id	sid	tid	token	lemma	pos
fcis-23012	1	1	frontiers	frontier	NOUN
fcis-23012	1	2	in	in	ADP
fcis-23012	1	3	computing	computing	NOUN
fcis-23012	1	4	and	and	CCONJ
fcis-23012	1	5	intelligent	intelligent	ADJ
fcis-23012	1	6	systems	system	NOUN
fcis-23012	1	7	issn	issn	VERB
fcis-23012	1	8	:	:	PUNCT
fcis-23012	1	9	2832	2832	NUM
fcis-23012	1	10	-	-	SYM
fcis-23012	1	11	6024	6024	NUM
fcis-23012	1	12	|	|	NOUN
fcis-23012	1	13	vol	vol	NOUN
fcis-23012	1	14	.	.	PROPN
fcis-23012	2	1	8	8	NUM
fcis-23012	2	2	,	,	PUNCT
fcis-23012	2	3	no	no	INTJ
fcis-23012	2	4	.	.	NOUN
fcis-23012	2	5	3	3	NUM
fcis-23012	2	6	,	,	PUNCT
fcis-23012	2	7	2024	2024	NUM
fcis-23012	2	8	70	70	NUM
fcis-23012	2	9	emotion	emotion	NOUN
fcis-23012	2	10	recognition	recognition	NOUN
fcis-23012	2	11	in	in	ADP
fcis-23012	2	12	lhasa	lhasa	PROPN
fcis-23012	2	13	tibetan	tibetan	PROPN
fcis-23012	2	14	speech	speech	NOUN
fcis-23012	2	15	based	base	VERB
fcis-23012	2	16	on	on	ADP
fcis-23012	2	17	bi‐lstm	bi‐lstm	NOUN
fcis-23012	2	18	graph	graph	NOUN
fcis-23012	2	19	convolutional	convolutional	ADJ
fcis-23012	2	20	networks	network	NOUN
fcis-23012	2	21	ang	ang	PROPN
fcis-23012	2	22	chen	chen	PROPN
fcis-23012	2	23	,	,	PUNCT
fcis-23012	2	24	rongzhao	rongzhao	PROPN
fcis-23012	2	25	huang	huang	PROPN
fcis-23012	2	26	,	,	PUNCT
fcis-23012	2	27	tong	tong	PROPN
fcis-23012	2	28	xi	xi	PROPN
fcis-23012	2	29	,	,	PUNCT
fcis-23012	2	30	liang	liang	PROPN
fcis-23012	2	31	wu	wu	PROPN
fcis-23012	2	32	,	,	PUNCT
fcis-23012	2	33	wangdui	wangdui	PROPN
fcis-23012	2	34	bianba	bianba	PROPN
fcis-23012	2	35	*	*	PUNCT
fcis-23012	2	36	school	school	NOUN
fcis-23012	2	37	of	of	ADP
fcis-23012	2	38	information	information	NOUN
fcis-23012	2	39	science	science	NOUN
fcis-23012	2	40	and	and	CCONJ
fcis-23012	2	41	technology	technology	NOUN
fcis-23012	2	42	,	,	PUNCT
fcis-23012	2	43	tibet	tibet	PROPN
fcis-23012	2	44	university	university	PROPN
fcis-23012	2	45	,	,	PUNCT
fcis-23012	2	46	lhasa	lhasa	PROPN
fcis-23012	2	47	,	,	PUNCT
fcis-23012	2	48	tibet	tibet	PROPN
fcis-23012	2	49	850000	850000	NUM
fcis-23012	2	50	,	,	PUNCT
fcis-23012	2	51	china	china	PROPN
fcis-23012	2	52	*	*	PUNCT
fcis-23012	2	53	corresponding	correspond	VERB
fcis-23012	2	54	author	author	NOUN
fcis-23012	2	55	:	:	PUNCT
fcis-23012	2	56	wangdui	wangdui	PROPN
fcis-23012	2	57	bianba	bianba	PROPN
fcis-23012	2	58	abstract	abstract	NOUN
fcis-23012	2	59	:	:	PUNCT
fcis-23012	2	60	speech	speech	NOUN
fcis-23012	2	61	emotion	emotion	NOUN
fcis-23012	2	62	recognition	recognition	NOUN
fcis-23012	2	63	(	(	PUNCT
fcis-23012	2	64	ser	ser	NOUN
fcis-23012	2	65	)	)	PUNCT
fcis-23012	2	66	is	be	AUX
fcis-23012	2	67	a	a	DET
fcis-23012	2	68	crucial	crucial	ADJ
fcis-23012	2	69	component	component	NOUN
fcis-23012	2	70	in	in	ADP
fcis-23012	2	71	the	the	DET
fcis-23012	2	72	field	field	NOUN
fcis-23012	2	73	of	of	ADP
fcis-23012	2	74	human	human	ADJ
fcis-23012	2	75	-	-	PUNCT
fcis-23012	2	76	computer	computer	NOUN
fcis-23012	2	77	interaction	interaction	NOUN
fcis-23012	2	78	(	(	PUNCT
fcis-23012	2	79	hci	hci	NOUN
fcis-23012	2	80	)	)	PUNCT
fcis-23012	2	81	,	,	PUNCT
fcis-23012	2	82	with	with	ADP
fcis-23012	2	83	significant	significant	ADJ
fcis-23012	2	84	research	research	NOUN
fcis-23012	2	85	and	and	CCONJ
fcis-23012	2	86	practical	practical	ADJ
fcis-23012	2	87	application	application	NOUN
fcis-23012	2	88	implications	implication	NOUN
fcis-23012	2	89	.	.	PUNCT
fcis-23012	3	1	however	however	ADV
fcis-23012	3	2	,	,	PUNCT
fcis-23012	3	3	due	due	ADP
fcis-23012	3	4	to	to	ADP
fcis-23012	3	5	the	the	DET
fcis-23012	3	6	complexity	complexity	NOUN
fcis-23012	3	7	of	of	ADP
fcis-23012	3	8	the	the	DET
fcis-23012	3	9	tibetan	tibetan	ADJ
fcis-23012	3	10	language	language	NOUN
fcis-23012	3	11	and	and	CCONJ
fcis-23012	3	12	the	the	DET
fcis-23012	3	13	scarcity	scarcity	NOUN
fcis-23012	3	14	of	of	ADP
fcis-23012	3	15	datasets	dataset	NOUN
fcis-23012	3	16	caused	cause	VERB
fcis-23012	3	17	by	by	ADP
fcis-23012	3	18	the	the	DET
fcis-23012	3	19	difficulty	difficulty	NOUN
fcis-23012	3	20	in	in	ADP
fcis-23012	3	21	collecting	collect	VERB
fcis-23012	3	22	various	various	ADJ
fcis-23012	3	23	dialects	dialect	NOUN
fcis-23012	3	24	,	,	PUNCT
fcis-23012	3	25	there	there	PRON
fcis-23012	3	26	are	be	VERB
fcis-23012	3	27	not	not	PART
fcis-23012	3	28	many	many	ADJ
fcis-23012	3	29	research	research	NOUN
fcis-23012	3	30	achievements	achievement	NOUN
fcis-23012	3	31	in	in	ADP
fcis-23012	3	32	tibetan	tibetan	ADJ
fcis-23012	3	33	speech	speech	NOUN
fcis-23012	3	34	recognition	recognition	NOUN
fcis-23012	3	35	.	.	PUNCT
fcis-23012	4	1	based	base	VERB
fcis-23012	4	2	on	on	ADP
fcis-23012	4	3	the	the	DET
fcis-23012	4	4	foundation	foundation	NOUN
fcis-23012	4	5	of	of	ADP
fcis-23012	4	6	constructing	construct	VERB
fcis-23012	4	7	a	a	DET
fcis-23012	4	8	tbls1	tbls1	ADJ
fcis-23012	4	9	dataset	dataset	NOUN
fcis-23012	4	10	containing	contain	VERB
fcis-23012	4	11	6,000	6,000	NUM
fcis-23012	4	12	tibetan	tibetan	ADJ
fcis-23012	4	13	-	-	PUNCT
fcis-23012	4	14	language	language	NOUN
fcis-23012	4	15	speech	speech	NOUN
fcis-23012	4	16	samples	sample	NOUN
fcis-23012	4	17	,	,	PUNCT
fcis-23012	4	18	an	an	DET
fcis-23012	4	19	approach	approach	NOUN
fcis-23012	4	20	was	be	AUX
fcis-23012	4	21	devised	devise	VERB
fcis-23012	4	22	for	for	ADP
fcis-23012	4	23	tibetan	tibetan	ADJ
fcis-23012	4	24	speech	speech	NOUN
fcis-23012	4	25	emotion	emotion	NOUN
fcis-23012	4	26	recognition	recognition	NOUN
fcis-23012	4	27	.	.	PUNCT
fcis-23012	5	1	this	this	DET
fcis-23012	5	2	approach	approach	NOUN
fcis-23012	5	3	leverages	leverage	VERB
fcis-23012	5	4	mfcc	mfcc	NOUN
fcis-23012	5	5	features	feature	NOUN
fcis-23012	5	6	and	and	CCONJ
fcis-23012	5	7	incorporates	incorporate	VERB
fcis-23012	5	8	a	a	DET
fcis-23012	5	9	bi	bi	ADJ
fcis-23012	5	10	-	-	ADJ
fcis-23012	5	11	directional	directional	ADJ
fcis-23012	5	12	long	long	ADJ
fcis-23012	5	13	short	short	ADJ
fcis-23012	5	14	-	-	PUNCT
fcis-23012	5	15	term	term	NOUN
fcis-23012	5	16	memory	memory	NOUN
fcis-23012	5	17	(	(	PUNCT
fcis-23012	5	18	bi	bi	NOUN
fcis-23012	5	19	-	-	ADJ
fcis-23012	5	20	lstm	lstm	ADJ
fcis-23012	5	21	)	)	PUNCT
fcis-23012	5	22	network	network	NOUN
fcis-23012	5	23	within	within	ADP
fcis-23012	5	24	a	a	DET
fcis-23012	5	25	graph	graph	NOUN
fcis-23012	5	26	convolutional	convolutional	ADJ
fcis-23012	5	27	neural	neural	ADJ
fcis-23012	5	28	network	network	NOUN
fcis-23012	5	29	.	.	PUNCT
fcis-23012	6	1	finally	finally	ADV
fcis-23012	6	2	,	,	PUNCT
fcis-23012	6	3	by	by	ADP
fcis-23012	6	4	comparing	compare	VERB
fcis-23012	6	5	the	the	DET
fcis-23012	6	6	performance	performance	NOUN
fcis-23012	6	7	of	of	ADP
fcis-23012	6	8	different	different	ADJ
fcis-23012	6	9	models	model	NOUN
fcis-23012	6	10	on	on	ADP
fcis-23012	6	11	this	this	DET
fcis-23012	6	12	dataset	dataset	NOUN
fcis-23012	6	13	,	,	PUNCT
fcis-23012	6	14	we	we	PRON
fcis-23012	6	15	demonstrated	demonstrate	VERB
fcis-23012	6	16	the	the	DET
fcis-23012	6	17	feasibility	feasibility	NOUN
fcis-23012	6	18	of	of	ADP
fcis-23012	6	19	our	our	PRON
fcis-23012	6	20	model	model	NOUN
fcis-23012	6	21	for	for	ADP
fcis-23012	6	22	tibetan	tibetan	ADJ
fcis-23012	6	23	speech	speech	NOUN
fcis-23012	6	24	emotion	emotion	NOUN
fcis-23012	6	25	recognition	recognition	NOUN
fcis-23012	6	26	.	.	PUNCT
fcis-23012	7	1	keywords	keyword	NOUN
fcis-23012	7	2	:	:	PUNCT
fcis-23012	7	3	tibetan	tibetan	ADJ
fcis-23012	7	4	speech	speech	NOUN
fcis-23012	7	5	emotion	emotion	NOUN
fcis-23012	7	6	recognition	recognition	NOUN
fcis-23012	7	7	;	;	PUNCT
fcis-23012	7	8	mel	mel	PROPN
fcis-23012	7	9	-	-	PUNCT
fcis-23012	7	10	frequency	frequency	ADJ
fcis-23012	7	11	cepstral	cepstral	ADJ
fcis-23012	7	12	coefficients	coefficient	NOUN
fcis-23012	7	13	(	(	PUNCT
fcis-23012	7	14	mfcc	mfcc	NOUN
fcis-23012	7	15	)	)	PUNCT
fcis-23012	7	16	;	;	PUNCT
fcis-23012	7	17	bidirectional	bidirectional	NOUN
fcis-23012	7	18	long	long	ADJ
fcis-23012	7	19	shortterm	shortterm	PROPN
fcis-23012	7	20	memory	memory	NOUN
fcis-23012	7	21	(	(	PUNCT
fcis-23012	7	22	bi	bi	NOUN
fcis-23012	7	23	-	-	NOUN
fcis-23012	7	24	lstm	lstm	ADJ
fcis-23012	7	25	)	)	PUNCT
fcis-23012	7	26	;	;	PUNCT
fcis-23012	7	27	graph	graph	NOUN
fcis-23012	7	28	convolution	convolution	NOUN
fcis-23012	7	29	network	network	NOUN
fcis-23012	7	30	(	(	PUNCT
fcis-23012	7	31	gcn	gcn	NOUN
fcis-23012	7	32	)	)	PUNCT
fcis-23012	7	33	.	.	PUNCT
fcis-23012	8	1	1	1	X
fcis-23012	8	2	.	.	X
fcis-23012	8	3	introduction	introduction	NOUN
fcis-23012	8	4	in	in	ADP
fcis-23012	8	5	today	today	NOUN
fcis-23012	8	6	’s	’s	PART
fcis-23012	8	7	increasingly	increasingly	ADV
fcis-23012	8	8	intelligent	intelligent	ADJ
fcis-23012	8	9	and	and	CCONJ
fcis-23012	8	10	automated	automated	ADJ
fcis-23012	8	11	technological	technological	ADJ
fcis-23012	8	12	context	context	NOUN
fcis-23012	8	13	,	,	PUNCT
fcis-23012	8	14	speech	speech	NOUN
fcis-23012	8	15	emotion	emotion	NOUN
fcis-23012	8	16	recognition	recognition	NOUN
fcis-23012	8	17	plays	play	VERB
fcis-23012	8	18	an	an	DET
fcis-23012	8	19	important	important	ADJ
fcis-23012	8	20	role	role	NOUN
fcis-23012	8	21	in	in	ADP
fcis-23012	8	22	human	human	ADJ
fcis-23012	8	23	-	-	PUNCT
fcis-23012	8	24	computer	computer	NOUN
fcis-23012	8	25	interaction	interaction	NOUN
fcis-23012	8	26	and	and	CCONJ
fcis-23012	8	27	is	be	AUX
fcis-23012	8	28	an	an	DET
fcis-23012	8	29	important	important	ADJ
fcis-23012	8	30	way	way	NOUN
fcis-23012	8	31	for	for	SCONJ
fcis-23012	8	32	artificial	artificial	ADJ
fcis-23012	8	33	intelligence	intelligence	NOUN
fcis-23012	8	34	to	to	PART
fcis-23012	8	35	understand	understand	VERB
fcis-23012	8	36	human	human	ADJ
fcis-23012	8	37	emotions	emotion	NOUN
fcis-23012	8	38	[	[	X
fcis-23012	8	39	1	1	NUM
fcis-23012	8	40	]	]	PUNCT
fcis-23012	8	41	.	.	PUNCT
fcis-23012	9	1	compared	compare	VERB
fcis-23012	9	2	to	to	ADP
fcis-23012	9	3	other	other	ADJ
fcis-23012	9	4	languages	language	NOUN
fcis-23012	9	5	,	,	PUNCT
fcis-23012	9	6	the	the	DET
fcis-23012	9	7	research	research	NOUN
fcis-23012	9	8	on	on	ADP
fcis-23012	9	9	tibetan	tibetan	ADJ
fcis-23012	9	10	speech	speech	NOUN
fcis-23012	9	11	emotion	emotion	NOUN
fcis-23012	9	12	recognition	recognition	NOUN
fcis-23012	9	13	started	start	VERB
fcis-23012	9	14	later	later	ADV
fcis-23012	9	15	,	,	PUNCT
fcis-23012	9	16	with	with	ADP
fcis-23012	9	17	fewer	few	ADJ
fcis-23012	9	18	participating	participate	VERB
fcis-23012	9	19	scholars	scholar	NOUN
fcis-23012	9	20	.	.	PUNCT
fcis-23012	10	1	moreover	moreover	ADV
fcis-23012	10	2	,	,	PUNCT
fcis-23012	10	3	due	due	ADP
fcis-23012	10	4	to	to	ADP
fcis-23012	10	5	the	the	DET
fcis-23012	10	6	large	large	ADJ
fcis-23012	10	7	dialectal	dialectal	ADJ
fcis-23012	10	8	differences	difference	NOUN
fcis-23012	10	9	in	in	ADP
fcis-23012	10	10	different	different	ADJ
fcis-23012	10	11	regions	region	NOUN
fcis-23012	10	12	of	of	ADP
fcis-23012	10	13	tibet	tibet	PROPN
fcis-23012	10	14	and	and	CCONJ
fcis-23012	10	15	inconvenient	inconvenient	ADJ
fcis-23012	10	16	transportation	transportation	NOUN
fcis-23012	10	17	,	,	PUNCT
fcis-23012	10	18	it	it	PRON
fcis-23012	10	19	is	be	AUX
fcis-23012	10	20	difficult	difficult	ADJ
fcis-23012	10	21	to	to	PART
fcis-23012	10	22	collect	collect	VERB
fcis-23012	10	23	datasets	dataset	NOUN
fcis-23012	10	24	,	,	PUNCT
fcis-23012	10	25	resulting	result	VERB
fcis-23012	10	26	in	in	ADP
fcis-23012	10	27	fewer	few	ADJ
fcis-23012	10	28	overall	overall	ADJ
fcis-23012	10	29	research	research	NOUN
fcis-23012	10	30	results	result	NOUN
fcis-23012	10	31	[	[	X
fcis-23012	10	32	2	2	NUM
fcis-23012	10	33	]	]	PUNCT
fcis-23012	10	34	.	.	PUNCT
fcis-23012	11	1	among	among	ADP
fcis-23012	11	2	the	the	DET
fcis-23012	11	3	tibetan	tibetan	NOUN
fcis-23012	11	4	-	-	PUNCT
fcis-23012	11	5	speaking	speak	VERB
fcis-23012	11	6	regions	region	NOUN
fcis-23012	11	7	,	,	PUNCT
fcis-23012	11	8	lhasa	lhasa	PROPN
fcis-23012	11	9	,	,	PUNCT
fcis-23012	11	10	as	as	ADP
fcis-23012	11	11	a	a	DET
fcis-23012	11	12	populous	populous	ADJ
fcis-23012	11	13	area	area	NOUN
fcis-23012	11	14	in	in	ADP
fcis-23012	11	15	tibet	tibet	PROPN
fcis-23012	11	16	,	,	PUNCT
fcis-23012	11	17	has	have	VERB
fcis-23012	11	18	a	a	DET
fcis-23012	11	19	larger	large	ADJ
fcis-23012	11	20	base	base	NOUN
fcis-23012	11	21	of	of	ADP
fcis-23012	11	22	speakers	speaker	NOUN
fcis-23012	11	23	of	of	ADP
fcis-23012	11	24	the	the	DET
fcis-23012	11	25	lhasa	lhasa	PROPN
fcis-23012	11	26	dialect	dialect	NOUN
fcis-23012	11	27	than	than	ADP
fcis-23012	11	28	other	other	ADJ
fcis-23012	11	29	tibetan	tibetan	ADJ
fcis-23012	11	30	dialects	dialect	NOUN
fcis-23012	11	31	.	.	PUNCT
fcis-23012	12	1	it	it	PRON
fcis-23012	12	2	is	be	AUX
fcis-23012	12	3	widely	widely	ADV
fcis-23012	12	4	spoken	speak	VERB
fcis-23012	12	5	,	,	PUNCT
fcis-23012	12	6	and	and	CCONJ
fcis-23012	12	7	it	it	PRON
fcis-23012	12	8	is	be	AUX
fcis-23012	12	9	relatively	relatively	ADV
fcis-23012	12	10	easy	easy	ADJ
fcis-23012	12	11	to	to	PART
fcis-23012	12	12	collect	collect	VERB
fcis-23012	12	13	data	datum	NOUN
fcis-23012	12	14	,	,	PUNCT
fcis-23012	12	15	making	make	VERB
fcis-23012	12	16	it	it	PRON
fcis-23012	12	17	easier	easy	ADJ
fcis-23012	12	18	to	to	PART
fcis-23012	12	19	collect	collect	VERB
fcis-23012	12	20	highquality	highquality	NOUN
fcis-23012	12	21	speech	speech	NOUN
fcis-23012	12	22	data	datum	NOUN
fcis-23012	12	23	.	.	PUNCT
fcis-23012	13	1	therefore	therefore	ADV
fcis-23012	13	2	,	,	PUNCT
fcis-23012	13	3	this	this	DET
fcis-23012	13	4	paper	paper	NOUN
fcis-23012	13	5	mainly	mainly	ADV
fcis-23012	13	6	focuses	focus	VERB
fcis-23012	13	7	on	on	ADP
fcis-23012	13	8	the	the	DET
fcis-23012	13	9	emotion	emotion	NOUN
fcis-23012	13	10	recognition	recognition	NOUN
fcis-23012	13	11	of	of	ADP
fcis-23012	13	12	the	the	DET
fcis-23012	13	13	lhasa	lhasa	PROPN
fcis-23012	13	14	dialect	dialect	NOUN
fcis-23012	13	15	.	.	PUNCT
fcis-23012	14	1	datasets	dataset	NOUN
fcis-23012	14	2	due	due	ADP
fcis-23012	14	3	to	to	ADP
fcis-23012	14	4	the	the	DET
fcis-23012	14	5	lack	lack	NOUN
fcis-23012	14	6	of	of	ADP
fcis-23012	14	7	large	large	ADJ
fcis-23012	14	8	-	-	PUNCT
fcis-23012	14	9	scale	scale	NOUN
fcis-23012	14	10	,	,	PUNCT
fcis-23012	14	11	high	high	ADJ
fcis-23012	14	12	-	-	PUNCT
fcis-23012	14	13	quality	quality	NOUN
fcis-23012	14	14	public	public	NOUN
fcis-23012	14	15	corpora	corpora	PROPN
fcis-23012	14	16	in	in	ADP
fcis-23012	14	17	tibetan	tibetan	PROPN
fcis-23012	14	18	,	,	PUNCT
fcis-23012	14	19	we	we	PRON
fcis-23012	14	20	built	build	VERB
fcis-23012	14	21	a	a	DET
fcis-23012	14	22	lhasa	lhasa	PROPN
fcis-23012	14	23	dialect	dialect	NOUN
fcis-23012	14	24	emotion	emotion	NOUN
fcis-23012	14	25	recognition	recognition	NOUN
fcis-23012	14	26	dataset	dataset	VERB
fcis-23012	14	27	for	for	ADP
fcis-23012	14	28	this	this	DET
fcis-23012	14	29	experiment	experiment	NOUN
fcis-23012	14	30	.	.	PUNCT
fcis-23012	15	1	this	this	DET
fcis-23012	15	2	dataset	dataset	NOUN
fcis-23012	15	3	was	be	AUX
fcis-23012	15	4	recorded	record	VERB
fcis-23012	15	5	by	by	ADP
fcis-23012	15	6	12	12	NUM
fcis-23012	15	7	tibetan	tibetan	ADJ
fcis-23012	15	8	students	student	NOUN
fcis-23012	15	9	,	,	PUNCT
fcis-23012	15	10	including	include	VERB
fcis-23012	15	11	6	6	NUM
fcis-23012	15	12	males	male	NOUN
fcis-23012	15	13	and	and	CCONJ
fcis-23012	15	14	6	6	NUM
fcis-23012	15	15	females	female	NOUN
fcis-23012	15	16	.	.	PUNCT
fcis-23012	16	1	we	we	PRON
fcis-23012	16	2	selected	select	VERB
fcis-23012	16	3	some	some	DET
fcis-23012	16	4	commonly	commonly	ADV
fcis-23012	16	5	used	use	VERB
