522	178	2			PROPN
fcis-30522	178	3			NOUN
fcis-30522	179	1			PROPN
fcis-30522	180	1			PROPN
fcis-30522	180	2			PROPN
fcis-30522	181	1			ADJ
fcis-30522	181	2			PROPN
fcis-30522	181	3			PROPN
fcis-30522	181	4			NOUN
fcis-30522	181	5	(	(	PUNCT
fcis-30522	181	6	4	4	NUM
fcis-30522	181	7	)	)	PUNCT
fcis-30522	181	8	according	accord	VERB
fcis-30522	181	9	to	to	ADP
fcis-30522	181	10	general	general	ADJ
fcis-30522	181	11	signal	signal	NOUN
fcis-30522	181	12	processing	processing	NOUN
fcis-30522	181	13	principles	principle	NOUN
fcis-30522	181	14	,	,	PUNCT
fcis-30522	181	15	the	the	DET
fcis-30522	181	16	method	method	NOUN
fcis-30522	181	17	of	of	ADP
fcis-30522	181	18	padding	padding	NOUN
fcis-30522	181	19	with	with	ADP
fcis-30522	181	20	zero	zero	NUM
fcis-30522	181	21	vectors	vector	NOUN
fcis-30522	181	22	can	can	AUX
fcis-30522	181	23	lead	lead	VERB
fcis-30522	181	24	to	to	ADP
fcis-30522	181	25	the	the	DET
fcis-30522	181	26	sparsification	sparsification	NOUN
fcis-30522	181	27	of	of	ADP
fcis-30522	181	28	text	text	NOUN
fcis-30522	181	29	data	datum	NOUN
fcis-30522	181	30	in	in	ADP
fcis-30522	181	31	text	text	NOUN
fcis-30522	181	32	classification	classification	NOUN
fcis-30522	181	33	,	,	PUNCT
fcis-30522	181	34	thereby	thereby	ADV
fcis-30522	181	35	reducing	reduce	VERB
fcis-30522	181	36	classification	classification	NOUN
fcis-30522	181	37	efficiency	efficiency	NOUN
fcis-30522	181	38	.	.	PUNCT
fcis-30522	182	1	in	in	ADP
fcis-30522	182	2	other	other	ADJ
fcis-30522	182	3	words	word	NOUN
fcis-30522	182	4	,	,	PUNCT
fcis-30522	182	5	the	the	DET
fcis-30522	182	6	zero	zero	NUM
fcis-30522	182	7	vectors	vector	NOUN
fcis-30522	182	8	added	add	VERB
fcis-30522	182	9	here	here	ADV
fcis-30522	182	10	are	be	AUX
fcis-30522	182	11	essentially	essentially	ADV
fcis-30522	182	12	data	datum	NOUN
fcis-30522	182	13	white	white	ADJ
fcis-30522	182	14	noise	noise	NOUN
fcis-30522	182	15	,	,	PUNCT
fcis-30522	182	16	and	and	CCONJ
fcis-30522	182	17	the	the	DET
fcis-30522	182	18	whiter	whiter	NOUN
fcis-30522	182	19	noise	noise	NOUN
fcis-30522	182	20	there	there	PRON
fcis-30522	182	21	is	be	VERB
fcis-30522	182	22	,	,	PUNCT
fcis-30522	182	23	the	the	PRON
fcis-30522	182	24	more	more	ADV
fcis-30522	182	25	negatively	negatively	ADV
fcis-30522	182	26	it	it	PRON
fcis-30522	182	27	impacts	impact	VERB
fcis-30522	182	28	subsequent	subsequent	ADJ
fcis-30522	182	29	classification	classification	NOUN
fcis-30522	182	30	processing	processing	NOUN
fcis-30522	182	31	.	.	PUNCT
fcis-30522	183	1	in	in	ADP
fcis-30522	183	2	summary	summary	NOUN
fcis-30522	183	3	,	,	PUNCT
fcis-30522	183	4	the	the	DET
fcis-30522	183	5	choice	choice	NOUN
fcis-30522	183	6	of	of	ADP
fcis-30522	183	7	x*will	x*will	NOUN
fcis-30522	183	8	significantly	significantly	ADV
fcis-30522	183	9	affect	affect	VERB
fcis-30522	183	10	the	the	DET
fcis-30522	183	11	model	model	NOUN
fcis-30522	183	12	's	's	PART
fcis-30522	183	13	performance	performance	NOUN
fcis-30522	183	14	.	.	PUNCT
fcis-30522	184	1	if	if	SCONJ
fcis-30522	184	2	x	x	NOUN
fcis-30522	184	3	*	*	VERB
fcis-30522	184	4	is	be	AUX
fcis-30522	184	5	greater	great	ADJ
fcis-30522	184	6	than	than	ADP
fcis-30522	184	7	the	the	DET
fcis-30522	184	8	text	text	NOUN
fcis-30522	184	9	data	datum	NOUN
fcis-30522	184	10	length	length	NOUN
fcis-30522	184	11	and	and	CCONJ
fcis-30522	184	12	the	the	DET
fcis-30522	184	13	difference	difference	NOUN
fcis-30522	184	14	is	be	AUX
fcis-30522	184	15	large	large	ADJ
fcis-30522	184	16	,	,	PUNCT
fcis-30522	184	17	it	it	PRON
fcis-30522	184	18	will	will	AUX
fcis-30522	184	19	result	result	VERB
fcis-30522	184	20	in	in	ADP
fcis-30522	184	21	too	too	ADV
fcis-30522	184	22	much	much	ADJ
fcis-30522	184	23	content	content	NOUN
fcis-30522	184	24	being	be	AUX
fcis-30522	184	25	deleted	delete	VERB
fcis-30522	184	26	.	.	PUNCT
fcis-30522	185	1	similarly	similarly	ADV
fcis-30522	185	2	,	,	PUNCT
fcis-30522	185	3	if	if	SCONJ
fcis-30522	185	4	x*is	x*is	PROPN
fcis-30522	185	5	less	less	ADJ
fcis-30522	185	6	than	than	ADP
fcis-30522	185	7	the	the	DET
fcis-30522	185	8	text	text	NOUN
fcis-30522	185	9	data	datum	NOUN
fcis-30522	185	10	length	length	NOUN
fcis-30522	185	11	and	and	CCONJ
fcis-30522	185	12	the	the	DET
fcis-30522	185	13	difference	difference	NOUN
fcis-30522	185	14	is	be	AUX
fcis-30522	185	15	large	large	ADJ
fcis-30522	185	16	,	,	PUNCT
fcis-30522	185	17	it	it	PRON
fcis-30522	185	18	will	will	AUX
fcis-30522	185	19	result	result	VERB
fcis-30522	185	20	in	in	ADP
fcis-30522	185	21	more	more	ADJ
fcis-30522	185	22	white	white	ADJ
fcis-30522	185	23	noise	noise	NOUN
fcis-30522	185	24	being	be	AUX
fcis-30522	185	25	added	add	VERB
fcis-30522	185	26	.	.	PUNCT
fcis-30522	186	1	both	both	DET
fcis-30522	186	2	situations	situation	NOUN
fcis-30522	186	3	will	will	AUX
fcis-30522	186	4	cause	cause	VERB
fcis-30522	186	5	the	the	DET
fcis-30522	186	6	model	model	NOUN
fcis-30522	186	7	to	to	PART
fcis-30522	186	8	have	have	VERB
fcis-30522	186	9	a	a	DET
fcis-30522	186	10	biased	biased	ADJ
fcis-30522	186	11	understanding	understanding	NOUN
fcis-30522	186	12	of	of	ADP
fcis-30522	186	13	the	the	DET
fcis-30522	186	14	original	original	ADJ
fcis-30522	186	15	meaning	meaning	NOUN
fcis-30522	186	16	,	,	PUNCT
fcis-30522	186	17	resulting	result	VERB
fcis-30522	186	18	in	in	ADP
fcis-30522	186	19	poor	poor	ADJ
fcis-30522	186	20	performance	performance	NOUN
fcis-30522	186	21	.	.	PUNCT
fcis-30522	187	1	(	(	PUNCT
fcis-30522	187	2	fig	fig	NOUN
fcis-30522	187	3	2	2	X
fcis-30522	187	4	)	)	PUNCT
fcis-30522	187	5	shows	show	VERB
fcis-30522	187	6	the	the	DET
fcis-30522	187	7	skewed	skewed	ADJ
fcis-30522	187	8	distribution	distribution	NOUN
fcis-30522	187	9	of	of	ADP
fcis-30522	187	10	the	the	DET
fcis-30522	187	11	number	number	NOUN
fcis-30522	187	12	of	of	ADP
fcis-30522	187	13	samples	sample	NOUN
fcis-30522	187	14	across	across	ADP
fcis-30522	187	15	different	different	ADJ
fcis-30522	187	16	word	word	NOUN
fcis-30522	187	17	ranges	range	VERB
fcis-30522	187	18	for	for	ADP
fcis-30522	187	19	each	each	DET
fcis-30522	187	20	sample	sample	NOUN
fcis-30522	187	21	type	type	NOUN
fcis-30522	187	22	.	.	PUNCT
fcis-30522	188	1	from	from	ADP
fcis-30522	188	2	the	the	DET
fcis-30522	188	3	distribution	distribution	NOUN
fcis-30522	188	4	in	in	ADP
fcis-30522	188	5	the	the	DET
fcis-30522	188	6	fig	fig	NOUN
fcis-30522	188	7	,	,	PUNCT
fcis-30522	188	8	it	it	PRON
fcis-30522	188	9	can	can	AUX
fcis-30522	188	10	be	be	AUX
fcis-30522	188	11	seen	see	VERB
fcis-30522	188	12	that	that	SCONJ
fcis-30522	188	13	the	the	DET
fcis-30522	188	14	characteristics	characteristic	NOUN
fcis-30522	188	15	of	of	ADP
fcis-30522	188	16	the	the	DET
fcis-30522	188	17	three	three	NUM
fcis-30522	188	18	types	type	NOUN
fcis-30522	188	19	of	of	ADP
fcis-30522	188	20	data	datum	NOUN
fcis-30522	188	21	are	be	AUX
fcis-30522	188	22	all	all	PRON
fcis-30522	188	23	rightskewed	rightskewe	VERB
fcis-30522	188	24	—	—	PUNCT
fcis-30522	188	25	that	that	ADV
fcis-30522	188	26	is	is	ADV
fcis-30522	188	27	,	,	PUNCT
fcis-30522	188	28	the	the	DET
fcis-30522	188	29	right	right	ADJ
fcis-30522	188	30	tail	tail	NOUN
fcis-30522	188	31	of	of	ADP
fcis-30522	188	32	the	the	DET
fcis-30522	188	33	data	datum	NOUN
fcis-30522	188	34	is	be	AUX
fcis-30522	188	35	longer	long	ADJ
fcis-30522	188	36	,	,	PUNCT
fcis-30522	188	37	and	and	CCONJ
fcis-30522	188	38	most	most	ADJ
fcis-30522	188	39	of	of	ADP
fcis-30522	188	40	the	the	DET
fcis-30522	188	41	data	datum	NOUN
fcis-30522	188	42	is	be	AUX
fcis-30522	188	43	concentrated	concentrate	VERB
fcis-30522	188	44	on	on	ADP
fcis-30522	188	45	the	the	DET
fcis-30522	188	46	left	left	NOUN
fcis-30522	188	47	.	.	PUNCT
fcis-30522	189	1	additionally	additionally	ADV
fcis-30522	189	2	,	,	PUNCT
fcis-30522	189	3	from	from	ADP
fcis-30522	189	4	the	the	DET
fcis-30522	189	5	distribution	distribution	NOUN
fcis-30522	189	6	of	of	ADP
fcis-30522	189	7	the	the	DET
fcis-30522	189	8	text	text	NOUN
fcis-30522	189	9	in	in	ADP
fcis-30522	189	10	different	different	ADJ
fcis-30522	189	11	intervals	interval	NOUN
fcis-30522	189	12	,	,	PUNCT
fcis-30522	189	13	it	it	PRON
fcis-30522	189	14	can	can	AUX
fcis-30522	189	15	be	be	AUX
fcis-30522	189	16	seen	see	VERB
fcis-30522	189	17	that	that	SCONJ
fcis-30522	189	18	,	,	PUNCT
fcis-30522	189	19	for	for	ADP
fcis-30522	189	20	the	the	DET
fcis-30522	189	21	three	three	NUM
fcis-30522	189	22	sample	sample	NOUN
fcis-30522	189	23	types	type	NOUN
fcis-30522	189	24	,	,	PUNCT
fcis-30522	189	25	the	the	DET
fcis-30522	189	26	text	text	NOUN
fcis-30522	189	27	word	word	NOUN
fcis-30522	189	28	range	range	NOUN
fcis-30522	189	29	of	of	ADP
fcis-30522	189	30	100–300	100–300	NUM
fcis-30522	189	31	accounts	account	NOUN
fcis-30522	189	32	for	for	ADP
fcis-30522	189	33	the	the	DET
fcis-30522	189	34	largest	large	ADJ
fcis-30522	189	35	proportion	proportion	NOUN
fcis-30522	189	36	of	of	ADP
fcis-30522	189	37	the	the	DET
fcis-30522	189	38	overall	overall	ADJ
fcis-30522	189	39	samples	sample	NOUN
fcis-30522	189	40	.	.	PUNCT
fcis-30522	190	1	taking	take	VERB
fcis-30522	190	2	the	the	DET
fcis-30522	190	3	original	original	ADJ
fcis-30522	190	4	sample	sample	NOUN
fcis-30522	190	5	as	as	ADP
fcis-30522	190	6	an	an	DET
fcis-30522	190	7	example	example	NOUN
fcis-30522	190	8	,	,	PUNCT
fcis-30522	190	9	the	the	DET
fcis-30522	190	10	total	total	ADJ
fcis-30522	190	11	number	number	NOUN
fcis-30522	190	12	of	of	ADP
fcis-30522	190	13	samples	sample	NOUN
fcis-30522	190	14	with	with	ADP
fcis-30522	190	15	100–300	100–300	NUM
fcis-30522	190	16	words	word	NOUN
fcis-30522	190	17	is	be	AUX
fcis-30522	190	18	1,806	1,806	NUM
fcis-30522	190	19	,	,	PUNCT
fcis-30522	190	20	accounting	account	VERB
fcis-30522	190	21	for	for	ADP
fcis-30522	190	22	71	71	NUM
fcis-30522	190	23	%	%	NOUN
fcis-30522	190	24	of	of	ADP
fcis-30522	190	25	the	the	DET
fcis-30522	190	26	total	total	ADJ
fcis-30522	190	27	samples	sample	NOUN
fcis-30522	190	28	.	.	PUNCT
fcis-30522	191	1	therefore	therefore	ADV
fcis-30522	191	2	,	,	PUNCT
fcis-30522	191	3	when	when	SCONJ
fcis-30522	191	4	selecting	select	VERB
fcis-30522	191	5	the	the	DET
fcis-30522	191	6	word	word	NOUN
fcis-30522	191	7	length	length	NOUN
fcis-30522	191	8	,	,	PUNCT
fcis-30522	191	9	we	we	PRON
fcis-30522	191	10	choose	choose	VERB
fcis-30522	191	11	the	the	DET
fcis-30522	191	12	middle	middle	ADJ
fcis-30522	191	13	word	word	NOUN
fcis-30522	191	14	count	count	NOUN
fcis-30522	191	15	of	of	ADP
fcis-30522	191	16	200	200	NUM
fcis-30522	191	17	words	word	NOUN
fcis-30522	191	18	as	as	ADP
fcis-30522	191	19	the	the	DET
fcis-30522	191	20	benchmark	benchmark	NOUN
fcis-30522	191	21	and	and	CCONJ
fcis-30522	191	22	use	use	VERB
fcis-30522	191	23	a	a	DET
fcis-30522	191	24	range	range	NOUN
fcis-30522	191	25	of	of	ADP
fcis-30522	191	26	±50	±50	NOUN
fcis-30522	191	27	words	word	NOUN
fcis-30522	191	28	to	to	PART
fcis-30522	191	29	conduct	conduct	VERB
fcis-30522	191	30	model	model	NOUN
fcis-30522	191	31	testing	testing	NOUN
fcis-30522	191	32	.	.	PUNCT
fcis-30522	192	1	it	it	PRON
fcis-30522	192	2	is	be	AUX
fcis-30522	192	3	found	find	VERB
fcis-30522	192	4	that	that	SCONJ
fcis-30522	192	5	when	when	SCONJ
fcis-30522	192	6	the	the	DET
fcis-30522	192	7	truncation	truncation	NOUN
fcis-30522	192	8	length	length	NOUN
fcis-30522	192	9	is	be	AUX
fcis-30522	192	10	150–200	150–200	NUM
fcis-30522	192	11	words	word	NOUN
fcis-30522	192	12	,	,	PUNCT
fcis-30522	192	13	the	the	DET
fcis-30522	192	14	model	model	NOUN
fcis-30522	192	15	performance	performance	NOUN
fcis-30522	192	16	is	be	AUX
fcis-30522	192	17	generally	generally	ADV
fcis-30522	192	18	low	low	ADJ
fcis-30522	192	19	.	.	PUNCT
fcis-30522	193	1	this	this	PRON
fcis-30522	193	2	is	be	AUX
fcis-30522	193	3	because	because	SCONJ
fcis-30522	193	4	the	the	DET
fcis-30522	193	5	negative	negative	ADJ
fcis-30522	193	6	impact	impact	NOUN
fcis-30522	193	7	of	of	ADP
fcis-30522	193	8	excessive	excessive	ADJ
fcis-30522	193	9	content	content	NOUN
fcis-30522	193	10	deletion	deletion	NOUN
