id	sid	tid	token	lemma	pos
fbem-10171	1	1	frontiers	frontier	NOUN
fbem-10171	1	2	in	in	ADP
fbem-10171	1	3	business	business	NOUN
fbem-10171	1	4	,	,	PUNCT
fbem-10171	1	5	economics	economic	NOUN
fbem-10171	1	6	and	and	CCONJ
fbem-10171	1	7	management	management	NOUN
fbem-10171	1	8	issn	issn	PROPN
fbem-10171	1	9	:	:	PUNCT
fbem-10171	1	10	2766	2766	NUM
fbem-10171	1	11	-	-	PUNCT
fbem-10171	1	12	824x	824x	NUM
fbem-10171	1	13	|	|	ADJ
fbem-10171	1	14	vol	vol	NOUN
fbem-10171	1	15	.	.	PROPN
fbem-10171	2	1	10	10	NUM
fbem-10171	2	2	,	,	PUNCT
fbem-10171	2	3	no	no	INTJ
fbem-10171	2	4	.	.	NOUN
fbem-10171	2	5	1	1	NUM
fbem-10171	2	6	,	,	PUNCT
fbem-10171	2	7	2023	2023	NUM
fbem-10171	2	8	58	58	NUM
fbem-10171	2	9	a	a	DET
fbem-10171	2	10	review	review	NOUN
fbem-10171	2	11	of	of	ADP
fbem-10171	2	12	text	text	NOUN
fbem-10171	2	13	sentiment	sentiment	NOUN
fbem-10171	2	14	analysis	analysis	NOUN
fbem-10171	2	15	methods	method	NOUN
fbem-10171	2	16	and	and	CCONJ
fbem-10171	2	17	applications	application	NOUN
fbem-10171	2	18	yuxin	yuxin	PROPN
fbem-10171	2	19	jin1	jin1	PROPN
fbem-10171	2	20	,	,	PUNCT
fbem-10171	2	21	2	2	NUM
fbem-10171	2	22	,	,	PUNCT
fbem-10171	2	23	kui	kui	PROPN
fbem-10171	2	24	cheng2	cheng2	PROPN
fbem-10171	2	25	and	and	CCONJ
fbem-10171	2	26	xinjie	xinjie	PROPN
fbem-10171	2	27	wang1	wang1	PROPN
fbem-10171	2	28	,	,	PUNCT
fbem-10171	2	29	2	2	NUM
fbem-10171	2	30	,	,	PUNCT
fbem-10171	2	31	lecai	lecai	PROPN
fbem-10171	2	32	cai2	cai2	PROPN
fbem-10171	2	33	,	,	PUNCT
fbem-10171	2	34	*	*	PUNCT
fbem-10171	2	35	1	1	NUM
fbem-10171	2	36	school	school	NOUN
fbem-10171	2	37	of	of	ADP
fbem-10171	2	38	automation	automation	NOUN
fbem-10171	2	39	and	and	CCONJ
fbem-10171	2	40	information	information	NOUN
fbem-10171	2	41	engineering	engineering	NOUN
fbem-10171	2	42	,	,	PUNCT
fbem-10171	2	43	sichuan	sichuan	PROPN
fbem-10171	2	44	university	university	PROPN
fbem-10171	2	45	of	of	ADP
fbem-10171	2	46	science	science	PROPN
fbem-10171	2	47	&	&	CCONJ
fbem-10171	2	48	engineering	engineering	PROPN
fbem-10171	2	49	,	,	PUNCT
fbem-10171	2	50	yibin	yibin	PROPN
fbem-10171	2	51	644000	644000	NUM
fbem-10171	2	52	,	,	PUNCT
fbem-10171	2	53	china	china	PROPN
fbem-10171	2	54	2	2	NUM
fbem-10171	2	55	sanjiang	sanjiang	NOUN
fbem-10171	2	56	artificial	artificial	ADJ
fbem-10171	2	57	intelligence	intelligence	NOUN
fbem-10171	2	58	and	and	CCONJ
fbem-10171	2	59	robot	robot	PROPN
fbem-10171	2	60	research	research	PROPN
fbem-10171	2	61	institute	institute	PROPN
fbem-10171	2	62	,	,	PUNCT
fbem-10171	2	63	yibin	yibin	PROPN
fbem-10171	2	64	644000	644000	NUM
fbem-10171	2	65	,	,	PUNCT
fbem-10171	2	66	china	china	PROPN
fbem-10171	2	67	*	*	PUNCT
fbem-10171	2	68	corresponding	correspond	VERB
fbem-10171	2	69	author	author	NOUN
fbem-10171	2	70	abstract	abstract	NOUN
fbem-10171	2	71	:	:	PUNCT
fbem-10171	2	72	this	this	DET
fbem-10171	2	73	paper	paper	NOUN
fbem-10171	2	74	reviews	review	VERB
fbem-10171	2	75	the	the	DET
fbem-10171	2	76	application	application	NOUN
fbem-10171	2	77	of	of	ADP
fbem-10171	2	78	natural	natural	ADJ
fbem-10171	2	79	language	language	NOUN
fbem-10171	2	80	processing	processing	NOUN
fbem-10171	2	81	in	in	ADP
fbem-10171	2	82	sentiment	sentiment	NOUN
fbem-10171	2	83	analysis	analysis	NOUN
fbem-10171	2	84	.	.	PUNCT
fbem-10171	3	1	sentiment	sentiment	NOUN
fbem-10171	3	2	analysis	analysis	NOUN
fbem-10171	3	3	is	be	AUX
fbem-10171	3	4	an	an	DET
fbem-10171	3	5	important	important	ADJ
fbem-10171	3	6	task	task	NOUN
fbem-10171	3	7	aimed	aim	VERB
fbem-10171	3	8	at	at	ADP
fbem-10171	3	9	automatically	automatically	ADV
fbem-10171	3	10	identifying	identify	VERB
fbem-10171	3	11	and	and	CCONJ
fbem-10171	3	12	inferring	infer	VERB
fbem-10171	3	13	sentiment	sentiment	NOUN
fbem-10171	3	14	tendencies	tendency	NOUN
fbem-10171	3	15	and	and	CCONJ
fbem-10171	3	16	sentiment	sentiment	NOUN
fbem-10171	3	17	intensity	intensity	NOUN
fbem-10171	3	18	in	in	ADP
fbem-10171	3	19	texts	text	NOUN
fbem-10171	3	20	.	.	PUNCT
fbem-10171	4	1	this	this	DET
fbem-10171	4	2	paper	paper	NOUN
fbem-10171	4	3	first	first	ADV
fbem-10171	4	4	introduces	introduce	VERB
fbem-10171	4	5	the	the	DET
fbem-10171	4	6	application	application	NOUN
fbem-10171	4	7	areas	area	NOUN
fbem-10171	4	8	of	of	ADP
fbem-10171	4	9	sentiment	sentiment	NOUN
fbem-10171	4	10	analysis	analysis	NOUN
fbem-10171	4	11	,	,	PUNCT
fbem-10171	4	12	including	include	VERB
fbem-10171	4	13	practical	practical	ADJ