fcis-23012	16	6	phrases	phrase	NOUN
fcis-23012	16	7	and	and	CCONJ
fcis-23012	16	8	words	word	NOUN
fcis-23012	16	9	in	in	ADP
fcis-23012	16	10	their	their	PRON
fcis-23012	16	11	daily	daily	ADJ
fcis-23012	16	12	lives	life	NOUN
fcis-23012	16	13	for	for	ADP
fcis-23012	16	14	recording	recording	NOUN
fcis-23012	16	15	.	.	PUNCT
fcis-23012	17	1	recording	recording	NOUN
fcis-23012	17	2	was	be	AUX
fcis-23012	17	3	done	do	VERB
fcis-23012	17	4	using	use	VERB
fcis-23012	17	5	adobe	adobe	PROPN
fcis-23012	17	6	audition	audition	PROPN
fcis-23012	17	7	22.0	22.0	NUM
fcis-23012	17	8	.	.	PUNCT
fcis-23012	18	1	each	each	DET
fcis-23012	18	2	student	student	NOUN
fcis-23012	18	3	recorded	record	VERB
fcis-23012	18	4	voices	voice	NOUN
fcis-23012	18	5	of	of	ADP
fcis-23012	18	6	5	5	NUM
fcis-23012	18	7	different	different	ADJ
fcis-23012	18	8	emotions	emotion	NOUN
fcis-23012	18	9	(	(	PUNCT
fcis-23012	18	10	anger	anger	NOUN
fcis-23012	18	11	,	,	PUNCT
fcis-23012	18	12	fear	fear	NOUN
fcis-23012	18	13	,	,	PUNCT
fcis-23012	18	14	happiness	happiness	NOUN
fcis-23012	18	15	,	,	PUNCT
fcis-23012	18	16	neutral	neutral	ADJ
fcis-23012	18	17	,	,	PUNCT
fcis-23012	18	18	sadness	sadness	NOUN
fcis-23012	18	19	)	)	PUNCT
fcis-23012	18	20	.	.	PUNCT
fcis-23012	19	1	each	each	DET
fcis-23012	19	2	emotion	emotion	NOUN
fcis-23012	19	3	includes	include	VERB
fcis-23012	19	4	100	100	NUM
fcis-23012	19	5	voice	voice	NOUN
fcis-23012	19	6	recordings	recording	NOUN
fcis-23012	19	7	,	,	PUNCT
fcis-23012	19	8	totaling	total	VERB
fcis-23012	19	9	6000	6000	NUM
fcis-23012	19	10	voice	voice	NOUN
fcis-23012	19	11	recordings	recording	NOUN
fcis-23012	19	12	,	,	PUNCT
fcis-23012	19	13	which	which	PRON
fcis-23012	19	14	make	make	VERB
fcis-23012	19	15	up	up	ADP
fcis-23012	19	16	the	the	DET
fcis-23012	19	17	dataset	dataset	ADJ
fcis-23012	19	18	tbls1	tbls1	NOUN
fcis-23012	19	19	used	use	VERB
fcis-23012	19	20	in	in	ADP
fcis-23012	19	21	this	this	DET
fcis-23012	19	22	experiment	experiment	NOUN
fcis-23012	19	23	.	.	PUNCT
fcis-23012	20	1	the	the	DET
fcis-23012	20	2	specific	specific	ADJ
fcis-23012	20	3	recording	recording	NOUN
fcis-23012	20	4	parameters	parameter	NOUN
fcis-23012	20	5	are	be	AUX
fcis-23012	20	6	:	:	PUNCT
fcis-23012	20	7	sampling	sample	VERB
fcis-23012	20	8	rate	rate	NOUN
fcis-23012	20	9	:	:	PUNCT
fcis-23012	20	10	16khz	16khz	NOUN
fcis-23012	20	11	,	,	PUNCT
fcis-23012	20	12	quantization	quantization	NOUN
fcis-23012	20	13	level	level	NOUN
fcis-23012	20	14	:	:	PUNCT
fcis-23012	20	15	16bit	16bit	NOUN
fcis-23012	20	16	,	,	PUNCT
fcis-23012	20	17	single	single	ADJ
fcis-23012	20	18	channel	channel	NOUN
fcis-23012	20	19	.	.	PUNCT
fcis-23012	21	1	the	the	DET
fcis-23012	21	2	participants	participant	NOUN
fcis-23012	21	3	in	in	ADP
fcis-23012	21	4	the	the	DET
fcis-23012	21	5	recording	recording	NOUN
fcis-23012	21	6	are	be	AUX
fcis-23012	21	7	not	not	PART
fcis-23012	21	8	professional	professional	ADJ
fcis-23012	21	9	actors	actor	NOUN
fcis-23012	21	10	,	,	PUNCT
fcis-23012	21	11	so	so	SCONJ
fcis-23012	21	12	there	there	PRON
fcis-23012	21	13	may	may	AUX
fcis-23012	21	14	be	be	AUX
fcis-23012	21	15	a	a	DET
fcis-23012	21	16	slight	slight	ADJ
fcis-23012	21	17	lack	lack	NOUN
fcis-23012	21	18	of	of	ADP
fcis-23012	21	19	authenticity	authenticity	NOUN
fcis-23012	21	20	in	in	ADP
fcis-23012	21	21	the	the	DET
fcis-23012	21	22	voice	voice	NOUN
fcis-23012	21	23	data	datum	NOUN
fcis-23012	21	24	.	.	PUNCT
fcis-23012	22	1	however	however	ADV
fcis-23012	22	2	,	,	PUNCT
fcis-23012	22	3	this	this	DET
fcis-23012	22	4	dataset	dataset	NOUN
fcis-23012	22	5	is	be	AUX
fcis-23012	22	6	balanced	balance	VERB
fcis-23012	22	7	in	in	ADP
fcis-23012	22	8	terms	term	NOUN
fcis-23012	22	9	of	of	ADP
fcis-23012	22	10	gender	gender	NOUN
fcis-23012	22	11	and	and	CCONJ
fcis-23012	22	12	the	the	DET
fcis-23012	22	13	number	number	NOUN
fcis-23012	22	14	of	of	ADP
fcis-23012	22	15	data	datum	NOUN
fcis-23012	22	16	points	point	NOUN
fcis-23012	22	17	for	for	ADP
fcis-23012	22	18	each	each	DET
fcis-23012	22	19	emotion	emotion	NOUN
fcis-23012	22	20	,	,	PUNCT
fcis-23012	22	21	so	so	CCONJ
fcis-23012	22	22	it	it	PRON
fcis-23012	22	23	is	be	AUX
fcis-23012	22	24	unlikely	unlikely	ADJ
fcis-23012	22	25	to	to	PART
fcis-23012	22	26	be	be	AUX
fcis-23012	22	27	affected	affect	VERB
fcis-23012	22	28	by	by	ADP
fcis-23012	22	29	the	the	DET
fcis-23012	22	30	imbalance	imbalance	NOUN
fcis-23012	22	31	of	of	ADP
fcis-23012	22	32	voice	voice	NOUN
fcis-23012	22	33	data	datum	NOUN
fcis-23012	22	34	volume	volume	NOUN
fcis-23012	22	35	from	from	ADP
fcis-23012	22	36	a	a	DET
fcis-23012	22	37	certain	certain	ADJ
fcis-23012	22	38	participant	participant	NOUN
fcis-23012	22	39	or	or	CCONJ
fcis-23012	22	40	emotion	emotion	NOUN
fcis-23012	22	41	.	.	PUNCT
fcis-23012	23	1	therefore	therefore	ADV
fcis-23012	23	2	,	,	PUNCT
fcis-23012	23	3	it	it	PRON
fcis-23012	23	4	can	can	AUX
fcis-23012	23	5	be	be	AUX
fcis-23012	23	6	considered	consider	VERB
fcis-23012	23	7	that	that	SCONJ
fcis-23012	23	8	the	the	DET
fcis-23012	23	9	model	model	NOUN
fcis-23012	23	10	’s	’s	PART
fcis-23012	23	11	performance	performance	NOUN
fcis-23012	23	12	on	on	ADP
fcis-23012	23	13	this	this	DET
fcis-23012	23	14	dataset	dataset	NOUN
fcis-23012	23	15	is	be	AUX
fcis-23012	23	16	highly	highly	ADV
fcis-23012	23	17	credible	credible	ADJ
fcis-23012	23	18	.	.	PUNCT
fcis-23012	24	1	2	2	X
fcis-23012	24	2	.	.	X
fcis-23012	24	3	tibetan	tibetan	ADJ
fcis-23012	24	4	speech	speech	NOUN
fcis-23012	24	5	emotion	emotion	NOUN
fcis-23012	24	6	feature	feature	NOUN
fcis-23012	24	7	extraction	extraction	NOUN
fcis-23012	24	8	2.1	2.1	NUM
fcis-23012	24	9	.	.	PUNCT
fcis-23012	25	1	mfcc	mfcc	PROPN
fcis-23012	25	2	mel	mel	PROPN
fcis-23012	25	3	-	-	PUNCT
fcis-23012	25	4	frequency	frequency	ADJ
fcis-23012	25	5	cepstral	cepstral	ADJ
fcis-23012	25	6	coefficients	coefficient	NOUN
fcis-23012	25	7	(	(	PUNCT
fcis-23012	25	8	mfcc	mfcc	NOUN
fcis-23012	25	9	)	)	PUNCT
fcis-23012	25	10	is	be	AUX
fcis-23012	25	11	a	a	DET
fcis-23012	25	12	very	very	ADV
fcis-23012	25	13	widely	widely	ADV
fcis-23012	25	14	used	use	VERB
fcis-23012	25	15	speech	speech	NOUN
fcis-23012	25	16	feature	feature	NOUN
fcis-23012	25	17	in	in	ADP
fcis-23012	25	18	the	the	DET
fcis-23012	25	19	field	field	NOUN
fcis-23012	25	20	of	of	ADP
fcis-23012	25	21	speech	speech	NOUN
fcis-23012	25	22	emotion	emotion	NOUN
fcis-23012	25	23	recognition[3	recognition[3	NOUN
fcis-23012	25	24	]	]	PUNCT
fcis-23012	25	25	.	.	PUNCT
fcis-23012	26	1	the	the	DET
fcis-23012	26	2	mel	mel	PROPN
fcis-23012	26	3	frequency	frequency	NOUN
fcis-23012	26	4	is	be	AUX
fcis-23012	26	5	a	a	DET
fcis-23012	26	6	frequency	frequency	NOUN
fcis-23012	26	7	scale	scale	NOUN
fcis-23012	26	8	established	establish	VERB
fcis-23012	26	9	based	base	VERB
fcis-23012	26	10	on	on	ADP
fcis-23012	26	11	the	the	DET
fcis-23012	26	12	human	human	ADJ
fcis-23012	26	13	auditory	auditory	ADJ
fcis-23012	26	14	system	system	NOUN
fcis-23012	26	15	’s	’s	PART
fcis-23012	26	16	different	different	ADJ
fcis-23012	26	17	acceptance	acceptance	NOUN
fcis-23012	26	18	levels	level	NOUN
fcis-23012	26	19	for	for	ADP
fcis-23012	26	20	sound	sound	ADJ
fcis-23012	26	21	signals	signal	NOUN
fcis-23012	26	22	at	at	ADP
fcis-23012	26	23	different	different	ADJ
fcis-23012	26	24	frequency	frequency	NOUN
fcis-23012	26	25	bands	band	NOUN
fcis-23012	26	26	,	,	PUNCT
fcis-23012	26	27	and	and	CCONJ
fcis-23012	26	28	it	it	PRON
fcis-23012	26	29	has	have	VERB
fcis-23012	26	30	a	a	DET
fcis-23012	26	31	nonlinear	nonlinear	ADJ
fcis-23012	26	32	relationship	relationship	NOUN
fcis-23012	26	33	with	with	ADP
fcis-23012	26	34	frequency	frequency	NOUN
fcis-23012	26	35	.	.	PUNCT
fcis-23012	27	1	the	the	DET
fcis-23012	27	2	correspondence	correspondence	NOUN
fcis-23012	27	3	between	between	ADP
fcis-23012	27	4	the	the	DET
fcis-23012	27	5	mel	mel	PROPN
fcis-23012	27	6	frequency	frequency	PROPN
fcis-23012	27	7	scale	scale	NOUN
fcis-23012	27	8	and	and	CCONJ
fcis-23012	27	9	the	the	DET
fcis-23012	27	10	normal	normal	ADJ
fcis-23012	27	11	frequency	frequency	NOUN
fcis-23012	27	12	is	be	AUX
fcis-23012	27	13	as	as	SCONJ
fcis-23012	27	14	follows	follow	VERB
fcis-23012	27	15	:	:	PUNCT
fcis-23012	27	16	)	)	PUNCT
fcis-23012	28	1	100/1(log*2595	100/1(log*2595	NUM
fcis-23012	28	2	)	)	PUNCT
fcis-23012	28	3	(	(	PUNCT
fcis-23012	28	4	10	10	NUM
fcis-23012	28	5	ffmel	ffmel	NOUN
fcis-23012	28	6			ADJ
fcis-23012	28	7	(	(	PUNCT
fcis-23012	28	8	1	1	NUM
fcis-23012	28	9	)	)	PUNCT
fcis-23012	28	10	figure	figure	NOUN
fcis-23012	28	11	1	1	NUM
fcis-23012	28	12	.	.	PUNCT
fcis-23012	28	13	mfcc	mfcc	NOUN
fcis-23012	28	14	features	feature	VERB
fcis-23012	28	15	extraction	extraction	NOUN
fcis-23012	28	16	mfcc	mfcc	NOUN
fcis-23012	28	17	features	feature	NOUN
fcis-23012	28	18	are	be	AUX
fcis-23012	28	19	the	the	DET
fcis-23012	28	20	cepstral	cepstral	ADJ
fcis-23012	28	21	coefficients	coefficient	NOUN
fcis-23012	28	22	obtained	obtain	VERB
fcis-23012	28	23	by	by	ADP
fcis-23012	28	24	taking	take	VERB
fcis-23012	28	25	the	the	DET
fcis-23012	28	26	logarithm	logarithm	NOUN
fcis-23012	28	27	of	of	ADP
fcis-23012	28	28	the	the	DET
fcis-23012	28	29	mel	mel	PROPN
fcis-23012	28	30	spectrum	spectrum	PROPN
fcis-23012	28	31	and	and	CCONJ
fcis-23012	28	32	then	then	ADV
fcis-23012	28	33	performing	perform	VERB
fcis-23012	28	34	an	an	DET
fcis-23012	28	35	inverse	inverse	NOUN
fcis-23012	28	36	transformation	transformation	NOUN
fcis-23012	28	37	.	.	PUNCT
fcis-23012	29	1	the	the	DET
fcis-23012	29	2	process	process	NOUN
fcis-23012	29	3	of	of	ADP
fcis-23012	29	4	extracting	extract	VERB
fcis-23012	29	5	mfcc	mfcc	NOUN
fcis-23012	29	6	usually	usually	ADV
fcis-23012	29	7	includes	include	VERB
fcis-23012	29	8	pre	pre	ADJ
fcis-23012	29	9	-	-	NOUN
fcis-23012	29	10	emphasis	emphasis	ADJ
fcis-23012	29	11	,	,	PUNCT
fcis-23012	29	12	framing	framing	NOUN
fcis-23012	29	13	,	,	PUNCT
fcis-23012	29	14	windowing	windowing	NOUN
fcis-23012	29	15	,	,	PUNCT
fcis-23012	29	16	etc	etc	X
fcis-23012	29	17	.	.	X
fcis-23012	29	18	preemphasis	preemphasis	NOUN
fcis-23012	29	19	is	be	AUX
fcis-23012	29	20	to	to	PART
fcis-23012	29	21	enhance	enhance	VERB
fcis-23012	29	22	the	the	DET
fcis-23012	29	23	high	high	ADJ
fcis-23012	29	24	-	-	PUNCT
fcis-23012	29	25	frequency	frequency	NOUN
fcis-23012	29	26	part	part	NOUN
fcis-23012	29	27	of	of	ADP
fcis-23012	29	28	the	the	DET
fcis-23012	29	29	speech	speech	NOUN
fcis-23012	29	30	signal	signal	NOUN
fcis-23012	29	31	;	;	PUNCT
fcis-23012	29	32	while	while	SCONJ
fcis-23012	29	33	framing	frame	VERB
fcis-23012	29	34	and	and	CCONJ
fcis-23012	29	35	windowing	windowing	NOUN
fcis-23012	29	36	are	be	AUX
fcis-23012	29	37	to	to	PART
fcis-23012	29	38	perform	perform	VERB
fcis-23012	29	39	local	local	ADJ
fcis-23012	29	40	analysis	analysis	NOUN
fcis-23012	29	41	of	of	ADP
fcis-23012	29	42	the	the	DET
fcis-23012	29	43	signal	signal	NOUN
fcis-23012	29	44	in	in	ADP
fcis-23012	29	45	the	the	DET
fcis-23012	29	46	time	time	NOUN
fcis-23012	29	47	domain	domain	NOUN
fcis-23012	29	48	.	.	PUNCT
fcis-23012	30	1	subsequently	subsequently	ADV
fcis-23012	30	2	,	,	PUNCT
fcis-23012	30	3	the	the	DET
fcis-23012	30	4	data	data	NOUN
fcis-23012	30	5	is	be	AUX
fcis-23012	30	6	subjected	subject	VERB
fcis-23012	30	7	to	to	ADP
fcis-23012	30	8	fft	fft	PROPN
fcis-23012	30	9	,	,	PUNCT
fcis-23012	30	10	passed	pass	VERB
fcis-23012	30	11	through	through	ADP
fcis-23012	30	12	a	a	DET
fcis-23012	30	13	mel	mel	PROPN
fcis-23012	30	14	filter	filter	NOUN
fcis-23012	30	15	,	,	PUNCT
fcis-23012	30	16	and	and	CCONJ
fcis-23012	30	17	the	the	DET
fcis-23012	30	18	71	71	NUM
fcis-23012	30	19	logarithm	logarithm	NOUN
fcis-23012	30	20	is	be	AUX
fcis-23012	30	21	taken	take	VERB
fcis-23012	30	22	and	and	CCONJ
fcis-23012	30	23	then	then	ADV
fcis-23012	30	24	the	the	DET
fcis-23012	30	25	required	require	VERB
fcis-23012	30	26	cepstral	cepstral	ADJ
fcis-23012	30	27	coefficients	coefficient	NOUN
fcis-23012	30	28	are	be	AUX
fcis-23012	30	29	obtained	obtain	VERB
fcis-23012	30	30	through	through	ADP
fcis-23012	30	31	the	the	DET
fcis-23012	30	32	discrete	discrete	ADJ
fcis-23012	30	33	cosine	cosine	NOUN
fcis-23012	30	34	transform	transform	NOUN
fcis-23012	30	35	(	(	PUNCT
fcis-23012	30	36	dct)[4	dct)[4	X
fcis-23012	30	37	]	]	PUNCT
fcis-23012	30	38	.	.	PUNCT
fcis-23012	31	1	this	this	DET
fcis-23012	31	2	experiment	experiment	NOUN
fcis-23012	31	3	extracted	extract	VERB
fcis-23012	31	4	the	the	DET
fcis-23012	31	5	first	first	ADJ
fcis-23012	31	6	12	12	NUM
fcis-23012	31	7	mfcc	mfcc	NOUN
fcis-23012	31	8	coefficients	coefficient	NOUN
fcis-23012	31	9	,	,	PUNCT
fcis-23012	31	10	because	because	SCONJ
fcis-23012	31	11	the	the	DET
fcis-23012	31	12	first	first	ADJ
fcis-23012	31	13	12	12	NUM
fcis-23012	31	14	cepstral	cepstral	ADJ
fcis-23012	31	15	coefficients	coefficient	NOUN
fcis-23012	31	16	correspond	correspond	VERB
fcis-23012	31	17	to	to	ADP
fcis-23012	31	18	the	the	DET
fcis-23012	31	19	low	low	ADJ
fcis-23012	31	20	-	-	PUNCT
fcis-23012	31	21	frequency	frequency	NOUN
fcis-23012	31	22	band	band	NOUN
fcis-23012	31	23	,	,	PUNCT
fcis-23012	31	24	which	which	PRON
fcis-23012	31	25	contains	contain	VERB
fcis-23012	31	26	more	more	ADJ
fcis-23012	31	27	emotion	emotion	NOUN
fcis-23012	31	28	-	-	PUNCT
fcis-23012	31	29	related	relate	VERB
fcis-23012	31	30	information	information	NOUN
fcis-23012	31	31	compared	compare	VERB
fcis-23012	31	32	to	to	ADP
fcis-23012	31	33	the	the	DET
fcis-23012	31	34	high	high	ADJ
fcis-23012	31	35	-	-	PUNCT
fcis-23012	31	36	frequency	frequency	NOUN
fcis-23012	31	37	band	band	NOUN
fcis-23012	31	38	[	[	X
fcis-23012	31	39	5	5	NUM
fcis-23012	31	40	]	]	PUNCT
fcis-23012	31	41	.	.	PUNCT
fcis-23012	32	1	in	in	ADP
fcis-23012	32	2	addition	addition	NOUN
fcis-23012	32	3	,	,	PUNCT
fcis-23012	32	4	this	this	DET
fcis-23012	32	5	experiment	experiment	NOUN
fcis-23012	32	6	combines	combine	VERB
fcis-23012	32	7	the	the	DET
fcis-23012	32	8	mfcc	mfcc	NOUN
fcis-23012	32	9	features	feature	VERB
fcis-23012	32	10	with	with	ADP
fcis-23012	32	11	its	its	PRON
fcis-23012	32	12	first	first	ADJ
fcis-23012	32	13	-	-	PUNCT
fcis-23012	32	14	order	order	NOUN
fcis-23012	32	15	difference	difference	NOUN
fcis-23012	32	16	,	,	PUNCT
fcis-23012	32	17	zero	zero	NUM
fcis-23012	32	18	-	-	PUNCT
fcis-23012	32	19	crossing	crossing	NOUN
fcis-23012	32	20	rate	rate	NOUN
fcis-23012	32	21	,	,	PUNCT
fcis-23012	32	22	fundamental	fundamental	ADJ
fcis-23012	32	23	frequency	frequency	NOUN
fcis-23012	32	24	,	,	PUNCT
fcis-23012	32	25	frame	frame	NOUN
fcis-23012	32	26	energy	energy	NOUN
fcis-23012	32	27	(	(	PUNCT
fcis-23012	32	28	short	short	ADJ
fcis-23012	32	29	-	-	PUNCT
fcis-23012	32	30	time	time	NOUN
fcis-23012	32	31	energy	energy	NOUN
fcis-23012	32	32	of	of	ADP
fcis-23012	32	33	speech	speech	NOUN
fcis-23012	32	34	)	)	PUNCT
fcis-23012	32	35	,	,	PUNCT
fcis-23012	32	36	and	and	CCONJ
fcis-23012	32	37	other	other	ADJ
fcis-23012	32	38	features	feature	NOUN
fcis-23012	32	39	as	as	ADP
fcis-23012	32	40	the	the	DET
fcis-23012	32	41	final	final	ADJ
fcis-23012	32	42	input	input	NOUN
fcis-23012	32	43	features	feature	NOUN
fcis-23012	32	44	.	.	PUNCT
fcis-23012	33	1	the	the	DET
fcis-23012	33	2	final	final	ADJ
fcis-23012	33	3	feature	feature	NOUN
fcis-23012	33	4	set	set	NOUN
fcis-23012	33	5	contains	contain	VERB
fcis-23012	33	6	35	35	NUM
fcis-23012	33	7	features	feature	NOUN
fcis-23012	33	8	for	for	ADP
fcis-23012	33	9	each	each	DET
fcis-23012	33	10	frame	frame	NOUN
fcis-23012	33	11	of	of	ADP
fcis-23012	33	12	speech	speech	NOUN