fcis-30522	193	11	is	be	AUX
fcis-30522	193	12	often	often	ADV
fcis-30522	193	13	greater	great	ADJ
fcis-30522	193	14	than	than	ADP
fcis-30522	193	15	the	the	DET
fcis-30522	193	16	impact	impact	NOUN
fcis-30522	193	17	of	of	ADP
fcis-30522	193	18	adding	add	VERB
fcis-30522	193	19	white	white	ADJ
fcis-30522	193	20	noise	noise	NOUN
fcis-30522	193	21	.	.	PUNCT
fcis-30522	194	1	the	the	DET
fcis-30522	194	2	accuracy	accuracy	NOUN
fcis-30522	194	3	of	of	ADP
fcis-30522	194	4	the	the	DET
fcis-30522	194	5	model	model	NOUN
fcis-30522	194	6	is	be	AUX
fcis-30522	194	7	higher	high	ADJ
fcis-30522	194	8	when	when	SCONJ
fcis-30522	194	9	the	the	DET
fcis-30522	194	10	truncation	truncation	NOUN
fcis-30522	194	11	length	length	NOUN
fcis-30522	194	12	is	be	AUX
fcis-30522	194	13	200–250	200–250	NUM
fcis-30522	194	14	words	word	NOUN
fcis-30522	194	15	,	,	PUNCT
fcis-30522	194	16	reaching	reach	VERB
fcis-30522	194	17	its	its	PRON
fcis-30522	194	18	highest	high	ADJ
fcis-30522	194	19	at	at	ADP
fcis-30522	194	20	230–250	230–250	NUM
fcis-30522	194	21	words	word	NOUN
fcis-30522	194	22	,	,	PUNCT
fcis-30522	194	23	generally	generally	ADV
fcis-30522	194	24	exceeding	exceed	VERB
fcis-30522	194	25	90	90	NUM
fcis-30522	194	26	%	%	NOUN
fcis-30522	194	27	.	.	PUNCT
fcis-30522	195	1	within	within	ADP
fcis-30522	195	2	this	this	DET
fcis-30522	195	3	range	range	NOUN
fcis-30522	195	4	,	,	PUNCT
fcis-30522	195	5	we	we	PRON
fcis-30522	195	6	choose	choose	VERB
fcis-30522	195	7	5	5	NUM
fcis-30522	195	8	as	as	ADP
fcis-30522	195	9	the	the	DET
fcis-30522	195	10	step	step	NOUN
fcis-30522	195	11	size	size	NOUN
fcis-30522	195	12	to	to	PART
fcis-30522	195	13	divide	divide	VERB
fcis-30522	195	14	the	the	DET
fcis-30522	195	15	interval	interval	NOUN
fcis-30522	195	16	,	,	PUNCT
fcis-30522	195	17	and	and	CCONJ
fcis-30522	195	18	we	we	PRON
fcis-30522	195	19	select	select	VERB
fcis-30522	195	20	truncation	truncation	NOUN
fcis-30522	195	21	lengths	length	NOUN
fcis-30522	195	22	of	of	ADP
fcis-30522	195	23	230,235,240,245	230,235,240,245	NUM
fcis-30522	195	24	and	and	CCONJ
fcis-30522	195	25	250	250	NUM
fcis-30522	195	26	words	word	NOUN
fcis-30522	195	27	to	to	PART
fcis-30522	195	28	test	test	VERB
fcis-30522	195	29	and	and	CCONJ
fcis-30522	195	30	determine	determine	VERB
fcis-30522	195	31	the	the	DET
fcis-30522	195	32	final	final	ADJ
fcis-30522	195	33	word	word	NOUN
fcis-30522	195	34	length	length	NOUN
fcis-30522	195	35	.	.	PUNCT
fcis-30522	196	1	table	table	NOUN
fcis-30522	196	2	4	4	NUM
fcis-30522	196	3	and	and	CCONJ
fcis-30522	196	4	(	(	PUNCT
fcis-30522	196	5	fig	fig	NOUN
fcis-30522	196	6	3	3	X
fcis-30522	196	7	)	)	PUNCT
fcis-30522	196	8	show	show	VERB
fcis-30522	196	9	the	the	DET
fcis-30522	196	10	test	test	NOUN
fcis-30522	196	11	results	result	NOUN
fcis-30522	196	12	and	and	CCONJ
fcis-30522	196	13	corresponding	correspond	VERB
fcis-30522	196	14	training	training	NOUN
fcis-30522	196	15	metrics	metric	NOUN
fcis-30522	196	16	.	.	PUNCT
fcis-30522	197	1	fig	fig	NOUN
fcis-30522	197	2	2	2	NUM
fcis-30522	197	3	.	.	PUNCT
fcis-30522	197	4	three	three	NUM
fcis-30522	197	5	samples	sample	NOUN
fcis-30522	197	6	number	number	NOUN
fcis-30522	197	7	of	of	ADP
fcis-30522	197	8	samples	sample	NOUN
fcis-30522	197	9	in	in	ADP
fcis-30522	197	10	different	different	ADJ
fcis-30522	197	11	sample	sample	NOUN
fcis-30522	197	12	word	word	NOUN
fcis-30522	197	13	ranges	range	VERB
fcis-30522	197	14	74	74	NUM
fcis-30522	197	15	table	table	NOUN
fcis-30522	197	16	4	4	NUM
fcis-30522	197	17	.	.	PUNCT
fcis-30522	197	18	test	test	NOUN
fcis-30522	197	19	results	result	NOUN
fcis-30522	197	20	for	for	ADP
fcis-30522	197	21	different	different	ADJ
fcis-30522	197	22	word	word	NOUN
fcis-30522	197	23	length	length	NOUN
fcis-30522	197	24	models	model	NOUN
fcis-30522	197	25	test	test	NOUN
fcis-30522	197	26	model	model	PROPN
fcis-30522	197	27	macbert	macbert	PROPN
fcis-30522	197	28	-	-	PUNCT
fcis-30522	197	29	sa	sa	NOUN
fcis-30522	197	30	-	-	PUNCT
fcis-30522	197	31	bilstm	bilstm	NOUN
fcis-30522	197	32	-	-	PUNCT
fcis-30522	197	33	textcnn	textcnn	NOUN
fcis-30522	197	34	-	-	PUNCT
fcis-30522	197	35	sa	sa	PROPN
fcis-30522	197	36	word	word	NOUN
fcis-30522	197	37	length	length	NOUN
fcis-30522	197	38	accuracy	accuracy	NOUN
fcis-30522	197	39	f1	f1	NOUN
fcis-30522	197	40	230	230	NUM
fcis-30522	197	41	94.09	94.09	NUM
fcis-30522	197	42	%	%	NOUN
fcis-30522	197	43	94.09	94.09	NUM
fcis-30522	197	44	%	%	NOUN
fcis-30522	197	45	235	235	NUM
fcis-30522	197	46	94.82	94.82	NUM
fcis-30522	197	47	%	%	NOUN
fcis-30522	197	48	94.80	94.80	NUM
fcis-30522	197	49	%	%	NOUN
fcis-30522	197	50	240	240	NUM
fcis-30522	197	51	93.08	93.08	NUM
fcis-30522	197	52	%	%	NOUN
fcis-30522	197	53	93.08	93.08	NUM
fcis-30522	197	54	%	%	NOUN
fcis-30522	197	55	245	245	NUM
fcis-30522	197	56	94.26	94.26	NUM
fcis-30522	197	57	%	%	NOUN
fcis-30522	197	58	94.22	94.22	NUM
fcis-30522	197	59	%	%	NOUN
fcis-30522	197	60	250	250	NUM
fcis-30522	197	61	94.05	94.05	NUM
fcis-30522	197	62	%	%	NOUN
fcis-30522	197	63	94.05	94.05	NUM
fcis-30522	197	64	%	%	NOUN
fcis-30522	197	65	fig	fig	NOUN
fcis-30522	197	66	3	3	NUM
fcis-30522	197	67	.	.	PUNCT
fcis-30522	197	68	performance	performance	NOUN
fcis-30522	197	69	trend	trend	NOUN
fcis-30522	197	70	diagram	diagram	NOUN
fcis-30522	197	71	for	for	ADP
fcis-30522	197	72	different	different	ADJ
fcis-30522	197	73	word	word	NOUN
fcis-30522	197	74	length	length	NOUN
fcis-30522	197	75	models	model	NOUN
fcis-30522	197	76	the	the	DET
fcis-30522	197	77	test	test	NOUN
fcis-30522	197	78	results	result	NOUN
fcis-30522	197	79	show	show	VERB
fcis-30522	197	80	that	that	SCONJ
fcis-30522	197	81	when	when	SCONJ
fcis-30522	197	82	the	the	DET
fcis-30522	197	83	word	word	NOUN
fcis-30522	197	84	length	length	NOUN
fcis-30522	197	85	is	be	AUX
fcis-30522	197	86	235	235	NUM
fcis-30522	197	87	,	,	PUNCT
fcis-30522	197	88	both	both	DET
fcis-30522	197	89	the	the	DET
fcis-30522	197	90	accuracy	accuracy	NOUN
fcis-30522	197	91	and	and	CCONJ
fcis-30522	197	92	f1	f1	NOUN
fcis-30522	197	93	values	value	NOUN
fcis-30522	197	94	reach	reach	VERB
fcis-30522	197	95	their	their	PRON
fcis-30522	197	96	highest	high	ADJ
fcis-30522	197	97	levels	level	NOUN
fcis-30522	197	98	,	,	PUNCT
fcis-30522	197	99	with	with	ADP
fcis-30522	197	100	an	an	DET
fcis-30522	197	101	accuracy	accuracy	NOUN
fcis-30522	197	102	rate	rate	NOUN
fcis-30522	197	103	of	of	ADP
fcis-30522	197	104	94.82	94.82	NUM
fcis-30522	197	105	%	%	NOUN
fcis-30522	197	106	.	.	PUNCT
fcis-30522	198	1	a	a	DET
fcis-30522	198	2	high	high	ADJ
fcis-30522	198	3	accuracy	accuracy	NOUN
fcis-30522	198	4	and	and	CCONJ
fcis-30522	198	5	f1	f1	NOUN
fcis-30522	198	6	value	value	NOUN
fcis-30522	198	7	mean	mean	VERB
fcis-30522	198	8	that	that	SCONJ
fcis-30522	198	9	the	the	DET
fcis-30522	198	10	model	model	NOUN
fcis-30522	198	11	performs	perform	VERB
fcis-30522	198	12	well	well	ADV
fcis-30522	198	13	,	,	PUNCT
fcis-30522	198	14	but	but	CCONJ
fcis-30522	198	15	it	it	PRON
fcis-30522	198	16	is	be	AUX
fcis-30522	198	17	also	also	ADV
fcis-30522	198	18	important	important	ADJ
fcis-30522	198	19	to	to	PART
fcis-30522	198	20	pay	pay	VERB
fcis-30522	198	21	attention	attention	NOUN
fcis-30522	198	22	to	to	ADP
fcis-30522	198	23	overfitting	overfitte	VERB
fcis-30522	198	24	.	.	PUNCT
fcis-30522	199	1	if	if	SCONJ
fcis-30522	199	2	the	the	DET
fcis-30522	199	3	training	training	NOUN
fcis-30522	199	4	loss	loss	NOUN
fcis-30522	199	5	is	be	AUX
fcis-30522	199	6	much	much	ADV
fcis-30522	199	7	lower	low	ADJ
fcis-30522	199	8	than	than	ADP
fcis-30522	199	9	the	the	DET
fcis-30522	199	10	test	test	NOUN
fcis-30522	199	11	loss	loss	NOUN
fcis-30522	199	12	,	,	PUNCT
fcis-30522	199	13	it	it	PRON
fcis-30522	199	14	may	may	AUX
fcis-30522	199	15	indicate	indicate	VERB
fcis-30522	199	16	that	that	SCONJ
fcis-30522	199	17	the	the	DET
fcis-30522	199	18	model	model	NOUN
fcis-30522	199	19	performs	perform	VERB
fcis-30522	199	20	well	well	ADV
fcis-30522	199	21	on	on	ADP
fcis-30522	199	22	the	the	DET
fcis-30522	199	23	training	training	NOUN
fcis-30522	199	24	data	datum	NOUN
fcis-30522	199	25	but	but	CCONJ
fcis-30522	199	26	may	may	AUX
fcis-30522	199	27	not	not	PART
fcis-30522	199	28	generalize	generalize	VERB
fcis-30522	199	29	well	well	ADV
fcis-30522	199	30	to	to	ADP
fcis-30522	199	31	new	new	ADJ
fcis-30522	199	32	,	,	PUNCT
fcis-30522	199	33	unseen	unseen	ADJ
fcis-30522	199	34	data	datum	NOUN
fcis-30522	199	35	.	.	PUNCT
fcis-30522	200	1	the	the	DET
fcis-30522	200	2	bottom	bottom	ADJ
fcis-30522	200	3	half	half	NOUN
fcis-30522	200	4	of	of	ADP
fcis-30522	200	5	(	(	PUNCT
fcis-30522	200	6	fig	fig	NOUN
fcis-30522	200	7	3	3	NUM
fcis-30522	200	8	)	)	PUNCT
fcis-30522	200	9	shows	show	VERB
fcis-30522	200	10	the	the	DET
fcis-30522	200	11	values	value	NOUN
fcis-30522	200	12	of	of	ADP
fcis-30522	200	13	training	training	NOUN
fcis-30522	200	14	loss	loss	NOUN
fcis-30522	200	15	and	and	CCONJ
fcis-30522	200	16	test	test	NOUN
fcis-30522	200	17	loss	loss	NOUN
fcis-30522	200	18	when	when	SCONJ
fcis-30522	200	19	the	the	DET
fcis-30522	200	20	word	word	NOUN
fcis-30522	200	21	length	length	NOUN
fcis-30522	200	22	is	be	AUX
fcis-30522	200	23	235	235	NUM
fcis-30522	200	24	.	.	PUNCT
fcis-30522	200	25	from	from	ADP
fcis-30522	200	26	this	this	DET
fcis-30522	200	27	fig	fig	NOUN
fcis-30522	200	28	,	,	PUNCT
fcis-30522	200	29	it	it	PRON
fcis-30522	200	30	can	can	AUX
fcis-30522	200	31	be	be	AUX
fcis-30522	200	32	seen	see	VERB
fcis-30522	200	33	that	that	SCONJ
fcis-30522	200	34	the	the	DET
fcis-30522	200	35	difference	difference	NOUN
fcis-30522	200	36	between	between	ADP
fcis-30522	200	37	the	the	DET
fcis-30522	200	38	training	training	NOUN
fcis-30522	200	39	loss	loss	NOUN
fcis-30522	200	40	and	and	CCONJ
fcis-30522	200	41	test	test	NOUN
fcis-30522	200	42	loss	loss	NOUN
fcis-30522	200	43	is	be	AUX
fcis-30522	200	44	very	very	ADV
fcis-30522	200	45	small	small	ADJ
fcis-30522	200	46	and	and	CCONJ
fcis-30522	200	47	they	they	PRON
fcis-30522	200	48	essentially	essentially	ADV
fcis-30522	200	49	overlap	overlap	VERB
fcis-30522	200	50	,	,	PUNCT
fcis-30522	200	51	which	which	PRON
fcis-30522	200	52	indicates	indicate	VERB
fcis-30522	200	53	that	that	SCONJ
fcis-30522	200	54	the	the	DET
fcis-30522	200	55	experimental	experimental	ADJ
fcis-30522	200	56	results	result	NOUN
fcis-30522	200	57	are	be	AUX
fcis-30522	200	58	good	good	ADJ
fcis-30522	200	59	and	and	CCONJ
fcis-30522	200	60	there	there	PRON
fcis-30522	200	61	is	be	VERB
fcis-30522	200	62	no	no	DET
fcis-30522	200	63	overfitting	overfitte	VERB
fcis-30522	200	64	problem	problem	NOUN
fcis-30522	200	65	.	.	PUNCT
fcis-30522	201	1	3.2	3.2	NUM
fcis-30522	201	2	.	.	PUNCT
fcis-30522	201	3	model	model	NOUN
fcis-30522	201	4	comparison	comparison	NOUN
fcis-30522	201	5	experiment	experiment	NOUN
fcis-30522	201	6	and	and	CCONJ
fcis-30522	201	7	result	result	VERB
fcis-30522	201	8	analysis	analysis	NOUN
fcis-30522	201	9	the	the	DET
fcis-30522	201	10	experiment	experiment	NOUN
fcis-30522	201	11	aims	aim	VERB
fcis-30522	201	12	first	first	ADV
fcis-30522	201	13	to	to	PART
fcis-30522	201	14	verify	verify	VERB
fcis-30522	201	15	the	the	DET
fcis-30522	201	16	performance	performance	NOUN
fcis-30522	201	17	of	of	ADP
fcis-30522	201	18	the	the	DET
fcis-30522	201	19	basic	basic	ADJ
fcis-30522	201	20	model	model	NOUN
fcis-30522	201	21	macbert	macbert	PROPN
fcis-30522	201	22	-	-	PUNCT
fcis-30522	201	23	sa	sa	NOUN
fcis-30522	201	24	-	-	PUNCT
fcis-30522	201	25	textcnn	textcnn	ADJ
fcis-30522	201	26	-	-	PUNCT
fcis-30522	201	27	bilstm	bilstm	NOUN
fcis-30522	201	28	-	-	PUNCT
fcis-30522	201	29	sa	sa	NOUN
fcis-30522	201	30	constructed	construct	VERB
fcis-30522	201	31	in	in	ADP
fcis-30522	201	32	this	this	DET
fcis-30522	201	33	paper	paper	NOUN
fcis-30522	201	34	,	,	PUNCT
fcis-30522	201	35	and	and	CCONJ
fcis-30522	201	36	second	second	ADV
fcis-30522	201	37	to	to	PART
fcis-30522	201	38	validate	validate	VERB
fcis-30522	201	39	the	the	DET