fbem-10171	4	14	applications	application	NOUN
fbem-10171	4	15	of	of	ADP
fbem-10171	4	16	text	text	NOUN
fbem-10171	4	17	sentiment	sentiment	NOUN
fbem-10171	4	18	analysis	analysis	NOUN
fbem-10171	4	19	.	.	PUNCT
fbem-10171	5	1	then	then	ADV
fbem-10171	5	2	,	,	PUNCT
fbem-10171	5	3	text	text	NOUN
fbem-10171	5	4	pre	pre	ADJ
fbem-10171	5	5	-	-	ADJ
fbem-10171	5	6	processing	processing	ADJ
fbem-10171	5	7	techniques	technique	NOUN
fbem-10171	5	8	such	such	ADJ
fbem-10171	5	9	as	as	ADP
fbem-10171	5	10	word	word	NOUN
fbem-10171	5	11	separation	separation	NOUN
fbem-10171	5	12	,	,	PUNCT
fbem-10171	5	13	deactivation	deactivation	NOUN
fbem-10171	5	14	removal	removal	NOUN
fbem-10171	5	15	and	and	CCONJ
fbem-10171	5	16	punctuation	punctuation	NOUN
fbem-10171	5	17	processing	processing	NOUN
fbem-10171	5	18	are	be	AUX
fbem-10171	5	19	discussed	discuss	VERB
fbem-10171	5	20	.	.	PUNCT
fbem-10171	6	1	then	then	ADV
fbem-10171	6	2	,	,	PUNCT
fbem-10171	6	3	feature	feature	NOUN
fbem-10171	6	4	extraction	extraction	NOUN
fbem-10171	6	5	and	and	CCONJ
fbem-10171	6	6	representation	representation	NOUN
fbem-10171	6	7	methods	method	NOUN
fbem-10171	6	8	are	be	AUX
fbem-10171	6	9	explored	explore	VERB
fbem-10171	6	10	,	,	PUNCT
fbem-10171	6	11	including	include	VERB
fbem-10171	6	12	bag	bag	NOUN
fbem-10171	6	13	-	-	PUNCT
fbem-10171	6	14	of	of	ADP
fbem-10171	6	15	-	-	PUNCT
fbem-10171	6	16	words	word	NOUN
fbem-10171	6	17	model	model	NOUN
fbem-10171	6	18	,	,	PUNCT
fbem-10171	6	19	tf	tf	PROPN
fbem-10171	6	20	-	-	PUNCT
fbem-10171	6	21	idf	idf	NOUN
fbem-10171	6	22	,	,	PUNCT
fbem-10171	6	23	word	word	NOUN
fbem-10171	6	24	embedding	embed	VERB
fbem-10171	6	25	and	and	CCONJ
fbem-10171	6	26	word2vec	word2vec	ADJ
fbem-10171	6	27	,	,	PUNCT
fbem-10171	6	28	attention	attention	NOUN
fbem-10171	6	29	mechanism	mechanism	NOUN
fbem-10171	6	30	and	and	CCONJ
fbem-10171	6	31	transformer	transformer	NOUN
fbem-10171	6	32	.	.	PUNCT
fbem-10171	7	1	in	in	ADP
fbem-10171	7	2	addition	addition	NOUN
fbem-10171	7	3	,	,	PUNCT
fbem-10171	7	4	methods	method	NOUN
fbem-10171	7	5	for	for	ADP
fbem-10171	7	6	sentiment	sentiment	NOUN
fbem-10171	7	7	analysis	analysis	NOUN
fbem-10171	7	8	,	,	PUNCT
fbem-10171	7	9	such	such	ADJ
fbem-10171	7	10	as	as	ADP
fbem-10171	7	11	sentiment	sentiment	NOUN
fbem-10171	7	12	dictionaries	dictionary	NOUN
fbem-10171	7	13	and	and	CCONJ
fbem-10171	7	14	rule	rule	NOUN
fbem-10171	7	15	-	-	PUNCT
fbem-10171	7	16	based	base	VERB
fbem-10171	7	17	methods	method	NOUN
fbem-10171	7	18	,	,	PUNCT
fbem-10171	7	19	traditional	traditional	ADJ
fbem-10171	7	20	machine	machine	NOUN
fbem-10171	7	21	learning	learning	NOUN
fbem-10171	7	22	methods	method	NOUN
fbem-10171	7	23	,	,	PUNCT
fbem-10171	7	24	and	and	CCONJ
fbem-10171	7	25	deep	deep	ADJ
fbem-10171	7	26	learning	learning	NOUN
fbem-10171	7	27	-	-	PUNCT
fbem-10171	7	28	based	base	VERB
fbem-10171	7	29	methods	method	NOUN
fbem-10171	7	30	,	,	PUNCT
fbem-10171	7	31	are	be	AUX
fbem-10171	7	32	presented	present	VERB
fbem-10171	7	33	.	.	PUNCT
fbem-10171	8	1	finally	finally	ADV
fbem-10171	8	2	,	,	PUNCT
fbem-10171	8	3	the	the	DET
fbem-10171	8	4	application	application	NOUN
fbem-10171	8	5	areas	area	NOUN
fbem-10171	8	6	of	of	ADP
fbem-10171	8	7	sentiment	sentiment	NOUN
fbem-10171	8	8	analysis	analysis	NOUN
fbem-10171	8	9	are	be	AUX
fbem-10171	8	10	discussed	discuss	VERB
fbem-10171	8	11	and	and	CCONJ
fbem-10171	8	12	conclusions	conclusion	NOUN
fbem-10171	8	13	are	be	AUX
fbem-10171	8	14	given	give	VERB
fbem-10171	8	15	.	.	PUNCT
fbem-10171	9	1	the	the	DET
fbem-10171	9	2	review	review	NOUN
fbem-10171	9	3	in	in	ADP
fbem-10171	9	4	this	this	DET
fbem-10171	9	5	paper	paper	NOUN
fbem-10171	9	6	will	will	AUX
fbem-10171	9	7	help	help	VERB
fbem-10171	9	8	readers	reader	NOUN
fbem-10171	9	9	understand	understand	VERB
fbem-10171	9	10	the	the	DET
fbem-10171	9	11	current	current	ADJ
fbem-10171	9	12	status	status	NOUN
fbem-10171	9	13	and	and	CCONJ
fbem-10171	9	14	development	development	NOUN
fbem-10171	9	15	trend	trend	NOUN
fbem-10171	9	16	of	of	ADP
fbem-10171	9	17	natural	natural	ADJ
fbem-10171	9	18	language	language	NOUN
fbem-10171	9	19	processing	processing	NOUN
fbem-10171	9	20	applications	application	NOUN
fbem-10171	9	21	in	in	ADP
fbem-10171	9	22	sentiment	sentiment	NOUN
fbem-10171	9	23	analysis	analysis	NOUN
fbem-10171	9	24	,	,	PUNCT
fbem-10171	9	25	as	as	ADV
fbem-10171	9	26	well	well	ADV