fcis-23012	33	13	.	.	PUNCT
fcis-23012	34	1	2.2	2.2	NUM
fcis-23012	34	2	.	.	PUNCT
fcis-23012	35	1	other	other	ADJ
fcis-23012	35	2	features	feature	VERB
fcis-23012	35	3	the	the	DET
fcis-23012	35	4	zero	zero	NUM
fcis-23012	35	5	-	-	PUNCT
fcis-23012	35	6	crossing	crossing	NOUN
fcis-23012	35	7	rate	rate	NOUN
fcis-23012	35	8	is	be	AUX
fcis-23012	35	9	the	the	DET
fcis-23012	35	10	number	number	NOUN
fcis-23012	35	11	of	of	ADP
fcis-23012	35	12	times	time	NOUN
fcis-23012	35	13	the	the	DET
fcis-23012	35	14	sample	sample	NOUN
fcis-23012	35	15	values	value	NOUN
fcis-23012	35	16	in	in	ADP
fcis-23012	35	17	each	each	DET
fcis-23012	35	18	frame	frame	NOUN
fcis-23012	35	19	of	of	ADP
fcis-23012	35	20	the	the	DET
fcis-23012	35	21	sound	sound	ADJ
fcis-23012	35	22	signal	signal	NOUN
fcis-23012	35	23	pass	pass	VERB
fcis-23012	35	24	through	through	ADP
fcis-23012	35	25	zero	zero	NUM
fcis-23012	35	26	.	.	PUNCT
fcis-23012	36	1	since	since	SCONJ
fcis-23012	36	2	the	the	DET
fcis-23012	36	3	zero	zero	NUM
fcis-23012	36	4	-	-	PUNCT
fcis-23012	36	5	crossing	crossing	NOUN
fcis-23012	36	6	rates	rate	NOUN
fcis-23012	36	7	of	of	ADP
fcis-23012	36	8	unvoiced	unvoiced	ADJ
fcis-23012	36	9	and	and	CCONJ
fcis-23012	36	10	voiced	voice	VERB
fcis-23012	36	11	sounds	sound	NOUN
fcis-23012	36	12	differ	differ	VERB
fcis-23012	36	13	greatly	greatly	ADV
fcis-23012	36	14	,	,	PUNCT
fcis-23012	36	15	the	the	DET
fcis-23012	36	16	zero	zero	NUM
fcis-23012	36	17	-	-	PUNCT
fcis-23012	36	18	crossing	crossing	NOUN
fcis-23012	36	19	rate	rate	NOUN
fcis-23012	36	20	is	be	AUX
fcis-23012	36	21	generally	generally	ADV
fcis-23012	36	22	introduced	introduce	VERB
fcis-23012	36	23	as	as	ADP
fcis-23012	36	24	a	a	DET
fcis-23012	36	25	distinguishing	distinguish	VERB
fcis-23012	36	26	feature	feature	NOUN
fcis-23012	36	27	in	in	ADP
fcis-23012	36	28	the	the	DET
fcis-23012	36	29	discrimination	discrimination	NOUN
fcis-23012	36	30	of	of	ADP
fcis-23012	36	31	these	these	DET
fcis-23012	36	32	two	two	NUM
fcis-23012	36	33	.	.	PUNCT
fcis-23012	37	1	the	the	DET
fcis-23012	37	2	zero	zero	NUM
fcis-23012	37	3	-	-	PUNCT
fcis-23012	37	4	crossing	crossing	NOUN
fcis-23012	37	5	rate	rate	NOUN
fcis-23012	37	6	of	of	ADP
fcis-23012	37	7	unvoiced	unvoiced	ADJ
fcis-23012	37	8	sounds	sound	NOUN
fcis-23012	37	9	is	be	AUX
fcis-23012	37	10	generally	generally	ADV
fcis-23012	37	11	significantly	significantly	ADV
fcis-23012	37	12	greater	great	ADJ
fcis-23012	37	13	than	than	ADP
fcis-23012	37	14	that	that	PRON
fcis-23012	37	15	of	of	ADP
fcis-23012	37	16	voiced	voice	VERB
fcis-23012	37	17	sounds	sound	NOUN
fcis-23012	37	18	.	.	PUNCT
fcis-23012	38	1	it	it	PRON
fcis-23012	38	2	should	should	AUX
fcis-23012	38	3	be	be	AUX
fcis-23012	38	4	noted	note	VERB
fcis-23012	38	5	that	that	SCONJ
fcis-23012	38	6	the	the	DET
fcis-23012	38	7	zero	zero	NUM
fcis-23012	38	8	-	-	PUNCT
fcis-23012	38	9	crossing	crossing	NOUN
fcis-23012	38	10	rate	rate	NOUN
fcis-23012	38	11	of	of	ADP
fcis-23012	38	12	noise	noise	NOUN
fcis-23012	38	13	is	be	AUX
fcis-23012	38	14	close	close	ADJ
fcis-23012	38	15	to	to	ADP
fcis-23012	38	16	that	that	PRON
fcis-23012	38	17	of	of	ADP
fcis-23012	38	18	unvoiced	unvoiced	ADJ
fcis-23012	38	19	sounds	sound	NOUN
fcis-23012	38	20	,	,	PUNCT
fcis-23012	38	21	and	and	CCONJ
fcis-23012	38	22	the	the	DET
fcis-23012	38	23	discrimination	discrimination	NOUN
fcis-23012	38	24	between	between	ADP
fcis-23012	38	25	unvoiced	unvoiced	ADJ
fcis-23012	38	26	sounds	sound	NOUN
fcis-23012	38	27	and	and	CCONJ
fcis-23012	38	28	noise	noise	NOUN
fcis-23012	38	29	needs	need	NOUN
fcis-23012	38	30	to	to	PART
fcis-23012	38	31	be	be	AUX
fcis-23012	38	32	done	do	VERB
fcis-23012	38	33	through	through	ADP
fcis-23012	38	34	other	other	ADJ
fcis-23012	38	35	features	feature	NOUN
fcis-23012	38	36	.	.	PUNCT
fcis-23012	39	1	the	the	DET
fcis-23012	39	2	calculation	calculation	NOUN
fcis-23012	39	3	process	process	NOUN
fcis-23012	39	4	is	be	AUX
fcis-23012	39	5	as	as	SCONJ
fcis-23012	39	6	follows	follow	NOUN
fcis-23012	39	7	,	,	PUNCT
fcis-23012	39	8	ω(n	ω(n	NUM
fcis-23012	39	9	)	)	PUNCT
fcis-23012	39	10	is	be	AUX
fcis-23012	39	11	the	the	DET
fcis-23012	39	12	window	window	NOUN
fcis-23012	39	13	function	function	NOUN
fcis-23012	39	14	,	,	PUNCT
fcis-23012	39	15	we	we	PRON
fcis-23012	39	16	chose	choose	VERB
fcis-23012	39	17	hamming	hamming	NOUN
fcis-23012	39	18	window	window	NOUN
fcis-23012	39	19	:	:	PUNCT
fcis-23012	39	20	)	)	PUNCT
fcis-23012	39	21	(	(	PUNCT
fcis-23012	39	22	*	*	NOUN
fcis-23012	39	23	|)]1(sgn[)](sgn[|	|)]1(sgn[)](sgn[|	X
fcis-23012	39	24	mnmxmxz	mnmxmxz	NOUN
fcis-23012	39	25	m	m	VERB
fcis-23012	39	26	n	n	PRON
fcis-23012	39	27			NOUN
fcis-23012	39	28			X
fcis-23012	39	29			ADJ
fcis-23012	39	30			PROPN
fcis-23012	39	31			X
fcis-23012	39	32	(	(	PUNCT
fcis-23012	39	33	2	2	NUM
fcis-23012	39	34	)	)	PUNCT
fcis-23012	39	35	the	the	DET
fcis-23012	39	36	fundamental	fundamental	ADJ
fcis-23012	39	37	frequency	frequency	NOUN
fcis-23012	39	38	contains	contain	VERB
fcis-23012	39	39	a	a	DET
fcis-23012	39	40	large	large	ADJ
fcis-23012	39	41	number	number	NOUN
fcis-23012	39	42	of	of	ADP
fcis-23012	39	43	basic	basic	ADJ
fcis-23012	39	44	features	feature	NOUN
fcis-23012	39	45	of	of	ADP
fcis-23012	39	46	the	the	DET
fcis-23012	39	47	sound	sound	NOUN
fcis-23012	39	48	signal	signal	NOUN
fcis-23012	39	49	.	.	PUNCT
fcis-23012	40	1	the	the	DET
fcis-23012	40	2	fundamental	fundamental	ADJ
fcis-23012	40	3	frequency	frequency	NOUN
fcis-23012	40	4	is	be	AUX
fcis-23012	40	5	the	the	DET
fcis-23012	40	6	lowest	low	ADJ
fcis-23012	40	7	frequency	frequency	NOUN
fcis-23012	40	8	component	component	NOUN
fcis-23012	40	9	in	in	ADP
fcis-23012	40	10	the	the	DET
fcis-23012	40	11	sound	sound	NOUN
fcis-23012	40	12	signal	signal	NOUN
fcis-23012	40	13	,	,	PUNCT
fcis-23012	40	14	corresponding	correspond	VERB
fcis-23012	40	15	to	to	ADP
fcis-23012	40	16	the	the	DET
fcis-23012	40	17	period	period	NOUN
fcis-23012	40	18	of	of	ADP
fcis-23012	40	19	vocal	vocal	ADJ
fcis-23012	40	20	cord	cord	NOUN
fcis-23012	40	21	vibration	vibration	NOUN
fcis-23012	40	22	.	.	PUNCT
fcis-23012	41	1	it	it	PRON
fcis-23012	41	2	is	be	AUX
fcis-23012	41	3	closely	closely	ADV
fcis-23012	41	4	related	relate	VERB
fcis-23012	41	5	to	to	ADP
fcis-23012	41	6	the	the	DET
fcis-23012	41	7	speaker	speaker	NOUN
fcis-23012	41	8	’s	’s	PART
fcis-23012	41	9	tone	tone	NOUN
fcis-23012	41	10	,	,	PUNCT
fcis-23012	41	11	pitch	pitch	NOUN
fcis-23012	41	12	,	,	PUNCT
fcis-23012	41	13	and	and	CCONJ
fcis-23012	41	14	emotional	emotional	ADJ
fcis-23012	41	15	state	state	NOUN
fcis-23012	41	16	.	.	PUNCT
fcis-23012	42	1	generally	generally	ADV
fcis-23012	42	2	speaking	speak	VERB
fcis-23012	42	3	,	,	PUNCT
fcis-23012	42	4	a	a	DET
fcis-23012	42	5	higher	high	ADJ
fcis-23012	42	6	fundamental	fundamental	ADJ
fcis-23012	42	7	frequency	frequency	NOUN
fcis-23012	42	8	usually	usually	ADV
fcis-23012	42	9	corresponds	correspond	VERB
fcis-23012	42	10	to	to	ADP
fcis-23012	42	11	a	a	DET
fcis-23012	42	12	higher	high	ADJ
fcis-23012	42	13	pitch	pitch	NOUN
fcis-23012	42	14	,	,	PUNCT
fcis-23012	42	15	while	while	SCONJ
fcis-23012	42	16	a	a	DET
fcis-23012	42	17	lower	low	ADJ
fcis-23012	42	18	fundamental	fundamental	ADJ
fcis-23012	42	19	frequency	frequency	NOUN
fcis-23012	42	20	corresponds	correspond	VERB
fcis-23012	42	21	to	to	ADP
fcis-23012	42	22	a	a	DET
fcis-23012	42	23	lower	low	ADJ
fcis-23012	42	24	pitch	pitch	NOUN
fcis-23012	42	25	.	.	PUNCT
fcis-23012	43	1	changes	change	NOUN
fcis-23012	43	2	in	in	ADP
fcis-23012	43	3	tone	tone	NOUN
fcis-23012	43	4	can	can	AUX
fcis-23012	43	5	convey	convey	VERB
fcis-23012	43	6	emotional	emotional	ADJ
fcis-23012	43	7	information	information	NOUN
fcis-23012	43	8	,	,	PUNCT
fcis-23012	43	9	such	such	ADJ
fcis-23012	43	10	as	as	ADP
fcis-23012	43	11	happiness	happiness	NOUN
fcis-23012	43	12	,	,	PUNCT
fcis-23012	43	13	excitement	excitement	NOUN
fcis-23012	43	14	,	,	PUNCT
fcis-23012	43	15	or	or	CCONJ
fcis-23012	43	16	depression	depression	NOUN
fcis-23012	43	17	.	.	PUNCT
fcis-23012	44	1	short	short	ADJ
fcis-23012	44	2	-	-	PUNCT
fcis-23012	44	3	time	time	NOUN
fcis-23012	44	4	energy	energy	NOUN
fcis-23012	44	5	of	of	ADP
fcis-23012	44	6	speech	speech	NOUN
fcis-23012	44	7	also	also	ADV
fcis-23012	44	8	has	have	VERB
fcis-23012	44	9	certain	certain	ADJ
fcis-23012	44	10	value	value	NOUN
fcis-23012	44	11	in	in	ADP
fcis-23012	44	12	speech	speech	NOUN
fcis-23012	44	13	emotion	emotion	NOUN
fcis-23012	44	14	recognition	recognition	NOUN
fcis-23012	44	15	.	.	PUNCT
fcis-23012	45	1	generally	generally	ADV
fcis-23012	45	2	,	,	PUNCT
fcis-23012	45	3	the	the	DET
fcis-23012	45	4	energy	energy	NOUN
fcis-23012	45	5	of	of	ADP
fcis-23012	45	6	the	the	DET
fcis-23012	45	7	speech	speech	NOUN
fcis-23012	45	8	signal	signal	NOUN
fcis-23012	45	9	is	be	AUX
fcis-23012	45	10	calculated	calculate	VERB
fcis-23012	45	11	frame	frame	NOUN
fcis-23012	45	12	by	by	ADP
fcis-23012	45	13	frame	frame	NOUN
fcis-23012	45	14	.	.	PUNCT
fcis-23012	46	1	this	this	DET
fcis-23012	46	2	feature	feature	NOUN
fcis-23012	46	3	can	can	AUX
fcis-23012	46	4	better	well	ADV
fcis-23012	46	5	distinguish	distinguish	VERB
fcis-23012	46	6	between	between	ADP
fcis-23012	46	7	unvoiced	unvoiced	ADJ
fcis-23012	46	8	and	and	CCONJ
fcis-23012	46	9	voiced	voice	VERB
fcis-23012	46	10	sounds	sound	NOUN
fcis-23012	46	11	,	,	PUNCT
fcis-23012	46	12	with	with	ADP
fcis-23012	46	13	the	the	DET
fcis-23012	46	14	short	short	ADJ
fcis-23012	46	15	-	-	PUNCT
fcis-23012	46	16	time	time	NOUN
fcis-23012	46	17	energy	energy	NOUN
fcis-23012	46	18	of	of	ADP
fcis-23012	46	19	voiced	voice	VERB
fcis-23012	46	20	sounds	sound	NOUN
fcis-23012	46	21	generally	generally	ADV
fcis-23012	46	22	much	much	ADV
fcis-23012	46	23	higher	high	ADJ
fcis-23012	46	24	than	than	ADP
fcis-23012	46	25	that	that	PRON
fcis-23012	46	26	of	of	ADP
fcis-23012	46	27	unvoiced	unvoiced	ADJ
fcis-23012	46	28	sounds	sound	NOUN
fcis-23012	46	29	.	.	PUNCT
fcis-23012	47	1	in	in	ADP
fcis-23012	47	2	terms	term	NOUN
fcis-23012	47	3	of	of	ADP
fcis-23012	47	4	emotion	emotion	NOUN
fcis-23012	47	5	recognition	recognition	NOUN
fcis-23012	47	6	,	,	PUNCT
fcis-23012	47	7	the	the	DET
fcis-23012	47	8	energy	energy	NOUN
fcis-23012	47	9	of	of	ADP
fcis-23012	47	10	intense	intense	ADJ
fcis-23012	47	11	emotions	emotion	NOUN
fcis-23012	47	12	such	such	ADJ
fcis-23012	47	13	as	as	ADP
fcis-23012	47	14	anger	anger	NOUN
fcis-23012	47	15	and	and	CCONJ
fcis-23012	47	16	happiness	happiness	NOUN
fcis-23012	47	17	are	be	AUX
fcis-23012	47	18	usually	usually	ADV
fcis-23012	47	19	higher	high	ADJ
fcis-23012	47	20	than	than	ADP
fcis-23012	47	21	that	that	PRON
fcis-23012	47	22	of	of	ADP
fcis-23012	47	23	neutral	neutral	ADJ
fcis-23012	47	24	and	and	CCONJ
fcis-23012	47	25	sad	sad	ADJ
fcis-23012	47	26	emotions	emotion	NOUN
fcis-23012	47	27	.	.	PUNCT
fcis-23012	48	1	the	the	DET
fcis-23012	48	2	calculation	calculation	NOUN
fcis-23012	48	3	of	of	ADP
fcis-23012	48	4	the	the	DET
fcis-23012	48	5	short	short	ADJ
fcis-23012	48	6	-	-	PUNCT
fcis-23012	48	7	time	time	NOUN
fcis-23012	48	8	energy	energy	NOUN
fcis-23012	48	9	en	en	ADP
fcis-23012	48	10	corresponding	correspond	VERB
fcis-23012	48	11	to	to	ADP
fcis-23012	48	12	the	the	DET
fcis-23012	48	13	window	window	NOUN
fcis-23012	48	14	function	function	NOUN
fcis-23012	48	15	starting	start	VERB
fcis-23012	48	16	from	from	ADP
fcis-23012	48	17	the	the	DET
fcis-23012	48	18	nth	nth	NOUN
fcis-23012	48	19	point	point	NOUN
fcis-23012	48	20	is	be	AUX
fcis-23012	48	21	as	as	SCONJ
fcis-23012	48	22	follows	follow	VERB
fcis-23012	48	23	:	:	PUNCT
fcis-23012	48	24	ω(n	ω(n	NUM
fcis-23012	48	25	)	)	PUNCT
fcis-23012	48	26	is	be	AUX
fcis-23012	48	27	the	the	DET
fcis-23012	48	28	window	window	NOUN
fcis-23012	48	29	function	function	NOUN
fcis-23012	48	30	,	,	PUNCT
fcis-23012	48	31	h(n)=ω(n)2	h(n)=ω(n)2	PROPN
fcis-23012	48	32	;	;	PUNCT
fcis-23012	48	33	)	)	PUNCT
fcis-23012	48	34	(	(	PUNCT
fcis-23012	48	35	*	*	NOUN
fcis-23012	48	36	)	)	PUNCT
fcis-23012	48	37	(	(	PUNCT
fcis-23012	48	38	)	)	PUNCT
fcis-23012	48	39	(	(	PUNCT
fcis-23012	48	40	)	)	PUNCT
fcis-23012	48	41	(	(	PUNCT
fcis-23012	48	42	)	)	PUNCT
fcis-23012	48	43	]	]	X
fcis-23012	48	44	(	(	PUNCT
fcis-23012	48	45	)	)	PUNCT
fcis-23012	48	46	(	(	PUNCT
fcis-23012	48	47	[	[	PUNCT
fcis-23012	48	48	222	222	NUM
fcis-23012	48	49	nhnxmnhmxmnmxe	nhnxmnhmxmnmxe	ADJ
fcis-23012	48	50	mm	mm	NOUN
fcis-23012	48	51	n	n	PRON
fcis-23012	48	52			NOUN
fcis-23012	48	53			ADP
fcis-23012	48	54			VERB
fcis-23012	48	55			PROPN
fcis-23012	48	56			VERB
fcis-23012	48	57			PROPN
fcis-23012	48	58			X
fcis-23012	48	59	(	(	PUNCT
fcis-23012	48	60	3	3	X
fcis-23012	48	61	)	)	PUNCT
fcis-23012	48	62	the	the	DET
fcis-23012	48	63	features	feature	NOUN
fcis-23012	48	64	are	be	AUX
fcis-23012	48	65	concatenated	concatenate	VERB
fcis-23012	48	66	to	to	PART
fcis-23012	48	67	obtain	obtain	VERB
fcis-23012	48	68	the	the	DET
fcis-23012	48	69	final	final	ADJ
fcis-23012	48	70	feature	feature	NOUN
fcis-23012	48	71	set	set	NOUN
fcis-23012	48	72	.	.	PUNCT
fcis-23012	49	1	3	3	X
fcis-23012	49	2	.	.	NOUN
fcis-23012	49	3	speech	speech	NOUN
fcis-23012	49	4	emotion	emotion	NOUN
fcis-23012	49	5	recognition	recognition	NOUN
fcis-23012	49	6	net	net	ADJ
fcis-23012	49	7	3.1	3.1	NUM
fcis-23012	49	8	.	.	PUNCT
fcis-23012	50	1	graph	graph	NOUN
fcis-23012	50	2	convolution	convolution	NOUN
fcis-23012	50	3	neural	neural	ADJ
fcis-23012	50	4	networks	network	NOUN
fcis-23012	50	5	compared	compare	VERB
fcis-23012	50	6	to	to	ADP
fcis-23012	50	7	traditional	traditional	ADJ
fcis-23012	50	8	neural	neural	ADJ
fcis-23012	50	9	networks	network	NOUN
fcis-23012	50	10	,	,	PUNCT
fcis-23012	50	11	graph	graph	NOUN
fcis-23012	50	12	neural	neural	ADJ
fcis-23012	50	13	networks	network	NOUN
fcis-23012	50	14	(	(	PUNCT
fcis-23012	50	15	gnns	gnns	NOUN
fcis-23012	50	16	)	)	PUNCT
fcis-23012	50	17	have	have	VERB
fcis-23012	50	18	a	a	DET
fcis-23012	50	19	unique	unique	ADJ
fcis-23012	50	20	structure	structure	NOUN
fcis-23012	50	21	and	and	CCONJ
fcis-23012	50	22	inter	inter	ADJ
fcis-23012	50	23	-	-	ADJ
fcis-23012	50	24	data	data	ADJ
fcis-23012	50	25	relationship	relationship	NOUN
fcis-23012	50	26	,	,	PUNCT
fcis-23012	50	27	and	and	CCONJ
fcis-23012	50	28	they	they	PRON
fcis-23012	50	29	have	have	AUX
fcis-23012	50	30	shown	show	VERB
fcis-23012	50	31	impressive	impressive	ADJ
fcis-23012	50	32	performance	performance	NOUN
fcis-23012	50	33	in	in	ADP
fcis-23012	50	34	the	the	DET
fcis-23012	50	35	field	field	NOUN
fcis-23012	50	36	of	of	ADP
fcis-23012	50	37	speech	speech	NOUN
fcis-23012	50	38	emotion	emotion	NOUN
fcis-23012	50	39	recognition	recognition	NOUN
fcis-23012	50	40	.	.	PUNCT
fcis-23012	51	1	they	they	PRON
fcis-23012	51	2	first	first	ADV
fcis-23012	51	3	need	need	VERB
fcis-23012	51	4	to	to	PART
fcis-23012	51	5	convert	convert	VERB
fcis-23012	51	6	the	the	DET
fcis-23012	51	7	input	input	NOUN
fcis-23012	51	8	data	datum	NOUN
fcis-23012	51	9	into	into	ADP
fcis-23012	51	10	a	a	DET
fcis-23012	51	11	graph	graph	NOUN
fcis-23012	51	12	structure	structure	NOUN
fcis-23012	51	13	,	,	PUNCT
fcis-23012	51	14	where	where	SCONJ
fcis-23012	51	15	the	the	DET
fcis-23012	51	16	smallest	small	ADJ
fcis-23012	51	17	specific	specific	ADJ
fcis-23012	51	18	data	datum	NOUN
fcis-23012	51	19	units	unit	NOUN
fcis-23012	51	20	in	in	ADP
fcis-23012	51	21	the	the	DET
fcis-23012	51	22	graph	graph	NOUN
fcis-23012	51	23	serve	serve	VERB
fcis-23012	51	24	as	as	ADP