fcis-30522	201	40	rationale	rationale	NOUN
fcis-30522	201	41	of	of	ADP
fcis-30522	201	42	the	the	DET
fcis-30522	201	43	summary	summary	NOUN
fcis-30522	201	44	text	text	NOUN
fcis-30522	201	45	classification	classification	NOUN
fcis-30522	201	46	model	model	NOUN
fcis-30522	201	47	through	through	ADP
fcis-30522	201	48	comparative	comparative	ADJ
fcis-30522	201	49	experiments	experiment	NOUN
fcis-30522	201	50	.	.	PUNCT
fcis-30522	202	1	table	table	NOUN
fcis-30522	202	2	5	5	NUM
fcis-30522	202	3	presents	present	VERB
fcis-30522	202	4	a	a	DET
fcis-30522	202	5	comparison	comparison	NOUN
fcis-30522	202	6	of	of	ADP
fcis-30522	202	7	the	the	DET
fcis-30522	202	8	results	result	NOUN
fcis-30522	202	9	of	of	ADP
fcis-30522	202	10	each	each	DET
fcis-30522	202	11	evaluated	evaluate	VERB
fcis-30522	202	12	model	model	NOUN
fcis-30522	202	13	index	index	NOUN
fcis-30522	202	14	,	,	PUNCT
fcis-30522	202	15	and	and	CCONJ
fcis-30522	202	16	(	(	PUNCT
fcis-30522	202	17	fig	fig	NOUN
fcis-30522	202	18	4	4	NUM
fcis-30522	202	19	)	)	PUNCT
fcis-30522	202	20	shows	show	VERB
fcis-30522	202	21	a	a	DET
fcis-30522	202	22	comparison	comparison	NOUN
fcis-30522	202	23	chart	chart	NOUN
fcis-30522	202	24	of	of	ADP
fcis-30522	202	25	the	the	DET
fcis-30522	202	26	performance	performance	NOUN
fcis-30522	202	27	of	of	ADP
fcis-30522	202	28	the	the	DET
fcis-30522	202	29	three	three	NUM
fcis-30522	202	30	samples	sample	NOUN
fcis-30522	202	31	under	under	ADP
fcis-30522	202	32	the	the	DET
fcis-30522	202	33	accuracy	accuracy	NOUN
fcis-30522	202	34	metric	metric	NOUN
fcis-30522	202	35	.	.	PUNCT
fcis-30522	203	1	as	as	SCONJ
fcis-30522	203	2	shown	show	VERB
fcis-30522	203	3	in	in	ADP
fcis-30522	203	4	(	(	PUNCT
fcis-30522	203	5	fig	fig	NOUN
fcis-30522	203	6	4	4	NUM
fcis-30522	203	7	)	)	PUNCT
fcis-30522	203	8	,	,	PUNCT
fcis-30522	203	9	comparing	compare	VERB
fcis-30522	203	10	the	the	DET
fcis-30522	203	11	results	result	NOUN
fcis-30522	203	12	of	of	ADP
fcis-30522	203	13	aeda	aeda	NOUN
fcis-30522	203	14	and	and	CCONJ
fcis-30522	203	15	eda	eda	PROPN
fcis-30522	203	16	with	with	ADP
fcis-30522	203	17	the	the	DET
fcis-30522	203	18	original	original	ADJ
fcis-30522	203	19	sample	sample	NOUN
fcis-30522	203	20	data	datum	NOUN
fcis-30522	203	21	reveals	reveal	VERB
fcis-30522	203	22	that	that	SCONJ
fcis-30522	203	23	aeda	aeda	NOUN
fcis-30522	203	24	has	have	VERB
fcis-30522	203	25	a	a	DET
fcis-30522	203	26	greater	great	ADJ
fcis-30522	203	27	effect	effect	NOUN
fcis-30522	203	28	on	on	ADP
fcis-30522	203	29	improving	improve	VERB
fcis-30522	203	30	model	model	NOUN
fcis-30522	203	31	performance	performance	NOUN
fcis-30522	203	32	,	,	PUNCT
fcis-30522	203	33	with	with	ADP
fcis-30522	203	34	improvements	improvement	NOUN
fcis-30522	203	35	exceeding	exceed	VERB
fcis-30522	203	36	10	10	NUM
fcis-30522	203	37	%	%	NOUN
fcis-30522	203	38	in	in	ADP
fcis-30522	203	39	all	all	DET
fcis-30522	203	40	cases	case	NOUN
fcis-30522	203	41	.	.	PUNCT
fcis-30522	204	1	this	this	PRON
fcis-30522	204	2	indicates	indicate	VERB
fcis-30522	204	3	that	that	SCONJ
fcis-30522	204	4	when	when	SCONJ
fcis-30522	204	5	using	use	VERB
fcis-30522	204	6	a	a	DET
fcis-30522	204	7	pre	pre	ADJ
fcis-30522	204	8	-	-	ADJ
fcis-30522	204	9	trained	train	VERB
fcis-30522	204	10	language	language	NOUN
fcis-30522	204	11	model	model	NOUN
fcis-30522	204	12	,	,	PUNCT
fcis-30522	204	13	aeda	aeda	ADJ
fcis-30522	204	14	data	data	NOUN
fcis-30522	204	15	augmentation	augmentation	NOUN
fcis-30522	204	16	technology	technology	NOUN
fcis-30522	204	17	can	can	AUX
fcis-30522	204	18	effectively	effectively	ADV
fcis-30522	204	19	enhance	enhance	VERB
fcis-30522	204	20	model	model	NOUN
fcis-30522	204	21	performance	performance	NOUN
fcis-30522	204	22	.	.	PUNCT
fcis-30522	205	1	a	a	DET
fcis-30522	205	2	comparison	comparison	NOUN
fcis-30522	205	3	of	of	ADP
fcis-30522	205	4	aeda	aeda	NOUN
fcis-30522	205	5	and	and	CCONJ
fcis-30522	205	6	eda	eda	PROPN
fcis-30522	205	7	shows	show	VERB
fcis-30522	205	8	that	that	SCONJ
fcis-30522	205	9	although	although	SCONJ
fcis-30522	205	10	eda	eda	PROPN
fcis-30522	205	11	achieves	achieve	VERB
fcis-30522	205	12	greater	great	ADJ
fcis-30522	205	13	improvement	improvement	NOUN
fcis-30522	205	14	than	than	ADP
fcis-30522	205	15	the	the	DET
fcis-30522	205	16	original	original	ADJ
fcis-30522	205	17	sample	sample	NOUN
fcis-30522	205	18	,	,	PUNCT
fcis-30522	205	19	its	its	PRON
fcis-30522	205	20	effectiveness	effectiveness	NOUN
fcis-30522	205	21	is	be	AUX
fcis-30522	205	22	not	not	PART
fcis-30522	205	23	as	as	ADV
fcis-30522	205	24	pronounced	pronounce	VERB
fcis-30522	205	25	as	as	ADP
fcis-30522	205	26	that	that	PRON
fcis-30522	205	27	of	of	ADP
fcis-30522	205	28	aeda	aeda	NOUN
fcis-30522	205	29	across	across	ADP
fcis-30522	205	30	all	all	DET
fcis-30522	205	31	models	model	NOUN
fcis-30522	205	32	,	,	PUNCT
fcis-30522	205	33	and	and	CCONJ
fcis-30522	205	34	the	the	DET
fcis-30522	205	35	aeda	aeda	ADJ
fcis-30522	205	36	method	method	NOUN
fcis-30522	205	37	does	do	AUX
fcis-30522	205	38	not	not	PART
fcis-30522	205	39	significantly	significantly	ADV
fcis-30522	205	40	impact	impact	VERB
fcis-30522	205	41	the	the	DET
fcis-30522	205	42	semantics	semantic	NOUN
fcis-30522	205	43	of	of	ADP
fcis-30522	205	44	the	the	DET
fcis-30522	205	45	original	original	ADJ
fcis-30522	205	46	text	text	NOUN
fcis-30522	205	47	.	.	PUNCT
fcis-30522	206	1	(	(	PUNCT
fcis-30522	206	2	fig	fig	NOUN
fcis-30522	206	3	5	5	NUM
fcis-30522	206	4	)	)	PUNCT
fcis-30522	206	5	compares	compare	VERB
fcis-30522	206	6	experimental	experimental	ADJ
fcis-30522	206	7	results	result	NOUN
fcis-30522	206	8	.	.	PUNCT
fcis-30522	207	1	the	the	DET
fcis-30522	207	2	aedaenhanced	aedaenhance	VERB
fcis-30522	207	3	data	datum	NOUN
fcis-30522	207	4	in	in	ADP
fcis-30522	207	5	comparison	comparison	NOUN
fcis-30522	207	6	group	group	NOUN
fcis-30522	207	7	3	3	NUM
fcis-30522	207	8	reveals	reveal	VERB
fcis-30522	207	9	a	a	DET
fcis-30522	207	10	significant	significant	ADJ
fcis-30522	207	11	performance	performance	NOUN
fcis-30522	207	12	gap	gap	NOUN
fcis-30522	207	13	between	between	ADP
fcis-30522	207	14	textcnn	textcnn	NOUN
fcis-30522	207	15	and	and	CCONJ
fcis-30522	207	16	bilstm	bilstm	NOUN
fcis-30522	207	17	,	,	PUNCT
fcis-30522	207	18	with	with	ADP
fcis-30522	207	19	textcnn	textcnn	NOUN
fcis-30522	207	20	alone	alone	ADV
fcis-30522	207	21	performing	perform	VERB
fcis-30522	207	22	worst	bad	ADJ
fcis-30522	207	23	overall	overall	ADJ
fcis-30522	207	24	.	.	PUNCT
fcis-30522	208	1	this	this	PRON
fcis-30522	208	2	occurs	occur	VERB
fcis-30522	208	3	because	because	SCONJ
fcis-30522	208	4	literature	literature	NOUN
fcis-30522	208	5	abstracts	abstract	NOUN
fcis-30522	208	6	contain	contain	VERB
fcis-30522	208	7	complex	complex	ADJ
fcis-30522	208	8	semantics	semantic	NOUN
fcis-30522	208	9	and	and	CCONJ
fcis-30522	208	10	strong	strong	ADJ
fcis-30522	208	11	contextual	contextual	ADJ
fcis-30522	208	12	connections	connection	NOUN
fcis-30522	208	13	.	.	PUNCT
fcis-30522	209	1	while	while	SCONJ
fcis-30522	209	2	textcnn	textcnn	PROPN
fcis-30522	209	3	effectively	effectively	ADV
fcis-30522	209	4	captures	capture	VERB
fcis-30522	209	5	local	local	ADJ
fcis-30522	209	6	features	feature	NOUN
fcis-30522	209	7	,	,	PUNCT
fcis-30522	209	8	it	it	PRON
fcis-30522	209	9	struggles	struggle	VERB
fcis-30522	209	10	with	with	ADP
fcis-30522	209	11	long	long	ADJ
fcis-30522	209	12	-	-	PUNCT
fcis-30522	209	13	distance	distance	NOUN
fcis-30522	209	14	information	information	NOUN
fcis-30522	209	15	and	and	CCONJ
fcis-30522	209	16	contextual	contextual	ADJ
fcis-30522	209	17	connections	connection	NOUN
fcis-30522	209	18	,	,	PUNCT
fcis-30522	209	19	leading	lead	VERB
fcis-30522	209	20	to	to	ADP
fcis-30522	209	21	insufficient	insufficient	ADJ
fcis-30522	209	22	information	information	NOUN
fcis-30522	209	23	capture	capture	NOUN
fcis-30522	209	24	.	.	PUNCT
fcis-30522	210	1	comparison	comparison	NOUN
fcis-30522	210	2	group	group	NOUN
fcis-30522	210	3	4	4	NUM
fcis-30522	210	4	shows	show	VERB
fcis-30522	210	5	that	that	SCONJ
fcis-30522	210	6	combining	combine	VERB
fcis-30522	210	7	textcnn	textcnn	NOUN
fcis-30522	210	8	with	with	ADP
fcis-30522	210	9	bilstm	bilstm	NOUN
fcis-30522	210	10	compensates	compensate	NOUN
fcis-30522	210	11	for	for	ADP
fcis-30522	210	12	bilstm	bilstm	NOUN
fcis-30522	210	13	's	's	PART
fcis-30522	210	14	weakness	weakness	NOUN
fcis-30522	210	15	in	in	ADP
fcis-30522	210	16	capturing	capture	VERB
fcis-30522	210	17	local	local	ADJ
fcis-30522	210	18	features	feature	NOUN
fcis-30522	210	19	,	,	PUNCT
fcis-30522	210	20	improving	improve	VERB
fcis-30522	210	21	overall	overall	ADJ
fcis-30522	210	22	performance	performance	NOUN
fcis-30522	210	23	by	by	ADP
fcis-30522	210	24	0.98	0.98	NUM
fcis-30522	210	25	%	%	NOUN
fcis-30522	210	26	.	.	PUNCT
fcis-30522	211	1	75	75	NUM
fcis-30522	211	2	table	table	NOUN
fcis-30522	211	3	5	5	NUM
fcis-30522	211	4	.	.	PUNCT
fcis-30522	211	5	indicator	indicator	NOUN
fcis-30522	211	6	results	result	VERB
fcis-30522	211	7	for	for	ADP
fcis-30522	211	8	the	the	DET
fcis-30522	211	9	base	base	NOUN
fcis-30522	211	10	model	model	NOUN
fcis-30522	211	11	and	and	CCONJ
fcis-30522	211	12	each	each	PRON
fcis-30522	211	13	of	of	ADP
fcis-30522	211	14	the	the	DET
fcis-30522	211	15	comparison	comparison	NOUN
fcis-30522	211	16	models	model	NOUN
fcis-30522	211	17	models	model	VERB
fcis-30522	211	18	type	type	NOUN
fcis-30522	211	19	accuracy	accuracy	NOUN
fcis-30522	211	20	f1	f1	PROPN
fcis-30522	211	21	macbert	macbert	PROPN
fcis-30522	211	22	-	-	PUNCT
fcis-30522	211	23	sa	sa	NOUN
fcis-30522	211	24	-	-	PUNCT
fcis-30522	211	25	textcnn	textcnn	ADJ
fcis-30522	211	26	-	-	PUNCT
fcis-30522	211	27	bilstm	bilstm	NOUN
fcis-30522	211	28	-	-	PUNCT
fcis-30522	211	29	sa	sa	NOUN
fcis-30522	211	30	aeda	aeda	NOUN
fcis-30522	211	31	94.82	94.82	NUM
fcis-30522	211	32	%	%	NOUN
fcis-30522	211	33	94.80	94.80	NUM
fcis-30522	211	34	%	%	NOUN
fcis-30522	211	35	eda	eda	PROPN
fcis-30522	211	36	92.14	92.14	NUM
fcis-30522	211	37	%	%	NOUN
fcis-30522	211	38	92.13	92.13	NUM
fcis-30522	211	39	%	%	NOUN
fcis-30522	211	40	original	original	ADJ
fcis-30522	211	41	sample	sample	NOUN
fcis-30522	211	42	81.43	81.43	NUM
fcis-30522	211	43	%	%	NOUN
fcis-30522	211	44	80.70	80.70	NUM
fcis-30522	211	45	%	%	NOUN
fcis-30522	211	46	bert	bert	PROPN
fcis-30522	211	47	-	-	PUNCT
fcis-30522	211	48	sa	sa	NOUN
fcis-30522	211	49	-	-	PUNCT
fcis-30522	211	50	bilstm	bilstm	NOUN
fcis-30522	211	51	-	-	PUNCT
fcis-30522	211	52	textcnn	textcnn	NOUN
fcis-30522	211	53	-	-	PUNCT
fcis-30522	211	54	sa	sa	PROPN
fcis-30522	211	55	(	(	PUNCT
fcis-30522	211	56	bert	bert	PROPN
fcis-30522	211	57	-	-	PUNCT
fcis-30522	211	58	chinese	chinese	ADJ
fcis-30522	211	59	)	)	PUNCT
fcis-30522	211	60	aeda	aeda	NOUN
fcis-30522	211	61	94.46	94.46	NUM
fcis-30522	211	62	%	%	NOUN
fcis-30522	211	63	94.78	94.78	NUM
fcis-30522	211	64	%	%	NOUN
fcis-30522	211	65	eda	eda	PROPN
fcis-30522	211	66	90.82	90.82	NUM
fcis-30522	211	67	%	%	NOUN
fcis-30522	211	68	90.82	90.82	NUM
fcis-30522	211	69	%	%	NOUN
fcis-30522	211	70	original	original	ADJ
fcis-30522	211	71	sample	sample	NOUN
fcis-30522	211	72	80.61	80.61	NUM
fcis-30522	211	73	%	%	NOUN
fcis-30522	211	74	79.05	79.05	NUM
fcis-30522	211	75	%	%	NOUN
fcis-30522	211	76	macbert	macbert	NOUN
fcis-30522	211	77	-	-	PUNCT
fcis-30522	211	78	textcnn	textcnn	ADJ
fcis-30522	211	79	-	-	PUNCT
fcis-30522	211	80	bilstm	bilstm	NOUN
fcis-30522	211	81	aeda	aeda	NOUN
fcis-30522	211	82	93.58	93.58	NUM
fcis-30522	211	83	%	%	NOUN
fcis-30522	211	84	93.12	93.12	NUM
fcis-30522	211	85	%	%	NOUN
fcis-30522	211	86	eda	eda	PROPN
fcis-30522	211	87	90.64	90.64	NUM
fcis-30522	211	88	%	%	NOUN
fcis-30522	211	89	90.59	90.59	NUM
fcis-30522	211	90	%	%	NOUN
fcis-30522	211	91	original	original	ADJ
fcis-30522	211	92	sample	sample	NOUN
fcis-30522	211	93	78.42	78.42	NUM
fcis-30522	211	94	%	%	NOUN
fcis-30522	211	95	78.12	78.12	NUM
fcis-30522	211	96	%	%	NOUN