fbem-10171	9	27	as	as	ADP
fbem-10171	9	28	the	the	DET
fbem-10171	9	29	advantages	advantage	NOUN
fbem-10171	9	30	and	and	CCONJ
fbem-10171	9	31	disadvantages	disadvantage	NOUN
fbem-10171	9	32	of	of	ADP
fbem-10171	9	33	different	different	ADJ
fbem-10171	9	34	methods	method	NOUN
fbem-10171	9	35	in	in	ADP
fbem-10171	9	36	sentiment	sentiment	NOUN
fbem-10171	9	37	analysis	analysis	NOUN
fbem-10171	9	38	.	.	PUNCT
fbem-10171	10	1	keywords	keyword	NOUN
fbem-10171	10	2	:	:	PUNCT
fbem-10171	10	3	natural	natural	ADJ
fbem-10171	10	4	language	language	NOUN
fbem-10171	10	5	processing	processing	NOUN
fbem-10171	10	6	,	,	PUNCT
fbem-10171	10	7	sentiment	sentiment	NOUN
fbem-10171	10	8	analysis	analysis	NOUN
fbem-10171	10	9	,	,	PUNCT
fbem-10171	10	10	machine	machine	NOUN
fbem-10171	10	11	learning	learning	NOUN
fbem-10171	10	12	,	,	PUNCT
fbem-10171	10	13	deep	deep	ADJ
fbem-10171	10	14	learning	learning	NOUN
fbem-10171	10	15	.	.	PUNCT
fbem-10171	11	1	1	1	X
fbem-10171	11	2	.	.	X
fbem-10171	11	3	introduction	introduction	NOUN
fbem-10171	11	4	natural	natural	ADJ
fbem-10171	11	5	language	language	NOUN
fbem-10171	11	6	processing	processing	NOUN
fbem-10171	11	7	(	(	PUNCT
fbem-10171	11	8	nlp	nlp	NOUN
fbem-10171	11	9	)	)	PUNCT
fbem-10171	11	10	is	be	AUX
fbem-10171	11	11	an	an	DET
fbem-10171	11	12	important	important	ADJ
fbem-10171	11	13	research	research	NOUN
fbem-10171	11	14	direction	direction	NOUN
fbem-10171	11	15	in	in	ADP
fbem-10171	11	16	computer	computer	NOUN
fbem-10171	11	17	science	science	NOUN
fbem-10171	11	18	and	and	CCONJ
fbem-10171	11	19	artificial	artificial	ADJ
fbem-10171	11	20	intelligence	intelligence	NOUN
fbem-10171	11	21	,	,	PUNCT
fbem-10171	11	22	aiming	aim	VERB
fbem-10171	11	23	to	to	PART
fbem-10171	11	24	enable	enable	VERB
fbem-10171	11	25	computers	computer	NOUN
fbem-10171	11	26	to	to	PART
fbem-10171	11	27	understand	understand	VERB
fbem-10171	11	28	,	,	PUNCT
fbem-10171	11	29	process	process	NOUN
fbem-10171	11	30	and	and	CCONJ
fbem-10171	11	31	generate	generate	VERB
fbem-10171	11	32	human	human	ADJ
fbem-10171	11	33	language	language	NOUN
fbem-10171	11	34	.	.	PUNCT
fbem-10171	12	1	text	text	NOUN
fbem-10171	12	2	sentiment	sentiment	NOUN
fbem-10171	12	3	analysis	analysis	NOUN
fbem-10171	12	4	is	be	AUX
fbem-10171	12	5	an	an	DET
fbem-10171	12	6	important	important	ADJ
fbem-10171	12	7	task	task	NOUN
fbem-10171	12	8	in	in	ADP
fbem-10171	12	9	nlp	nlp	NOUN
fbem-10171	12	10	,	,	PUNCT
fbem-10171	12	11	which	which	PRON
fbem-10171	12	12	automatically	automatically	ADV
fbem-10171	12	13	identifies	identify	VERB
fbem-10171	12	14	and	and	CCONJ
fbem-10171	12	15	inferred	infer	VERB
fbem-10171	12	16	the	the	DET
fbem-10171	12	17	sentiment	sentiment	NOUN
fbem-10171	12	18	tendency	tendency	NOUN
fbem-10171	12	19	and	and	CCONJ
fbem-10171	12	20	intensity	intensity	NOUN
fbem-10171	12	21	in	in	ADP
fbem-10171	12	22	text	text	NOUN
fbem-10171	12	23	by	by	ADP
fbem-10171	12	24	extracting	extract	VERB
fbem-10171	12	25	information	information	NOUN
fbem-10171	12	26	from	from	ADP
fbem-10171	12	27	user	user	NOUN
fbem-10171	12	28	opinions	opinion	NOUN
fbem-10171	12	29	to	to	PART
fbem-10171	12	30	obtain	obtain	VERB
fbem-10171	12	31	key	key	ADJ
fbem-10171	12	32	sentiment	sentiment	NOUN
fbem-10171	12	33	information[1	information[1	PROPN
fbem-10171	12	34	]	]	X
fbem-10171	12	35	.	.	PUNCT
fbem-10171	13	1	with	with	ADP
fbem-10171	13	2	the	the	DET
fbem-10171	13	3	increase	increase	NOUN
fbem-10171	13	4	of	of	ADP
fbem-10171	13	5	text	text	NOUN
fbem-10171	13	6	data	datum	NOUN
fbem-10171	13	7	such	such	ADJ
fbem-10171	13	8	as	as	ADP
fbem-10171	13	9	social	social	ADJ
fbem-10171	13	10	media	medium	NOUN
fbem-10171	13	11	and	and	CCONJ
fbem-10171	13	12	online	online	ADJ
fbem-10171	13	13	comments	comment	NOUN
fbem-10171	13	14	,	,	PUNCT
fbem-10171	13	15	sentiment	sentiment	NOUN
fbem-10171	13	16	analysis	analysis	NOUN
fbem-10171	13	17	is	be	AUX
fbem-10171	13	18	crucial	crucial	ADJ
fbem-10171	13	19	to	to	PART
fbem-10171	13	20	understand	understand	VERB
fbem-10171	13	21	user	user	NOUN
fbem-10171	13	22	needs	need	NOUN
fbem-10171	13	23	,	,	PUNCT
fbem-10171	13	24	market	market	NOUN
fbem-10171	13	25	dynamics	dynamic	NOUN
fbem-10171	13	26	and	and	CCONJ
fbem-10171	13	27	changes	change	NOUN
fbem-10171	13	28	in	in	ADP
fbem-10171	13	29	public	public	ADJ
fbem-10171	13	30	opinion[2	opinion[2	PROPN
fbem-10171	13	31	]	]	PUNCT