fcis-23012	51	25	nodes	node	NOUN
fcis-23012	51	26	.	.	PUNCT
fcis-23012	52	1	different	different	ADJ
fcis-23012	52	2	nodes	node	NOUN
fcis-23012	52	3	are	be	AUX
fcis-23012	52	4	connected	connect	VERB
fcis-23012	52	5	by	by	ADP
fcis-23012	52	6	edges	edge	NOUN
fcis-23012	52	7	,	,	PUNCT
fcis-23012	52	8	which	which	PRON
fcis-23012	52	9	reflect	reflect	VERB
fcis-23012	52	10	the	the	DET
fcis-23012	52	11	relationships	relationship	NOUN
fcis-23012	52	12	between	between	ADP
fcis-23012	52	13	different	different	ADJ
fcis-23012	52	14	nodes	node	NOUN
fcis-23012	52	15	.	.	PUNCT
fcis-23012	53	1	the	the	DET
fcis-23012	53	2	construction	construction	NOUN
fcis-23012	53	3	of	of	ADP
fcis-23012	53	4	the	the	DET
fcis-23012	53	5	graph	graph	NOUN
fcis-23012	53	6	can	can	AUX
fcis-23012	53	7	be	be	AUX
fcis-23012	53	8	divided	divide	VERB
fcis-23012	53	9	into	into	ADP
fcis-23012	53	10	cyclic	cyclic	ADJ
fcis-23012	53	11	graphs	graph	NOUN
fcis-23012	53	12	and	and	CCONJ
fcis-23012	53	13	chain	chain	NOUN
fcis-23012	53	14	graphs	graph	NOUN
fcis-23012	53	15	(	(	PUNCT
fcis-23012	53	16	closed	closed	ADJ
fcis-23012	53	17	-	-	PUNCT
fcis-23012	53	18	loop	loop	NOUN
fcis-23012	53	19	and	and	CCONJ
fcis-23012	53	20	open	open	ADJ
fcis-23012	53	21	-	-	PUNCT
fcis-23012	53	22	loop	loop	NOUN
fcis-23012	53	23	)	)	PUNCT
fcis-23012	53	24	.	.	PUNCT
fcis-23012	54	1	the	the	DET
fcis-23012	54	2	difference	difference	NOUN
fcis-23012	54	3	is	be	AUX
fcis-23012	54	4	that	that	SCONJ
fcis-23012	54	5	in	in	ADP
fcis-23012	54	6	cyclic	cyclic	ADJ
fcis-23012	54	7	graphs	graph	NOUN
fcis-23012	54	8	,	,	PUNCT
fcis-23012	54	9	there	there	PRON
fcis-23012	54	10	is	be	VERB
fcis-23012	54	11	an	an	DET
fcis-23012	54	12	edge	edge	NOUN
fcis-23012	54	13	connecting	connect	VERB
fcis-23012	54	14	the	the	DET
fcis-23012	54	15	first	first	ADJ
fcis-23012	54	16	and	and	CCONJ
fcis-23012	54	17	last	last	ADJ
fcis-23012	54	18	nodes	node	NOUN
fcis-23012	54	19	,	,	PUNCT
fcis-23012	54	20	which	which	PRON
fcis-23012	54	21	is	be	AUX
fcis-23012	54	22	often	often	ADV
fcis-23012	54	23	used	use	VERB
fcis-23012	54	24	to	to	PART
fcis-23012	54	25	handle	handle	VERB
fcis-23012	54	26	data	datum	NOUN
fcis-23012	54	27	with	with	ADP
fcis-23012	54	28	cyclic	cyclic	ADJ
fcis-23012	54	29	relationships	relationship	NOUN
fcis-23012	54	30	,	,	PUNCT
fcis-23012	54	31	and	and	CCONJ
fcis-23012	54	32	information	information	NOUN
fcis-23012	54	33	can	can	AUX
fcis-23012	54	34	propagate	propagate	VERB
fcis-23012	54	35	in	in	ADP
fcis-23012	54	36	the	the	DET
fcis-23012	54	37	loop	loop	NOUN
fcis-23012	54	38	.	.	PUNCT
fcis-23012	55	1	on	on	ADP
fcis-23012	55	2	the	the	DET
fcis-23012	55	3	contrary	contrary	NOUN
fcis-23012	55	4	,	,	PUNCT
fcis-23012	55	5	in	in	ADP
fcis-23012	55	6	chain	chain	NOUN
fcis-23012	55	7	graphs	graph	NOUN
fcis-23012	55	8	,	,	PUNCT
fcis-23012	55	9	the	the	DET
fcis-23012	55	10	data	datum	NOUN
fcis-23012	55	11	can	can	AUX
fcis-23012	55	12	not	not	PART
fcis-23012	55	13	cycle	cycle	VERB
fcis-23012	55	14	from	from	ADP
fcis-23012	55	15	the	the	DET
fcis-23012	55	16	end	end	NOUN
fcis-23012	55	17	to	to	ADP
fcis-23012	55	18	the	the	DET
fcis-23012	55	19	beginning	beginning	NOUN
fcis-23012	55	20	,	,	PUNCT
fcis-23012	55	21	and	and	CCONJ
fcis-23012	55	22	the	the	DET
fcis-23012	55	23	information	information	NOUN
fcis-23012	55	24	propagation	propagation	NOUN
fcis-23012	55	25	is	be	AUX
fcis-23012	55	26	unidirectional	unidirectional	ADJ
fcis-23012	55	27	[	[	X
fcis-23012	55	28	6	6	NUM
fcis-23012	55	29	]	]	PUNCT
fcis-23012	55	30	.	.	PUNCT
fcis-23012	56	1	figure	figure	NOUN
fcis-23012	56	2	2	2	NUM
fcis-23012	56	3	.	.	NOUN
fcis-23012	56	4	cycle	cycle	NOUN
fcis-23012	56	5	graph	graph	NOUN
fcis-23012	56	6	and	and	CCONJ
fcis-23012	56	7	line	line	NOUN
fcis-23012	56	8	graph	graph	NOUN
fcis-23012	56	9	in	in	ADP
fcis-23012	56	10	this	this	DET
fcis-23012	56	11	experiment	experiment	NOUN
fcis-23012	56	12	,	,	PUNCT
fcis-23012	56	13	a	a	DET
fcis-23012	56	14	simple	simple	ADJ
fcis-23012	56	15	frame	frame	NOUN
fcis-23012	56	16	-	-	PUNCT
fcis-23012	56	17	to	to	ADP
fcis-23012	56	18	-	-	PUNCT
fcis-23012	56	19	node	node	ADJ
fcis-23012	56	20	conversion	conversion	NOUN
fcis-23012	56	21	is	be	AUX
fcis-23012	56	22	used	use	VERB
fcis-23012	56	23	,	,	PUNCT
fcis-23012	56	24	and	and	CCONJ
fcis-23012	56	25	each	each	DET
fcis-23012	56	26	frame	frame	NOUN
fcis-23012	56	27	node	node	NOUN
fcis-23012	56	28	is	be	AUX
fcis-23012	56	29	connected	connect	VERB
fcis-23012	56	30	by	by	ADP
fcis-23012	56	31	a	a	DET
fcis-23012	56	32	unidirectional	unidirectional	ADJ
fcis-23012	56	33	edge	edge	NOUN
fcis-23012	56	34	according	accord	VERB
fcis-23012	56	35	to	to	ADP
fcis-23012	56	36	the	the	DET
fcis-23012	56	37	time	time	NOUN
fcis-23012	56	38	series	series	PROPN
fcis-23012	56	39	.	.	PUNCT
fcis-23012	57	1	since	since	SCONJ
fcis-23012	57	2	the	the	DET
fcis-23012	57	3	speech	speech	NOUN
fcis-23012	57	4	data	datum	NOUN
fcis-23012	57	5	is	be	AUX
fcis-23012	57	6	not	not	PART
fcis-23012	57	7	a	a	DET
fcis-23012	57	8	cyclic	cyclic	ADJ
fcis-23012	57	9	data	data	NOUN
fcis-23012	57	10	structure	structure	NOUN
fcis-23012	57	11	from	from	ADP
fcis-23012	57	12	beginning	begin	VERB
fcis-23012	57	13	to	to	ADP
fcis-23012	57	14	end	end	NOUN
fcis-23012	57	15	,	,	PUNCT
fcis-23012	57	16	this	this	DET
fcis-23012	57	17	experiment	experiment	NOUN
fcis-23012	57	18	constructs	construct	VERB
fcis-23012	57	19	the	the	DET
fcis-23012	57	20	speech	speech	NOUN
fcis-23012	57	21	data	datum	NOUN
fcis-23012	57	22	as	as	ADP
fcis-23012	57	23	a	a	DET
fcis-23012	57	24	chain	chain	NOUN
fcis-23012	57	25	graph	graph	NOUN
fcis-23012	57	26	.	.	PUNCT
fcis-23012	58	1	each	each	DET
fcis-23012	58	2	speech	speech	NOUN
fcis-23012	58	3	is	be	AUX
fcis-23012	58	4	an	an	DET
fcis-23012	58	5	independent	independent	ADJ
fcis-23012	58	6	chain	chain	NOUN
fcis-23012	58	7	graph	graph	NOUN
fcis-23012	58	8	,	,	PUNCT
fcis-23012	58	9	with	with	ADP
fcis-23012	58	10	frames	frame	NOUN
fcis-23012	58	11	as	as	ADP
fcis-23012	58	12	nodes	node	NOUN
fcis-23012	58	13	,	,	PUNCT
fcis-23012	58	14	connected	connect	VERB
fcis-23012	58	15	by	by	ADP
fcis-23012	58	16	edges	edge	NOUN
fcis-23012	58	17	in	in	ADP
fcis-23012	58	18	the	the	DET
fcis-23012	58	19	order	order	NOUN
fcis-23012	58	20	of	of	ADP
fcis-23012	58	21	the	the	DET
fcis-23012	58	22	time	time	NOUN
fcis-23012	58	23	series	series	NOUN
fcis-23012	58	24	,	,	PUNCT
fcis-23012	58	25	and	and	CCONJ
fcis-23012	58	26	each	each	DET
fcis-23012	58	27	node	node	NOUN
fcis-23012	58	28	is	be	AUX
fcis-23012	58	29	associated	associate	VERB
fcis-23012	58	30	with	with	ADP
fcis-23012	58	31	a	a	DET
fcis-23012	58	32	node	node	ADJ
fcis-23012	58	33	feature	feature	NOUN
fcis-23012	58	34	vector[7	vector[7	ADV
fcis-23012	58	35	]	]	PUNCT
fcis-23012	58	36	,	,	PUNCT
fcis-23012	58	37	the	the	DET
fcis-23012	58	38	node	node	ADJ
fcis-23012	58	39	feature	feature	NOUN
fcis-23012	58	40	vector	vector	NOUN
fcis-23012	58	41	contains	contain	VERB
fcis-23012	58	42	the	the	DET
fcis-23012	58	43	35	35	NUM
fcis-23012	58	44	-	-	PUNCT
fcis-23012	58	45	dimensional	dimensional	ADJ
fcis-23012	58	46	feature	feature	NOUN
fcis-23012	58	47	information	information	NOUN
fcis-23012	58	48	on	on	ADP
fcis-23012	58	49	that	that	DET
fcis-23012	58	50	node	node	NOUN
fcis-23012	58	51	frame	frame	NOUN
fcis-23012	58	52	.	.	PUNCT
fcis-23012	59	1	3.2	3.2	NUM
fcis-23012	59	2	.	.	PUNCT
fcis-23012	60	1	graph	graph	NOUN
fcis-23012	60	2	convolution	convolution	NOUN
fcis-23012	60	3	figure	figure	NOUN
fcis-23012	60	4	3	3	NUM
fcis-23012	60	5	.	.	PUNCT
fcis-23012	60	6	node	node	ADJ
fcis-23012	60	7	aggregation	aggregation	NOUN
fcis-23012	60	8	method	method	NOUN
fcis-23012	60	9	in	in	ADP
fcis-23012	60	10	graph	graph	NOUN
fcis-23012	60	11	convolution	convolution	NOUN
fcis-23012	60	12	layer	layer	NOUN
fcis-23012	60	13	in	in	ADP
fcis-23012	60	14	the	the	DET
fcis-23012	60	15	graph	graph	NOUN
fcis-23012	60	16	structure	structure	NOUN
fcis-23012	60	17	,	,	PUNCT
fcis-23012	60	18	the	the	DET
fcis-23012	60	19	behavior	behavior	NOUN
fcis-23012	60	20	of	of	ADP
fcis-23012	60	21	a	a	DET
fcis-23012	60	22	node	node	ADJ
fcis-23012	60	23	aggregating	aggregate	VERB
fcis-23012	60	24	information	information	NOUN
fcis-23012	60	25	from	from	ADP
fcis-23012	60	26	adjacent	adjacent	ADJ
fcis-23012	60	27	nodes	node	NOUN
fcis-23012	60	28	with	with	ADP
fcis-23012	60	29	its	its	PRON
fcis-23012	60	30	own	own	ADJ
fcis-23012	60	31	information	information	NOUN
fcis-23012	60	32	is	be	AUX
fcis-23012	60	33	called	call	VERB
fcis-23012	60	34	graph	graph	NOUN
fcis-23012	60	35	convolution	convolution	NOUN
fcis-23012	60	36	.	.	PUNCT
fcis-23012	61	1	aggregation	aggregation	NOUN
fcis-23012	61	2	methods	method	NOUN
fcis-23012	61	3	usually	usually	ADV
fcis-23012	61	4	include	include	VERB
fcis-23012	61	5	taking	take	VERB
fcis-23012	61	6	the	the	DET
fcis-23012	61	7	average	average	ADJ
fcis-23012	61	8	,	,	PUNCT
fcis-23012	61	9	sum	sum	NOUN
fcis-23012	61	10	,	,	PUNCT
fcis-23012	61	11	or	or	CCONJ
fcis-23012	61	12	maximum	maximum	NOUN
fcis-23012	61	13	.	.	PUNCT
fcis-23012	62	1	this	this	DET
fcis-23012	62	2	paper	paper	NOUN
fcis-23012	62	3	uses	use	VERB
fcis-23012	62	4	the	the	DET
fcis-23012	62	5	sum	sum	NOUN
fcis-23012	62	6	aggregation	aggregation	NOUN
fcis-23012	62	7	method	method	NOUN
fcis-23012	62	8	.	.	PUNCT
fcis-23012	63	1	after	after	ADP
fcis-23012	63	2	performing	perform	VERB
fcis-23012	63	3	a	a	DET
fcis-23012	63	4	convolution	convolution	NOUN
fcis-23012	63	5	,	,	PUNCT
fcis-23012	63	6	the	the	DET
fcis-23012	63	7	node	node	NOUN
fcis-23012	63	8	contains	contain	VERB
fcis-23012	63	9	information	information	NOUN
fcis-23012	63	10	from	from	ADP
fcis-23012	63	11	adjacent	adjacent	ADJ
fcis-23012	63	12	nodes	node	NOUN
fcis-23012	63	13	.	.	PUNCT
fcis-23012	64	1	based	base	VERB
fcis-23012	64	2	on	on	ADP
fcis-23012	64	3	this	this	PRON
fcis-23012	64	4	,	,	PUNCT
fcis-23012	64	5	repeating	repeat	VERB
fcis-23012	64	6	the	the	DET
fcis-23012	64	7	graph	graph	NOUN
fcis-23012	64	8	convolution	convolution	NOUN
fcis-23012	64	9	operation	operation	NOUN
fcis-23012	64	10	m	m	PROPN
fcis-23012	64	11	times	time	NOUN
fcis-23012	64	12	can	can	AUX
fcis-23012	64	13	achieve	achieve	VERB
fcis-23012	64	14	the	the	DET
fcis-23012	64	15	effect	effect	NOUN
fcis-23012	64	16	of	of	ADP
fcis-23012	64	17	aggregating	aggregate	VERB
fcis-23012	64	18	the	the	DET
fcis-23012	64	19	information	information	NOUN
fcis-23012	64	20	of	of	ADP
fcis-23012	64	21	nodes	node	NOUN
fcis-23012	64	22	at	at	ADP
fcis-23012	64	23	a	a	DET
fcis-23012	64	24	stride	stride	NOUN
fcis-23012	64	25	of	of	ADP
fcis-23012	64	26	m	m	PRON
fcis-23012	64	27	to	to	ADP
fcis-23012	64	28	one	one	NUM
fcis-23012	64	29	node	node	NOUN
fcis-23012	64	30	[	[	X
fcis-23012	64	31	8	8	NUM
fcis-23012	64	32	]	]	PUNCT
fcis-23012	64	33	,	,	PUNCT
fcis-23012	64	34	therefore	therefore	ADV
fcis-23012	64	35	,	,	PUNCT
fcis-23012	64	36	the	the	DET
fcis-23012	64	37	graph	graph	NOUN
fcis-23012	64	38	convolution	convolution	NOUN
fcis-23012	64	39	structure	structure	NOUN
fcis-23012	64	40	has	have	VERB
fcis-23012	64	41	the	the	DET
fcis-23012	64	42	advantage	advantage	NOUN
fcis-23012	64	43	of	of	ADP
fcis-23012	64	44	being	be	AUX
fcis-23012	64	45	able	able	ADJ
fcis-23012	64	46	to	to	PART
fcis-23012	64	47	capture	capture	VERB
fcis-23012	64	48	local	local	ADJ
fcis-23012	64	49	information	information	NOUN
fcis-23012	64	50	correlations	correlation	NOUN
fcis-23012	64	51	well	well	INTJ
fcis-23012	65	1	[	[	X
fcis-23012	65	2	9	9	NUM
fcis-23012	65	3	]	]	PUNCT
fcis-23012	65	4	.	.	PUNCT
fcis-23012	66	1	if	if	SCONJ
fcis-23012	66	2	the	the	DET
fcis-23012	66	3	number	number	NOUN
fcis-23012	66	4	of	of	ADP
fcis-23012	66	5	layers	layer	NOUN
fcis-23012	66	6	in	in	ADP
fcis-23012	66	7	the	the	DET
fcis-23012	66	8	graph	graph	NOUN
fcis-23012	66	9	convolution	convolution	NOUN
fcis-23012	66	10	is	be	AUX
fcis-23012	66	11	too	too	ADV
fcis-23012	66	12	high	high	ADJ
fcis-23012	66	13	,	,	PUNCT
fcis-23012	66	14	it	it	PRON
fcis-23012	66	15	can	can	AUX
fcis-23012	66	16	lead	lead	VERB
fcis-23012	66	17	to	to	ADP
fcis-23012	66	18	the	the	DET
fcis-23012	66	19	node	node	NOUN
fcis-23012	66	20	aggregating	aggregate	VERB
fcis-23012	66	21	too	too	ADV
fcis-23012	66	22	much	much	ADJ
fcis-23012	66	23	information	information	NOUN
fcis-23012	66	24	from	from	ADP
fcis-23012	66	25	neighboring	neighboring	NOUN
fcis-23012	66	26	nodes	node	NOUN
fcis-23012	66	27	and	and	CCONJ
fcis-23012	66	28	diluting	dilute	VERB
fcis-23012	66	29	its	its	PRON
fcis-23012	66	30	own	own	ADJ
fcis-23012	66	31	information	information	NOUN
fcis-23012	66	32	.	.	PUNCT
fcis-23012	67	1	generally	generally	ADV
fcis-23012	67	2	,	,	PUNCT
fcis-23012	67	3	the	the	DET
fcis-23012	67	4	effect	effect	NOUN
fcis-23012	67	5	is	be	AUX
fcis-23012	67	6	best	good	ADJ
fcis-23012	67	7	with	with	ADP
fcis-23012	67	8	2	2	NUM
fcis-23012	67	9	to	to	PART
fcis-23012	67	10	3	3	NUM
fcis-23012	67	11	layers	layer	NOUN
fcis-23012	67	12	.	.	PUNCT
fcis-23012	68	1	in	in	ADP
fcis-23012	68	2	this	this	DET
fcis-23012	68	3	experiment	experiment	NOUN
fcis-23012	68	4	,	,	PUNCT
fcis-23012	68	5	a	a	DET
fcis-23012	68	6	2	2	NUM
fcis-23012	68	7	-	-	PUNCT
fcis-23012	68	8	layer	layer	NOUN
fcis-23012	68	9	graph	graph	NOUN
fcis-23012	68	10	convolution	convolution	NOUN
fcis-23012	68	11	structure	structure	NOUN
fcis-23012	68	12	was	be	AUX
fcis-23012	68	13	set	set	VERB
fcis-23012	68	14	up	up	ADP
fcis-23012	68	15	.	.	PUNCT
fcis-23012	69	1	72	72	NUM
fcis-23012	69	2	the	the	DET
fcis-23012	69	3	specific	specific	ADJ
fcis-23012	69	4	process	process	NOUN
fcis-23012	69	5	is	be	AUX
fcis-23012	69	6	shown	show	VERB
fcis-23012	69	7	in	in	ADP
fcis-23012	69	8	the	the	DET
fcis-23012	69	9	figure	figure	NOUN
fcis-23012	69	10	below	below	ADV
fcis-23012	69	11	,	,	PUNCT
fcis-23012	69	12	where	where	SCONJ
fcis-23012	69	13	in	in	ADP
fcis-23012	69	14	the	the	DET
fcis-23012	69	15	same	same	ADJ
fcis-23012	69	16	convolution	convolution	NOUN
fcis-23012	69	17	layer	layer	NOUN
fcis-23012	69	18	,	,	PUNCT
fcis-23012	69	19	the	the	DET
fcis-23012	69	20	aggregation	aggregation	NOUN
fcis-23012	69	21	operations	operation	NOUN
fcis-23012	69	22	of	of	ADP
fcis-23012	69	23	each	each	DET
fcis-23012	69	24	node	node	NOUN
fcis-23012	69	25	are	be	AUX
fcis-23012	69	26	performed	perform	VERB
fcis-23012	69	27	simultaneously	simultaneously	ADV
fcis-23012	69	28	.	.	PUNCT
fcis-23012	70	1	for	for	ADP
fcis-23012	70	2	the	the	DET
fcis-23012	70	3	constructed	construct	VERB
fcis-23012	70	4	graph	graph	NOUN
fcis-23012	70	5	structure	structure	NOUN
fcis-23012	70	6	data	datum	NOUN
fcis-23012	70	7	,	,	PUNCT
fcis-23012	70	8	suppose	suppose	VERB
fcis-23012	70	9	there	there	PRON
fcis-23012	70	10	are	be	VERB
fcis-23012	70	11	n	n	DET
fcis-23012	70	12	nodes	node	NOUN
fcis-23012	70	13	,	,	PUNCT
fcis-23012	70	14	and	and	CCONJ
fcis-23012	70	15	the	the	DET
fcis-23012	70	16	dimension	dimension	NOUN
fcis-23012	70	17	of	of	ADP
fcis-23012	70	18	the	the	DET
fcis-23012	70	19	feature	feature	NOUN
fcis-23012	70	20	vector	vector	NOUN
fcis-23012	70	21	of	of	ADP
fcis-23012	70	22	each	each	DET
fcis-23012	70	23	node	node	NOUN
fcis-23012	70	24	is	be	AUX
fcis-23012	70	25	h.	h.	PROPN
fcis-23012	70	26	all	all	DET
fcis-23012	70	27	node	node	ADJ
fcis-23012	70	28	features	feature	NOUN
fcis-23012	70	29	can	can	AUX
fcis-23012	70	30	be	be	AUX
fcis-23012	70	31	composed	compose	VERB
fcis-23012	70	32	into	into	ADP
fcis-23012	70	33	an	an	DET
fcis-23012	70	34	n×h	n×h	PROPN