fcis-30522	211	97	macbert	macbert	NOUN
fcis-30522	211	98	-	-	PUNCT
fcis-30522	211	99	textcnn	textcnn	VERB
fcis-30522	211	100	aeda	aeda	NOUN
fcis-30522	211	101	85.73	85.73	NUM
fcis-30522	211	102	%	%	NOUN
fcis-30522	211	103	83.46	83.46	NUM
fcis-30522	211	104	%	%	NOUN
fcis-30522	211	105	eda	eda	PROPN
fcis-30522	211	106	78.60	78.60	NUM
fcis-30522	211	107	%	%	NOUN
fcis-30522	211	108	71.39	71.39	NUM
fcis-30522	211	109	%	%	NOUN
fcis-30522	211	110	original	original	ADJ
fcis-30522	211	111	sample	sample	NOUN
fcis-30522	211	112	75.42	75.42	NUM
fcis-30522	211	113	%	%	NOUN
fcis-30522	211	114	74.05	74.05	NUM
fcis-30522	211	115	%	%	NOUN
fcis-30522	211	116	macbert	macbert	NOUN
fcis-30522	211	117	-	-	PUNCT
fcis-30522	211	118	bilstm	bilstm	NOUN
fcis-30522	211	119	aeda	aeda	NOUN
fcis-30522	211	120	92.60	92.60	NUM
fcis-30522	211	121	%	%	NOUN
fcis-30522	211	122	92.63	92.63	NUM
fcis-30522	211	123	%	%	NOUN
fcis-30522	211	124	eda	eda	PROPN
fcis-30522	211	125	89.88	89.88	NUM
fcis-30522	211	126	%	%	NOUN
fcis-30522	211	127	89.85	89.85	NUM
fcis-30522	211	128	%	%	NOUN
fcis-30522	211	129	original	original	ADJ
fcis-30522	211	130	sample	sample	NOUN
fcis-30522	211	131	79.59	79.59	NUM
fcis-30522	211	132	%	%	NOUN
fcis-30522	211	133	80.61	80.61	NUM
fcis-30522	211	134	%	%	NOUN
fcis-30522	211	135	macbert	macbert	NOUN
fcis-30522	211	136	-	-	PUNCT
fcis-30522	211	137	textcnn	textcnn	ADJ
fcis-30522	211	138	-	-	PUNCT
fcis-30522	211	139	bilstm	bilstm	NOUN
fcis-30522	211	140	-	-	PUNCT
fcis-30522	211	141	sa	sa	NOUN
fcis-30522	211	142	aeda	aeda	NOUN
fcis-30522	211	143	94.11	94.11	NUM
fcis-30522	211	144	%	%	NOUN
fcis-30522	211	145	94.32	94.32	NUM
fcis-30522	211	146	%	%	NOUN
fcis-30522	211	147	eda	eda	PROPN
fcis-30522	211	148	91.57	91.57	NUM
fcis-30522	211	149	%	%	NOUN
fcis-30522	211	150	91.52	91.52	NUM
fcis-30522	211	151	%	%	NOUN
fcis-30522	211	152	original	original	ADJ
fcis-30522	211	153	sample	sample	NOUN
fcis-30522	211	154	81.46	81.46	NUM
fcis-30522	211	155	%	%	NOUN
fcis-30522	211	156	80.59	80.59	NUM
fcis-30522	211	157	%	%	NOUN
fcis-30522	211	158	fig	fig	NOUN
fcis-30522	211	159	4	4	NUM
fcis-30522	211	160	.	.	PUNCT
fcis-30522	211	161	comparison	comparison	NOUN
fcis-30522	211	162	of	of	ADP
fcis-30522	211	163	model	model	NOUN
fcis-30522	211	164	performance	performance	NOUN
fcis-30522	211	165	under	under	ADP
fcis-30522	211	166	three	three	NUM
fcis-30522	211	167	sample	sample	NOUN
fcis-30522	211	168	conditions	condition	NOUN
fcis-30522	211	169	in	in	ADP
fcis-30522	211	170	comparison	comparison	NOUN
fcis-30522	211	171	group	group	NOUN
fcis-30522	211	172	5	5	NUM
fcis-30522	211	173	,	,	PUNCT
fcis-30522	211	174	adding	add	VERB
fcis-30522	211	175	a	a	DET
fcis-30522	211	176	self	self	NOUN
fcis-30522	211	177	-	-	PUNCT
fcis-30522	211	178	attention	attention	NOUN
fcis-30522	211	179	mechanism	mechanism	NOUN
fcis-30522	211	180	at	at	ADP
fcis-30522	211	181	the	the	DET
fcis-30522	211	182	end	end	NOUN
fcis-30522	211	183	of	of	ADP
fcis-30522	211	184	macbert	macbert	NOUN
fcis-30522	211	185	-	-	PUNCT
fcis-30522	211	186	textcnn	textcnn	NOUN
fcis-30522	211	187	-	-	PUNCT
fcis-30522	211	188	bilstm	bilstm	NOUN
fcis-30522	211	189	to	to	ADP
fcis-30522	211	190	weight	weight	NOUN
fcis-30522	211	191	combined	combine	VERB
fcis-30522	211	192	feature	feature	NOUN
fcis-30522	211	193	representations	representation	NOUN
fcis-30522	211	194	before	before	SCONJ
fcis-30522	211	195	classification	classification	NOUN
fcis-30522	211	196	significantly	significantly	ADV
fcis-30522	211	197	enhances	enhance	VERB
fcis-30522	211	198	performance	performance	NOUN
fcis-30522	211	199	by	by	ADP
fcis-30522	211	200	0.53	0.53	NUM
fcis-30522	211	201	%	%	NOUN
fcis-30522	211	202	.	.	PUNCT
fcis-30522	212	1	comparison	comparison	NOUN
fcis-30522	212	2	group	group	NOUN
fcis-30522	212	3	2	2	NUM
fcis-30522	212	4	demonstrates	demonstrate	VERB
fcis-30522	212	5	that	that	SCONJ
fcis-30522	212	6	adding	add	VERB
fcis-30522	212	7	a	a	DET
fcis-30522	212	8	self	self	NOUN
fcis-30522	212	9	-	-	PUNCT
fcis-30522	212	10	attention	attention	NOUN
fcis-30522	212	11	mechanism	mechanism	NOUN
fcis-30522	212	12	at	at	ADP
fcis-30522	212	13	the	the	DET
fcis-30522	212	14	end	end	NOUN
fcis-30522	212	15	of	of	ADP
fcis-30522	212	16	the	the	DET
fcis-30522	212	17	pre	pre	ADJ
fcis-30522	212	18	-	-	ADJ
fcis-30522	212	19	trained	train	VERB
fcis-30522	212	20	language	language	NOUN
fcis-30522	212	21	model	model	NOUN
fcis-30522	212	22	improves	improve	VERB
fcis-30522	212	23	performance	performance	NOUN
fcis-30522	212	24	by	by	ADP
fcis-30522	212	25	0.71	0.71	NUM
fcis-30522	212	26	%	%	NOUN
fcis-30522	212	27	by	by	ADP
fcis-30522	212	28	enabling	enable	VERB
fcis-30522	212	29	better	well	ADJ
fcis-30522	212	30	capture	capture	NOUN
fcis-30522	212	31	of	of	ADP
fcis-30522	212	32	longdistance	longdistance	NOUN
fcis-30522	212	33	dependencies	dependency	NOUN
fcis-30522	212	34	.	.	PUNCT
fcis-30522	213	1	comparison	comparison	NOUN
fcis-30522	213	2	group	group	NOUN
fcis-30522	213	3	1	1	NUM
fcis-30522	213	4	shows	show	VERB
fcis-30522	213	5	macbert	macbert	NOUN
fcis-30522	213	6	outperforms	outperform	VERB
fcis-30522	213	7	traditional	traditional	ADJ
fcis-30522	213	8	bert	bert	NOUN
fcis-30522	213	9	in	in	ADP
fcis-30522	213	10	processing	processing	NOUN
fcis-30522	213	11	summary	summary	NOUN
fcis-30522	213	12	texts	text	NOUN
fcis-30522	213	13	by	by	ADP
fcis-30522	213	14	0.36	0.36	NUM
fcis-30522	213	15	%	%	NOUN
fcis-30522	213	16	.	.	PUNCT
fcis-30522	214	1	the	the	DET
fcis-30522	214	2	macbert	macbert	PROPN
fcis-30522	214	3	-	-	PUNCT
fcis-30522	214	4	sa	sa	NOUN
fcis-30522	214	5	-	-	PUNCT
fcis-30522	214	6	textcnnbilstm	textcnnbilstm	VERB
fcis-30522	214	7	-	-	PUNCT
fcis-30522	214	8	sa	sa	PROPN
fcis-30522	214	9	model	model	NOUN
fcis-30522	214	10	achieves	achieve	VERB
fcis-30522	214	11	94.82	94.82	NUM
fcis-30522	214	12	%	%	NOUN
fcis-30522	214	13	accuracy	accuracy	NOUN
fcis-30522	214	14	,	,	PUNCT
fcis-30522	214	15	outperforming	outperform	VERB
fcis-30522	214	16	all	all	DET
fcis-30522	214	17	comparison	comparison	NOUN
fcis-30522	214	18	models	model	NOUN
fcis-30522	214	19	.	.	PUNCT
fcis-30522	215	1	the	the	DET
fcis-30522	215	2	confusion	confusion	NOUN
fcis-30522	215	3	matrix	matrix	NOUN
fcis-30522	215	4	(	(	PUNCT
fcis-30522	215	5	fig	fig	NOUN
fcis-30522	215	6	6	6	NUM
fcis-30522	215	7	)	)	PUNCT
fcis-30522	215	8	shows	show	VERB
fcis-30522	215	9	few	few	ADJ
fcis-30522	215	10	incorrect	incorrect	ADJ
fcis-30522	215	11	predictions	prediction	NOUN
fcis-30522	215	12	with	with	ADP
fcis-30522	215	13	no	no	DET
fcis-30522	215	14	instances	instance	NOUN
fcis-30522	215	15	of	of	ADP
fcis-30522	215	16	completely	completely	ADV
fcis-30522	215	17	wrong	wrong	ADJ
fcis-30522	215	18	category	category	NOUN
fcis-30522	215	19	predictions	prediction	NOUN
fcis-30522	215	20	.	.	PUNCT
fcis-30522	216	1	by	by	ADP
fcis-30522	216	2	introducing	introduce	VERB
fcis-30522	216	3	multi	multi	ADJ
fcis-30522	216	4	-	-	ADJ
fcis-30522	216	5	layer	layer	ADJ
fcis-30522	216	6	dropout	dropout	NOUN
fcis-30522	216	7	and	and	CCONJ
fcis-30522	216	8	optimizing	optimize	VERB
fcis-30522	216	9	feature	feature	NOUN
fcis-30522	216	10	fusion	fusion	NOUN
fcis-30522	216	11	,	,	PUNCT
fcis-30522	216	12	we	we	PRON
fcis-30522	216	13	developed	develop	VERB
fcis-30522	216	14	the	the	DET
fcis-30522	216	15	mbcresattnet	mbcresattnet	NOUN
fcis-30522	216	16	model	model	NOUN
fcis-30522	216	17	for	for	ADP
fcis-30522	216	18	small	small	ADJ
fcis-30522	216	19	-	-	PUNCT
fcis-30522	216	20	sample	sample	NOUN
fcis-30522	216	21	text	text	NOUN
fcis-30522	216	22	summary	summary	NOUN
fcis-30522	216	23	classification	classification	NOUN
fcis-30522	216	24	.	.	PUNCT
fcis-30522	217	1	table	table	NOUN
fcis-30522	217	2	8	8	NUM
fcis-30522	217	3	presents	present	VERB
fcis-30522	217	4	mbcresattnet	mbcresattnet	NOUN
fcis-30522	217	5	's	's	PART
fcis-30522	217	6	performance	performance	NOUN
fcis-30522	217	7	compared	compare	VERB
fcis-30522	217	8	to	to	ADP
fcis-30522	217	9	macbert	macbert	VERB
fcis-30522	217	10	-	-	PUNCT
fcis-30522	217	11	sa	sa	NOUN
fcis-30522	217	12	-	-	PUNCT
fcis-30522	217	13	textcnnbilstm	textcnnbilstm	NOUN
fcis-30522	217	14	-	-	PUNCT
fcis-30522	217	15	sa	sa	NOUN
fcis-30522	217	16	across	across	ADP
fcis-30522	217	17	three	three	NUM
fcis-30522	217	18	samples	sample	NOUN
fcis-30522	217	19	.	.	PUNCT
fcis-30522	218	1	(	(	PUNCT
fcis-30522	218	2	fig	fig	NOUN
fcis-30522	218	3	7	7	NUM
fcis-30522	218	4	)	)	PUNCT
fcis-30522	218	5	shows	show	VERB
fcis-30522	218	6	the	the	DET
fcis-30522	218	7	training	training	NOUN
fcis-30522	218	8	metrics	metric	NOUN
fcis-30522	218	9	of	of	ADP
fcis-30522	218	10	the	the	DET
fcis-30522	218	11	mbcresattnet	mbcresattnet	NOUN
fcis-30522	218	12	model	model	NOUN
fcis-30522	218	13	.	.	PUNCT
fcis-30522	219	1	the	the	DET
fcis-30522	219	2	top	top	ADJ
fcis-30522	219	3	half	half	NOUN
fcis-30522	219	4	is	be	AUX
fcis-30522	219	5	a	a	DET
fcis-30522	219	6	locally	locally	ADV
fcis-30522	219	7	zoomed	zoom	VERB
fcis-30522	219	8	-	-	PUNCT
fcis-30522	219	9	in	in	ADP
fcis-30522	219	10	plot	plot	NOUN
fcis-30522	219	11	of	of	ADP
fcis-30522	219	12	the	the	DET
fcis-30522	219	13	accuracy	accuracy	NOUN
fcis-30522	219	14	and	and	CCONJ
fcis-30522	219	15	f1	f1	NOUN
fcis-30522	219	16	values	value	NOUN
fcis-30522	219	17	,	,	PUNCT
fcis-30522	219	18	and	and	CCONJ
fcis-30522	219	19	the	the	DET
fcis-30522	219	20	bottom	bottom	ADJ
fcis-30522	219	21	half	half	NOUN
fcis-30522	219	22	is	be	AUX
fcis-30522	219	23	a	a	DET
fcis-30522	219	24	line	line	NOUN
fcis-30522	219	25	graph	graph	NOUN
fcis-30522	219	26	of	of	ADP
fcis-30522	219	27	the	the	DET
fcis-30522	219	28	complete	complete	ADJ
fcis-30522	219	29	training	training	NOUN
fcis-30522	219	30	data	datum	NOUN
fcis-30522	219	31	containing	contain	VERB
fcis-30522	219	32	the	the	DET
fcis-30522	219	33	training	training	NOUN
fcis-30522	219	34	loss	loss	NOUN
fcis-30522	219	35	and	and	CCONJ
fcis-30522	219	36	test	test	NOUN
fcis-30522	219	37	loss	loss	NOUN
fcis-30522	219	38	.	.	PUNCT
fcis-30522	220	1	this	this	DET
fcis-30522	220	2	fig	fig	NOUN
fcis-30522	220	3	shows	show	VERB
fcis-30522	220	4	that	that	SCONJ
fcis-30522	220	5	the	the	DET
fcis-30522	220	6	difference	difference	NOUN
fcis-30522	220	7	between	between	ADP
fcis-30522	220	8	the	the	DET
fcis-30522	220	9	training	training	NOUN
fcis-30522	220	10	loss	loss	NOUN
fcis-30522	220	11	and	and	CCONJ
fcis-30522	220	12	test	test	NOUN
fcis-30522	220	13	loss	loss	NOUN
fcis-30522	220	14	is	be	AUX
fcis-30522	220	15	very	very	ADV
fcis-30522	220	16	small	small	ADJ
fcis-30522	220	17	and	and	CCONJ
fcis-30522	220	18	essentially	essentially	ADV
fcis-30522	220	19	overlaps	overlap	VERB
fcis-30522	220	20	,	,	PUNCT
fcis-30522	220	21	indicating	indicate	VERB
fcis-30522	220	22	that	that	SCONJ
fcis-30522	220	23	the	the	DET
fcis-30522	220	24	model	model	NOUN
fcis-30522	220	25	has	have	VERB
fcis-30522	220	26	good	good	ADJ
fcis-30522	220	27	performance	performance	NOUN
fcis-30522	220	28	and	and	CCONJ
fcis-30522	220	29	76	76	NUM
fcis-30522	220	30	does	do	AUX
fcis-30522	220	31	not	not	PART
fcis-30522	220	32	have	have	VERB
fcis-30522	220	33	overfitting	overfitte	VERB
fcis-30522	220	34	problems	problem	NOUN
fcis-30522	220	35	.	.	PUNCT
fcis-30522	221	1	fig	fig	NOUN
fcis-30522	221	2	5	5	NUM
fcis-30522	221	3	.	.	PUNCT
fcis-30522	221	4	comparison	comparison	NOUN
fcis-30522	221	5	of	of	ADP
fcis-30522	221	6	models	model	NOUN
fcis-30522	221	7	fig	fig	VERB
fcis-30522	221	8	6	6	NUM
fcis-30522	221	9	.	.	PUNCT
fcis-30522	222	1	macbert	macbert	PROPN
fcis-30522	222	2	-	-	PUNCT
fcis-30522	222	3	sa	sa	NOUN
fcis-30522	222	4	-	-	PUNCT
fcis-30522	222	5	textcnn	textcnn	ADJ
fcis-30522	222	6	-	-	PUNCT
fcis-30522	222	7	bilstm	bilstm	NOUN
fcis-30522	222	8	-	-	PUNCT
fcis-30522	222	9	sa	sa	NOUN
fcis-30522	222	10	confusion	confusion	NOUN
fcis-30522	222	11	matrix	matrix	NOUN
fcis-30522	222	12	table	table	NOUN
fcis-30522	222	13	6	6	NUM
fcis-30522	222	14	.	.	PUNCT