fbem-10171	13	32	.	.	PUNCT
fbem-10171	14	1	2	2	X
fbem-10171	14	2	.	.	X
fbem-10171	14	3	applications	application	NOUN
fbem-10171	14	4	in	in	ADP
fbem-10171	14	5	text	text	NOUN
fbem-10171	14	6	sentiment	sentiment	NOUN
fbem-10171	14	7	analysis	analysis	NOUN
fbem-10171	14	8	2.1	2.1	NUM
fbem-10171	14	9	.	.	PUNCT
fbem-10171	15	1	text	text	NOUN
fbem-10171	15	2	pre	pre	ADJ
fbem-10171	15	3	-	-	ADJ
fbem-10171	15	4	processing	processing	ADJ
fbem-10171	15	5	technology	technology	NOUN
fbem-10171	15	6	before	before	ADP
fbem-10171	15	7	text	text	NOUN
fbem-10171	15	8	sentiment	sentiment	NOUN
fbem-10171	15	9	analysis	analysis	NOUN
fbem-10171	15	10	,	,	PUNCT
fbem-10171	15	11	text	text	NOUN
fbem-10171	15	12	pre	pre	ADJ
fbem-10171	15	13	-	-	ADJ
fbem-10171	15	14	processing	processing	NOUN
fbem-10171	15	15	is	be	AUX
fbem-10171	15	16	a	a	DET
fbem-10171	15	17	key	key	ADJ
fbem-10171	15	18	step	step	NOUN
fbem-10171	15	19	that	that	PRON
fbem-10171	15	20	involves	involve	VERB
fbem-10171	15	21	cleaning	clean	VERB
fbem-10171	15	22	,	,	PUNCT
fbem-10171	15	23	normalizing	normalize	VERB
fbem-10171	15	24	and	and	CCONJ
fbem-10171	15	25	transforming	transform	VERB
fbem-10171	15	26	the	the	DET
fbem-10171	15	27	raw	raw	ADJ
fbem-10171	15	28	text	text	NOUN
fbem-10171	15	29	data	datum	NOUN
fbem-10171	15	30	for	for	ADP
fbem-10171	15	31	subsequent	subsequent	ADJ
fbem-10171	15	32	feature	feature	NOUN
fbem-10171	15	33	extraction	extraction	NOUN
fbem-10171	15	34	and	and	CCONJ
fbem-10171	15	35	analysis[3	analysis[3	NUM
fbem-10171	15	36	]	]	PUNCT
fbem-10171	15	37	.	.	PUNCT
fbem-10171	16	1	the	the	DET
fbem-10171	16	2	specific	specific	ADJ
fbem-10171	16	3	text	text	NOUN
fbem-10171	16	4	preprocessing	preprocessing	NOUN
fbem-10171	16	5	process	process	NOUN
fbem-10171	16	6	is	be	AUX
fbem-10171	16	7	shown	show	VERB
fbem-10171	16	8	in	in	ADP
fbem-10171	16	9	figure	figure	NOUN
fbem-10171	16	10	1	1	NUM
fbem-10171	16	11	.	.	PUNCT
fbem-10171	17	1	figure	figure	NOUN
fbem-10171	17	2	1	1	NUM
fbem-10171	17	3	.	.	PUNCT
fbem-10171	18	1	data	datum	NOUN
fbem-10171	18	2	pre	pre	ADJ
fbem-10171	18	3	-	-	ADJ
fbem-10171	18	4	processing	processing	ADJ
fbem-10171	18	5	flow	flow	NOUN
fbem-10171	18	6	chart	chart	NOUN
fbem-10171	18	7	2.1.1	2.1.1	NUM
fbem-10171	18	8	.	.	PUNCT
fbem-10171	19	1	participle	participle	NOUN
fbem-10171	19	2	and	and	CCONJ
fbem-10171	19	3	word	word	NOUN
fbem-10171	19	4	marking	mark	VERB
fbem-10171	19	5	phrase	phrase	NOUN
fbem-10171	19	6	splitting	splitting	NOUN
fbem-10171	19	7	slices	slice	VERB
fbem-10171	19	8	a	a	DET
fbem-10171	19	9	text	text	NOUN
fbem-10171	19	10	sequence	sequence	NOUN
fbem-10171	19	11	into	into	ADP
fbem-10171	19	12	words	word	NOUN
fbem-10171	19	13	to	to	PART
fbem-10171	19	14	understand	understand	VERB
fbem-10171	19	15	the	the	DET
fbem-10171	19	16	sentiment	sentiment	NOUN
fbem-10171	19	17	tendency	tendency	NOUN
fbem-10171	19	18	of	of	ADP
fbem-10171	19	19	each	each	DET
fbem-10171	19	20	word[4	word[4	NOUN
fbem-10171	19	21	]	]	PUNCT
fbem-10171	19	22	.	.	PUNCT
fbem-10171	20	1	commonly	commonly	ADV
fbem-10171	20	2	used	use	VERB
fbem-10171	20	3	word	word	NOUN
fbem-10171	20	4	separation	separation	NOUN
fbem-10171	20	5	algorithms	algorithm	NOUN
fbem-10171	20	6	include	include	VERB
fbem-10171	20	7	lexical	lexical	ADJ
fbem-10171	20	8	,	,	PUNCT
fbem-10171	20	9	rule	rule	NOUN
fbem-10171	20	10	-	-	PUNCT
fbem-10171	20	11	based	base	VERB
fbem-10171	20	12	and	and	CCONJ
fbem-10171	20	13	statistical	statistical	ADJ
fbem-10171	20	14	approaches[5	approaches[5	PROPN
fbem-10171	20	15	]	]	PUNCT
fbem-10171	20	16	.	.	PUNCT
fbem-10171	21	1	chinese	chinese	ADJ
fbem-10171	21	2	word	word	NOUN
fbem-10171	21	3	separation	separation	NOUN
fbem-10171	21	4	is	be	AUX
fbem-10171	21	5	an	an	DET
fbem-10171	21	6	important	important	ADJ
fbem-10171	21	7	pre	pre	ADJ
fbem-10171	21	8	-	-	ADJ
fbem-10171	21	9	processing	processing	ADJ
fbem-10171	21	10	step	step	NOUN
fbem-10171	21	11	,	,	PUNCT
fbem-10171	21	12	and	and	CCONJ
fbem-10171	21	13	commonly	commonly	ADV
fbem-10171	21	14	used	use	VERB
fbem-10171	21	15	tools	tool	NOUN
fbem-10171	21	16	include	include	VERB
fbem-10171	21	17	jieba[6	jieba[6	ADV
fbem-10171	21	18	]	]	PUNCT
fbem-10171	21	19	and	and	CCONJ
fbem-10171	21	20	thulac[7	thulac[7	PROPN
fbem-10171	21	21	]	]	PUNCT
fbem-10171	21	22	.	.	PUNCT
fbem-10171	22	1	english	english	PROPN
fbem-10171	22	2	can	can	AUX
fbem-10171	22	3	be	be	AUX