fcis-23012	70	35	matrix	matrix	NOUN
fcis-23012	70	36	x.	x.	PUNCT
fcis-23012	71	1	the	the	DET
fcis-23012	71	2	connection	connection	NOUN
fcis-23012	71	3	relationship	relationship	NOUN
fcis-23012	71	4	between	between	ADP
fcis-23012	71	5	each	each	DET
fcis-23012	71	6	node	node	NOUN
fcis-23012	71	7	is	be	AUX
fcis-23012	71	8	an	an	DET
fcis-23012	71	9	n×n	n×n	PROPN
fcis-23012	71	10	relationship	relationship	NOUN
fcis-23012	71	11	matrix	matrix	NOUN
fcis-23012	71	12	,	,	PUNCT
fcis-23012	71	13	that	that	ADV
fcis-23012	71	14	is	is	ADV
fcis-23012	71	15	,	,	PUNCT
fcis-23012	71	16	the	the	DET
fcis-23012	71	17	adjacency	adjacency	NOUN
fcis-23012	71	18	matrix	matrix	NOUN
fcis-23012	71	19	,	,	PUNCT
fcis-23012	71	20	as	as	SCONJ
fcis-23012	71	21	follows	follow	VERB
fcis-23012	71	22	[	[	X
fcis-23012	71	23	10	10	NUM
fcis-23012	71	24	]	]	PUNCT
fcis-23012	71	25	,	,	PUNCT
fcis-23012	71	26	if	if	SCONJ
fcis-23012	71	27	the	the	DET
fcis-23012	71	28	points	point	NOUN
fcis-23012	71	29	are	be	AUX
fcis-23012	71	30	connected	connect	VERB
fcis-23012	71	31	,	,	PUNCT
fcis-23012	71	32	the	the	DET
fcis-23012	71	33	value	value	NOUN
fcis-23012	71	34	is	be	AUX
fcis-23012	71	35	1	1	NUM
fcis-23012	71	36	;	;	PUNCT
fcis-23012	71	37	if	if	SCONJ
fcis-23012	71	38	they	they	PRON
fcis-23012	71	39	are	be	AUX
fcis-23012	71	40	not	not	PART
fcis-23012	71	41	connected	connect	VERB
fcis-23012	71	42	,	,	PUNCT
fcis-23012	71	43	the	the	DET
fcis-23012	71	44	value	value	NOUN
fcis-23012	71	45	is	be	AUX
fcis-23012	71	46	0	0	NUM
fcis-23012	71	47	.	.	PUNCT
fcis-23012	72	1	since	since	SCONJ
fcis-23012	72	2	this	this	DET
fcis-23012	72	3	experiment	experiment	NOUN
fcis-23012	72	4	uses	use	VERB
fcis-23012	72	5	a	a	DET
fcis-23012	72	6	chain	chain	NOUN
fcis-23012	72	7	graph	graph	NOUN
fcis-23012	72	8	,	,	PUNCT
fcis-23012	72	9	the	the	DET
fcis-23012	72	10	first	first	ADJ
fcis-23012	72	11	and	and	CCONJ
fcis-23012	72	12	last	last	ADJ
fcis-23012	72	13	nodes	node	NOUN
fcis-23012	72	14	are	be	AUX
fcis-23012	72	15	not	not	PART
fcis-23012	72	16	connected	connect	VERB
fcis-23012	72	17	.	.	PUNCT
fcis-23012	73	1	a	a	DET
fcis-23012	73	2	0	0	NUM
fcis-23012	73	3	1	1	NUM
fcis-23012	73	4	0	0	NUM
fcis-23012	73	5	1	1	NUM
fcis-23012	73	6	0	0	NUM
fcis-23012	73	7	1	1	NUM
fcis-23012	73	8	0	0	NUM
fcis-23012	73	9	1	1	NUM
fcis-23012	73	10	0	0	NUM
fcis-23012	73	11	0	0	NUM
fcis-23012	73	12	⋯	⋯	ADP
fcis-23012	73	13	0	0	NUM
fcis-23012	73	14	0	0	NUM
fcis-23012	73	15	⋯	⋯	VERB
fcis-23012	73	16	0	0	NUM
fcis-23012	73	17	1	1	NUM
fcis-23012	73	18	⋯	⋯	ADP
fcis-23012	73	19	0	0	NUM
fcis-23012	73	20	0	0	NUM
fcis-23012	73	21	0	0	NUM
fcis-23012	73	22	1	1	NUM
fcis-23012	73	23	⋮	⋮	NOUN
fcis-23012	73	24	⋮	⋮	ADJ
fcis-23012	73	25	⋮	⋮	NOUN
fcis-23012	73	26	0	0	NUM
fcis-23012	73	27	0	0	NUM
fcis-23012	73	28	0	0	NUM
fcis-23012	73	29	0	0	NUM
fcis-23012	73	30	⋯	⋯	NOUN
fcis-23012	73	31	0	0	NUM
fcis-23012	73	32	⋮	⋮	NOUN
fcis-23012	73	33	⋱	⋱	PUNCT
fcis-23012	73	34	⋮	⋮	NOUN
fcis-23012	73	35	⋯	⋯	ADP
fcis-23012	73	36	1	1	NUM
fcis-23012	73	37	0	0	NUM
fcis-23012	73	38	let	let	VERB
fcis-23012	73	39	d	d	PRON
fcis-23012	73	40	be	be	AUX
fcis-23012	73	41	the	the	DET
fcis-23012	73	42	degree	degree	NOUN
fcis-23012	73	43	matrix	matrix	NOUN
fcis-23012	73	44	of	of	ADP
fcis-23012	73	45	this	this	DET
fcis-23012	73	46	graph	graph	NOUN
fcis-23012	73	47	,	,	PUNCT
fcis-23012	73	48	that	that	ADV
fcis-23012	73	49	is	is	ADV
fcis-23012	73	50	,	,	PUNCT
fcis-23012	73	51	a	a	DET
fcis-23012	73	52	diagonal	diagonal	ADJ
fcis-23012	73	53	matrix	matrix	NOUN
fcis-23012	73	54	whose	whose	DET
fcis-23012	73	55	diagonal	diagonal	ADJ
fcis-23012	73	56	elements	element	NOUN
fcis-23012	73	57	are	be	AUX
fcis-23012	73	58	the	the	DET
fcis-23012	73	59	number	number	NOUN
fcis-23012	73	60	of	of	ADP
fcis-23012	73	61	nodes	node	NOUN
fcis-23012	73	62	connected	connect	VERB
fcis-23012	73	63	to	to	ADP
fcis-23012	73	64	that	that	DET
fcis-23012	73	65	node	node	NOUN
fcis-23012	73	66	.	.	PUNCT
fcis-23012	74	1	the	the	DET
fcis-23012	74	2	laplacian	laplacian	ADJ
fcis-23012	74	3	matrix	matrix	NOUN
fcis-23012	74	4	adl	adl	PROPN
fcis-23012	74	5			PROPN
fcis-23012	74	6	is	be	AUX
fcis-23012	74	7	a	a	DET
fcis-23012	74	8	commonly	commonly	ADV
fcis-23012	74	9	used	use	VERB
fcis-23012	74	10	matrix	matrix	NOUN
fcis-23012	74	11	to	to	PART
fcis-23012	74	12	describe	describe	VERB
fcis-23012	74	13	the	the	DET
fcis-23012	74	14	overall	overall	ADJ
fcis-23012	74	15	structure	structure	NOUN
fcis-23012	74	16	of	of	ADP
fcis-23012	74	17	the	the	DET
fcis-23012	74	18	graph	graph	NOUN
fcis-23012	74	19	[	[	X
fcis-23012	74	20	11	11	NUM
fcis-23012	74	21	]	]	PUNCT
fcis-23012	74	22	.	.	PUNCT
fcis-23012	75	1	the	the	DET
fcis-23012	75	2	more	more	ADJ
fcis-23012	75	3	nodes	node	NOUN
fcis-23012	75	4	a	a	DET
fcis-23012	75	5	node	node	NOUN
fcis-23012	75	6	is	be	AUX
fcis-23012	75	7	connected	connect	VERB
fcis-23012	75	8	to	to	ADP
fcis-23012	75	9	,	,	PUNCT
fcis-23012	75	10	the	the	PRON
fcis-23012	75	11	lower	low	ADJ
fcis-23012	75	12	the	the	DET
fcis-23012	75	13	weight	weight	NOUN
fcis-23012	75	14	of	of	ADP
fcis-23012	75	15	the	the	DET
fcis-23012	75	16	information	information	NOUN
fcis-23012	75	17	passed	pass	VERB
fcis-23012	75	18	from	from	ADP
fcis-23012	75	19	each	each	DET
fcis-23012	75	20	connected	connect	VERB
fcis-23012	75	21	node	node	NOUN
fcis-23012	75	22	to	to	ADP
fcis-23012	75	23	that	that	DET
fcis-23012	75	24	node	node	NOUN
fcis-23012	75	25	during	during	ADP
fcis-23012	75	26	aggregation	aggregation	NOUN
fcis-23012	75	27	,	,	PUNCT
fcis-23012	75	28	therefore	therefore	ADV
fcis-23012	75	29	,	,	PUNCT
fcis-23012	75	30	use	use	VERB
fcis-23012	75	31	ld	ld	PROPN
fcis-23012	75	32	1	1	NUM
fcis-23012	75	33	to	to	PART
fcis-23012	75	34	represent	represent	VERB
fcis-23012	75	35	the	the	DET
fcis-23012	75	36	weight	weight	NOUN
fcis-23012	75	37	of	of	ADP
fcis-23012	75	38	information	information	NOUN
fcis-23012	75	39	each	each	DET
fcis-23012	75	40	node	node	NOUN
fcis-23012	75	41	should	should	AUX
fcis-23012	75	42	get	get	VERB
fcis-23012	75	43	from	from	ADP
fcis-23012	75	44	neighboring	neighboring	NOUN
fcis-23012	75	45	nodes	node	NOUN
fcis-23012	75	46	.	.	PUNCT
fcis-23012	76	1	because	because	SCONJ
fcis-23012	76	2	ld	ld	PROPN
fcis-23012	76	3	1	1	NUM
fcis-23012	76	4	will	will	AUX
fcis-23012	76	5	make	make	VERB
fcis-23012	76	6	the	the	DET
fcis-23012	76	7	feature	feature	NOUN
fcis-23012	76	8	value	value	NOUN
fcis-23012	76	9	of	of	ADP
fcis-23012	76	10	the	the	DET
fcis-23012	76	11	node	node	NOUN
fcis-23012	76	12	with	with	ADP
fcis-23012	76	13	more	more	ADJ
fcis-23012	76	14	neighboring	neighboring	NOUN
fcis-23012	76	15	nodes	node	NOUN
fcis-23012	76	16	greater	great	ADJ
fcis-23012	76	17	than	than	ADP
fcis-23012	76	18	that	that	PRON
fcis-23012	76	19	of	of	ADP
fcis-23012	76	20	the	the	DET
fcis-23012	76	21	node	node	NOUN
fcis-23012	76	22	with	with	ADP
fcis-23012	76	23	fewer	few	ADJ
fcis-23012	76	24	neighboring	neighboring	NOUN
fcis-23012	76	25	nodes	node	NOUN
fcis-23012	76	26	after	after	ADP
fcis-23012	76	27	aggregation	aggregation	NOUN
fcis-23012	76	28	,	,	PUNCT
fcis-23012	76	29	we	we	PRON
fcis-23012	76	30	use	use	VERB
fcis-23012	76	31	2	2	NUM
fcis-23012	76	32	1	1	NUM
fcis-23012	76	33	2	2	NUM
fcis-23012	76	34	1	1	NUM
fcis-23012	76	35			NOUN
fcis-23012	76	36	ldd	ldd	NOUN
fcis-23012	76	37	as	as	ADP
fcis-23012	76	38	the	the	DET
fcis-23012	76	39	node	node	ADJ
fcis-23012	76	40	aggregation	aggregation	NOUN
fcis-23012	76	41	matrix	matrix	NOUN
fcis-23012	76	42	after	after	ADP
fcis-23012	76	43	normalization	normalization	NOUN
fcis-23012	76	44	.	.	PUNCT
fcis-23012	77	1	therefore	therefore	ADV
fcis-23012	77	2	,	,	PUNCT
fcis-23012	77	3	for	for	ADP
fcis-23012	77	4	a	a	DET
fcis-23012	77	5	gcn	gcn	NOUN
fcis-23012	77	6	network	network	NOUN
fcis-23012	77	7	with	with	ADP
fcis-23012	77	8	i	i	PROPN
fcis-23012	77	9	layers	layer	NOUN
fcis-23012	77	10	,	,	PUNCT
fcis-23012	77	11	the	the	DET
fcis-23012	77	12	propagation	propagation	NOUN
fcis-23012	77	13	function	function	NOUN
fcis-23012	77	14	between	between	ADP
fcis-23012	77	15	each	each	DET
fcis-23012	77	16	layer	layer	NOUN
fcis-23012	77	17	is	be	AUX
fcis-23012	77	18	:	:	PUNCT
fcis-23012	77	19	)	)	PUNCT
fcis-23012	77	20	(	(	PUNCT
fcis-23012	77	21	)	)	PUNCT
fcis-23012	77	22	(	(	PUNCT
fcis-23012	77	23	)	)	PUNCT
fcis-23012	77	24	(	(	PUNCT
fcis-23012	77	25	)	)	PUNCT
fcis-23012	77	26	1	1	NUM
fcis-23012	77	27	(	(	PUNCT
fcis-23012	77	28	2	2	NUM
fcis-23012	77	29	1	1	NUM
fcis-23012	77	30	2	2	NUM
fcis-23012	77	31	1	1	NUM
fcis-23012	77	32	iii	iii	NUM
fcis-23012	77	33	wxlddx	wxlddx	NOUN
fcis-23012	77	34			NOUN
fcis-23012	77	35			NOUN
fcis-23012	77	36			X
fcis-23012	77	37	(	(	PUNCT
fcis-23012	77	38	4	4	NUM
fcis-23012	77	39	)	)	PUNCT
fcis-23012	77	40	)	)	PUNCT
fcis-23012	78	1	(	(	PUNCT
fcis-23012	78	2	iw	iw	X
fcis-23012	78	3	is	be	AUX
fcis-23012	78	4	the	the	DET
fcis-23012	78	5	weight	weight	NOUN
fcis-23012	78	6	matrix	matrix	NOUN
fcis-23012	78	7	of	of	ADP
fcis-23012	78	8	the	the	DET
fcis-23012	78	9	i	i	PROPN
fcis-23012	78	10	-	-	PUNCT
fcis-23012	78	11	th	th	X
fcis-23012	78	12	layer	layer	NOUN
fcis-23012	78	13	；	；	NOUN
fcis-23012	78	14	)	)	PUNCT
fcis-23012	78	15	(	(	PUNCT
fcis-23012	78	16			X
fcis-23012	78	17	is	be	AUX
fcis-23012	78	18	a	a	DET
fcis-23012	78	19	nonlinear	nonlinear	ADJ
fcis-23012	78	20	activation	activation	NOUN
fcis-23012	78	21	function	function	NOUN
fcis-23012	78	22	,	,	PUNCT
fcis-23012	78	23	and	and	CCONJ
fcis-23012	78	24	the	the	DET
fcis-23012	78	25	relu	relu	NOUN
fcis-23012	78	26	function	function	NOUN
fcis-23012	78	27	is	be	AUX
fcis-23012	78	28	used	use	VERB
fcis-23012	78	29	in	in	ADP
fcis-23012	78	30	this	this	DET
fcis-23012	78	31	experiment	experiment	NOUN
fcis-23012	78	32	;	;	PUNCT
fcis-23012	78	33	)	)	PUNCT
fcis-23012	78	34	(	(	PUNCT
fcis-23012	78	35	ix	ix	ADV
fcis-23012	78	36	represents	represent	VERB
fcis-23012	78	37	the	the	DET
fcis-23012	78	38	input	input	NOUN
fcis-23012	78	39	information	information	NOUN
fcis-23012	78	40	of	of	ADP
fcis-23012	78	41	the	the	DET
fcis-23012	78	42	i	i	PROPN
fcis-23012	78	43	-	-	PUNCT
fcis-23012	78	44	th	th	VERB
fcis-23012	78	45	layer	layer	NOUN
fcis-23012	78	46	.	.	PUNCT
fcis-23012	79	1	3.3	3.3	NUM
fcis-23012	79	2	.	.	PUNCT
fcis-23012	80	1	bi	bi	ADJ
fcis-23012	80	2	-	-	ADJ
fcis-23012	80	3	lstm	lstm	ADJ
fcis-23012	80	4	long	long	ADJ
fcis-23012	80	5	short	short	ADJ
fcis-23012	80	6	-	-	PUNCT
fcis-23012	80	7	term	term	NOUN
fcis-23012	80	8	memory	memory	NOUN
fcis-23012	80	9	(	(	PUNCT
fcis-23012	80	10	lstm	lstm	NOUN
fcis-23012	80	11	)	)	PUNCT
fcis-23012	80	12	is	be	AUX
fcis-23012	80	13	a	a	DET
fcis-23012	80	14	special	special	ADJ
fcis-23012	80	15	type	type	NOUN
fcis-23012	80	16	of	of	ADP
fcis-23012	80	17	recurrent	recurrent	ADJ
fcis-23012	80	18	neural	neural	ADJ
fcis-23012	80	19	network	network	NOUN
fcis-23012	80	20	that	that	PRON
fcis-23012	80	21	selectively	selectively	ADV
fcis-23012	80	22	allows	allow	VERB
fcis-23012	80	23	information	information	NOUN
fcis-23012	80	24	to	to	PART
fcis-23012	80	25	pass	pass	VERB
fcis-23012	80	26	through	through	ADP
fcis-23012	80	27	gate	gate	NOUN
fcis-23012	80	28	units	unit	NOUN
fcis-23012	80	29	,	,	PUNCT
fcis-23012	80	30	enabling	enable	VERB
fcis-23012	80	31	the	the	DET
fcis-23012	80	32	network	network	NOUN
fcis-23012	80	33	to	to	PART
fcis-23012	80	34	maintain	maintain	VERB
fcis-23012	80	35	long	long	ADJ
fcis-23012	80	36	-	-	PUNCT
fcis-23012	80	37	term	term	NOUN
fcis-23012	80	38	memory	memory	NOUN
fcis-23012	80	39	of	of	ADP
fcis-23012	80	40	overall	overall	ADJ
fcis-23012	80	41	sequence	sequence	NOUN
fcis-23012	80	42	features	feature	VERB
fcis-23012	80	43	[	[	X
fcis-23012	80	44	12	12	NUM
fcis-23012	80	45	]	]	PUNCT
fcis-23012	80	46	,	,	PUNCT
fcis-23012	80	47	it	it	PRON
fcis-23012	80	48	has	have	VERB
fcis-23012	80	49	good	good	ADJ
fcis-23012	80	50	performance	performance	NOUN
fcis-23012	80	51	in	in	ADP
fcis-23012	80	52	capturing	capture	VERB
fcis-23012	80	53	the	the	DET
fcis-23012	80	54	overall	overall	ADJ
fcis-23012	80	55	features	feature	NOUN
fcis-23012	80	56	of	of	ADP
fcis-23012	80	57	the	the	DET
fcis-23012	80	58	sequence[13	sequence[13	NOUN
fcis-23012	80	59	]	]	PUNCT
fcis-23012	80	60	.	.	PUNCT
fcis-23012	81	1	figure	figure	NOUN
fcis-23012	81	2	4	4	NUM
fcis-23012	81	3	.	.	PUNCT
fcis-23012	82	1	lstm	lstm	PROPN
fcis-23012	82	2	network	network	PROPN
fcis-23012	82	3	the	the	DET
fcis-23012	82	4	bi	bi	ADJ
fcis-23012	82	5	-	-	ADJ
fcis-23012	82	6	directional	directional	ADJ
fcis-23012	82	7	long	long	ADJ
fcis-23012	82	8	short	short	ADJ
fcis-23012	82	9	-	-	PUNCT
fcis-23012	82	10	term	term	NOUN
fcis-23012	82	11	memory	memory	NOUN
fcis-23012	82	12	(	(	PUNCT
fcis-23012	82	13	bi	bi	NOUN
fcis-23012	82	14	-	-	ADJ
fcis-23012	82	15	lstm	lstm	ADJ
fcis-23012	82	16	)	)	PUNCT
fcis-23012	82	17	consists	consist	VERB
fcis-23012	82	18	of	of	ADP
fcis-23012	82	19	two	two	NUM
fcis-23012	82	20	independent	independent	ADJ
fcis-23012	82	21	lstms	lstms	NOUN
fcis-23012	82	22	,	,	PUNCT
fcis-23012	82	23	which	which	PRON
fcis-23012	82	24	process	process	VERB
fcis-23012	82	25	the	the	DET
fcis-23012	82	26	forward	forward	ADJ
fcis-23012	82	27	and	and	CCONJ
fcis-23012	82	28	backward	backward	ADJ
fcis-23012	82	29	information	information	NOUN
fcis-23012	82	30	of	of	ADP
fcis-23012	82	31	the	the	DET
fcis-23012	82	32	data	datum	NOUN
fcis-23012	82	33	sequence	sequence	NOUN
fcis-23012	82	34	respectively	respectively	ADV
fcis-23012	82	35	.	.	PUNCT
fcis-23012	83	1	this	this	DET
fcis-23012	83	2	network	network	NOUN
fcis-23012	83	3	design	design	NOUN
fcis-23012	83	4	allows	allow	VERB
fcis-23012	83	5	the	the	DET
fcis-23012	83	6	model	model	NOUN
fcis-23012	83	7	to	to	PART
fcis-23012	83	8	learn	learn	VERB
fcis-23012	83	9	the	the	DET
fcis-23012	83	10	context	context	NOUN
fcis-23012	83	11	information	information	NOUN
fcis-23012	83	12	of	of	ADP
fcis-23012	83	13	the	the	DET
fcis-23012	83	14	past	past	NOUN
fcis-23012	83	15	and	and	CCONJ
fcis-23012	83	16	future	future	NOUN
fcis-23012	83	17	simultaneously	simultaneously	ADV
fcis-23012	83	18	[	[	X
fcis-23012	83	19	14	14	NUM
fcis-23012	83	20	]	]	PUNCT
fcis-23012	83	21	.	.	PUNCT
fcis-23012	84	1	through	through	ADP
fcis-23012	84	2	this	this	DET
fcis-23012	84	3	bi	bi	ADJ
fcis-23012	84	4	-	-	ADJ
fcis-23012	84	5	directional	directional	ADJ
fcis-23012	84	6	processing	processing	NOUN
fcis-23012	84	7	mechanism	mechanism	NOUN
fcis-23012	84	8	,	,	PUNCT
fcis-23012	84	9	bilstm	bilstm	NOUN
fcis-23012	84	10	enhances	enhance	VERB
fcis-23012	84	11	the	the	DET
fcis-23012	84	12	model	model	NOUN
fcis-23012	84	13	’s	’s	PART
fcis-23012	84	14	ability	ability	NOUN
fcis-23012	84	15	to	to	PART
fcis-23012	84	16	capture	capture	VERB
fcis-23012	84	17	information	information	NOUN
fcis-23012	84	18	in	in	ADP
fcis-23012	84	19	scenarios	scenario	NOUN
fcis-23012	84	20	where	where	SCONJ
fcis-23012	84	21	understanding	understand	VERB
fcis-23012	84	22	the	the	DET
fcis-23012	84	23	entire	entire	ADJ
fcis-23012	84	24	data	datum	NOUN
fcis-23012	84	25	stream	stream	NOUN
fcis-23012	84	26	is	be	AUX
fcis-23012	84	27	required	require	VERB
fcis-23012	84	28	.	.	PUNCT
fcis-23012	85	1	gcn	gcn	NOUN
fcis-23012	85	2	networks	network	NOUN
fcis-23012	85	3	are	be	AUX
fcis-23012	85	4	good	good	ADJ
fcis-23012	85	5	at	at	ADP
fcis-23012	85	6	aggregating	aggregate	VERB
fcis-23012	85	7	information	information	NOUN
fcis-23012	85	8	from	from	ADP
fcis-23012	85	9	neighboring	neighboring	NOUN
fcis-23012	85	10	nodes	node	NOUN