fcis-30522	223	1	mbcresattnet	mbcresattnet	NOUN
fcis-30522	223	2	different	different	ADJ
fcis-30522	223	3	sample	sample	NOUN
fcis-30522	223	4	performance	performance	NOUN
fcis-30522	223	5	test	test	NOUN
fcis-30522	223	6	results	result	NOUN
fcis-30522	223	7	models	model	NOUN
fcis-30522	223	8	type	type	NOUN
fcis-30522	223	9	accuracy	accuracy	NOUN
fcis-30522	223	10	f1	f1	NOUN
fcis-30522	223	11	mbcresattnet	mbcresattnet	NOUN
fcis-30522	223	12	aeda	aeda	NOUN
fcis-30522	223	13	96.17	96.17	NUM
fcis-30522	223	14	%	%	NOUN
fcis-30522	223	15	96.15	96.15	NUM
fcis-30522	223	16	%	%	NOUN
fcis-30522	223	17	eda	eda	PROPN
fcis-30522	223	18	93.19	93.19	NUM
fcis-30522	223	19	%	%	NOUN
fcis-30522	223	20	93.18	93.18	NUM
fcis-30522	223	21	%	%	NOUN
fcis-30522	223	22	original	original	ADJ
fcis-30522	223	23	sample	sample	NOUN
fcis-30522	223	24	82.76	82.76	NUM
fcis-30522	223	25	%	%	NOUN
fcis-30522	223	26	81.20	81.20	NUM
fcis-30522	223	27	%	%	NOUN
fcis-30522	223	28	macbert	macbert	NOUN
fcis-30522	223	29	-	-	PUNCT
fcis-30522	223	30	sa	sa	NOUN
fcis-30522	223	31	-	-	PUNCT
fcis-30522	223	32	textcnn	textcnn	ADJ
fcis-30522	223	33	-	-	PUNCT
fcis-30522	223	34	bilstm	bilstm	NOUN
fcis-30522	223	35	-	-	PUNCT
fcis-30522	223	36	sa	sa	NOUN
fcis-30522	223	37	aeda	aeda	NOUN
fcis-30522	223	38	94.82	94.82	NUM
fcis-30522	223	39	%	%	NOUN
fcis-30522	223	40	94.80	94.80	NUM
fcis-30522	223	41	%	%	NOUN
fcis-30522	223	42	eda	eda	PROPN
fcis-30522	223	43	92.14	92.14	NUM
fcis-30522	223	44	%	%	NOUN
fcis-30522	223	45	92.13	92.13	NUM
fcis-30522	223	46	%	%	NOUN
fcis-30522	223	47	original	original	ADJ
fcis-30522	223	48	sample	sample	NOUN
fcis-30522	223	49	81.46	81.46	NUM
fcis-30522	223	50	%	%	NOUN
fcis-30522	223	51	80.59	80.59	NUM
fcis-30522	223	52	%	%	NOUN
fcis-30522	223	53	77	77	NUM
fcis-30522	223	54	fig	fig	NOUN
fcis-30522	223	55	7	7	NUM
fcis-30522	223	56	.	.	PUNCT
fcis-30522	224	1	mbcresattnet	mbcresattnet	NOUN
fcis-30522	224	2	model	model	NOUN
fcis-30522	224	3	training	training	NOUN
fcis-30522	224	4	indicator	indicator	PROPN
fcis-30522	224	5	chart	chart	PROPN
fcis-30522	224	6	(	(	PUNCT
fcis-30522	224	7	fig	fig	NOUN
fcis-30522	224	8	8)	8)	NUM
fcis-30522	224	9	illustrates	illustrate	VERB
fcis-30522	224	10	the	the	DET
fcis-30522	224	11	results	result	NOUN
fcis-30522	224	12	of	of	ADP
fcis-30522	224	13	the	the	DET
fcis-30522	224	14	mbcresattnet	mbcresattnet	NOUN
fcis-30522	224	15	model	model	NOUN
fcis-30522	224	16	.	.	PUNCT
fcis-30522	225	1	the	the	DET
fcis-30522	225	2	left	left	ADJ
fcis-30522	225	3	side	side	NOUN
fcis-30522	225	4	shows	show	VERB
fcis-30522	225	5	the	the	DET
fcis-30522	225	6	corresponding	corresponding	ADJ
fcis-30522	225	7	confusion	confusion	NOUN
fcis-30522	225	8	matrix	matrix	NOUN
fcis-30522	225	9	,	,	PUNCT
fcis-30522	225	10	from	from	ADP
fcis-30522	225	11	which	which	PRON
fcis-30522	225	12	it	it	PRON
fcis-30522	225	13	can	can	AUX
fcis-30522	225	14	be	be	AUX
fcis-30522	225	15	seen	see	VERB
fcis-30522	225	16	that	that	SCONJ
fcis-30522	225	17	the	the	DET
fcis-30522	225	18	model	model	NOUN
fcis-30522	225	19	predicts	predict	VERB
fcis-30522	225	20	few	few	ADJ
fcis-30522	225	21	labels	label	NOUN
fcis-30522	225	22	incorrectly	incorrectly	ADV
fcis-30522	225	23	,	,	PUNCT
fcis-30522	225	24	indicating	indicate	VERB
fcis-30522	225	25	high	high	ADJ
fcis-30522	225	26	overall	overall	ADJ
fcis-30522	225	27	accuracy	accuracy	NOUN
fcis-30522	225	28	on	on	ADP
fcis-30522	225	29	the	the	DET
fcis-30522	225	30	classification	classification	NOUN
fcis-30522	225	31	task	task	NOUN
fcis-30522	225	32	.	.	PUNCT
fcis-30522	226	1	the	the	DET
fcis-30522	226	2	right	right	ADJ
fcis-30522	226	3	side	side	NOUN
fcis-30522	226	4	displays	display	VERB
fcis-30522	226	5	the	the	DET
fcis-30522	226	6	performance	performance	NOUN
fcis-30522	226	7	comparison	comparison	NOUN
fcis-30522	226	8	under	under	ADP
fcis-30522	226	9	different	different	ADJ
fcis-30522	226	10	samples	sample	NOUN
fcis-30522	226	11	,	,	PUNCT
fcis-30522	226	12	revealing	reveal	VERB
fcis-30522	226	13	that	that	SCONJ
fcis-30522	226	14	across	across	ADP
fcis-30522	226	15	the	the	DET
fcis-30522	226	16	three	three	NUM
fcis-30522	226	17	types	type	NOUN
fcis-30522	226	18	of	of	ADP
fcis-30522	226	19	samples	sample	NOUN
fcis-30522	226	20	,	,	PUNCT
fcis-30522	226	21	retaining	retain	VERB
fcis-30522	226	22	the	the	DET
fcis-30522	226	23	original	original	ADJ
fcis-30522	226	24	features	feature	NOUN
fcis-30522	226	25	by	by	ADP
fcis-30522	226	26	residual	residual	ADJ
fcis-30522	226	27	linkage	linkage	NOUN
fcis-30522	226	28	with	with	ADP
fcis-30522	226	29	the	the	DET
fcis-30522	226	30	weighted	weight	VERB
fcis-30522	226	31	features	feature	NOUN
fcis-30522	226	32	can	can	AUX
fcis-30522	226	33	effectively	effectively	ADV
fcis-30522	226	34	improve	improve	VERB
fcis-30522	226	35	the	the	DET
fcis-30522	226	36	model	model	NOUN
fcis-30522	226	37	's	's	PART
fcis-30522	226	38	performance	performance	NOUN
fcis-30522	226	39	by	by	ADP
fcis-30522	226	40	1.35	1.35	NUM
fcis-30522	226	41	%	%	NOUN
fcis-30522	226	42	under	under	ADP
fcis-30522	226	43	the	the	DET
fcis-30522	226	44	aeda	aeda	ADJ
fcis-30522	226	45	data	data	NOUN
fcis-30522	226	46	augmentation	augmentation	NOUN
fcis-30522	226	47	technique	technique	NOUN
fcis-30522	226	48	.	.	PUNCT
fcis-30522	227	1	fig	fig	NOUN
fcis-30522	227	2	8	8	NUM
fcis-30522	227	3	.	.	PUNCT
fcis-30522	228	1	comparison	comparison	NOUN
fcis-30522	228	2	of	of	ADP
fcis-30522	228	3	mbcresattnet	mbcresattnet	NOUN
fcis-30522	228	4	performance	performance	NOUN
fcis-30522	228	5	under	under	ADP
fcis-30522	228	6	three	three	NUM
fcis-30522	228	7	sample	sample	NOUN
fcis-30522	228	8	conditions	condition	NOUN
fcis-30522	228	9	and	and	CCONJ
fcis-30522	228	10	confusion	confusion	NOUN
fcis-30522	228	11	matrix	matrix	NOUN
fcis-30522	228	12	diagram	diagram	NOUN
fcis-30522	228	13	4	4	NUM
fcis-30522	228	14	.	.	PUNCT
fcis-30522	229	1	conclusion	conclusion	NOUN
fcis-30522	229	2	abstract	abstract	ADJ
fcis-30522	229	3	texts	text	NOUN
fcis-30522	229	4	with	with	ADP
fcis-30522	229	5	complex	complex	ADJ
fcis-30522	229	6	linguistic	linguistic	ADJ
fcis-30522	229	7	features	feature	NOUN
fcis-30522	229	8	often	often	ADV
fcis-30522	229	9	demand	demand	VERB
fcis-30522	229	10	sophisticated	sophisticated	ADJ
fcis-30522	229	11	models	model	NOUN
fcis-30522	229	12	and	and	CCONJ
fcis-30522	229	13	extensive	extensive	ADJ
fcis-30522	229	14	training	training	NOUN
fcis-30522	229	15	data	datum	NOUN
fcis-30522	229	16	.	.	PUNCT
fcis-30522	230	1	many	many	ADJ
fcis-30522	230	2	researchers	researcher	NOUN
fcis-30522	230	3	utilize	utilize	VERB
fcis-30522	230	4	pre	pre	ADJ
fcis-30522	230	5	-	-	ADJ
fcis-30522	230	6	trained	train	VERB
fcis-30522	230	7	language	language	NOUN
fcis-30522	230	8	models	model	NOUN
fcis-30522	230	9	like	like	SCONJ
fcis-30522	230	10	bert	bert	NOUN
fcis-30522	230	11	to	to	PART
fcis-30522	230	12	improve	improve	VERB
fcis-30522	230	13	performance	performance	NOUN
fcis-30522	230	14	.	.	PUNCT
fcis-30522	231	1	however	however	ADV
fcis-30522	231	2	,	,	PUNCT
fcis-30522	231	3	when	when	SCONJ
fcis-30522	231	4	applying	apply	VERB
fcis-30522	231	5	eda	eda	PROPN
fcis-30522	231	6	for	for	ADP
fcis-30522	231	7	data	datum	NOUN
fcis-30522	231	8	augmentation	augmentation	NOUN
fcis-30522	231	9	on	on	ADP
fcis-30522	231	10	small	small	ADJ
fcis-30522	231	11	samples	sample	NOUN
fcis-30522	231	12	,	,	PUNCT
fcis-30522	231	13	results	result	NOUN
fcis-30522	231	14	can	can	AUX
fcis-30522	231	15	deteriorate	deteriorate	VERB
fcis-30522	231	16	as	as	SCONJ
fcis-30522	231	17	original	original	ADJ
fcis-30522	231	18	word	word	NOUN
fcis-30522	231	19	order	order	NOUN
fcis-30522	231	20	may	may	AUX
fcis-30522	231	21	change	change	VERB
fcis-30522	231	22	.	.	PUNCT
fcis-30522	232	1	this	this	DET
fcis-30522	232	2	paper	paper	NOUN
fcis-30522	232	3	presents	present	VERB
fcis-30522	232	4	a	a	DET
fcis-30522	232	5	method	method	NOUN
fcis-30522	232	6	combining	combine	VERB
fcis-30522	232	7	aeda	aeda	NOUN
fcis-30522	232	8	and	and	CCONJ
fcis-30522	232	9	macbert	macbert	NOUN
fcis-30522	232	10	that	that	PRON
fcis-30522	232	11	preserves	preserve	VERB
fcis-30522	232	12	word	word	NOUN
fcis-30522	232	13	order	order	NOUN
fcis-30522	232	14	during	during	ADP
fcis-30522	232	15	augmentation	augmentation	NOUN
fcis-30522	232	16	.	.	PUNCT
fcis-30522	233	1	the	the	DET
fcis-30522	233	2	proposed	propose	VERB
fcis-30522	233	3	mbcresattnet	mbcresattnet	NOUN
fcis-30522	233	4	model	model	NOUN
fcis-30522	233	5	significantly	significantly	ADV
fcis-30522	233	6	outperforms	outperform	VERB
fcis-30522	233	7	alternatives	alternative	NOUN
fcis-30522	233	8	by	by	ADP
fcis-30522	233	9	integrating	integrate	VERB
fcis-30522	233	10	self	self	NOUN
fcis-30522	233	11	-	-	PUNCT
fcis-30522	233	12	attention	attention	NOUN
fcis-30522	233	13	mechanisms	mechanism	NOUN
fcis-30522	233	14	at	at	ADP
fcis-30522	233	15	multiple	multiple	ADJ
fcis-30522	233	16	levels	level	NOUN
fcis-30522	233	17	—	—	PUNCT
fcis-30522	233	18	after	after	ADP
fcis-30522	233	19	the	the	DET
fcis-30522	233	20	pre	pre	ADJ
fcis-30522	233	21	-	-	ADJ
fcis-30522	233	22	trained	train	VERB
fcis-30522	233	23	model	model	NOUN
fcis-30522	233	24	and	and	CCONJ
fcis-30522	233	25	after	after	ADP
fcis-30522	233	26	the	the	DET
fcis-30522	233	27	downstream	downstream	PROPN
fcis-30522	233	28	task	task	NOUN
fcis-30522	233	29	—	—	PUNCT
fcis-30522	233	30	enabling	enable	VERB
fcis-30522	233	31	better	well	ADJ
fcis-30522	233	32	comprehension	comprehension	NOUN
fcis-30522	233	33	of	of	ADP
fcis-30522	233	34	long	long	ADJ
fcis-30522	233	35	chinese	chinese	ADJ
fcis-30522	233	36	texts	text	NOUN
fcis-30522	233	37	with	with	ADP
fcis-30522	233	38	complex	complex	ADJ
fcis-30522	233	39	contexts	contexts	NOUN
fcis-30522	233	40	.	.	PUNCT
fcis-30522	234	1	adding	add	VERB
fcis-30522	234	2	dropout	dropout	NOUN
fcis-30522	234	3	layers	layer	NOUN
fcis-30522	234	4	prevents	prevent	VERB
fcis-30522	234	5	local	local	ADJ
fcis-30522	234	6	optima	optima	NOUN
fcis-30522	234	7	and	and	CCONJ
fcis-30522	234	8	overfitting	overfitting	NOUN
fcis-30522	234	9	.	.	PUNCT
fcis-30522	235	1	by	by	ADP
fcis-30522	235	2	retaining	retain	VERB
fcis-30522	235	3	and	and	CCONJ
fcis-30522	235	4	combining	combine	VERB
fcis-30522	235	5	original	original	ADJ
fcis-30522	235	6	features	feature	NOUN
fcis-30522	235	7	with	with	ADP
fcis-30522	235	8	weighted	weight	VERB
fcis-30522	235	9	features	feature	NOUN
fcis-30522	235	10	during	during	ADP
fcis-30522	235	11	fusion	fusion	NOUN
fcis-30522	235	12	,	,	PUNCT
fcis-30522	235	13	the	the	DET
fcis-30522	235	14	method	method	NOUN
fcis-30522	235	15	ensures	ensure	VERB
fcis-30522	235	16	complete	complete	ADJ
fcis-30522	235	17	utilization	utilization	NOUN
fcis-30522	235	18	of	of	ADP
fcis-30522	235	19	feature	feature	NOUN
fcis-30522	235	20	information	information	NOUN
fcis-30522	235	21	while	while	SCONJ
fcis-30522	235	22	avoiding	avoid	VERB
fcis-30522	235	23	information	information	NOUN
fcis-30522	235	24	loss	loss	NOUN
fcis-30522	235	25	between	between	ADP
fcis-30522	235	26	layers	layer	NOUN
fcis-30522	235	27	,	,	PUNCT
fcis-30522	235	28	achieving	achieve	VERB
fcis-30522	235	29	strong	strong	ADJ
fcis-30522	235	30	performance	performance	NOUN
fcis-30522	235	31	in	in	ADP
fcis-30522	235	32	multi	multi	ADJ
fcis-30522	235	33	-	-	ADJ
fcis-30522	235	34	label	label	ADJ
fcis-30522	235	35	small	small	ADJ
fcis-30522	235	36	-	-	PUNCT
fcis-30522	235	37	sample	sample	NOUN
fcis-30522	235	38	classification	classification	NOUN
fcis-30522	235	39	.	.	PUNCT
fcis-30522	236	1	future	future	ADJ
fcis-30522	236	2	improvements	improvement	NOUN
fcis-30522	236	3	could	could	AUX
fcis-30522	236	4	involve	involve	VERB
fcis-30522	236	5	domain	domain	NOUN
fcis-30522	236	6	-	-	PUNCT
fcis-30522	236	7	specific	specific	ADJ
fcis-30522	236	8	macbert	macbert	NOUN
fcis-30522	236	9	fine	fine	ADV
fcis-30522	236	10	-	-	PUNCT
fcis-30522	236	11	tuning	tuning	NOUN
fcis-30522	236	12	.	.	PUNCT