fbem-10171	22	4	simply	simply	ADV
fbem-10171	22	5	separated	separate	VERB
fbem-10171	22	6	by	by	ADP
fbem-10171	22	7	spaces	space	NOUN
fbem-10171	22	8	.	.	PUNCT
fbem-10171	23	1	lexical	lexical	ADJ
fbem-10171	23	2	annotation	annotation	NOUN
fbem-10171	23	3	determines	determine	VERB
fbem-10171	23	4	lexical	lexical	ADJ
fbem-10171	23	5	properties	property	NOUN
fbem-10171	23	6	for	for	ADP
fbem-10171	23	7	words	word	NOUN
fbem-10171	23	8	,	,	PUNCT
fbem-10171	23	9	such	such	ADJ
fbem-10171	23	10	as	as	ADP
fbem-10171	23	11	nouns	noun	NOUN
fbem-10171	23	12	,	,	PUNCT
fbem-10171	23	13	verbs	verbs	PROPN
fbem-10171	23	14	,	,	PUNCT
fbem-10171	23	15	adjectives	adjective	NOUN
fbem-10171	23	16	,	,	PUNCT
fbem-10171	23	17	etc	etc	X
fbem-10171	23	18	.	.	X
fbem-10171	23	19	lexical	lexical	ADJ
fbem-10171	23	20	annotation	annotation	NOUN
fbem-10171	23	21	provides	provide	VERB
fbem-10171	23	22	contextual	contextual	ADJ
fbem-10171	23	23	information	information	NOUN
fbem-10171	23	24	and	and	CCONJ
fbem-10171	23	25	helps	help	VERB
fbem-10171	23	26	to	to	PART
fbem-10171	23	27	identify	identify	VERB
fbem-10171	23	28	emotionally	emotionally	ADV
fbem-10171	23	29	colored	color	VERB
fbem-10171	23	30	words	word	NOUN
fbem-10171	23	31	.	.	PUNCT
fbem-10171	24	1	commonly	commonly	ADV
fbem-10171	24	2	used	use	VERB
fbem-10171	24	3	tools	tool	NOUN
fbem-10171	24	4	are	be	AUX
fbem-10171	24	5	hidden	hide	VERB
fbem-10171	24	6	markov	markov	NOUN
fbem-10171	24	7	model	model	NOUN
fbem-10171	24	8	,	,	PUNCT
fbem-10171	24	9	maximum	maximum	PROPN
fbem-10171	24	10	entropy	entropy	NOUN
fbem-10171	24	11	model	model	NOUN
fbem-10171	24	12	and	and	CCONJ
fbem-10171	24	13	conditional	conditional	ADJ
fbem-10171	24	14	random	random	ADJ
fbem-10171	24	15	field[8][9][10][11][12][13][14	field[8][9][10][11][12][13][14	NOUN
fbem-10171	24	16	]	]	PUNCT
fbem-10171	24	17	.	.	PUNCT
fbem-10171	25	1	2.1.2	2.1.2	X
fbem-10171	25	2	.	.	PUNCT
fbem-10171	25	3	deactivation	deactivation	NOUN
fbem-10171	25	4	and	and	CCONJ
fbem-10171	25	5	punctuation	punctuation	NOUN
fbem-10171	25	6	handling	handling	NOUN
fbem-10171	25	7	deactivation	deactivation	NOUN
fbem-10171	25	8	and	and	CCONJ
fbem-10171	25	9	punctuation	punctuation	NOUN
fbem-10171	25	10	processing	processing	NOUN
fbem-10171	25	11	are	be	AUX
fbem-10171	25	12	common	common	ADJ
fbem-10171	25	13	steps	step	NOUN
fbem-10171	25	14	in	in	ADP
fbem-10171	25	15	text	text	NOUN
fbem-10171	25	16	pre	pre	ADJ
fbem-10171	25	17	-	-	NOUN
fbem-10171	25	18	processing	processing	NOUN
fbem-10171	25	19	,	,	PUNCT
fbem-10171	25	20	which	which	PRON
fbem-10171	25	21	aim	aim	VERB
fbem-10171	25	22	to	to	PART
fbem-10171	25	23	cleanse	cleanse	VERB
fbem-10171	25	24	and	and	CCONJ
fbem-10171	25	25	purify	purify	VERB
fbem-10171	25	26	text	text	NOUN
fbem-10171	25	27	data	datum	NOUN
fbem-10171	25	28	by	by	ADP
fbem-10171	25	29	removing	remove	VERB
fbem-10171	25	30	words	word	NOUN
fbem-10171	25	31	and	and	CCONJ
fbem-10171	25	32	symbols	symbol	NOUN
fbem-10171	25	33	that	that	PRON
fbem-10171	25	34	are	be	AUX
fbem-10171	25	35	irrelevant	irrelevant	ADJ
fbem-10171	25	36	or	or	CCONJ
fbem-10171	25	37	distracting	distract	VERB
fbem-10171	25	38	to	to	ADP
fbem-10171	25	39	sentiment	sentiment	NOUN
fbem-10171	25	40	analysis[15	analysis[15	NOUN
fbem-10171	25	41	]	]	PUNCT
fbem-10171	25	42	.	.	PUNCT
fbem-10171	26	1	deactivated	deactivate	VERB
fbem-10171	26	2	words	word	NOUN
fbem-10171	26	3	are	be	AUX
fbem-10171	26	4	words	word	NOUN
fbem-10171	26	5	that	that	PRON
fbem-10171	26	6	appear	appear	VERB
fbem-10171	26	7	frequently	frequently	ADV
fbem-10171	26	8	in	in	ADP
fbem-10171	26	9	the	the	DET
fbem-10171	26	10	text	text	NOUN
fbem-10171	26	11	but	but	CCONJ
fbem-10171	26	12	have	have	VERB
fbem-10171	26	13	no	no	DET
fbem-10171	26	14	real	real	ADJ
fbem-10171	26	15	meaning	meaning	NOUN
fbem-10171	26	16	for	for	ADP
fbem-10171	26	17	the	the	DET
fbem-10171	26	18	task	task	NOUN
fbem-10171	26	19	of	of	ADP
fbem-10171	26	20	sentiment	sentiment	NOUN
fbem-10171	26	21	analysis	analysis	NOUN
fbem-10171	26	22	,	,	PUNCT
fbem-10171	26	23	such	such	ADJ
fbem-10171	26	24	as	as	ADP
fbem-10171	26	25	"	"	PUNCT
fbem-10171	26	26	of	of	ADP
fbem-10171	26	27	"	"	PUNCT
fbem-10171	26	28	,	,	PUNCT
fbem-10171	26	29	"	"	PUNCT
fbem-10171	26	30	had	have	AUX
fbem-10171	26	31	"	"	PUNCT
fbem-10171	26	32	,	,	PUNCT
fbem-10171	26	33	"	"	PUNCT
fbem-10171	26	34	is	be	AUX