fcis-23012	85	11	and	and	CCONJ
fcis-23012	85	12	capturing	capture	VERB
fcis-23012	85	13	the	the	DET
fcis-23012	85	14	connection	connection	NOUN
fcis-23012	85	15	of	of	ADP
fcis-23012	85	16	local	local	ADJ
fcis-23012	85	17	features	feature	NOUN
fcis-23012	85	18	.	.	PUNCT
fcis-23012	86	1	if	if	SCONJ
fcis-23012	86	2	you	you	PRON
fcis-23012	86	3	add	add	VERB
fcis-23012	86	4	too	too	ADV
fcis-23012	86	5	many	many	ADJ
fcis-23012	86	6	gcn	gcn	ADJ
fcis-23012	86	7	layers	layer	NOUN
fcis-23012	86	8	in	in	ADP
fcis-23012	86	9	an	an	DET
fcis-23012	86	10	attempt	attempt	NOUN
fcis-23012	86	11	to	to	PART
fcis-23012	86	12	aggregate	aggregate	VERB
fcis-23012	86	13	information	information	NOUN
fcis-23012	86	14	from	from	ADP
fcis-23012	86	15	distant	distant	ADJ
fcis-23012	86	16	nodes	node	NOUN
fcis-23012	86	17	to	to	PART
fcis-23012	86	18	obtain	obtain	VERB
fcis-23012	86	19	global	global	ADJ
fcis-23012	86	20	information	information	NOUN
fcis-23012	86	21	,	,	PUNCT
fcis-23012	86	22	it	it	PRON
fcis-23012	86	23	will	will	AUX
fcis-23012	86	24	instead	instead	ADV
fcis-23012	86	25	cause	cause	VERB
fcis-23012	86	26	the	the	DET
fcis-23012	86	27	node	node	ADJ
fcis-23012	86	28	information	information	NOUN
fcis-23012	86	29	to	to	PART
fcis-23012	86	30	be	be	AUX
fcis-23012	86	31	overly	overly	ADV
fcis-23012	86	32	aggregated	aggregate	VERB
fcis-23012	86	33	and	and	CCONJ
fcis-23012	86	34	lose	lose	VERB
fcis-23012	86	35	its	its	PRON
fcis-23012	86	36	original	original	ADJ
fcis-23012	86	37	features	feature	NOUN
fcis-23012	86	38	.	.	PUNCT
fcis-23012	87	1	therefore	therefore	ADV
fcis-23012	87	2	,	,	PUNCT
fcis-23012	87	3	using	use	VERB
fcis-23012	87	4	a	a	DET
fcis-23012	87	5	gcn	gcn	NOUN
fcis-23012	87	6	with	with	ADP
fcis-23012	87	7	fewer	few	ADJ
fcis-23012	87	8	layers	layer	NOUN
fcis-23012	87	9	to	to	PART
fcis-23012	87	10	capture	capture	VERB
fcis-23012	87	11	local	local	ADJ
fcis-23012	87	12	information	information	NOUN
fcis-23012	87	13	,	,	PUNCT
fcis-23012	87	14	while	while	SCONJ
fcis-23012	87	15	combining	combine	VERB
fcis-23012	87	16	it	it	PRON
fcis-23012	87	17	with	with	ADP
fcis-23012	87	18	a	a	DET
fcis-23012	87	19	bi	bi	ADJ
fcis-23012	87	20	-	-	ADJ
fcis-23012	87	21	lstm	lstm	ADJ
fcis-23012	87	22	network	network	NOUN
fcis-23012	87	23	to	to	PART
fcis-23012	87	24	obtain	obtain	VERB
fcis-23012	87	25	the	the	DET
fcis-23012	87	26	connection	connection	NOUN
fcis-23012	87	27	of	of	ADP
fcis-23012	87	28	global	global	ADJ
fcis-23012	87	29	features	feature	NOUN
fcis-23012	87	30	,	,	PUNCT
fcis-23012	87	31	can	can	AUX
fcis-23012	87	32	compensate	compensate	VERB
fcis-23012	87	33	for	for	ADP
fcis-23012	87	34	the	the	DET
fcis-23012	87	35	deficiency	deficiency	NOUN
fcis-23012	87	36	of	of	ADP
fcis-23012	87	37	the	the	DET
fcis-23012	87	38	gcn	gcn	NOUN
fcis-23012	87	39	network	network	NOUN
fcis-23012	87	40	’s	’s	PART
fcis-23012	87	41	receptive	receptive	ADJ
fcis-23012	87	42	field[15	field[15	NOUN
fcis-23012	87	43	]	]	PUNCT
fcis-23012	87	44	.	.	PUNCT
fcis-23012	88	1	figure	figure	NOUN
fcis-23012	88	2	5	5	NUM
fcis-23012	88	3	.	.	PUNCT
fcis-23012	89	1	overall	overall	ADJ
fcis-23012	89	2	network	network	NOUN
fcis-23012	89	3	structure	structure	NOUN
fcis-23012	89	4	3.4	3.4	NUM
fcis-23012	89	5	.	.	PUNCT
fcis-23012	90	1	loss	loss	NOUN
fcis-23012	90	2	function	function	NOUN
fcis-23012	90	3	in	in	ADP
fcis-23012	90	4	terms	term	NOUN
fcis-23012	90	5	of	of	ADP
fcis-23012	90	6	loss	loss	NOUN
fcis-23012	90	7	function	function	NOUN
fcis-23012	90	8	,	,	PUNCT
fcis-23012	90	9	previous	previous	ADJ
fcis-23012	90	10	speech	speech	NOUN
fcis-23012	90	11	emotion	emotion	NOUN
fcis-23012	90	12	recognition	recognition	NOUN
fcis-23012	90	13	tasks	task	NOUN
fcis-23012	90	14	often	often	ADV
fcis-23012	90	15	use	use	VERB
fcis-23012	90	16	cross	cross	NOUN
fcis-23012	90	17	entropy	entropy	PROPN
fcis-23012	90	18	loss	loss	PROPN
fcis-23012	90	19	(	(	PUNCT
fcis-23012	90	20	ce	ce	PROPN
fcis-23012	90	21	)	)	PUNCT
fcis-23012	90	22	,	,	PUNCT
fcis-23012	90	23	which	which	PRON
fcis-23012	90	24	has	have	AUX
fcis-23012	90	25	always	always	ADV
fcis-23012	90	26	performed	perform	VERB
fcis-23012	90	27	well	well	ADV
fcis-23012	90	28	in	in	ADP
fcis-23012	90	29	multi	multi	ADJ
fcis-23012	90	30	-	-	ADJ
fcis-23012	90	31	class	class	ADJ
fcis-23012	90	32	tasks	task	NOUN
fcis-23012	90	33	.	.	PUNCT
fcis-23012	91	1	in	in	ADP
fcis-23012	91	2	order	order	NOUN
fcis-23012	91	3	to	to	PART
fcis-23012	91	4	further	far	ADV
fcis-23012	91	5	capture	capture	VERB
fcis-23012	91	6	the	the	DET
fcis-23012	91	7	emotional	emotional	ADJ
fcis-23012	91	8	features	feature	NOUN
fcis-23012	91	9	of	of	ADP
fcis-23012	91	10	the	the	DET
fcis-23012	91	11	tibetan	tibetan	ADJ
fcis-23012	91	12	language	language	NOUN
fcis-23012	91	13	and	and	CCONJ
fcis-23012	91	14	more	more	ADV
fcis-23012	91	15	accurately	accurately	ADV
fcis-23012	91	16	direct	direct	VERB
fcis-23012	91	17	the	the	DET
fcis-23012	91	18	model	model	NOUN
fcis-23012	91	19	training	training	NOUN
fcis-23012	91	20	,	,	PUNCT
fcis-23012	91	21	a	a	DET
fcis-23012	91	22	joint	joint	ADJ
fcis-23012	91	23	loss	loss	NOUN
fcis-23012	91	24	function	function	NOUN
fcis-23012	91	25	of	of	ADP
fcis-23012	91	26	cross	cross	NOUN
fcis-23012	91	27	entropy	entropy	PROPN
fcis-23012	91	28	and	and	CCONJ
fcis-23012	91	29	center	center	ADJ
fcis-23012	91	30	loss	loss	NOUN
fcis-23012	91	31	is	be	AUX
fcis-23012	91	32	used	use	VERB
fcis-23012	91	33	.	.	PUNCT
fcis-23012	92	1	the	the	DET
fcis-23012	92	2	center	center	NOUN
fcis-23012	92	3	loss	loss	NOUN
fcis-23012	92	4	function	function	NOUN
fcis-23012	92	5	sets	set	VERB
fcis-23012	92	6	a	a	DET
fcis-23012	92	7	center	center	NOUN
fcis-23012	92	8	vector	vector	NOUN
fcis-23012	92	9	for	for	ADP
fcis-23012	92	10	each	each	DET
fcis-23012	92	11	emotion	emotion	NOUN
fcis-23012	92	12	label	label	NOUN
fcis-23012	92	13	.	.	PUNCT
fcis-23012	93	1	during	during	ADP
fcis-23012	93	2	the	the	DET
fcis-23012	93	3	training	training	NOUN
fcis-23012	93	4	process	process	NOUN
fcis-23012	93	5	,	,	PUNCT
fcis-23012	93	6	the	the	DET
fcis-23012	93	7	center	center	NOUN
fcis-23012	93	8	loss	loss	NOUN
fcis-23012	93	9	will	will	AUX
fcis-23012	93	10	calculate	calculate	VERB
fcis-23012	93	11	the	the	DET
fcis-23012	93	12	euclidean	euclidean	ADJ
fcis-23012	93	13	distance	distance	NOUN
fcis-23012	93	14	between	between	ADP
fcis-23012	93	15	the	the	DET
fcis-23012	93	16	current	current	ADJ
fcis-23012	93	17	sample	sample	NOUN
fcis-23012	93	18	feature	feature	NOUN
fcis-23012	93	19	vector	vector	NOUN
fcis-23012	93	20	and	and	CCONJ
fcis-23012	93	21	the	the	DET
fcis-23012	93	22	center	center	ADJ
fcis-23012	93	23	vector	vector	NOUN
fcis-23012	93	24	of	of	ADP
fcis-23012	93	25	its	its	PRON
fcis-23012	93	26	label	label	NOUN
fcis-23012	93	27	in	in	ADP
fcis-23012	93	28	each	each	DET
fcis-23012	93	29	iteration	iteration	NOUN
fcis-23012	93	30	,	,	PUNCT
fcis-23012	93	31	and	and	CCONJ
fcis-23012	93	32	update	update	VERB
fcis-23012	93	33	the	the	DET
fcis-23012	93	34	center	center	ADJ
fcis-23012	93	35	vector	vector	NOUN
fcis-23012	93	36	of	of	ADP
fcis-23012	93	37	the	the	DET
fcis-23012	93	38	label	label	NOUN
fcis-23012	93	39	through	through	ADP
fcis-23012	93	40	backpropagation	backpropagation	NOUN
fcis-23012	93	41	.	.	PUNCT
fcis-23012	94	1	the	the	DET
fcis-23012	94	2	inclusion	inclusion	NOUN
fcis-23012	94	3	of	of	ADP
fcis-23012	94	4	the	the	DET
fcis-23012	94	5	center	center	NOUN
fcis-23012	94	6	loss	loss	NOUN
fcis-23012	94	7	function	function	NOUN
fcis-23012	94	8	theoretically	theoretically	ADV
fcis-23012	94	9	helps	help	VERB
fcis-23012	94	10	the	the	DET
fcis-23012	94	11	model	model	NOUN
fcis-23012	94	12	better	well	ADV
fcis-23012	94	13	extract	extract	VERB
fcis-23012	94	14	the	the	DET
fcis-23012	94	15	unique	unique	ADJ
fcis-23012	94	16	features	feature	NOUN
fcis-23012	94	17	of	of	ADP
fcis-23012	94	18	each	each	DET
fcis-23012	94	19	emotion	emotion	NOUN
fcis-23012	94	20	.	.	PUNCT
fcis-23012	95	1	in	in	ADP
fcis-23012	95	2	this	this	DET
fcis-23012	95	3	experiment	experiment	NOUN
fcis-23012	95	4	,	,	PUNCT
fcis-23012	95	5	we	we	PRON
fcis-23012	95	6	add	add	VERB
fcis-23012	95	7	the	the	DET
fcis-23012	95	8	cross	cross	NOUN
fcis-23012	95	9	entropy	entropy	PROPN
fcis-23012	95	10	loss	loss	NOUN
fcis-23012	95	11	function	function	NOUN
fcis-23012	95	12	and	and	CCONJ
fcis-23012	95	13	73	73	NUM
fcis-23012	95	14	the	the	DET
fcis-23012	95	15	center	center	NOUN
fcis-23012	95	16	loss	loss	NOUN
fcis-23012	95	17	function	function	NOUN
fcis-23012	95	18	with	with	ADP
fcis-23012	95	19	weights	weight	NOUN
fcis-23012	95	20	to	to	PART
fcis-23012	95	21	construct	construct	VERB
fcis-23012	95	22	a	a	DET
fcis-23012	95	23	joint	joint	ADJ
fcis-23012	95	24	loss	loss	NOUN
fcis-23012	95	25	function	function	NOUN
fcis-23012	95	26	.	.	PUNCT
fcis-23012	96	1	4	4	X
fcis-23012	96	2	.	.	X
fcis-23012	96	3	experiment	experiment	NOUN
fcis-23012	96	4	and	and	CCONJ
fcis-23012	96	5	analysis	analysis	NOUN
fcis-23012	96	6	in	in	ADP
fcis-23012	96	7	this	this	DET
fcis-23012	96	8	experiment	experiment	NOUN
fcis-23012	96	9	,	,	PUNCT
fcis-23012	96	10	the	the	DET
fcis-23012	96	11	tibetan	tibetan	ADJ
fcis-23012	96	12	language	language	PROPN
fcis-23012	96	13	speech	speech	NOUN
fcis-23012	96	14	emotion	emotion	NOUN
fcis-23012	96	15	recognition	recognition	NOUN
fcis-23012	96	16	dataset	dataset	NOUN
fcis-23012	96	17	tbls1	tbls1	VERB
fcis-23012	96	18	is	be	AUX
fcis-23012	96	19	evenly	evenly	ADV
fcis-23012	96	20	divided	divide	VERB
fcis-23012	96	21	into	into	ADP
fcis-23012	96	22	5	5	NUM
fcis-23012	96	23	groups	group	NOUN
fcis-23012	96	24	,	,	PUNCT
fcis-23012	96	25	with	with	ADP
fcis-23012	96	26	one	one	NUM
fcis-23012	96	27	group	group	NOUN
fcis-23012	96	28	being	be	AUX
fcis-23012	96	29	used	use	VERB
fcis-23012	96	30	as	as	ADP
fcis-23012	96	31	the	the	DET
fcis-23012	96	32	validation	validation	NOUN
fcis-23012	96	33	set	set	VERB
fcis-23012	96	34	in	in	ADP
fcis-23012	96	35	turn	turn	NOUN
fcis-23012	96	36	,	,	PUNCT
fcis-23012	96	37	and	and	CCONJ
fcis-23012	96	38	the	the	DET
fcis-23012	96	39	rest	rest	NOUN
fcis-23012	96	40	being	be	AUX
fcis-23012	96	41	used	use	VERB
fcis-23012	96	42	as	as	ADP
fcis-23012	96	43	the	the	DET
fcis-23012	96	44	training	training	NOUN
fcis-23012	96	45	set	set	NOUN
fcis-23012	96	46	.	.	PUNCT
fcis-23012	97	1	the	the	DET
fcis-23012	97	2	early	early	ADJ
fcis-23012	97	3	stop	stop	NOUN
fcis-23012	97	4	function	function	NOUN
fcis-23012	97	5	is	be	AUX
fcis-23012	97	6	added	add	VERB
fcis-23012	97	7	,	,	PUNCT
fcis-23012	97	8	with	with	ADP
fcis-23012	97	9	a	a	DET
fcis-23012	97	10	patience	patience	NOUN
fcis-23012	97	11	index	index	NOUN
fcis-23012	97	12	set	set	VERB
fcis-23012	97	13	to	to	ADP
fcis-23012	97	14	10	10	NUM
fcis-23012	97	15	epochs	epoch	NOUN
fcis-23012	97	16	.	.	PUNCT
fcis-23012	98	1	each	each	DET
fcis-23012	98	2	epoch	epoch	NOUN
fcis-23012	98	3	includes	include	VERB
fcis-23012	98	4	50	50	NUM
fcis-23012	98	5	pieces	piece	NOUN
fcis-23012	98	6	of	of	ADP
fcis-23012	98	7	data	datum	NOUN
fcis-23012	98	8	.	.	PUNCT
fcis-23012	99	1	to	to	PART
fcis-23012	99	2	ensure	ensure	VERB
fcis-23012	99	3	the	the	DET
fcis-23012	99	4	amount	amount	NOUN
fcis-23012	99	5	of	of	ADP
fcis-23012	99	6	training	training	NOUN
fcis-23012	99	7	,	,	PUNCT
fcis-23012	99	8	it	it	PRON
fcis-23012	99	9	is	be	AUX
fcis-23012	99	10	set	set	VERB
fcis-23012	99	11	that	that	SCONJ
fcis-23012	99	12	the	the	DET
fcis-23012	99	13	final	final	ADJ
fcis-23012	99	14	result	result	NOUN
fcis-23012	99	15	can	can	AUX
fcis-23012	99	16	only	only	ADV
fcis-23012	99	17	be	be	AUX
fcis-23012	99	18	output	output	NOUN
fcis-23012	99	19	after	after	ADP
fcis-23012	99	20	at	at	ADV
fcis-23012	99	21	least	least	ADV
fcis-23012	99	22	300	300	NUM
fcis-23012	99	23	epochs	epoch	NOUN
fcis-23012	99	24	.	.	PUNCT
fcis-23012	100	1	the	the	DET
fcis-23012	100	2	performance	performance	NOUN
fcis-23012	100	3	of	of	ADP
fcis-23012	100	4	different	different	ADJ
fcis-23012	100	5	networks	network	NOUN
fcis-23012	100	6	on	on	ADP
fcis-23012	100	7	this	this	DET
fcis-23012	100	8	dataset	dataset	NOUN
fcis-23012	100	9	is	be	AUX
fcis-23012	100	10	as	as	SCONJ
fcis-23012	100	11	follows	follow	VERB
fcis-23012	100	12	:	:	PUNCT
fcis-23012	100	13	table	table	NOUN
fcis-23012	100	14	1	1	NUM
fcis-23012	100	15	.	.	PUNCT
fcis-23012	100	16	comparison	comparison	NOUN
fcis-23012	100	17	of	of	ADP
fcis-23012	100	18	performance	performance	NOUN
fcis-23012	100	19	of	of	ADP
fcis-23012	100	20	different	different	ADJ
fcis-23012	100	21	models	model	NOUN
fcis-23012	100	22	network	network	NOUN
fcis-23012	100	23	loss	loss	NOUN
fcis-23012	100	24	function	function	NOUN
fcis-23012	100	25	accuracy	accuracy	NOUN
fcis-23012	100	26	(	(	PUNCT
fcis-23012	100	27	%	%	INTJ
fcis-23012	100	28	)	)	PUNCT
fcis-23012	100	29	lstm	lstm	PROPN
fcis-23012	100	30	cross	cross	PROPN
fcis-23012	100	31	entropy	entropy	PROPN
fcis-23012	100	32	loss	loss	PROPN
fcis-23012	100	33	45.5726	45.5726	NUM
fcis-23012	100	34	crnn	crnn	PROPN
fcis-23012	100	35	cross	cross	PROPN
fcis-23012	100	36	entropy	entropy	PROPN
fcis-23012	100	37	loss	loss	PROPN
fcis-23012	100	38	50.2564	50.2564	NUM
fcis-23012	100	39	cnn	cnn	PROPN
fcis-23012	100	40	cross	cross	PROPN
fcis-23012	100	41	entropy	entropy	PROPN
fcis-23012	100	42	loss	loss	PROPN
fcis-23012	100	43	57.0854	57.0854	NUM
fcis-23012	100	44	cnn+lstm	cnn+lstm	PROPN
fcis-23012	100	45	cross	cross	PROPN
fcis-23012	100	46	entropy	entropy	PROPN
fcis-23012	100	47	loss	loss	VERB
fcis-23012	100	48	59.2897	59.2897	NUM
fcis-23012	100	49	gcn	gcn	PROPN
fcis-23012	100	50	cross	cross	PROPN
fcis-23012	100	51	entropy	entropy	PROPN
fcis-23012	100	52	loss	loss	VERB
fcis-23012	100	53	59.3970	59.3970	NUM
fcis-23012	100	54	gcn+bi	gcn+bi	PROPN
fcis-23012	100	55	-	-	PUNCT
fcis-23012	100	56	lstm	lstm	PROPN
fcis-23012	100	57	cross	cross	PROPN
fcis-23012	100	58	entropy	entropy	PROPN
fcis-23012	100	59	loss	loss	VERB
fcis-23012	100	60	61.7092	61.7092	NUM
fcis-23012	100	61	gcn+bi	gcn+bi	PROPN
fcis-23012	100	62	-	-	PUNCT
fcis-23012	100	63	lstm	lstm	PROPN
fcis-23012	100	64	cross	cross	NOUN
fcis-23012	100	65	entropy	entropy	PROPN
fcis-23012	100	66	loss	loss	PROPN
fcis-23012	100	67	+	+	CCONJ
fcis-23012	100	68	center	center	NOUN
fcis-23012	100	69	loss	loss	NOUN
fcis-23012	100	70	63.8897	63.8897	NUM
fcis-23012	100	71	it	it	PRON
fcis-23012	100	72	can	can	AUX
fcis-23012	100	73	be	be	AUX
fcis-23012	100	74	seen	see	VERB
fcis-23012	100	75	that	that	SCONJ
fcis-23012	100	76	compared	compare	VERB
fcis-23012	100	77	with	with	ADP
fcis-23012	100	78	the	the	DET
fcis-23012	100	79	previous	previous	ADJ
fcis-23012	100	80	convolutional	convolutional	ADJ
fcis-23012	100	81	models	model	NOUN
fcis-23012	100	82	,	,	PUNCT
fcis-23012	100	83	the	the	DET
fcis-23012	100	84	graph	graph	NOUN
fcis-23012	100	85	convolutional	convolutional	ADJ
fcis-23012	100	86	network	network	NOUN
fcis-23012	100	87	(	(	PUNCT
fcis-23012	100	88	gcn	gcn	NOUN
fcis-23012	100	89	)	)	PUNCT
fcis-23012	100	90	performs	perform	VERB
fcis-23012	100	91	better	well	ADV
fcis-23012	100	92	in	in	ADP
fcis-23012	100	93	the	the	DET
fcis-23012	100	94	task	task	NOUN
fcis-23012	100	95	of	of	ADP
fcis-23012	100	96	tibetan	tibetan	ADJ
fcis-23012	100	97	speech	speech	NOUN
fcis-23012	100	98	emotion	emotion	NOUN
fcis-23012	100	99	recognition	recognition	NOUN
fcis-23012	100	100	,	,	PUNCT
fcis-23012	100	101	and	and	CCONJ
fcis-23012	100	102	the	the	DET
fcis-23012	100	103	accuracy	accuracy	NOUN
fcis-23012	100	104	of	of	ADP