fcis-30522	237	1	acknowledgments	acknowledgment	NOUN
fcis-30522	237	2	i	i	PRON
fcis-30522	237	3	would	would	AUX
fcis-30522	237	4	like	like	VERB
fcis-30522	237	5	to	to	PART
fcis-30522	237	6	extend	extend	VERB
fcis-30522	237	7	special	special	ADJ
fcis-30522	237	8	thanks	thank	NOUN
fcis-30522	237	9	to	to	ADP
fcis-30522	237	10	the	the	DET
fcis-30522	237	11	graduate	graduate	NOUN
fcis-30522	237	12	innovation	innovation	NOUN
fcis-30522	237	13	project	project	NOUN
fcis-30522	237	14	of	of	ADP
fcis-30522	237	15	north	north	PROPN
fcis-30522	237	16	minzu	minzu	PROPN
fcis-30522	237	17	university	university	PROPN
fcis-30522	237	18	(	(	PUNCT
fcis-30522	237	19	project	project	NOUN
fcis-30522	237	20	grant	grant	VERB
fcis-30522	237	21	78	78	NUM
fcis-30522	237	22	no	no	NOUN
fcis-30522	237	23	.	.	PUNCT
fcis-30522	237	24	:	:	PUNCT
fcis-30522	237	25	ycx24177	ycx24177	PROPN
fcis-30522	237	26	)	)	PUNCT
fcis-30522	237	27	for	for	ADP
fcis-30522	237	28	funding	fund	VERB
fcis-30522	237	29	this	this	DET
fcis-30522	237	30	project	project	NOUN
fcis-30522	237	31	.	.	PUNCT
fcis-30522	238	1	references	reference	NOUN
fcis-30522	238	2	[	[	X
fcis-30522	238	3	1	1	X
fcis-30522	238	4	]	]	PUNCT
fcis-30522	238	5	kim	kim	PROPN
fcis-30522	238	6	y.	y.	PROPN
fcis-30522	238	7	convolutional	convolutional	ADJ
fcis-30522	238	8	neural	neural	ADJ
fcis-30522	238	9	networks	network	NOUN
fcis-30522	238	10	for	for	ADP
fcis-30522	238	11	sentence	sentence	NOUN
fcis-30522	238	12	classification	classification	NOUN
fcis-30522	238	13	.	.	PUNCT
fcis-30522	239	1	arxiv	arxiv	PROPN
fcis-30522	239	2	preprint	preprint	VERB
fcis-30522	239	3	arxiv:14085882	arxiv:14085882	NOUN
fcis-30522	239	4	.	.	PROPN
fcis-30522	239	5	2014	2014	NUM
fcis-30522	239	6	.	.	PUNCT
fcis-30522	240	1	[	[	X
fcis-30522	240	2	2	2	NUM
fcis-30522	240	3	]	]	PUNCT
fcis-30522	240	4	rawat	rawat	PROPN
fcis-30522	240	5	a	a	PRON
fcis-30522	240	6	,	,	PUNCT
fcis-30522	240	7	wani	wani	PROPN
fcis-30522	240	8	ma	ma	PROPN
fcis-30522	240	9	,	,	PUNCT
fcis-30522	240	10	elaffendi	elaffendi	PROPN
fcis-30522	240	11	m	m	PROPN
fcis-30522	240	12	,	,	PUNCT
fcis-30522	240	13	imran	imran	ADJ
fcis-30522	240	14	as	as	ADP
fcis-30522	240	15	,	,	PUNCT
fcis-30522	240	16	kastrati	kastrati	PROPN
fcis-30522	240	17	z	z	PROPN
fcis-30522	240	18	,	,	PUNCT
fcis-30522	240	19	daudpota	daudpota	PROPN
fcis-30522	240	20	sm	sm	PROPN
fcis-30522	240	21	.	.	PROPN
fcis-30522	240	22	drug	drug	PROPN
fcis-30522	240	23	adverse	adverse	ADJ
fcis-30522	240	24	event	event	NOUN
fcis-30522	240	25	detection	detection	NOUN
fcis-30522	240	26	using	use	VERB
fcis-30522	240	27	text	text	NOUN
fcis-30522	240	28	-	-	PUNCT
fcis-30522	240	29	based	base	VERB
fcis-30522	240	30	convolutional	convolutional	ADJ
fcis-30522	240	31	neural	neural	ADJ
fcis-30522	240	32	networks	network	NOUN
fcis-30522	240	33	(	(	PUNCT
fcis-30522	240	34	textcnn	textcnn	NOUN
fcis-30522	240	35	)	)	PUNCT
fcis-30522	240	36	technique	technique	NOUN
fcis-30522	240	37	.	.	PUNCT
fcis-30522	241	1	electronics	electronic	NOUN
fcis-30522	241	2	.	.	PUNCT
fcis-30522	242	1	2022;11(20):3336	2022;11(20):3336	X
fcis-30522	242	2	.	.	PUNCT
fcis-30522	243	1	[	[	X
fcis-30522	243	2	3	3	NUM
fcis-30522	243	3	]	]	X
fcis-30522	243	4	guo	guo	PROPN
fcis-30522	243	5	b	b	PROPN
fcis-30522	243	6	,	,	PUNCT
fcis-30522	243	7	zhang	zhang	PROPN
fcis-30522	243	8	c	c	PROPN
fcis-30522	243	9	,	,	PUNCT
fcis-30522	243	10	liu	liu	PROPN
fcis-30522	243	11	j	j	PROPN
fcis-30522	243	12	,	,	PUNCT
fcis-30522	243	13	ma	ma	PROPN
fcis-30522	243	14	x.	x.	NOUN
fcis-30522	243	15	improving	improve	VERB
fcis-30522	243	16	text	text	NOUN
fcis-30522	243	17	classification	classification	NOUN
fcis-30522	243	18	with	with	ADP
fcis-30522	243	19	weighted	weight	VERB
fcis-30522	243	20	word	word	NOUN
fcis-30522	243	21	embeddings	embedding	NOUN
fcis-30522	243	22	via	via	ADP
fcis-30522	243	23	a	a	DET
fcis-30522	243	24	multi	multi	ADJ
fcis-30522	243	25	-	-	ADJ
fcis-30522	243	26	channel	channel	ADJ
fcis-30522	243	27	textcnn	textcnn	PROPN
fcis-30522	243	28	model	model	PROPN
fcis-30522	243	29	.	.	PUNCT
fcis-30522	244	1	2019;363:366	2019;363:366	PROPN
fcis-30522	244	2	-	-	PUNCT
fcis-30522	244	3	74	74	NUM
fcis-30522	244	4	.	.	PUNCT
fcis-30522	245	1	[	[	X
fcis-30522	245	2	4	4	X
fcis-30522	245	3	]	]	X
fcis-30522	245	4	yang	yang	PROPN
fcis-30522	245	5	z	z	PROPN
fcis-30522	245	6	,	,	PUNCT
fcis-30522	245	7	emmert	emmert	PROPN
fcis-30522	245	8	-	-	PUNCT
fcis-30522	245	9	streib	streib	PROPN
fcis-30522	245	10	f.	f.	PROPN
fcis-30522	245	11	optimal	optimal	ADJ
fcis-30522	245	12	performance	performance	NOUN
fcis-30522	245	13	of	of	ADP
fcis-30522	245	14	binary	binary	ADJ
fcis-30522	245	15	relevance	relevance	NOUN
fcis-30522	245	16	cnn	cnn	PROPN
fcis-30522	245	17	in	in	ADP
fcis-30522	245	18	targeted	target	VERB
fcis-30522	245	19	multi	multi	ADJ
fcis-30522	245	20	-	-	ADJ
fcis-30522	245	21	label	label	ADJ
fcis-30522	245	22	text	text	NOUN
fcis-30522	245	23	classification	classification	NOUN
fcis-30522	245	24	.	.	PUNCT
fcis-30522	246	1	knowledge	knowledge	NOUN
fcis-30522	246	2	-	-	PUNCT
fcis-30522	246	3	based	base	VERB
fcis-30522	246	4	systems	system	NOUN
fcis-30522	246	5	.	.	PUNCT
fcis-30522	247	1	2024	2024	NUM
fcis-30522	247	2	;	;	PUNCT
fcis-30522	247	3	284:111286	284:111286	X
fcis-30522	247	4	.	.	PUNCT
fcis-30522	248	1	[	[	X
fcis-30522	248	2	5	5	NUM
fcis-30522	248	3	]	]	SYM
fcis-30522	248	4	thekkekara	thekkekara	NOUN
fcis-30522	248	5	jp	jp	PROPN
fcis-30522	248	6	,	,	PUNCT
fcis-30522	248	7	yongchareon	yongchareon	PROPN
fcis-30522	248	8	s	s	PROPN
fcis-30522	248	9	,	,	PUNCT
fcis-30522	248	10	liesaputra	liesaputra	ADJ
fcis-30522	248	11	v.	v.	ADP
fcis-30522	248	12	an	an	DET
fcis-30522	248	13	attentionbased	attentionbase	VERB
fcis-30522	248	14	cnn	cnn	PROPN
fcis-30522	248	15	-	-	PUNCT
fcis-30522	248	16	bilstm	bilstm	NOUN
fcis-30522	248	17	model	model	NOUN
fcis-30522	248	18	for	for	ADP
fcis-30522	248	19	depression	depression	NOUN
fcis-30522	248	20	detection	detection	NOUN
fcis-30522	248	21	on	on	ADP
fcis-30522	248	22	social	social	ADJ
fcis-30522	248	23	media	medium	NOUN
fcis-30522	248	24	text	text	NOUN
fcis-30522	248	25	.	.	PUNCT
fcis-30522	249	1	expert	expert	NOUN
fcis-30522	249	2	systems	system	NOUN
fcis-30522	249	3	with	with	ADP
fcis-30522	249	4	applications	application	NOUN
fcis-30522	249	5	.	.	PUNCT
fcis-30522	250	1	2024	2024	NUM
fcis-30522	250	2	;	;	PUNCT
fcis-30522	250	3	249	249	NUM
fcis-30522	250	4	:	:	SYM
fcis-30522	250	5	123834	123834	NUM
fcis-30522	250	6	.	.	PUNCT
fcis-30522	251	1	[	[	X
fcis-30522	251	2	6	6	NUM
fcis-30522	251	3	]	]	PUNCT
fcis-30522	251	4	devlin	devlin	PROPN
fcis-30522	251	5	j.	j.	PROPN
fcis-30522	251	6	bert	bert	PROPN
fcis-30522	251	7	:	:	PUNCT
fcis-30522	251	8	pre	pre	ADJ
fcis-30522	251	9	-	-	NOUN
fcis-30522	251	10	training	training	NOUN
fcis-30522	251	11	of	of	ADP
fcis-30522	251	12	deep	deep	ADJ
fcis-30522	251	13	bidirectional	bidirectional	ADJ
fcis-30522	251	14	transformers	transformer	NOUN
fcis-30522	251	15	for	for	ADP
fcis-30522	251	16	language	language	NOUN
fcis-30522	251	17	understanding	understanding	NOUN
fcis-30522	251	18	.	.	PUNCT
fcis-30522	252	1	arxiv	arxiv	PROPN
fcis-30522	252	2	preprint	preprint	VERB
fcis-30522	252	3	arxiv:181004805	arxiv:181004805	AUX
fcis-30522	252	4	.	.	PUNCT
fcis-30522	253	1	2018	2018	NUM
fcis-30522	253	2	.	.	PUNCT
fcis-30522	254	1	[	[	X
fcis-30522	254	2	7	7	X
fcis-30522	254	3	]	]	X
fcis-30522	254	4	deng	deng	PROPN
fcis-30522	254	5	s	s	PROPN
fcis-30522	254	6	,	,	PUNCT
fcis-30522	254	7	li	li	PROPN
fcis-30522	254	8	q	q	PROPN
fcis-30522	254	9	,	,	PUNCT
fcis-30522	254	10	dai	dai	PROPN
fcis-30522	254	11	r	r	PROPN
fcis-30522	254	12	,	,	PUNCT
fcis-30522	254	13	wei	wei	PROPN
fcis-30522	254	14	s	s	PROPN
fcis-30522	254	15	,	,	PUNCT
fcis-30522	255	1	wu	wu	PROPN
fcis-30522	255	2	d	d	PROPN
fcis-30522	255	3	,	,	PUNCT
fcis-30522	255	4	he	he	PRON
fcis-30522	255	5	y	y	PROPN
fcis-30522	255	6	,	,	PUNCT
fcis-30522	255	7	et	et	PROPN
fcis-30522	255	8	al	al	PROPN
fcis-30522	255	9	.	.	PUNCT
fcis-30522	256	1	a	a	DET
fcis-30522	256	2	chinese	chinese	ADJ
fcis-30522	256	3	power	power	NOUN
fcis-30522	256	4	text	text	NOUN
fcis-30522	256	5	classification	classification	NOUN
fcis-30522	256	6	algorithm	algorithm	NOUN
fcis-30522	256	7	based	base	VERB
fcis-30522	256	8	on	on	ADP
fcis-30522	256	9	deep	deep	ADJ
fcis-30522	256	10	active	active	ADJ
fcis-30522	256	11	learning	learning	NOUN
fcis-30522	256	12	.	.	PUNCT
fcis-30522	256	13	applied	apply	VERB
fcis-30522	256	14	soft	soft	ADJ
fcis-30522	256	15	computing	computing	NOUN
fcis-30522	256	16	.	.	PUNCT
fcis-30522	257	1	2024;150:111067	2024;150:111067	X
fcis-30522	257	2	.	.	PUNCT
fcis-30522	258	1	[	[	X
fcis-30522	258	2	8	8	NUM
fcis-30522	258	3	]	]	SYM
fcis-30522	258	4	nithya	nithya	PROPN
fcis-30522	258	5	k	k	PROPN
fcis-30522	258	6	,	,	PUNCT
fcis-30522	258	7	krishnamoorthi	krishnamoorthi	PROPN
fcis-30522	258	8	m	m	PROPN
fcis-30522	258	9	,	,	PUNCT
fcis-30522	258	10	easwaramoorthy	easwaramoorthy	PROPN
fcis-30522	258	11	sv	sv	PROPN
fcis-30522	258	12	,	,	PUNCT
fcis-30522	258	13	dhivyaa	dhivyaa	PROPN
fcis-30522	258	14	c	c	PROPN
fcis-30522	258	15	,	,	PUNCT
fcis-30522	258	16	yoo	yoo	PROPN
fcis-30522	258	17	s	s	PROPN
fcis-30522	258	18	,	,	PUNCT
fcis-30522	258	19	cho	cho	PROPN
fcis-30522	258	20	j.	j.	PROPN
fcis-30522	258	21	hybrid	hybrid	PROPN
fcis-30522	258	22	approach	approach	NOUN
fcis-30522	258	23	of	of	ADP
fcis-30522	258	24	deep	deep	ADJ
fcis-30522	258	25	feature	feature	NOUN
fcis-30522	258	26	extraction	extraction	NOUN
fcis-30522	258	27	using	use	VERB
fcis-30522	258	28	bert	bert	PROPN
fcis-30522	258	29	–	–	PUNCT
fcis-30522	258	30	opcnn	opcnn	PROPN
fcis-30522	258	31	&	&	CCONJ
fcis-30522	258	32	fiac	fiac	PROPN
fcis-30522	258	33	with	with	ADP
fcis-30522	258	34	customized	customized	ADJ
fcis-30522	258	35	bi	bi	NOUN
fcis-30522	258	36	-	-	NOUN
fcis-30522	258	37	lstm	lstm	NOUN
fcis-30522	258	38	for	for	ADP
fcis-30522	258	39	rumor	rumor	NOUN
fcis-30522	258	40	text	text	NOUN
fcis-30522	258	41	classification	classification	NOUN
fcis-30522	258	42	.	.	PUNCT
fcis-30522	259	1	alexandria	alexandria	PROPN
fcis-30522	259	2	engineering	engineering	PROPN
fcis-30522	259	3	journal	journal	PROPN
fcis-30522	259	4	.	.	PUNCT
fcis-30522	260	1	2024;90	2024;90	NUM
fcis-30522	260	2	:	:	PUNCT
fcis-30522	260	3	65	65	NUM
fcis-30522	260	4	-	-	SYM
fcis-30522	260	5	75	75	NUM
fcis-30522	260	6	.	.	PUNCT
fcis-30522	261	1	[	[	X
fcis-30522	261	2	9	9	NUM
fcis-30522	261	3	]	]	PUNCT
fcis-30522	261	4	wilkho	wilkho	ADJ
fcis-30522	261	5	rs	rs	PROPN
fcis-30522	261	6	,	,	PUNCT
fcis-30522	261	7	chang	chang	PROPN
fcis-30522	261	8	s	s	PROPN
fcis-30522	261	9	,	,	PUNCT
fcis-30522	261	10	gharaibeh	gharaibeh	NOUN
fcis-30522	261	11	ng	ng	PROPN
fcis-30522	261	12	.	.	PROPN
fcis-30522	262	1	ff	ff	NOUN
fcis-30522	262	2	-	-	PUNCT
fcis-30522	262	3	bert	bert	NOUN
fcis-30522	262	4	:	:	PUNCT
fcis-30522	262	5	a	a	DET
fcis-30522	262	6	bertbased	bertbase	VERB
fcis-30522	262	7	ensemble	ensemble	ADJ
fcis-30522	262	8	for	for	ADP
fcis-30522	262	9	automated	automate	VERB
fcis-30522	262	10	classification	classification	NOUN
fcis-30522	262	11	of	of	ADP
fcis-30522	262	12	web	web	NOUN
fcis-30522	262	13	-	-	PUNCT
fcis-30522	262	14	based	base	VERB
fcis-30522	262	15	text	text	NOUN
fcis-30522	262	16	on	on	ADP
fcis-30522	262	17	flash	flash	NOUN
fcis-30522	262	18	flood	flood	NOUN
fcis-30522	262	19	events	event	NOUN
fcis-30522	262	20	.	.	PUNCT
fcis-30522	263	1	advanced	advanced	ADJ
fcis-30522	263	2	engineering	engineering	NOUN
fcis-30522	263	3	informatics	informatic	NOUN
fcis-30522	263	4	.	.	PUNCT
fcis-30522	264	1	2024;59:102293	2024;59:102293	NUM
fcis-30522	264	2	.	.	PUNCT
fcis-30522	265	1	[	[	X
fcis-30522	265	2	10	10	NUM
fcis-30522	265	3	]	]	X
fcis-30522	265	4	hui	hui	PROPN
fcis-30522	265	5	y	y	PROPN
fcis-30522	265	6	,	,	PUNCT
fcis-30522	265	7	du	du	PROPN
fcis-30522	265	8	l	l	PROPN
fcis-30522	265	9	,	,	PUNCT
fcis-30522	265	10	lin	lin	PROPN