fbem-10171	26	35	"	"	PUNCT
fbem-10171	26	36	,	,	PUNCT
fbem-10171	26	37	and	and	CCONJ
fbem-10171	26	38	common	common	ADJ
fbem-10171	26	39	pronouns	pronoun	NOUN
fbem-10171	26	40	,	,	PUNCT
fbem-10171	26	41	prepositions	preposition	NOUN
fbem-10171	26	42	,	,	PUNCT
fbem-10171	26	43	conjunctions	conjunction	NOUN
fbem-10171	26	44	,	,	PUNCT
fbem-10171	26	45	etc	etc	X
fbem-10171	26	46	.	.	X
fbem-10171	27	1	the	the	DET
fbem-10171	27	2	presence	presence	NOUN
fbem-10171	27	3	of	of	ADP
fbem-10171	27	4	deactivated	deactivate	VERB
fbem-10171	27	5	words	word	NOUN
fbem-10171	27	6	adds	add	VERB
fbem-10171	27	7	to	to	ADP
fbem-10171	27	8	the	the	DET
fbem-10171	27	9	text	text	NOUN
fbem-10171	27	10	.	.	PUNCT
fbem-10171	28	1	the	the	DET
fbem-10171	28	2	presence	presence	NOUN
fbem-10171	28	3	of	of	ADP
fbem-10171	28	4	deactivated	deactivate	VERB
fbem-10171	28	5	words	word	NOUN
fbem-10171	28	6	increases	increase	VERB
fbem-10171	28	7	the	the	DET
fbem-10171	28	8	complexity	complexity	NOUN
fbem-10171	28	9	of	of	ADP
fbem-10171	28	10	text	text	NOUN
fbem-10171	28	11	processing	processing	NOUN
fbem-10171	28	12	and	and	CCONJ
fbem-10171	28	13	may	may	AUX
fbem-10171	28	14	interfere	interfere	VERB
fbem-10171	28	15	with	with	ADP
fbem-10171	28	16	the	the	DET
fbem-10171	28	17	results	result	NOUN
fbem-10171	28	18	of	of	ADP
fbem-10171	28	19	sentiment	sentiment	NOUN
fbem-10171	28	20	analysis	analysis	NOUN
fbem-10171	28	21	.	.	PUNCT
fbem-10171	29	1	in	in	ADP
fbem-10171	29	2	sentiment	sentiment	NOUN
fbem-10171	29	3	analysis	analysis	NOUN
fbem-10171	29	4	,	,	PUNCT
fbem-10171	29	5	removing	remove	VERB
fbem-10171	29	6	deactivators	deactivator	NOUN
fbem-10171	29	7	can	can	AUX
fbem-10171	29	8	reduce	reduce	VERB
fbem-10171	29	9	noise	noise	NOUN
fbem-10171	29	10	and	and	CCONJ
fbem-10171	29	11	redundant	redundant	ADJ
fbem-10171	29	12	information	information	NOUN
fbem-10171	29	13	,	,	PUNCT
fbem-10171	29	14	and	and	CCONJ
fbem-10171	29	15	improve	improve	VERB
fbem-10171	29	16	the	the	DET
fbem-10171	29	17	performance	performance	NOUN
fbem-10171	29	18	and	and	CCONJ
fbem-10171	29	19	effectiveness	effectiveness	NOUN
fbem-10171	29	20	of	of	ADP
fbem-10171	29	21	sentiment	sentiment	NOUN
fbem-10171	29	22	analysis	analysis	NOUN
fbem-10171	29	23	models[16	models[16	PROPN
fbem-10171	29	24	]	]	PUNCT
fbem-10171	29	25	.	.	PUNCT
fbem-10171	30	1	the	the	DET
fbem-10171	30	2	list	list	NOUN
fbem-10171	30	3	of	of	ADP
fbem-10171	30	4	commonly	commonly	ADV
fbem-10171	30	5	used	use	VERB
fbem-10171	30	6	deactivated	deactivate	VERB
fbem-10171	30	7	words	word	NOUN
fbem-10171	30	8	can	can	AUX
fbem-10171	30	9	be	be	AUX
fbem-10171	30	10	obtained	obtain	VERB
fbem-10171	30	11	from	from	ADP
fbem-10171	30	12	open	open	ADJ
fbem-10171	30	13	source	source	NOUN
fbem-10171	30	14	web	web	NOUN
fbem-10171	30	15	libraries	library	NOUN
fbem-10171	30	16	or	or	CCONJ
fbem-10171	30	17	custom	custom	NOUN
fbem-10171	30	18	vocabularies	vocabulary	NOUN
fbem-10171	30	19	,	,	PUNCT
fbem-10171	30	20	and	and	CCONJ
fbem-10171	30	21	the	the	DET
fbem-10171	30	22	removal	removal	NOUN
fbem-10171	30	23	of	of	ADP
fbem-10171	30	24	deactivated	deactivate	VERB
fbem-10171	30	25	words	word	NOUN
fbem-10171	30	26	can	can	AUX
fbem-10171	30	27	be	be	AUX
fbem-10171	30	28	done	do	VERB
fbem-10171	30	29	by	by	ADP
fbem-10171	30	30	simple	simple	ADJ
fbem-10171	30	31	rule	rule	NOUN
fbem-10171	30	32	-	-	PUNCT
fbem-10171	30	33	based	base	VERB
fbem-10171	30	34	methods	method	NOUN
fbem-10171	30	35	or	or	CCONJ
fbem-10171	30	36	based	base	VERB
fbem-10171	30	37	on	on	ADP
fbem-10171	30	38	a	a	DET
fbem-10171	30	39	list	list	NOUN
fbem-10171	30	40	of	of	ADP
fbem-10171	30	41	deactivated	deactivate	VERB
fbem-10171	30	42	words[17	words[17	VERB
fbem-10171	30	43	]	]	PUNCT
fbem-10171	30	44	.	.	PUNCT
fbem-10171	31	1	in	in	ADP
fbem-10171	31	2	addition	addition	NOUN
fbem-10171	31	3	,	,	PUNCT
fbem-10171	31	4	various	various	ADJ
fbem-10171	31	5	dictionaries	dictionary	NOUN
fbem-10171	31	6	exist	exist	VERB
fbem-10171	31	7	in	in	ADP
fbem-10171	31	8	different	different	ADJ
fbem-10171	31	9	domains	domain	NOUN
fbem-10171	31	10	,	,	PUNCT
fbem-10171	31	11	such	such	ADJ
fbem-10171	31	12	as	as	ADP
fbem-10171	31	13	the	the	DET
fbem-10171	31	14	u.s	u.s	PROPN
fbem-10171	31	15	.	.	PROPN
fbem-10171	31	16	army	army	PROPN
fbem-10171	31	17	military	military	PROPN
fbem-10171	31	18	dictionary[18	dictionary[18	PROPN
fbem-10171	31	19	]	]	PUNCT