fcis-23012	100	105	gcn	gcn	NOUN
fcis-23012	100	106	is	be	AUX
fcis-23012	100	107	more	more	ADJ
fcis-23012	100	108	than	than	ADP
fcis-23012	100	109	2	2	NUM
fcis-23012	100	110	percentage	percentage	NOUN
fcis-23012	100	111	points	point	NOUN
fcis-23012	100	112	higher	high	ADJ
fcis-23012	100	113	than	than	ADP
fcis-23012	100	114	that	that	PRON
fcis-23012	100	115	of	of	ADP
fcis-23012	100	116	traditional	traditional	ADJ
fcis-23012	100	117	convolutional	convolutional	ADJ
fcis-23012	100	118	networks	network	NOUN
fcis-23012	100	119	.	.	PUNCT
fcis-23012	101	1	moreover	moreover	ADV
fcis-23012	101	2	,	,	PUNCT
fcis-23012	101	3	the	the	DET
fcis-23012	101	4	lstm	lstm	NOUN
fcis-23012	101	5	module	module	NOUN
fcis-23012	101	6	can	can	AUX
fcis-23012	101	7	indeed	indeed	ADV
fcis-23012	101	8	compensate	compensate	VERB
fcis-23012	101	9	for	for	ADP
fcis-23012	101	10	the	the	DET
fcis-23012	101	11	deficiency	deficiency	NOUN
fcis-23012	101	12	of	of	ADP
fcis-23012	101	13	gcn	gcn	NOUN
fcis-23012	101	14	in	in	ADP
fcis-23012	101	15	extracting	extract	VERB
fcis-23012	101	16	global	global	ADJ
fcis-23012	101	17	features	feature	NOUN
fcis-23012	101	18	,	,	PUNCT
fcis-23012	101	19	thereby	thereby	ADV
fcis-23012	101	20	improving	improve	VERB
fcis-23012	101	21	the	the	DET
fcis-23012	101	22	performance	performance	NOUN
fcis-23012	101	23	of	of	ADP
fcis-23012	101	24	the	the	DET
fcis-23012	101	25	model	model	NOUN
fcis-23012	101	26	and	and	CCONJ
fcis-23012	101	27	increasing	increase	VERB
fcis-23012	101	28	the	the	DET
fcis-23012	101	29	accuracy	accuracy	NOUN
fcis-23012	101	30	by	by	ADP
fcis-23012	101	31	about	about	ADV
fcis-23012	101	32	2	2	NUM
fcis-23012	101	33	%	%	NOUN
fcis-23012	101	34	.	.	PUNCT
fcis-23012	102	1	in	in	ADP
fcis-23012	102	2	terms	term	NOUN
fcis-23012	102	3	of	of	ADP
fcis-23012	102	4	loss	loss	NOUN
fcis-23012	102	5	function	function	NOUN
fcis-23012	102	6	,	,	PUNCT
fcis-23012	102	7	the	the	DET
fcis-23012	102	8	joint	joint	ADJ
fcis-23012	102	9	loss	loss	NOUN
fcis-23012	102	10	function	function	NOUN
fcis-23012	102	11	of	of	ADP
fcis-23012	102	12	cross	cross	NOUN
fcis-23012	102	13	entropy	entropy	PROPN
fcis-23012	102	14	and	and	CCONJ
fcis-23012	102	15	center	center	NOUN
fcis-23012	102	16	loss	loss	NOUN
fcis-23012	102	17	captures	capture	VERB
fcis-23012	102	18	the	the	DET
fcis-23012	102	19	optimization	optimization	NOUN
fcis-23012	102	20	direction	direction	NOUN
fcis-23012	102	21	and	and	CCONJ
fcis-23012	102	22	features	feature	NOUN
fcis-23012	102	23	of	of	ADP
fcis-23012	102	24	tibetan	tibetan	ADJ
fcis-23012	102	25	speech	speech	NOUN
fcis-23012	102	26	emotion	emotion	NOUN
fcis-23012	102	27	recognition	recognition	NOUN
fcis-23012	102	28	better	well	ADV
fcis-23012	102	29	than	than	ADP
fcis-23012	102	30	the	the	DET
fcis-23012	102	31	single	single	PROPN
fcis-23012	102	32	cross	cross	PROPN
fcis-23012	102	33	entropy	entropy	PROPN
fcis-23012	102	34	loss	loss	NOUN
fcis-23012	102	35	function	function	NOUN
fcis-23012	102	36	used	use	VERB
fcis-23012	102	37	in	in	ADP
fcis-23012	102	38	the	the	DET
fcis-23012	102	39	past	past	NOUN
fcis-23012	102	40	.	.	PUNCT
fcis-23012	103	1	on	on	ADP
fcis-23012	103	2	the	the	DET
fcis-23012	103	3	best	well	ADV
fcis-23012	103	4	-	-	PUNCT
fcis-23012	103	5	performing	perform	VERB
fcis-23012	103	6	gcn+bi	gcn+bi	NOUN
fcis-23012	103	7	-	-	PUNCT
fcis-23012	103	8	lstm	lstm	NOUN
fcis-23012	103	9	network	network	NOUN
fcis-23012	103	10	,	,	PUNCT
fcis-23012	103	11	using	use	VERB
fcis-23012	103	12	the	the	DET
fcis-23012	103	13	joint	joint	ADJ
fcis-23012	103	14	loss	loss	NOUN
fcis-23012	103	15	function	function	NOUN
fcis-23012	103	16	can	can	AUX
fcis-23012	103	17	increase	increase	VERB
fcis-23012	103	18	the	the	DET
fcis-23012	103	19	accuracy	accuracy	NOUN
fcis-23012	103	20	by	by	ADP
fcis-23012	103	21	about	about	ADV
fcis-23012	103	22	2	2	NUM
fcis-23012	103	23	%	%	NOUN
fcis-23012	103	24	compared	compare	VERB
fcis-23012	103	25	to	to	ADP
fcis-23012	103	26	using	use	VERB
fcis-23012	103	27	single	single	ADJ
fcis-23012	103	28	cross	cross	NOUN
fcis-23012	103	29	entropy	entropy	NOUN
fcis-23012	103	30	.	.	PUNCT
fcis-23012	104	1	5	5	X
fcis-23012	104	2	.	.	X
fcis-23012	104	3	conclusion	conclusion	NOUN
fcis-23012	104	4	in	in	ADP
fcis-23012	104	5	the	the	DET
fcis-23012	104	6	task	task	NOUN
fcis-23012	104	7	of	of	ADP
fcis-23012	104	8	lhasa	lhasa	PROPN
fcis-23012	104	9	tibetan	tibetan	PROPN
fcis-23012	104	10	speech	speech	PROPN
fcis-23012	104	11	emotion	emotion	NOUN
fcis-23012	104	12	recognition	recognition	NOUN
fcis-23012	104	13	,	,	PUNCT
fcis-23012	104	14	the	the	DET
fcis-23012	104	15	gcn+bi	gcn+bi	PROPN
fcis-23012	104	16	-	-	PUNCT
fcis-23012	104	17	lstm	lstm	NOUN
fcis-23012	104	18	network	network	NOUN
fcis-23012	104	19	performs	perform	VERB
fcis-23012	104	20	better	well	ADJ
fcis-23012	104	21	than	than	ADP
fcis-23012	104	22	traditional	traditional	ADJ
fcis-23012	104	23	models	model	NOUN
fcis-23012	104	24	,	,	PUNCT
fcis-23012	104	25	sacrificing	sacrifice	VERB
fcis-23012	104	26	a	a	DET
fcis-23012	104	27	bit	bit	NOUN
fcis-23012	104	28	of	of	ADP
fcis-23012	104	29	convergence	convergence	NOUN
fcis-23012	104	30	time	time	NOUN
fcis-23012	104	31	to	to	PART
fcis-23012	104	32	achieve	achieve	VERB
fcis-23012	104	33	higher	high	ADJ
fcis-23012	104	34	accuracy	accuracy	NOUN
fcis-23012	104	35	.	.	PUNCT
fcis-23012	105	1	moreover	moreover	ADV
fcis-23012	105	2	,	,	PUNCT
fcis-23012	105	3	the	the	DET
fcis-23012	105	4	joint	joint	ADJ
fcis-23012	105	5	loss	loss	NOUN
fcis-23012	105	6	function	function	NOUN
fcis-23012	105	7	combining	combine	VERB
fcis-23012	105	8	crossentropy	crossentropy	NOUN
fcis-23012	105	9	loss	loss	NOUN
fcis-23012	105	10	and	and	CCONJ
fcis-23012	105	11	center	center	NOUN
fcis-23012	105	12	loss	loss	NOUN
fcis-23012	105	13	also	also	ADV
fcis-23012	105	14	performs	perform	VERB
fcis-23012	105	15	better	well	ADV
fcis-23012	105	16	in	in	ADP
fcis-23012	105	17	this	this	DET
fcis-23012	105	18	task	task	NOUN
fcis-23012	105	19	than	than	ADP
fcis-23012	105	20	when	when	SCONJ
fcis-23012	105	21	only	only	ADV
fcis-23012	105	22	using	use	VERB
fcis-23012	105	23	crossentropy	crossentropy	NOUN
fcis-23012	105	24	as	as	ADP
fcis-23012	105	25	the	the	DET
fcis-23012	105	26	loss	loss	NOUN
fcis-23012	105	27	function	function	NOUN
fcis-23012	105	28	.	.	PUNCT
fcis-23012	106	1	considering	consider	VERB
fcis-23012	106	2	that	that	SCONJ
fcis-23012	106	3	many	many	ADJ
fcis-23012	106	4	models	model	NOUN
fcis-23012	106	5	did	do	AUX
fcis-23012	106	6	not	not	PART
fcis-23012	106	7	achieve	achieve	VERB
fcis-23012	106	8	high	high	ADJ
fcis-23012	106	9	accuracy	accuracy	NOUN
fcis-23012	106	10	in	in	ADP
fcis-23012	106	11	this	this	DET
fcis-23012	106	12	experiment	experiment	NOUN
fcis-23012	106	13	,	,	PUNCT
fcis-23012	106	14	there	there	PRON
fcis-23012	106	15	are	be	VERB
fcis-23012	106	16	currently	currently	ADV
fcis-23012	106	17	several	several	ADJ
fcis-23012	106	18	areas	area	NOUN
fcis-23012	106	19	for	for	ADP
fcis-23012	106	20	improvement	improvement	NOUN
fcis-23012	106	21	based	base	VERB
fcis-23012	106	22	on	on	ADP
fcis-23012	106	23	the	the	DET
fcis-23012	106	24	analysis	analysis	NOUN
fcis-23012	106	25	:	:	PUNCT
fcis-23012	107	1	1	1	X
fcis-23012	107	2	.	.	X
fcis-23012	107	3	the	the	DET
fcis-23012	107	4	tibetan	tibetan	PROPN
fcis-23012	107	5	speech	speech	NOUN
fcis-23012	107	6	recognition	recognition	NOUN
fcis-23012	107	7	dataset	dataset	NOUN
fcis-23012	107	8	is	be	AUX
fcis-23012	107	9	insufficient	insufficient	ADJ
fcis-23012	107	10	,	,	PUNCT
fcis-23012	107	11	and	and	CCONJ
fcis-23012	107	12	more	more	ADV
fcis-23012	107	13	high	high	ADJ
fcis-23012	107	14	-	-	PUNCT
fcis-23012	107	15	quality	quality	NOUN
fcis-23012	107	16	speech	speech	NOUN
fcis-23012	107	17	data	datum	NOUN
fcis-23012	107	18	needs	need	VERB
fcis-23012	107	19	to	to	PART
fcis-23012	107	20	be	be	AUX
fcis-23012	107	21	collected	collect	VERB
fcis-23012	107	22	.	.	PUNCT
fcis-23012	108	1	2	2	X
fcis-23012	108	2	.	.	X
fcis-23012	108	3	tibetan	tibetan	PROPN
fcis-23012	108	4	,	,	PUNCT
fcis-23012	108	5	as	as	ADP
fcis-23012	108	6	a	a	DET
fcis-23012	108	7	language	language	NOUN
fcis-23012	108	8	with	with	ADP
fcis-23012	108	9	unique	unique	ADJ
fcis-23012	108	10	pronunciation	pronunciation	NOUN
fcis-23012	108	11	habits	habit	NOUN
fcis-23012	108	12	,	,	PUNCT
fcis-23012	108	13	has	have	AUX
fcis-23012	108	14	not	not	PART
fcis-23012	108	15	yet	yet	ADV
fcis-23012	108	16	found	find	VERB
fcis-23012	108	17	features	feature	NOUN
fcis-23012	108	18	that	that	PRON
fcis-23012	108	19	can	can	AUX
fcis-23012	108	20	efficiently	efficiently	ADV
fcis-23012	108	21	represent	represent	VERB
fcis-23012	108	22	emotions	emotion	NOUN
fcis-23012	108	23	.	.	PUNCT
fcis-23012	109	1	the	the	DET
fcis-23012	109	2	features	feature	NOUN
fcis-23012	109	3	extracted	extract	VERB
fcis-23012	109	4	are	be	AUX
fcis-23012	109	5	mostly	mostly	ADV
fcis-23012	109	6	commonly	commonly	ADV
fcis-23012	109	7	used	use	VERB
fcis-23012	109	8	in	in	ADP
fcis-23012	109	9	speech	speech	NOUN
fcis-23012	109	10	emotion	emotion	NOUN
fcis-23012	109	11	recognition	recognition	NOUN
fcis-23012	109	12	.	.	PUNCT
fcis-23012	110	1	3	3	X
fcis-23012	110	2	.	.	X
fcis-23012	110	3	there	there	PRON
fcis-23012	110	4	is	be	VERB
fcis-23012	110	5	still	still	ADV
fcis-23012	110	6	room	room	NOUN
fcis-23012	110	7	for	for	ADP
fcis-23012	110	8	improvement	improvement	NOUN
fcis-23012	110	9	in	in	ADP
fcis-23012	110	10	terms	term	NOUN
fcis-23012	110	11	of	of	ADP
fcis-23012	110	12	network	network	NOUN
fcis-23012	110	13	complexity	complexity	NOUN
fcis-23012	110	14	.	.	PUNCT
fcis-23012	111	1	“	"	PUNCT
fcis-23012	111	2	in	in	ADP
fcis-23012	111	3	future	future	ADJ
fcis-23012	111	4	tibetan	tibetan	ADJ
fcis-23012	111	5	speech	speech	NOUN
fcis-23012	111	6	emotion	emotion	NOUN
fcis-23012	111	7	recognition	recognition	NOUN
fcis-23012	111	8	tasks	task	NOUN
fcis-23012	111	9	,	,	PUNCT
fcis-23012	111	10	graph	graph	NOUN
fcis-23012	111	11	neural	neural	ADJ
fcis-23012	111	12	networks	network	NOUN
fcis-23012	111	13	may	may	AUX
fcis-23012	111	14	have	have	VERB
fcis-23012	111	15	more	more	ADJ
fcis-23012	111	16	applications	application	NOUN
fcis-23012	111	17	and	and	CCONJ
fcis-23012	111	18	development	development	NOUN
fcis-23012	111	19	.	.	PUNCT
fcis-23012	112	1	references	reference	NOUN
fcis-23012	112	2	[	[	X
fcis-23012	112	3	1	1	NUM
fcis-23012	112	4	]	]	X
fcis-23012	112	5	guzeyue	guzeyue	NOUN
fcis-23012	112	6	,	,	PUNCT
fcis-23012	112	7	bianbawangdui	bianbawangdui	PROPN
fcis-23012	112	8	,	,	PUNCT
fcis-23012	112	9	qi	qi	PROPN
fcis-23012	112	10	jindong	jindong	PROPN
fcis-23012	112	11	.	.	PUNCT
fcis-23012	113	1	tibetan	tibetan	ADJ
fcis-23012	113	2	speech	speech	PROPN
fcis-23012	113	3	emotion	emotion	NOUN
fcis-23012	113	4	recognition	recognition	NOUN
fcis-23012	113	5	based	base	VERB
fcis-23012	113	6	on	on	ADP
fcis-23012	113	7	multi	multi	ADJ
fcis-23012	113	8	-	-	ADJ
fcis-23012	113	9	feature	feature	ADJ
fcis-23012	113	10	fusion[j	fusion[j	NOUN
fcis-23012	113	11	]	]	PUNCT
fcis-23012	113	12	.	.	PUNCT
fcis-23012	114	1	modern	modern	ADJ
fcis-23012	114	2	electronic	electronic	ADJ
fcis-23012	114	3	technology	technology	NOUN
fcis-23012	114	4	,	,	PUNCT
fcis-23012	114	5	2023	2023	NUM
fcis-23012	114	6	,	,	PUNCT
fcis-23012	114	7	46(21	46(21	NUM
fcis-23012	114	8	):	):	PUNCT
fcis-23012	114	9	129	129	NUM
fcis-23012	114	10	-	-	SYM
fcis-23012	114	11	133	133	NUM
fcis-23012	114	12	.	.	PUNCT
fcis-23012	115	1	doi	doi	NOUN
fcis-23012	115	2	:	:	PUNCT
fcis-23012	115	3	10	10	NUM
fcis-23012	115	4	.	.	X
fcis-23012	115	5	16652	16652	NUM
fcis-23012	115	6	/	/	SYM
fcis-23012	115	7	j.issn.1004	j.issn.1004	NOUN
fcis-23012	115	8	-	-	PUNCT
fcis-23012	115	9	373x.2023.21.024	373x.2023.21.024	NOUN
fcis-23012	115	10	.	.	PUNCT
fcis-23012	116	1	[	[	X
fcis-23012	116	2	2	2	NUM
fcis-23012	116	3	]	]	X
fcis-23012	116	4	cai	cai	PROPN
fcis-23012	116	5	youxin	youxin	PROPN
fcis-23012	116	6	,	,	PUNCT
fcis-23012	116	7	bianbawangdui	bianbawangdui	PROPN
fcis-23012	116	8	.	.	PUNCT
fcis-23012	116	9	tibetan	tibetan	ADJ
fcis-23012	116	10	speech	speech	NOUN
fcis-23012	116	11	emotion	emotion	NOUN
fcis-23012	116	12	recognition	recognition	NOUN
fcis-23012	116	13	based	base	VERB
fcis-23012	116	14	on	on	ADP
fcis-23012	116	15	bidirectional	bidirectional	PROPN
fcis-23012	116	16	gru	gru	PROPN
fcis-23012	116	17	model[j	model[j	PROPN
fcis-23012	116	18	]	]	PUNCT
fcis-23012	116	19	.	.	PUNCT
fcis-23012	117	1	information	information	NOUN
fcis-23012	117	2	technology	technology	NOUN
fcis-23012	117	3	and	and	CCONJ
fcis-23012	117	4	informatization	informatization	NOUN
fcis-23012	117	5	,	,	PUNCT
fcis-23012	117	6	2023	2023	NUM
fcis-23012	117	7	,	,	PUNCT
fcis-23012	117	8	(	(	PUNCT
fcis-23012	117	9	10	10	NUM
fcis-23012	117	10	):	):	PUNCT
fcis-23012	117	11	209213	209213	NUM
fcis-23012	117	12	.	.	PUNCT
fcis-23012	118	1	[	[	X
fcis-23012	118	2	3	3	X
fcis-23012	118	3	]	]	X
fcis-23012	118	4	ding	ding	PROPN
fcis-23012	118	5	nan	nan	PROPN
fcis-23012	118	6	.	.	PUNCT
fcis-23012	119	1	research	research	NOUN
fcis-23012	119	2	on	on	ADP
fcis-23012	119	3	speech	speech	NOUN
fcis-23012	119	4	emotion	emotion	NOUN
fcis-23012	119	5	recognition	recognition	NOUN
fcis-23012	119	6	based	base	VERB
fcis-23012	119	7	on	on	ADP
fcis-23012	119	8	feature	feature	NOUN
fcis-23012	119	9	learning[d	learning[d	PROPN
fcis-23012	119	10	]	]	PUNCT
fcis-23012	119	11	.	.	PUNCT
fcis-23012	120	1	nanjing	nanjing	PROPN
fcis-23012	120	2	university	university	PROPN
fcis-23012	120	3	of	of	ADP
fcis-23012	120	4	posts	post	NOUN
fcis-23012	120	5	and	and	CCONJ
fcis-23012	120	6	telecommunications	telecommunication	NOUN
fcis-23012	120	7	,	,	PUNCT
fcis-23012	120	8	2023	2023	NUM
fcis-23012	120	9	.	.	PUNCT
fcis-23012	121	1	doi	doi	NOUN
fcis-23012	121	2	:	:	PUNCT
fcis-23012	121	3	10.27251	10.27251	NUM
fcis-23012	121	4	/	/	SYM
fcis-23012	121	5	d.cnki	d.cnki	NOUN
fcis-23012	121	6	.	.	PUNCT
fcis-23012	121	7	gnjdc	gnjdc	PROPN
fcis-23012	121	8	.	.	PUNCT
fcis-23012	122	1	2023	2023	NUM
fcis-23012	122	2	.	.	PUNCT
fcis-23012	123	1	001294	001294	NUM
fcis-23012	123	2	.	.	PUNCT
fcis-23012	124	1	[	[	X
fcis-23012	124	2	4	4	NUM
fcis-23012	124	3	]	]	X
fcis-23012	124	4	akçay	akçay	NOUN
fcis-23012	124	5	m	m	PROPN
fcis-23012	124	6	b	b	PROPN
fcis-23012	124	7	,	,	PUNCT
fcis-23012	124	8	oğuz	oğuz	PROPN
fcis-23012	124	9	k.	k.	PROPN
fcis-23012	124	10	speech	speech	PROPN
fcis-23012	124	11	emotion	emotion	NOUN
fcis-23012	124	12	recognition	recognition	NOUN
fcis-23012	124	13	:	:	PUNCT
fcis-23012	124	14	emotional	emotional	ADJ
fcis-23012	124	15	models	model	NOUN
fcis-23012	124	16	,	,	PUNCT
fcis-23012	124	17	databases	database	NOUN
fcis-23012	124	18	,	,	PUNCT
fcis-23012	124	19	features	feature	NOUN
fcis-23012	124	20	,	,	PUNCT
fcis-23012	124	21	preprocessing	preprocesse	VERB
fcis-23012	124	22	methods	method	NOUN
fcis-23012	124	23	,	,	PUNCT
fcis-23012	124	24	supporting	support	VERB
fcis-23012	124	25	modalities	modality	NOUN
fcis-23012	124	26	,	,	PUNCT
fcis-23012	124	27	and	and	CCONJ
fcis-23012	124	28	classifiers[j	classifiers[j	PROPN
fcis-23012	124	29	]	]	PUNCT
fcis-23012	124	30	.	.	PUNCT
fcis-23012	125	1	speech	speech	NOUN
fcis-23012	125	2	communication	communication	NOUN
fcis-23012	125	3	,	,	PUNCT
fcis-23012	125	4	2020	2020	NUM
fcis-23012	125	5	,	,	PUNCT
fcis-23012	125	6	116	116	NUM
fcis-23012	125	7	:	:	PUNCT
fcis-23012	125	8	56	56	NUM
fcis-23012	125	9	-	-	SYM
fcis-23012	125	10	76	76	NUM
fcis-23012	125	11	.	.	PUNCT