fcis-30522	265	11	s	s	PROPN
fcis-30522	265	12	,	,	PUNCT
fcis-30522	265	13	qu	qu	PROPN
fcis-30522	265	14	y	y	PROPN
fcis-30522	265	15	,	,	PUNCT
fcis-30522	265	16	cao	cao	PROPN
fcis-30522	266	1	d	d	PROPN
fcis-30522	266	2	,	,	PUNCT
fcis-30522	266	3	editors	editor	NOUN
fcis-30522	266	4	.	.	PUNCT
fcis-30522	267	1	extraction	extraction	NOUN
fcis-30522	267	2	and	and	CCONJ
fcis-30522	267	3	classification	classification	NOUN
fcis-30522	267	4	of	of	ADP
fcis-30522	267	5	tcm	tcm	PROPN
fcis-30522	267	6	medical	medical	PROPN
fcis-30522	267	7	records	record	NOUN
fcis-30522	267	8	based	base	VERB
fcis-30522	267	9	on	on	ADP
fcis-30522	267	10	bert	bert	PROPN
fcis-30522	267	11	and	and	CCONJ
fcis-30522	267	12	bi	bi	NOUN
fcis-30522	267	13	-	-	NOUN
fcis-30522	267	14	lstm	lstm	ADJ
fcis-30522	267	15	with	with	ADP
fcis-30522	267	16	attention	attention	NOUN
fcis-30522	267	17	mechanism	mechanism	NOUN
fcis-30522	267	18	.	.	PUNCT
fcis-30522	268	1	2020	2020	NUM
fcis-30522	268	2	ieee	ieee	PROPN
fcis-30522	268	3	international	international	ADJ
fcis-30522	268	4	conference	conference	NOUN
fcis-30522	268	5	on	on	ADP
fcis-30522	268	6	bioinformatics	bioinformatics	NOUN
fcis-30522	268	7	and	and	CCONJ
fcis-30522	268	8	biomedicine	biomedicine	NOUN
fcis-30522	268	9	(	(	PUNCT
fcis-30522	268	10	bibm	bibm	PROPN
fcis-30522	268	11	)	)	PUNCT
fcis-30522	268	12	;	;	PUNCT
fcis-30522	268	13	2020	2020	NUM
fcis-30522	268	14	:	:	PUNCT
fcis-30522	268	15	ieee	ieee	NOUN
fcis-30522	268	16	.	.	PUNCT
fcis-30522	269	1	[	[	X
fcis-30522	269	2	11	11	NUM
fcis-30522	269	3	]	]	X
fcis-30522	269	4	liu	liu	PROPN
fcis-30522	269	5	y.	y.	PROPN
fcis-30522	269	6	roberta	roberta	PROPN
fcis-30522	269	7	:	:	PUNCT
fcis-30522	269	8	a	a	DET
fcis-30522	269	9	robustly	robustly	ADV
fcis-30522	269	10	optimized	optimize	VERB
fcis-30522	269	11	bert	bert	NOUN
fcis-30522	269	12	pretraining	pretraine	VERB
fcis-30522	269	13	approach	approach	NOUN
fcis-30522	269	14	.	.	PUNCT
fcis-30522	270	1	arxiv	arxiv	PROPN
fcis-30522	270	2	preprint	preprint	PROPN
fcis-30522	270	3	arxiv:190711692	arxiv:190711692	PROPN
fcis-30522	270	4	.	.	PROPN
fcis-30522	270	5	2019	2019	NUM
fcis-30522	270	6	.	.	PUNCT
fcis-30522	271	1	[	[	X
fcis-30522	271	2	12	12	NUM
fcis-30522	271	3	]	]	X
fcis-30522	271	4	chi	chi	NOUN
fcis-30522	271	5	p	p	PROPN
fcis-30522	271	6	-	-	PUNCT
fcis-30522	271	7	h	h	NOUN
fcis-30522	271	8	,	,	PUNCT
fcis-30522	271	9	chung	chung	PROPN
fcis-30522	271	10	p	p	PROPN
fcis-30522	271	11	-	-	PUNCT
fcis-30522	271	12	h	h	NOUN
fcis-30522	271	13	,	,	PUNCT
fcis-30522	271	14	wu	wu	PROPN
fcis-30522	271	15	t	t	PROPN
fcis-30522	271	16	-	-	PUNCT
fcis-30522	271	17	h	h	PROPN
fcis-30522	271	18	,	,	PUNCT
fcis-30522	271	19	hsieh	hsieh	PROPN
fcis-30522	271	20	c	c	PROPN
fcis-30522	271	21	-	-	PUNCT
fcis-30522	271	22	c	c	PROPN
fcis-30522	271	23	,	,	PUNCT
fcis-30522	271	24	chen	chen	PROPN
fcis-30522	271	25	y	y	PROPN
fcis-30522	271	26	-	-	PUNCT
fcis-30522	271	27	h	h	PROPN
fcis-30522	271	28	,	,	PUNCT
fcis-30522	271	29	li	li	PROPN
fcis-30522	271	30	s	s	PROPN
fcis-30522	271	31	-	-	PUNCT
fcis-30522	271	32	w	w	PROPN
fcis-30522	271	33	,	,	PUNCT
fcis-30522	271	34	et	et	PROPN
fcis-30522	271	35	al	al	PROPN
fcis-30522	271	36	.	.	PROPN
fcis-30522	271	37	,	,	PUNCT
fcis-30522	271	38	editors	editor	NOUN
fcis-30522	271	39	.	.	PUNCT
fcis-30522	272	1	audio	audio	PROPN
fcis-30522	272	2	albert	albert	PROPN
fcis-30522	272	3	:	:	PUNCT
fcis-30522	272	4	a	a	DET
fcis-30522	272	5	lite	lite	ADJ
fcis-30522	272	6	bert	bert	NOUN
fcis-30522	272	7	for	for	ADP
fcis-30522	272	8	self	self	NOUN
fcis-30522	272	9	-	-	PUNCT
fcis-30522	272	10	supervised	supervise	VERB
fcis-30522	272	11	learning	learning	NOUN
fcis-30522	272	12	of	of	ADP
fcis-30522	272	13	audio	audio	NOUN
fcis-30522	272	14	representation	representation	NOUN
fcis-30522	272	15	.	.	PUNCT
fcis-30522	273	1	2021	2021	NUM
fcis-30522	273	2	ieee	ieee	NOUN
fcis-30522	273	3	spoken	speak	VERB
fcis-30522	273	4	language	language	NOUN
fcis-30522	273	5	technology	technology	NOUN
fcis-30522	273	6	workshop	workshop	NOUN
fcis-30522	273	7	(	(	PUNCT
fcis-30522	273	8	slt	slt	PROPN
fcis-30522	273	9	)	)	PUNCT
fcis-30522	273	10	;	;	PUNCT
fcis-30522	273	11	2021	2021	NUM
fcis-30522	273	12	:	:	PUNCT
fcis-30522	273	13	ieee	ieee	NOUN
fcis-30522	273	14	.	.	PUNCT
fcis-30522	274	1	[	[	X
fcis-30522	274	2	13	13	NUM
fcis-30522	274	3	]	]	PUNCT
fcis-30522	274	4	sanh	sanh	PROPN
fcis-30522	274	5	v.	v.	PROPN
fcis-30522	274	6	distilbert	distilbert	PROPN
fcis-30522	274	7	,	,	PUNCT
fcis-30522	274	8	a	a	DET
fcis-30522	274	9	distilled	distil	VERB
fcis-30522	274	10	version	version	NOUN
fcis-30522	274	11	of	of	ADP
fcis-30522	274	12	bert	bert	NOUN
fcis-30522	274	13	:	:	PUNCT
fcis-30522	274	14	smaller	small	ADJ
fcis-30522	274	15	,	,	PUNCT
fcis-30522	274	16	faster	fast	ADJ
fcis-30522	274	17	,	,	PUNCT
fcis-30522	274	18	cheaper	cheap	ADJ
fcis-30522	274	19	and	and	CCONJ
fcis-30522	274	20	lighter	light	ADJ
fcis-30522	274	21	.	.	PUNCT
fcis-30522	275	1	arxiv	arxiv	PROPN
fcis-30522	275	2	preprint	preprint	PROPN
fcis-30522	275	3	arxiv:191001108	arxiv:191001108	NOUN
fcis-30522	275	4	.	.	PUNCT
fcis-30522	275	5	2019	2019	NUM
fcis-30522	275	6	.	.	PUNCT
fcis-30522	276	1	[	[	X
fcis-30522	276	2	14	14	NUM
fcis-30522	276	3	]	]	X
fcis-30522	276	4	sun	sun	PROPN
fcis-30522	276	5	z	z	PROPN
fcis-30522	276	6	,	,	PUNCT
fcis-30522	276	7	li	li	PROPN
fcis-30522	276	8	x	x	PROPN
fcis-30522	276	9	,	,	PUNCT
fcis-30522	276	10	sun	sun	PROPN
fcis-30522	276	11	x	x	PROPN
fcis-30522	276	12	,	,	PUNCT
fcis-30522	276	13	meng	meng	PROPN
fcis-30522	276	14	y	y	PROPN
fcis-30522	276	15	,	,	PUNCT
fcis-30522	276	16	ao	ao	PROPN
fcis-30522	276	17	x	x	NOUN
fcis-30522	276	18	,	,	PUNCT
fcis-30522	276	19	he	he	PRON
fcis-30522	276	20	q	q	AUX
fcis-30522	276	21	,	,	PUNCT
fcis-30522	276	22	et	et	PROPN
fcis-30522	276	23	al	al	PROPN
fcis-30522	276	24	.	.	PUNCT
fcis-30522	276	25	chinesebert	chinesebert	PROPN
fcis-30522	276	26	:	:	PUNCT
fcis-30522	277	1	chinese	chinese	ADJ
fcis-30522	277	2	pretraining	pretraine	VERB
fcis-30522	277	3	enhanced	enhance	VERB
fcis-30522	277	4	by	by	ADP
fcis-30522	277	5	glyph	glyph	NOUN
fcis-30522	277	6	and	and	CCONJ
fcis-30522	277	7	pinyin	pinyin	PROPN
fcis-30522	277	8	information	information	NOUN
fcis-30522	277	9	.	.	PUNCT
fcis-30522	278	1	arxiv	arxiv	PROPN
fcis-30522	278	2	preprint	preprint	PROPN
fcis-30522	278	3	arxiv:210616038	arxiv:210616038	PROPN
fcis-30522	278	4	.	.	PROPN
fcis-30522	278	5	2021	2021	NUM
fcis-30522	278	6	.	.	PUNCT
fcis-30522	279	1	[	[	X
fcis-30522	279	2	15	15	NUM
fcis-30522	279	3	]	]	PUNCT
fcis-30522	279	4	cui	cui	X
fcis-30522	279	5	y	y	PROPN
fcis-30522	279	6	,	,	PUNCT
fcis-30522	279	7	che	che	PROPN
fcis-30522	279	8	w	w	PROPN
fcis-30522	279	9	,	,	PUNCT
fcis-30522	279	10	liu	liu	PROPN
fcis-30522	279	11	t	t	PROPN
fcis-30522	279	12	,	,	PUNCT
fcis-30522	279	13	qin	qin	PROPN
fcis-30522	279	14	b	b	PROPN
fcis-30522	279	15	,	,	PUNCT
fcis-30522	279	16	yang	yang	PROPN
fcis-30522	279	17	z.	z.	PROPN
fcis-30522	279	18	pre	pre	VERB
fcis-30522	279	19	-	-	VERB
fcis-30522	279	20	training	training	NOUN
fcis-30522	279	21	with	with	ADP
fcis-30522	279	22	whole	whole	ADJ
fcis-30522	279	23	word	word	NOUN
fcis-30522	279	24	masking	mask	VERB
fcis-30522	279	25	for	for	ADP
fcis-30522	279	26	chinese	chinese	ADJ
fcis-30522	279	27	bert	bert	PROPN
fcis-30522	279	28	.	.	PUNCT
fcis-30522	280	1	ieee	ieee	PROPN
fcis-30522	280	2	/	/	SYM
fcis-30522	280	3	acm	acm	PROPN
fcis-30522	280	4	transactions	transaction	NOUN
fcis-30522	280	5	on	on	ADP
fcis-30522	280	6	audio	audio	NOUN
fcis-30522	280	7	,	,	PUNCT
fcis-30522	280	8	speech	speech	NOUN
fcis-30522	280	9	,	,	PUNCT
fcis-30522	280	10	and	and	CCONJ
fcis-30522	280	11	language	language	NOUN
fcis-30522	280	12	processing	processing	NOUN
fcis-30522	280	13	.	.	PUNCT
fcis-30522	281	1	2021;29:3504	2021;29:3504	NUM
fcis-30522	281	2	-	-	SYM
fcis-30522	281	3	14	14	NUM
fcis-30522	281	4	.	.	PUNCT
fcis-30522	282	1	[	[	X
fcis-30522	282	2	16	16	NUM
fcis-30522	282	3	]	]	X
fcis-30522	282	4	minaee	minaee	PROPN
fcis-30522	282	5	s	s	PROPN
fcis-30522	282	6	,	,	PUNCT
fcis-30522	282	7	kalchbrenner	kalchbrenner	NOUN
fcis-30522	282	8	n	n	CCONJ
fcis-30522	282	9	,	,	PUNCT
fcis-30522	282	10	cambria	cambria	PROPN
fcis-30522	282	11	e	e	PROPN
fcis-30522	282	12	,	,	PUNCT
fcis-30522	282	13	nikzad	nikzad	PROPN
fcis-30522	282	14	n	n	NUM
fcis-30522	282	15	,	,	PUNCT
fcis-30522	282	16	chenaghlu	chenaghlu	PROPN
fcis-30522	282	17	m	m	PROPN
fcis-30522	282	18	,	,	PUNCT
fcis-30522	282	19	gao	gao	PROPN
fcis-30522	282	20	j.	j.	PROPN
fcis-30522	282	21	deep	deep	PROPN
fcis-30522	282	22	learning	learning	PROPN
fcis-30522	282	23	--	--	PUNCT
fcis-30522	282	24	based	base	VERB
fcis-30522	282	25	text	text	NOUN
fcis-30522	282	26	classification	classification	NOUN
fcis-30522	282	27	:	:	PUNCT
fcis-30522	282	28	a	a	DET
fcis-30522	282	29	comprehensive	comprehensive	ADJ
fcis-30522	282	30	review	review	NOUN
fcis-30522	282	31	.	.	PUNCT
fcis-30522	283	1	acm	acm	PROPN
fcis-30522	283	2	computing	computing	NOUN
fcis-30522	283	3	surveys	survey	NOUN
fcis-30522	283	4	(	(	PUNCT
fcis-30522	283	5	csur	csur	NOUN
fcis-30522	283	6	)	)	PUNCT
fcis-30522	283	7	.	.	PUNCT
fcis-30522	284	1	2021	2021	NUM
fcis-30522	284	2	;	;	PUNCT
fcis-30522	284	3	54(3):1	54(3):1	NUM
fcis-30522	284	4	-	-	SYM
fcis-30522	284	5	40	40	NUM
fcis-30522	284	6	.	.	PUNCT
fcis-30522	285	1	[	[	X
fcis-30522	285	2	17	17	NUM
fcis-30522	285	3	]	]	PUNCT
fcis-30522	285	4	reusens	reusens	PROPN
fcis-30522	285	5	m	m	PROPN
fcis-30522	285	6	,	,	PUNCT
fcis-30522	285	7	stevens	steven	VERB
fcis-30522	285	8	a	a	PRON
fcis-30522	285	9	,	,	PUNCT
fcis-30522	285	10	tonglet	tonglet	NOUN
fcis-30522	285	11	j	j	PROPN
fcis-30522	285	12	,	,	PUNCT
fcis-30522	285	13	de	de	PROPN
fcis-30522	285	14	smedt	smedt	PROPN
fcis-30522	285	15	j	j	PROPN
fcis-30522	285	16	,	,	PUNCT
fcis-30522	285	17	verbeke	verbeke	PROPN
fcis-30522	285	18	w	w	PROPN
fcis-30522	285	19	,	,	PUNCT
fcis-30522	285	20	vanden	vanden	PROPN
fcis-30522	285	21	broucke	broucke	PROPN
fcis-30522	285	22	s	s	PROPN
fcis-30522	285	23	,	,	PUNCT
fcis-30522	285	24	et	et	PROPN
fcis-30522	285	25	al	al	PROPN
fcis-30522	285	26	.	.	PUNCT
fcis-30522	286	1	evaluating	evaluate	VERB
fcis-30522	286	2	text	text	NOUN
fcis-30522	286	3	classification	classification	NOUN
fcis-30522	286	4	:	:	PUNCT
fcis-30522	286	5	a	a	DET
fcis-30522	286	6	benchmark	benchmark	ADJ
fcis-30522	286	7	study	study	NOUN
fcis-30522	286	8	.	.	PUNCT
fcis-30522	287	1	expert	expert	NOUN
fcis-30522	287	2	systems	system	NOUN
fcis-30522	287	3	with	with	ADP
fcis-30522	287	4	applications	application	NOUN
fcis-30522	287	5	.	.	PUNCT
fcis-30522	288	1	2024:124302	2024:124302	X
fcis-30522	288	2	.	.	PUNCT
fcis-30522	289	1	[	[	X
fcis-30522	289	2	18	18	NUM
fcis-30522	289	3	]	]	X
fcis-30522	289	4	taha	taha	PROPN
fcis-30522	289	5	k	k	PROPN
fcis-30522	289	6	,	,	PUNCT
fcis-30522	289	7	yoo	yoo	PROPN
fcis-30522	289	8	pd	pd	PROPN
fcis-30522	289	9	,	,	PUNCT
fcis-30522	289	10	yeun	yeun	PROPN
fcis-30522	289	11	c	c	PROPN
fcis-30522	289	12	,	,	PUNCT
fcis-30522	289	13	homouz	homouz	NOUN
fcis-30522	289	14	d	d	PROPN
fcis-30522	289	15	,	,	PUNCT
fcis-30522	289	16	taha	taha	PROPN
fcis-30522	289	17	a.	a.	PROPN
fcis-30522	289	18	a	a	DET
fcis-30522	289	19	comprehensive	comprehensive	ADJ
fcis-30522	289	20	survey	survey	NOUN
fcis-30522	289	21	of	of	ADP
fcis-30522	289	22	text	text	NOUN
fcis-30522	289	23	classification	classification	NOUN
fcis-30522	289	24	techniques	technique	NOUN
fcis-30522	289	25	and	and	CCONJ
fcis-30522	289	26	their	their	PRON
fcis-30522	289	27	research	research	NOUN
fcis-30522	289	28	applications	application	NOUN
fcis-30522	289	29	:	:	PUNCT
fcis-30522	289	30	observational	observational	ADJ
fcis-30522	289	31	and	and	CCONJ
fcis-30522	289	32	experimental	experimental	ADJ
fcis-30522	289	33	insights	insight	NOUN
fcis-30522	289	34	.	.	PUNCT
fcis-30522	290	1	computer	computer	NOUN
fcis-30522	290	2	science	science	NOUN