fbem-10171	31	20	,	,	PUNCT
fbem-10171	31	21	the	the	DET
fbem-10171	31	22	attitude	attitude	NOUN
fbem-10171	31	23	dictionary	dictionary	PROPN
fbem-10171	31	24	,	,	PUNCT
fbem-10171	31	25	the	the	DET
fbem-10171	31	26	extended	extended	ADJ
fbem-10171	31	27	sentiment	sentiment	NOUN
fbem-10171	31	28	dictionary	dictionary	NOUN
fbem-10171	31	29	,	,	PUNCT
fbem-10171	31	30	and	and	CCONJ
fbem-10171	31	31	the	the	DET
fbem-10171	31	32	sentiment	sentiment	NOUN
fbem-10171	31	33	dictionary	dictionary	NOUN
fbem-10171	31	34	for	for	ADP
fbem-10171	31	35	the	the	DET
fbem-10171	31	36	book	book	NOUN
fbem-10171	31	37	domain	domain	NOUN
fbem-10171	31	38	based	base	VERB
fbem-10171	31	39	on	on	ADP
fbem-10171	31	40	ultra	ultra	ADJ
fbem-10171	31	41	-	-	ADJ
fbem-10171	31	42	short	short	ADJ
fbem-10171	31	43	reviews	review	NOUN
fbem-10171	31	44	,	,	PUNCT
fbem-10171	31	45	developed	develop	VERB
fbem-10171	31	46	based	base	VERB
fbem-10171	31	47	on	on	ADP
fbem-10171	31	48	terminological	terminological	ADJ
fbem-10171	31	49	theory[19	theory[19	NOUN
fbem-10171	31	50	]	]	PUNCT
fbem-10171	31	51	.	.	PUNCT
fbem-10171	32	1	punctuation	punctuation	NOUN
fbem-10171	32	2	processing	processing	NOUN
fbem-10171	32	3	refers	refer	VERB
fbem-10171	32	4	to	to	ADP
fbem-10171	32	5	the	the	DET
fbem-10171	32	6	removal	removal	NOUN
fbem-10171	32	7	of	of	ADP
fbem-10171	32	8	59	59	NUM
fbem-10171	32	9	punctuation	punctuation	NOUN
fbem-10171	32	10	marks	mark	NOUN
fbem-10171	32	11	from	from	ADP
fbem-10171	32	12	text	text	NOUN
fbem-10171	32	13	or	or	CCONJ
fbem-10171	32	14	their	their	PRON
fbem-10171	32	15	processing[20	processing[20	NOUN
fbem-10171	32	16	]	]	PUNCT
fbem-10171	32	17	.	.	PUNCT
fbem-10171	33	1	punctuation	punctuation	NOUN
fbem-10171	33	2	usually	usually	ADV
fbem-10171	33	3	does	do	AUX
fbem-10171	33	4	not	not	PART
fbem-10171	33	5	carry	carry	VERB
fbem-10171	33	6	sentiment	sentiment	NOUN
fbem-10171	33	7	information	information	NOUN
fbem-10171	33	8	in	in	ADP
fbem-10171	33	9	sentiment	sentiment	NOUN
fbem-10171	33	10	analysis	analysis	NOUN
fbem-10171	33	11	,	,	PUNCT
fbem-10171	33	12	and	and	CCONJ
fbem-10171	33	13	it	it	PRON
fbem-10171	33	14	adds	add	VERB
fbem-10171	33	15	complexity	complexity	NOUN
fbem-10171	33	16	to	to	ADP
fbem-10171	33	17	the	the	DET
fbem-10171	33	18	text	text	NOUN
fbem-10171	33	19	,	,	PUNCT
fbem-10171	33	20	which	which	PRON
fbem-10171	33	21	may	may	AUX
fbem-10171	33	22	interfere	interfere	VERB
fbem-10171	33	23	with	with	ADP
fbem-10171	33	24	the	the	DET
fbem-10171	33	25	process	process	NOUN
fbem-10171	33	26	of	of	ADP
fbem-10171	33	27	feature	feature	NOUN
fbem-10171	33	28	extraction	extraction	NOUN
fbem-10171	33	29	and	and	CCONJ
fbem-10171	33	30	sentiment	sentiment	NOUN
fbem-10171	33	31	classification	classification	NOUN
fbem-10171	33	32	.	.	PUNCT
fbem-10171	34	1	therefore	therefore	ADV
fbem-10171	34	2	,	,	PUNCT
fbem-10171	34	3	removing	remove	VERB
fbem-10171	34	4	punctuation	punctuation	NOUN
fbem-10171	34	5	marks	mark	NOUN
fbem-10171	34	6	can	can	AUX
fbem-10171	34	7	simplify	simplify	VERB
fbem-10171	34	8	the	the	DET
fbem-10171	34	9	structure	structure	NOUN
fbem-10171	34	10	of	of	ADP
fbem-10171	34	11	text	text	NOUN
fbem-10171	34	12	and	and	CCONJ
fbem-10171	34	13	improve	improve	VERB
fbem-10171	34	14	the	the	DET
fbem-10171	34	15	accuracy	accuracy	NOUN
fbem-10171	34	16	of	of	ADP
fbem-10171	34	17	sentiment	sentiment	NOUN
fbem-10171	34	18	analysis	analysis	NOUN
fbem-10171	34	19	.	.	PUNCT
fbem-10171	35	1	common	common	ADJ
fbem-10171	35	2	punctuation	punctuation	NOUN
fbem-10171	35	3	marks	mark	NOUN
fbem-10171	35	4	include	include	VERB
fbem-10171	35	5	periods	period	NOUN
fbem-10171	35	6	,	,	PUNCT
fbem-10171	35	7	commas	comma	NOUN
fbem-10171	35	8	,	,	PUNCT
fbem-10171	35	9	exclamation	exclamation	NOUN
fbem-10171	35	10	marks	mark	NOUN
fbem-10171	35	11	,	,	PUNCT
fbem-10171	35	12	question	question	NOUN
fbem-10171	35	13	marks	mark	NOUN
fbem-10171	35	14	,	,	PUNCT
fbem-10171	35	15	etc	etc	X
fbem-10171	35	16	.	.	X
fbem-10171	36	1	2.1.3	2.1.3	NUM
fbem-10171	36	2	.	.	PUNCT
fbem-10171	36	3	morphological	morphological	ADJ
fbem-10171	36	4	reduction	reduction	NOUN
fbem-10171	36	5	and	and	CCONJ
fbem-10171	36	6	stemming	stem	VERB
fbem-10171	36	7	word	word	NOUN
fbem-10171	36	8	form	form	NOUN
fbem-10171	36	9	reduction	reduction	NOUN
fbem-10171	36	10	and	and	CCONJ
fbem-10171	36	11	stem	stem	NOUN