fcis-23012	126	1	[	[	X
fcis-23012	126	2	5	5	X
fcis-23012	126	3	]	]	X
fcis-23012	126	4	huang	huang	PROPN
fcis-23012	126	5	xiyang	xiyang	PROPN
fcis-23012	126	6	,	,	PUNCT
fcis-23012	126	7	du	du	PROPN
fcis-23012	126	8	qingzhi	qingzhi	PROPN
fcis-23012	126	9	,	,	PUNCT
fcis-23012	126	10	long	long	PROPN
fcis-23012	126	11	hua	hua	PROPN
fcis-23012	126	12	,	,	PUNCT
fcis-23012	126	13	et	et	PROPN
fcis-23012	126	14	al	al	PROPN
fcis-23012	126	15	.	.	PROPN
fcis-23012	126	16	speech	speech	NOUN
fcis-23012	126	17	emotion	emotion	NOUN
fcis-23012	126	18	recognition	recognition	NOUN
fcis-23012	126	19	algorithm	algorithm	NOUN
fcis-23012	126	20	based	base	VERB
fcis-23012	126	21	on	on	ADP
fcis-23012	126	22	mfcc	mfcc	NOUN
fcis-23012	126	23	feature	feature	NOUN
fcis-23012	126	24	fusion[j	fusion[j	PROPN
fcis-23012	126	25	]	]	PUNCT
fcis-23012	126	26	.	.	PUNCT
fcis-23012	127	1	journal	journal	PROPN
fcis-23012	127	2	of	of	ADP
fcis-23012	127	3	shaanxi	shaanxi	PROPN
fcis-23012	127	4	university	university	PROPN
fcis-23012	127	5	of	of	ADP
fcis-23012	127	6	technology	technology	NOUN
fcis-23012	127	7	(	(	PUNCT
fcis-23012	127	8	natural	natural	ADJ
fcis-23012	127	9	science	science	NOUN
fcis-23012	127	10	edition	edition	NOUN
fcis-23012	127	11	)	)	PUNCT
fcis-23012	127	12	,	,	PUNCT
fcis-23012	127	13	2023	2023	NUM
fcis-23012	127	14	,	,	PUNCT
fcis-23012	127	15	39(04	39(04	NUM
fcis-23012	127	16	):	):	PUNCT
fcis-23012	127	17	17	17	NUM
fcis-23012	127	18	-	-	SYM
fcis-23012	127	19	25	25	NUM
fcis-23012	127	20	.	.	PUNCT
fcis-23012	128	1	[	[	X
fcis-23012	128	2	6	6	NUM
fcis-23012	128	3	]	]	PUNCT
fcis-23012	128	4	shirian	shirian	ADJ
fcis-23012	128	5	a	a	X
fcis-23012	128	6	,	,	PUNCT
fcis-23012	128	7	guha	guha	PROPN
fcis-23012	128	8	t.	t.	PROPN
fcis-23012	128	9	compact	compact	ADJ
fcis-23012	128	10	graph	graph	NOUN
fcis-23012	128	11	architecture	architecture	NOUN
fcis-23012	128	12	for	for	ADP
fcis-23012	128	13	speech	speech	NOUN
fcis-23012	128	14	emotion	emotion	NOUN
fcis-23012	128	15	recognition[c]//icassp	recognition[c]//icassp	PROPN
fcis-23012	128	16	2021	2021	NUM
fcis-23012	128	17	-	-	SYM
fcis-23012	128	18	2021	2021	NUM
fcis-23012	128	19	ieee	ieee	NOUN
fcis-23012	128	20	international	international	ADJ
fcis-23012	128	21	conference	conference	NOUN
fcis-23012	128	22	on	on	ADP
fcis-23012	128	23	acoustics	acoustic	NOUN
fcis-23012	128	24	,	,	PUNCT
fcis-23012	128	25	speech	speech	NOUN
fcis-23012	128	26	and	and	CCONJ
fcis-23012	128	27	signal	signal	NOUN
fcis-23012	128	28	processing	processing	NOUN
fcis-23012	128	29	(	(	PUNCT
fcis-23012	128	30	icassp	icassp	PROPN
fcis-23012	128	31	)	)	PUNCT
fcis-23012	128	32	.	.	PUNCT
fcis-23012	129	1	ieee	ieee	NOUN
fcis-23012	129	2	,	,	PUNCT
fcis-23012	129	3	2021	2021	NUM
fcis-23012	129	4	:	:	PUNCT
fcis-23012	129	5	6284	6284	NUM
fcis-23012	129	6	-	-	SYM
fcis-23012	129	7	6288	6288	NUM
fcis-23012	129	8	.	.	PUNCT
fcis-23012	130	1	[	[	X
fcis-23012	130	2	7	7	X
fcis-23012	130	3	]	]	X
fcis-23012	130	4	liu	liu	PROPN
fcis-23012	130	5	j	j	PROPN
fcis-23012	130	6	,	,	PUNCT
fcis-23012	130	7	wang	wang	PROPN
fcis-23012	130	8	h.	h.	PROPN
fcis-23012	130	9	graph	graph	PROPN
fcis-23012	130	10	isomorphism	isomorphism	NOUN
fcis-23012	130	11	network	network	NOUN
fcis-23012	130	12	for	for	ADP
fcis-23012	130	13	speech	speech	NOUN
fcis-23012	130	14	emotion	emotion	NOUN
fcis-23012	130	15	recognition[c]//interspeech	recognition[c]//interspeech	NOUN
fcis-23012	130	16	.	.	PUNCT
fcis-23012	131	1	2021	2021	NUM
fcis-23012	131	2	:	:	PUNCT
fcis-23012	131	3	3405	3405	NUM
fcis-23012	131	4	-	-	SYM
fcis-23012	131	5	3409	3409	NUM
fcis-23012	131	6	.	.	PUNCT
fcis-23012	132	1	[	[	X
fcis-23012	132	2	8	8	NUM
fcis-23012	132	3	]	]	X
fcis-23012	132	4	shirian	shirian	ADJ
fcis-23012	132	5	a	a	X
fcis-23012	132	6	,	,	PUNCT
fcis-23012	132	7	guha	guha	PROPN
fcis-23012	132	8	t.	t.	PROPN
fcis-23012	132	9	compact	compact	ADJ
fcis-23012	132	10	graph	graph	NOUN
fcis-23012	132	11	architecture	architecture	NOUN
fcis-23012	132	12	for	for	ADP
fcis-23012	132	13	speech	speech	NOUN
fcis-23012	132	14	emotion	emotion	NOUN
fcis-23012	132	15	recognition[c]//icassp	recognition[c]//icassp	PROPN
fcis-23012	132	16	2021	2021	NUM
fcis-23012	132	17	-	-	SYM
fcis-23012	132	18	2021	2021	NUM
fcis-23012	132	19	ieee	ieee	NOUN
fcis-23012	132	20	international	international	ADJ
fcis-23012	132	21	conference	conference	NOUN
fcis-23012	132	22	on	on	ADP
fcis-23012	132	23	acoustics	acoustic	NOUN
fcis-23012	132	24	,	,	PUNCT
fcis-23012	132	25	speech	speech	NOUN
fcis-23012	132	26	and	and	CCONJ
fcis-23012	132	27	signal	signal	NOUN
fcis-23012	132	28	processing	processing	NOUN
fcis-23012	132	29	(	(	PUNCT
fcis-23012	132	30	icassp	icassp	PROPN
fcis-23012	132	31	)	)	PUNCT
fcis-23012	132	32	.	.	PUNCT
fcis-23012	133	1	ieee	ieee	NOUN
fcis-23012	133	2	,	,	PUNCT
fcis-23012	133	3	2021	2021	NUM
fcis-23012	133	4	:	:	PUNCT
fcis-23012	133	5	6284	6284	NUM
fcis-23012	133	6	-	-	SYM
fcis-23012	133	7	6288	6288	NUM
fcis-23012	133	8	.	.	PUNCT
fcis-23012	134	1	[	[	X
fcis-23012	134	2	9	9	NUM
fcis-23012	134	3	]	]	X
fcis-23012	134	4	liu	liu	PROPN
fcis-23012	134	5	j	j	PROPN
fcis-23012	134	6	,	,	PUNCT
fcis-23012	134	7	wang	wang	PROPN
fcis-23012	134	8	h	h	PROPN
fcis-23012	134	9	,	,	PUNCT
fcis-23012	134	10	sun	sun	PROPN
fcis-23012	134	11	m	m	PROPN
fcis-23012	134	12	,	,	PUNCT
fcis-23012	134	13	et	et	PROPN
fcis-23012	134	14	al	al	PROPN
fcis-23012	134	15	.	.	PROPN
fcis-23012	134	16	graph	graph	PROPN
fcis-23012	134	17	based	base	VERB
fcis-23012	134	18	emotion	emotion	NOUN
fcis-23012	134	19	recognition	recognition	NOUN
fcis-23012	134	20	with	with	ADP
fcis-23012	134	21	attention	attention	NOUN
fcis-23012	134	22	pooling	pool	VERB
fcis-23012	134	23	for	for	ADP
fcis-23012	134	24	variable	variable	ADJ
fcis-23012	134	25	-	-	PUNCT
fcis-23012	134	26	length	length	NOUN
fcis-23012	134	27	utterances[j	utterances[j	PROPN
fcis-23012	134	28	]	]	PUNCT
fcis-23012	134	29	.	.	PUNCT
fcis-23012	135	1	neurocomputing	neurocomputing	NOUN
fcis-23012	135	2	,	,	PUNCT
fcis-23012	135	3	2022	2022	NUM
fcis-23012	135	4	,	,	PUNCT
fcis-23012	135	5	496	496	NUM
fcis-23012	135	6	:	:	SYM
fcis-23012	135	7	46	46	NUM
fcis-23012	135	8	-	-	SYM
fcis-23012	135	9	55	55	NUM
fcis-23012	135	10	.	.	PUNCT
fcis-23012	136	1	[	[	X
fcis-23012	136	2	10	10	NUM
fcis-23012	136	3	]	]	X
fcis-23012	136	4	li	li	PROPN
fcis-23012	136	5	zijing	zijing	PROPN
fcis-23012	136	6	,	,	PUNCT
fcis-23012	136	7	chen	chen	PROPN
fcis-23012	136	8	ning	ning	PROPN
fcis-23012	136	9	.	.	PUNCT
fcis-23012	136	10	speech	speech	NOUN
fcis-23012	136	11	emotion	emotion	NOUN
fcis-23012	136	12	recognition	recognition	NOUN
fcis-23012	136	13	model	model	NOUN
fcis-23012	136	14	based	base	VERB
fcis-23012	136	15	on	on	ADP
fcis-23012	136	16	multi	multi	ADJ
fcis-23012	136	17	-	-	ADJ
fcis-23012	136	18	modal	modal	ADJ
fcis-23012	136	19	fusion	fusion	NOUN
fcis-23012	136	20	of	of	ADP
fcis-23012	136	21	graph	graph	NOUN
fcis-23012	136	22	neural	neural	ADJ
fcis-23012	136	23	network[j	network[j	PROPN
fcis-23012	136	24	]	]	PUNCT
fcis-23012	136	25	.	.	PUNCT
fcis-23012	137	1	computer	computer	NOUN
fcis-23012	137	2	application	application	NOUN
fcis-23012	137	3	research	research	NOUN
fcis-23012	137	4	,	,	PUNCT
fcis-23012	137	5	2023	2023	NUM
fcis-23012	137	6	,	,	PUNCT
fcis-23012	137	7	40(08	40(08	NUM
fcis-23012	137	8	):	):	PUNCT
fcis-23012	137	9	22862291	22862291	NUM
fcis-23012	137	10	+	+	SYM
fcis-23012	137	11	2310	2310	NUM
fcis-23012	137	12	.	.	PUNCT
fcis-23012	138	1	doi	doi	NOUN
fcis-23012	138	2	:	:	PUNCT
fcis-23012	138	3	10.19734	10.19734	NUM
fcis-23012	138	4	/	/	SYM
fcis-23012	138	5	j.issn.1001	j.issn.1001	NOUN
fcis-23012	138	6	-	-	PUNCT
fcis-23012	138	7	3695.2023.01.0002	3695.2023.01.0002	NOUN
fcis-23012	138	8	.	.	PUNCT
fcis-23012	139	1	[	[	X
fcis-23012	139	2	11	11	NUM
fcis-23012	139	3	]	]	X
fcis-23012	139	4	zhang	zhang	PROPN
fcis-23012	139	5	s	s	PROPN
fcis-23012	139	6	,	,	PUNCT
fcis-23012	139	7	tong	tong	PROPN
fcis-23012	139	8	h	h	NOUN
fcis-23012	139	9	,	,	PUNCT
fcis-23012	139	10	xu	xu	PROPN
fcis-23012	139	11	j	j	PROPN
fcis-23012	139	12	,	,	PUNCT
fcis-23012	139	13	et	et	PROPN
fcis-23012	139	14	al	al	PROPN
fcis-23012	139	15	.	.	PROPN
fcis-23012	140	1	graph	graph	VERB
fcis-23012	140	2	convolutional	convolutional	ADJ
fcis-23012	140	3	networks	network	NOUN
fcis-23012	140	4	:	:	PUNCT
fcis-23012	140	5	algorithms	algorithm	NOUN
fcis-23012	140	6	,	,	PUNCT
fcis-23012	140	7	applications	application	NOUN
fcis-23012	140	8	and	and	CCONJ
fcis-23012	140	9	open	open	ADJ
fcis-23012	140	10	challenges[c]//	challenges[c]//	PROPN
fcis-23012	140	11	computational	computational	ADJ
fcis-23012	140	12	data	datum	NOUN
fcis-23012	140	13	and	and	CCONJ
fcis-23012	140	14	social	social	ADJ
fcis-23012	140	15	networks	network	NOUN
fcis-23012	140	16	:	:	PUNCT
fcis-23012	140	17	7th	7th	ADJ
fcis-23012	140	18	international	international	ADJ
fcis-23012	140	19	conference	conference	NOUN
fcis-23012	140	20	,	,	PUNCT
fcis-23012	140	21	csonet	csonet	NOUN
fcis-23012	140	22	2018	2018	NUM
fcis-23012	140	23	,	,	PUNCT
fcis-23012	140	24	shanghai	shanghai	PROPN
fcis-23012	140	25	,	,	PUNCT
fcis-23012	140	26	china	china	PROPN
fcis-23012	140	27	,	,	PUNCT
fcis-23012	140	28	december	december	PROPN
fcis-23012	140	29	18–20	18–20	NUM
fcis-23012	140	30	,	,	PUNCT
fcis-23012	140	31	2018	2018	NUM
fcis-23012	140	32	,	,	PUNCT
fcis-23012	140	33	proceedings	proceeding	NOUN
fcis-23012	140	34	7	7	NUM
fcis-23012	140	35	.	.	PUNCT
fcis-23012	140	36	springer	springer	NOUN
fcis-23012	140	37	international	international	ADJ
fcis-23012	140	38	publishing	publishing	NOUN
fcis-23012	140	39	,	,	PUNCT
fcis-23012	140	40	2018	2018	NUM
fcis-23012	140	41	:	:	PUNCT
fcis-23012	140	42	79	79	NUM
fcis-23012	140	43	-	-	SYM
fcis-23012	140	44	91	91	NUM
fcis-23012	140	45	.	.	PUNCT
fcis-23012	141	1	[	[	X
fcis-23012	141	2	12	12	NUM
fcis-23012	141	3	]	]	PUNCT
fcis-23012	141	4	xu	xu	PROPN
fcis-23012	141	5	huanan	huanan	PROPN
fcis-23012	141	6	,	,	PUNCT
fcis-23012	141	7	zhou	zhou	PROPN
fcis-23012	141	8	xiaoyan	xiaoyan	PROPN
fcis-23012	141	9	,	,	PUNCT
fcis-23012	141	10	jiang	jiang	PROPN
fcis-23012	141	11	wan	wan	PROPN
fcis-23012	141	12	,	,	PUNCT
fcis-23012	142	1	et	et	PROPN
fcis-23012	142	2	al	al	PROPN
fcis-23012	142	3	.	.	PROPN
fcis-23012	142	4	speech	speech	NOUN
fcis-23012	142	5	emotion	emotion	NOUN
fcis-23012	142	6	recognition	recognition	NOUN
fcis-23012	142	7	algorithm	algorithm	NOUN
fcis-23012	142	8	based	base	VERB
fcis-23012	142	9	on	on	ADP
fcis-23012	142	10	self	self	NOUN
fcis-23012	142	11	-	-	PUNCT
fcis-23012	142	12	attention	attention	NOUN
fcis-23012	142	13	temporal	temporal	ADJ
fcis-23012	142	14	and	and	CCONJ
fcis-23012	142	15	spatial	spatial	ADJ
fcis-23012	142	16	features[j	features[j	NOUN
fcis-23012	142	17	]	]	PUNCT
fcis-23012	142	18	.	.	PUNCT
fcis-23012	143	1	acoustic	acoustic	ADJ
fcis-23012	143	2	technology	technology	NOUN
fcis-23012	143	3	,	,	PUNCT
fcis-23012	143	4	2021	2021	NUM
fcis-23012	143	5	,	,	PUNCT
fcis-23012	143	6	40(06	40(06	NUM
fcis-23012	143	7	):	):	PUNCT
fcis-23012	143	8	807814	807814	NUM
fcis-23012	143	9	.	.	PUNCT
fcis-23012	144	1	doi	doi	NOUN
fcis-23012	144	2	:	:	PUNCT
fcis-23012	144	3	10.16300	10.16300	NUM
fcis-23012	144	4	/	/	SYM
fcis-23012	144	5	j.cnki.1000	j.cnki.1000	NOUN
fcis-23012	144	6	-	-	PUNCT
fcis-23012	144	7	3630.2021.06.011	3630.2021.06.011	PROPN
fcis-23012	144	8	.	.	PUNCT
fcis-23012	145	1	[	[	X
fcis-23012	145	2	13	13	NUM
fcis-23012	145	3	]	]	X
fcis-23012	145	4	peng	peng	PROPN
fcis-23012	145	5	z	z	PROPN
fcis-23012	145	6	,	,	PUNCT
fcis-23012	145	7	dang	dang	PROPN
fcis-23012	145	8	j	j	PROPN
fcis-23012	145	9	,	,	PUNCT
fcis-23012	145	10	unoki	unoki	PROPN
fcis-23012	145	11	m	m	PROPN
fcis-23012	145	12	,	,	PUNCT
fcis-23012	145	13	et	et	PROPN
fcis-23012	145	14	al	al	PROPN
fcis-23012	145	15	.	.	PUNCT
fcis-23012	145	16	multi	multi	ADJ
fcis-23012	145	17	-	-	ADJ
fcis-23012	145	18	resolution	resolution	NOUN
fcis-23012	145	19	modulationfiltered	modulationfiltere	VERB
fcis-23012	145	20	cochleagram	cochleagram	PROPN
fcis-23012	145	21	feature	feature	NOUN
fcis-23012	145	22	for	for	ADP
fcis-23012	145	23	lstm	lstm	NOUN
fcis-23012	145	24	-	-	PUNCT
fcis-23012	145	25	based	base	VERB
fcis-23012	145	26	dimensional	dimensional	ADJ
fcis-23012	145	27	emotion	emotion	NOUN
fcis-23012	145	28	recognition	recognition	NOUN
fcis-23012	145	29	from	from	ADP
fcis-23012	145	30	speech[j	speech[j	PROPN
fcis-23012	145	31	]	]	PUNCT
fcis-23012	145	32	.	.	PUNCT
fcis-23012	146	1	neural	neural	ADJ
fcis-23012	146	2	networks	network	NOUN
fcis-23012	146	3	,	,	PUNCT
fcis-23012	146	4	2021	2021	NUM
fcis-23012	146	5	,	,	PUNCT
fcis-23012	146	6	140	140	NUM
fcis-23012	146	7	:	:	SYM
fcis-23012	146	8	261	261	NUM
fcis-23012	146	9	-	-	SYM
fcis-23012	146	10	273	273	NUM
fcis-23012	146	11	.	.	PUNCT
fcis-23012	147	1	[	[	X
fcis-23012	147	2	14	14	NUM
fcis-23012	147	3	]	]	X
fcis-23012	147	4	li	li	PROPN
fcis-23012	147	5	y	y	PROPN
fcis-23012	147	6	,	,	PUNCT
fcis-23012	147	7	wang	wang	PROPN
fcis-23012	147	8	y	y	PROPN
fcis-23012	147	9	,	,	PUNCT
fcis-23012	147	10	yang	yang	PROPN
fcis-23012	147	11	x	x	PROPN
fcis-23012	147	12	,	,	PUNCT
fcis-23012	147	13	et	et	PROPN
fcis-23012	147	14	al	al	PROPN
fcis-23012	147	15	.	.	PROPN
fcis-23012	147	16	speech	speech	NOUN
fcis-23012	147	17	emotion	emotion	NOUN
fcis-23012	147	18	recognition	recognition	NOUN
fcis-23012	147	19	based	base	VERB
fcis-23012	147	20	on	on	ADP
fcis-23012	147	21	graph	graph	NOUN
fcis-23012	147	22	-	-	PUNCT
fcis-23012	147	23	lstm	lstm	NOUN
fcis-23012	147	24	neural	neural	ADJ
fcis-23012	147	25	network[j	network[j	PROPN
fcis-23012	147	26	]	]	PUNCT
fcis-23012	147	27	.	.	PUNCT
fcis-23012	148	1	eurasip	eurasip	PROPN
fcis-23012	148	2	journal	journal	PROPN
fcis-23012	148	3	on	on	ADP
fcis-23012	148	4	audio	audio	NOUN
fcis-23012	148	5	,	,	PUNCT
fcis-23012	148	6	speech	speech	NOUN
fcis-23012	148	7	,	,	PUNCT
fcis-23012	148	8	and	and	CCONJ
fcis-23012	148	9	music	music	NOUN
fcis-23012	148	10	processing	processing	NOUN
fcis-23012	148	11	,	,	PUNCT
fcis-23012	148	12	2023	2023	NUM
fcis-23012	148	13	,	,	PUNCT
fcis-23012	148	14	2023(1	2023(1	NUM
fcis-23012	148	15	):	):	PUNCT
fcis-23012	148	16	40	40	NUM
fcis-23012	148	17	.	.	PUNCT
fcis-23012	149	1	[	[	X
fcis-23012	149	2	15	15	NUM
fcis-23012	149	3	]	]	X
fcis-23012	149	4	su	su	PROPN
fcis-23012	149	5	b	b	PROPN
fcis-23012	149	6	h	h	PROPN
fcis-23012	149	7	,	,	PUNCT
fcis-23012	149	8	chang	chang	PROPN
fcis-23012	149	9	c	c	PROPN
fcis-23012	149	10	m	m	PROPN
fcis-23012	149	11	,	,	PUNCT
fcis-23012	149	12	lin	lin	PROPN
fcis-23012	149	13	y	y	PROPN
fcis-23012	149	14	s	s	PROPN
fcis-23012	149	15	,	,	PUNCT
fcis-23012	149	16	et	et	PROPN
fcis-23012	149	17	al	al	PROPN
fcis-23012	149	18	.	.	PUNCT
fcis-23012	149	19	improving	improve	VERB
fcis-23012	149	20	speech	speech	NOUN
fcis-23012	149	21	emotion	emotion	NOUN
fcis-23012	149	22	recognition	recognition	NOUN
fcis-23012	149	23	using	use	VERB
fcis-23012	149	24	graph	graph	NOUN
fcis-23012	149	25	attentive	attentive	ADJ
fcis-23012	149	26	bi	bi	ADJ
fcis-23012	149	27	-	-	ADJ
fcis-23012	149	28	directional	directional	ADJ
fcis-23012	149	29	gated	gate	VERB
fcis-23012	149	30	recurrent	recurrent	ADJ
fcis-23012	149	31	unit	unit	NOUN
fcis-23012	149	32	network[c]//interspeech	network[c]//interspeech	PROPN
fcis-23012	149	33	.	.	PUNCT
fcis-23012	150	1	2020	2020	NUM
fcis-23012	150	2	:	:	PUNCT
fcis-23012	150	3	506	506	NUM
fcis-23012	150	4	-	-	SYM
fcis-23012	150	5	510	510	NUM
fcis-23012	150	6	.	.	PUNCT