fcis-30522	290	3	review	review	PROPN
fcis-30522	290	4	.	.	PUNCT
fcis-30522	291	1	2024;54:100664	2024;54:100664	NUM
fcis-30522	291	2	.	.	PUNCT
fcis-30522	292	1	[	[	X
fcis-30522	292	2	19	19	NUM
fcis-30522	292	3	]	]	X
fcis-30522	292	4	shobana	shobana	PROPN
fcis-30522	292	5	j	j	PROPN
fcis-30522	292	6	,	,	PUNCT
fcis-30522	292	7	murali	murali	PROPN
fcis-30522	292	8	m.	m.	PROPN
fcis-30522	292	9	an	an	DET
fcis-30522	292	10	improved	improved	ADJ
fcis-30522	292	11	self	self	NOUN
fcis-30522	292	12	attention	attention	NOUN
fcis-30522	292	13	mechanism	mechanism	NOUN
fcis-30522	292	14	based	base	VERB
fcis-30522	292	15	on	on	ADP
fcis-30522	292	16	optimized	optimize	VERB
fcis-30522	292	17	bert	bert	NOUN
fcis-30522	292	18	-	-	PUNCT
fcis-30522	292	19	bilstm	bilstm	NOUN
fcis-30522	292	20	model	model	NOUN
fcis-30522	292	21	for	for	ADP
fcis-30522	292	22	accurate	accurate	ADJ
fcis-30522	292	23	polarity	polarity	NOUN
fcis-30522	292	24	prediction	prediction	NOUN
fcis-30522	292	25	.	.	PUNCT
fcis-30522	293	1	the	the	DET
fcis-30522	293	2	computer	computer	NOUN
fcis-30522	293	3	journal	journal	NOUN
fcis-30522	293	4	.	.	PUNCT
fcis-30522	294	1	2023;66(5):127994	2023;66(5):127994	NUM
fcis-30522	294	2	.	.	PUNCT
fcis-30522	295	1	[	[	X
fcis-30522	295	2	20	20	NUM
fcis-30522	295	3	]	]	X
fcis-30522	295	4	jia	jia	PROPN
fcis-30522	295	5	c	c	X
fcis-30522	295	6	,	,	PUNCT
fcis-30522	295	7	he	he	PRON
fcis-30522	295	8	h	h	VERB
fcis-30522	295	9	,	,	PUNCT
fcis-30522	295	10	zhou	zhou	PROPN
fcis-30522	295	11	j	j	PROPN
fcis-30522	295	12	,	,	PUNCT
fcis-30522	295	13	li	li	PROPN
fcis-30522	295	14	k	k	PROPN
fcis-30522	295	15	,	,	PUNCT
fcis-30522	295	16	li	li	PROPN
fcis-30522	295	17	j	j	PROPN
fcis-30522	295	18	,	,	PUNCT
fcis-30522	295	19	wei	wei	PROPN
fcis-30522	295	20	z.	z.	PROPN
fcis-30522	295	21	a	a	DET
fcis-30522	295	22	performance	performance	NOUN
fcis-30522	295	23	degradation	degradation	NOUN
fcis-30522	295	24	prediction	prediction	NOUN
fcis-30522	295	25	model	model	NOUN
fcis-30522	295	26	for	for	ADP
fcis-30522	295	27	pemfc	pemfc	NOUN
fcis-30522	295	28	based	base	VERB
fcis-30522	295	29	on	on	ADP
fcis-30522	295	30	bidirectional	bidirectional	ADJ
fcis-30522	295	31	long	long	ADJ
fcis-30522	295	32	short	short	ADJ
fcis-30522	295	33	-	-	PUNCT
fcis-30522	295	34	term	term	NOUN
fcis-30522	295	35	memory	memory	NOUN
fcis-30522	295	36	and	and	CCONJ
fcis-30522	295	37	multi	multi	ADJ
fcis-30522	295	38	-	-	ADJ
fcis-30522	295	39	head	head	ADJ
fcis-30522	295	40	selfattention	selfattention	NOUN
fcis-30522	295	41	mechanism	mechanism	NOUN
fcis-30522	295	42	.	.	PUNCT
fcis-30522	296	1	international	international	ADJ
fcis-30522	296	2	journal	journal	PROPN
fcis-30522	296	3	of	of	ADP
fcis-30522	296	4	hydrogen	hydrogen	NOUN
fcis-30522	296	5	energy	energy	NOUN
fcis-30522	296	6	.	.	PUNCT
fcis-30522	297	1	2024;60:133	2024;60:133	NUM
fcis-30522	297	2	-	-	SYM
fcis-30522	297	3	46	46	NUM
fcis-30522	297	4	.	.	PUNCT
fcis-30522	298	1	[	[	X
fcis-30522	298	2	21	21	NUM
fcis-30522	298	3	]	]	X
fcis-30522	298	4	cai	cai	PROPN
fcis-30522	298	5	z	z	PROPN
fcis-30522	298	6	,	,	PUNCT
fcis-30522	298	7	zhang	zhang	PROPN
fcis-30522	298	8	h	h	PROPN
fcis-30522	298	9	,	,	PUNCT
fcis-30522	298	10	zhan	zhan	PROPN
fcis-30522	298	11	p	p	PROPN
fcis-30522	298	12	,	,	PUNCT
fcis-30522	298	13	jia	jia	PROPN
fcis-30522	298	14	x	x	PROPN
fcis-30522	298	15	,	,	PUNCT
fcis-30522	298	16	yan	yan	PROPN
fcis-30522	298	17	y	y	PROPN
fcis-30522	298	18	,	,	PUNCT
fcis-30522	298	19	song	song	NOUN
fcis-30522	298	20	x	x	NOUN
fcis-30522	298	21	,	,	PUNCT
fcis-30522	298	22	et	et	PROPN
fcis-30522	298	23	al	al	PROPN
fcis-30522	298	24	.	.	PROPN
fcis-30522	298	25	multischema	multischema	PROPN
fcis-30522	298	26	prompting	prompt	VERB
fcis-30522	298	27	powered	power	VERB
fcis-30522	298	28	token	token	VERB
fcis-30522	298	29	-	-	PUNCT
fcis-30522	298	30	feature	feature	NOUN
fcis-30522	298	31	woven	weave	VERB
fcis-30522	298	32	attention	attention	NOUN
fcis-30522	298	33	network	network	NOUN
fcis-30522	298	34	for	for	ADP
fcis-30522	298	35	short	short	ADJ
fcis-30522	298	36	text	text	NOUN
fcis-30522	298	37	classification	classification	NOUN
fcis-30522	298	38	.	.	PUNCT
fcis-30522	299	1	pattern	pattern	NOUN
fcis-30522	299	2	recognition	recognition	NOUN
fcis-30522	299	3	.	.	PUNCT
fcis-30522	300	1	2024;156:110782	2024;156:110782	X
fcis-30522	300	2	.	.	PUNCT
fcis-30522	301	1	[	[	X
fcis-30522	301	2	22	22	NUM
fcis-30522	301	3	]	]	X
fcis-30522	301	4	li	li	PROPN
fcis-30522	301	5	g	g	PROPN
fcis-30522	301	6	,	,	PUNCT
fcis-30522	301	7	zhao	zhao	PROPN
fcis-30522	301	8	x	x	PROPN
fcis-30522	301	9	,	,	PUNCT
fcis-30522	301	10	wang	wang	PROPN
fcis-30522	301	11	x.	x.	PROPN
fcis-30522	301	12	quantum	quantum	PROPN
fcis-30522	301	13	self	self	NOUN
fcis-30522	301	14	-	-	PUNCT
fcis-30522	301	15	attention	attention	NOUN
fcis-30522	301	16	neural	neural	ADJ
fcis-30522	301	17	networks	network	NOUN
fcis-30522	301	18	for	for	ADP
fcis-30522	301	19	text	text	NOUN
fcis-30522	301	20	classification	classification	NOUN
fcis-30522	301	21	.	.	PUNCT
fcis-30522	302	1	science	science	PROPN
fcis-30522	302	2	china	china	PROPN
fcis-30522	302	3	information	information	PROPN
fcis-30522	302	4	sciences	sciences	PROPN
fcis-30522	302	5	.	.	PUNCT
fcis-30522	303	1	2024;67(4):142501	2024;67(4):142501	NUM
fcis-30522	303	2	.	.	PUNCT
fcis-30522	304	1	[	[	X
fcis-30522	304	2	23	23	NUM
fcis-30522	304	3	]	]	X
fcis-30522	304	4	liu	liu	PROPN
fcis-30522	304	5	c	c	PROPN
fcis-30522	304	6	,	,	PUNCT
fcis-30522	304	7	xu	xu	PROPN
fcis-30522	304	8	x.	x.	PROPN
fcis-30522	305	1	amff	amff	PROPN
fcis-30522	305	2	:	:	PUNCT
fcis-30522	305	3	a	a	DET
fcis-30522	305	4	new	new	ADJ
fcis-30522	305	5	attention	attention	NOUN
fcis-30522	305	6	-	-	PUNCT
fcis-30522	305	7	based	base	VERB
fcis-30522	305	8	multi	multi	ADJ
fcis-30522	305	9	-	-	ADJ
fcis-30522	305	10	feature	feature	ADJ
fcis-30522	305	11	fusion	fusion	NOUN
fcis-30522	305	12	method	method	NOUN
fcis-30522	305	13	for	for	ADP
fcis-30522	305	14	intention	intention	NOUN
fcis-30522	305	15	recognition	recognition	NOUN
fcis-30522	305	16	.	.	PUNCT
fcis-30522	306	1	knowledge	knowledge	NOUN
fcis-30522	306	2	-	-	PUNCT
fcis-30522	306	3	based	base	VERB
fcis-30522	306	4	systems	system	NOUN
fcis-30522	306	5	.	.	PUNCT
fcis-30522	307	1	2021;233:107525	2021;233:107525	NUM
fcis-30522	307	2	.	.	PUNCT
fcis-30522	308	1	[	[	X
fcis-30522	308	2	24	24	NUM
fcis-30522	308	3	]	]	X
fcis-30522	308	4	liu	liu	PROPN
fcis-30522	308	5	j	j	PROPN
fcis-30522	308	6	,	,	PUNCT
fcis-30522	309	1	li	li	PROPN
fcis-30522	309	2	d	d	PROPN
fcis-30522	309	3	,	,	PUNCT
fcis-30522	309	4	shan	shan	PROPN
fcis-30522	309	5	w	w	PROPN
fcis-30522	309	6	,	,	PUNCT
fcis-30522	309	7	liu	liu	PROPN
fcis-30522	309	8	s.	s.	PROPN
fcis-30522	309	9	a	a	DET
fcis-30522	309	10	feature	feature	NOUN
fcis-30522	309	11	selection	selection	NOUN
fcis-30522	309	12	method	method	NOUN
fcis-30522	309	13	based	base	VERB
fcis-30522	309	14	on	on	ADP
fcis-30522	309	15	multiple	multiple	ADJ
fcis-30522	309	16	feature	feature	NOUN
fcis-30522	309	17	subsets	subset	NOUN
fcis-30522	309	18	extraction	extraction	NOUN
fcis-30522	309	19	and	and	CCONJ
fcis-30522	309	20	result	result	VERB
fcis-30522	309	21	fusion	fusion	NOUN
fcis-30522	309	22	for	for	ADP
fcis-30522	309	23	improving	improve	VERB
fcis-30522	309	24	classification	classification	NOUN
fcis-30522	309	25	performance	performance	NOUN
fcis-30522	309	26	.	.	PUNCT
fcis-30522	310	1	applied	apply	VERB
fcis-30522	310	2	soft	soft	ADJ
fcis-30522	310	3	computing	computing	NOUN
fcis-30522	310	4	.	.	PUNCT
fcis-30522	311	1	2024	2024	NUM
fcis-30522	311	2	;	;	PUNCT
fcis-30522	311	3	150:111018	150:111018	NUM
fcis-30522	311	4	.	.	PUNCT
fcis-30522	312	1	[	[	X
fcis-30522	312	2	25	25	NUM
fcis-30522	312	3	]	]	PUNCT
fcis-30522	312	4	shorten	shorten	PROPN
fcis-30522	312	5	c	c	NOUN
fcis-30522	312	6	,	,	PUNCT
fcis-30522	312	7	khoshgoftaar	khoshgoftaar	PROPN
fcis-30522	312	8	tm	tm	PROPN
fcis-30522	312	9	,	,	PUNCT
fcis-30522	312	10	furht	furht	PROPN
fcis-30522	312	11	b.	b.	PROPN
fcis-30522	312	12	text	text	NOUN
fcis-30522	312	13	data	datum	NOUN
fcis-30522	312	14	augmentation	augmentation	NOUN
fcis-30522	312	15	for	for	ADP
fcis-30522	312	16	deep	deep	ADJ
fcis-30522	312	17	learning	learning	NOUN
fcis-30522	312	18	.	.	PUNCT
fcis-30522	313	1	journal	journal	NOUN
fcis-30522	313	2	of	of	ADP
fcis-30522	313	3	big	big	ADJ
fcis-30522	313	4	data	datum	NOUN
fcis-30522	313	5	.	.	PUNCT
fcis-30522	314	1	2021;8(1):101	2021;8(1):101	X
fcis-30522	314	2	.	.	PUNCT
fcis-30522	315	1	[	[	X
fcis-30522	315	2	26	26	NUM
fcis-30522	315	3	]	]	X
fcis-30522	315	4	wei	wei	PROPN
fcis-30522	315	5	j	j	PROPN
fcis-30522	315	6	,	,	PUNCT
fcis-30522	315	7	zou	zou	PROPN
fcis-30522	315	8	k.	k.	PROPN
fcis-30522	315	9	eda	eda	PROPN
fcis-30522	315	10	:	:	PUNCT
fcis-30522	315	11	easy	easy	ADJ
fcis-30522	315	12	data	datum	NOUN
fcis-30522	315	13	augmentation	augmentation	NOUN
fcis-30522	315	14	techniques	technique	NOUN
fcis-30522	315	15	for	for	ADP
fcis-30522	315	16	boosting	boost	VERB
fcis-30522	315	17	performance	performance	NOUN
fcis-30522	315	18	on	on	ADP
fcis-30522	315	19	text	text	NOUN
fcis-30522	315	20	classification	classification	NOUN
fcis-30522	315	21	tasks	task	NOUN
fcis-30522	315	22	.	.	PUNCT
fcis-30522	316	1	arxiv	arxiv	PROPN
fcis-30522	316	2	preprint	preprint	VERB
fcis-30522	316	3	arxiv:190111196	arxiv:190111196	PROPN
fcis-30522	316	4	.	.	PROPN
fcis-30522	316	5	2019	2019	NUM
fcis-30522	316	6	.	.	PUNCT
fcis-30522	317	1	[	[	X
fcis-30522	317	2	27	27	NUM
fcis-30522	317	3	]	]	PUNCT
fcis-30522	317	4	xia	xia	PROPN
fcis-30522	317	5	f	f	PROPN
fcis-30522	317	6	,	,	PUNCT
fcis-30522	317	7	weng	weng	PROPN
fcis-30522	317	8	y	y	PROPN
fcis-30522	317	9	,	,	PUNCT
fcis-30522	317	10	xia	xia	PROPN
fcis-30522	317	11	m	m	PROPN
fcis-30522	317	12	,	,	PUNCT
fcis-30522	317	13	yu	yu	PROPN
fcis-30522	317	14	q	q	PROPN
fcis-30522	317	15	,	,	PUNCT
fcis-30522	317	16	he	he	PRON
fcis-30522	317	17	s	s	PROPN
fcis-30522	317	18	,	,	PUNCT
fcis-30522	317	19	liu	liu	PROPN
fcis-30522	317	20	k	k	PROPN
fcis-30522	317	21	,	,	PUNCT
fcis-30522	317	22	et	et	PROPN
fcis-30522	317	23	al	al	PROPN
fcis-30522	317	24	.	.	PROPN
fcis-30522	317	25	,	,	PUNCT
fcis-30522	317	26	editors	editor	NOUN
fcis-30522	317	27	.	.	PUNCT
fcis-30522	318	1	does	do	AUX
fcis-30522	318	2	bert	bert	PROPN
fcis-30522	318	3	know	know	VERB
fcis-30522	318	4	which	which	PRON
fcis-30522	318	5	answer	answer	VERB
fcis-30522	318	6	beyond	beyond	ADP
fcis-30522	318	7	the	the	DET
fcis-30522	318	8	question	question	NOUN
fcis-30522	318	9	?	?	PUNCT
fcis-30522	319	1	china	china	PROPN
fcis-30522	319	2	conference	conference	PROPN
fcis-30522	319	3	on	on	ADP
fcis-30522	319	4	knowledge	knowledge	NOUN
fcis-30522	319	5	graph	graph	NOUN
fcis-30522	319	6	and	and	CCONJ
fcis-30522	319	7	semantic	semantic	ADJ
fcis-30522	319	8	computing	computing	NOUN
fcis-30522	319	9	;	;	PUNCT
fcis-30522	319	10	2021	2021	NUM
fcis-30522	319	11	:	:	PUNCT
fcis-30522	319	12	springer	springer	NOUN
fcis-30522	319	13	.	.	PUNCT
fcis-30522	320	1	[	[	X
fcis-30522	320	2	28	28	NUM
fcis-30522	320	3	]	]	X
fcis-30522	320	4	karimi	karimi	PROPN
fcis-30522	320	5	a	a	PROPN
fcis-30522	320	6	,	,	PUNCT
fcis-30522	320	7	rossi	rossi	ADJ
fcis-30522	320	8	l	l	PROPN
fcis-30522	320	9	,	,	PUNCT
fcis-30522	320	10	prati	prati	PROPN
fcis-30522	320	11	a.	a.	PROPN
fcis-30522	320	12	aeda	aeda	PROPN
fcis-30522	320	13	:	:	PUNCT
fcis-30522	320	14	an	an	DET
fcis-30522	320	15	easier	easy	ADJ
fcis-30522	320	16	data	datum	NOUN
fcis-30522	320	17	augmentation	augmentation	NOUN
fcis-30522	320	18	technique	technique	NOUN
fcis-30522	320	19	for	for	ADP
fcis-30522	320	20	text	text	NOUN
fcis-30522	320	21	classification	classification	NOUN
fcis-30522	320	22	.	.	PUNCT
fcis-30522	321	1	arxiv	arxiv	PROPN
fcis-30522	321	2	preprint	preprint	PROPN
fcis-30522	321	3	arxiv:210813230	arxiv:210813230	NOUN
fcis-30522	321	4	.	.	PROPN
fcis-30522	321	5	2021	2021	NUM
fcis-30522	321	6	.	.	PUNCT