fbem-10171	36	12	extraction	extraction	NOUN
fbem-10171	36	13	in	in	ADP
fbem-10171	36	14	sentiment	sentiment	NOUN
fbem-10171	36	15	analysis	analysis	NOUN
fbem-10171	36	16	serve	serve	VERB
fbem-10171	36	17	to	to	PART
fbem-10171	36	18	reduce	reduce	VERB
fbem-10171	36	19	words	word	NOUN
fbem-10171	36	20	to	to	ADP
fbem-10171	36	21	their	their	PRON
fbem-10171	36	22	original	original	ADJ
fbem-10171	36	23	forms	form	NOUN
fbem-10171	36	24	or	or	CCONJ
fbem-10171	36	25	to	to	PART
fbem-10171	36	26	extract	extract	VERB
fbem-10171	36	27	their	their	PRON
fbem-10171	36	28	stems	stem	NOUN
fbem-10171	36	29	in	in	ADP
fbem-10171	36	30	order	order	NOUN
fbem-10171	36	31	to	to	PART
fbem-10171	36	32	reduce	reduce	VERB
fbem-10171	36	33	variations	variation	NOUN
fbem-10171	36	34	in	in	ADP
fbem-10171	36	35	the	the	DET
fbem-10171	36	36	surface	surface	NOUN
fbem-10171	36	37	forms	form	NOUN
fbem-10171	36	38	of	of	ADP
fbem-10171	36	39	words	word	NOUN
fbem-10171	36	40	and	and	CCONJ
fbem-10171	36	41	to	to	PART
fbem-10171	36	42	capture	capture	VERB
fbem-10171	36	43	core	core	PROPN
fbem-10171	36	44	meaning[21	meaning[21	PROPN
fbem-10171	36	45	]	]	PUNCT
fbem-10171	36	46	.	.	PUNCT
fbem-10171	37	1	morphological	morphological	ADJ
fbem-10171	37	2	reduction	reduction	NOUN
fbem-10171	37	3	reduces	reduce	VERB
fbem-10171	37	4	words	word	NOUN
fbem-10171	37	5	to	to	ADP
fbem-10171	37	6	their	their	PRON
fbem-10171	37	7	basic	basic	ADJ
fbem-10171	37	8	form	form	NOUN
fbem-10171	37	9	by	by	ADP
fbem-10171	37	10	considering	consider	VERB
fbem-10171	37	11	information	information	NOUN
fbem-10171	37	12	such	such	ADJ
fbem-10171	37	13	as	as	ADP
fbem-10171	37	14	root	root	NOUN
fbem-10171	37	15	,	,	PUNCT
fbem-10171	37	16	lexical	lexical	ADJ
fbem-10171	37	17	nature	nature	NOUN
fbem-10171	37	18	and	and	CCONJ
fbem-10171	37	19	context	context	NOUN
fbem-10171	37	20	.	.	PUNCT
fbem-10171	38	1	stem	stem	NOUN
fbem-10171	38	2	extraction	extraction	NOUN
fbem-10171	38	3	,	,	PUNCT
fbem-10171	38	4	on	on	ADP
fbem-10171	38	5	the	the	DET
fbem-10171	38	6	other	other	ADJ
fbem-10171	38	7	hand	hand	NOUN
fbem-10171	38	8	,	,	PUNCT
fbem-10171	38	9	extracts	extract	NOUN
fbem-10171	38	10	word	word	NOUN
fbem-10171	38	11	stems	stem	VERB
fbem-10171	38	12	and	and	CCONJ
fbem-10171	38	13	reduces	reduce	VERB
fbem-10171	38	14	the	the	DET
fbem-10171	38	15	dimensionality	dimensionality	NOUN
fbem-10171	38	16	of	of	ADP
fbem-10171	38	17	the	the	DET
fbem-10171	38	18	feature	feature	NOUN
fbem-10171	38	19	space	space	NOUN
fbem-10171	38	20	.	.	PUNCT
fbem-10171	39	1	this	this	PRON
fbem-10171	39	2	better	well	ADJ
fbem-10171	39	3	captures	capture	VERB
fbem-10171	39	4	the	the	DET
fbem-10171	39	5	true	true	ADJ
fbem-10171	39	6	meaning	meaning	NOUN
fbem-10171	39	7	of	of	ADP
fbem-10171	39	8	words	word	NOUN
fbem-10171	39	9	and	and	CCONJ
fbem-10171	39	10	reduces	reduce	VERB
fbem-10171	39	11	redundancy	redundancy	NOUN
fbem-10171	39	12	.	.	PUNCT
fbem-10171	40	1	commonly	commonly	ADV
fbem-10171	40	2	used	use	VERB
fbem-10171	40	3	techniques	technique	NOUN
fbem-10171	40	4	include	include	AUX
fbem-10171	40	5	dictionarybased	dictionarybase	VERB
fbem-10171	40	6	and	and	CCONJ
fbem-10171	40	7	rule	rule	NOUN
fbem-10171	40	8	-	-	PUNCT
fbem-10171	40	9	based	base	VERB
fbem-10171	40	10	approaches[22	approaches[22	PROPN
fbem-10171	40	11	]	]	PUNCT
fbem-10171	40	12	.	.	PUNCT
fbem-10171	41	1	word	word	NOUN
fbem-10171	41	2	form	form	NOUN
fbem-10171	41	3	reduction	reduction	NOUN
fbem-10171	41	4	and	and	CCONJ
fbem-10171	41	5	stemming	stem	VERB
fbem-10171	41	6	extraction	extraction	NOUN
fbem-10171	41	7	provide	provide	VERB
fbem-10171	41	8	more	more	ADV
fbem-10171	41	9	accurate	accurate	ADJ
fbem-10171	41	10	and	and	CCONJ
fbem-10171	41	11	consistent	consistent	ADJ
fbem-10171	41	12	feature	feature	NOUN
fbem-10171	41	13	representation	representation	NOUN
fbem-10171	41	14	,	,	PUNCT
fbem-10171	41	15	eliminate	eliminate	VERB
fbem-10171	41	16	feature	feature	NOUN
fbem-10171	41	17	redundancy	redundancy	NOUN
fbem-10171	41	18	and	and	CCONJ
fbem-10171	41	19	noise	noise	NOUN
fbem-10171	41	20	,	,	PUNCT
fbem-10171	41	21	and	and	CCONJ
fbem-10171	41	22	improve	improve	VERB
fbem-10171	41	23	model	model	NOUN
fbem-10171	41	24	accuracy	accuracy	NOUN
fbem-10171	41	25	and	and	CCONJ
fbem-10171	41	26	ability	ability	NOUN
fbem-10171	41	27	to	to	PART
fbem-10171	41	28	capture	capture	VERB
fbem-10171	41	29	sentiment	sentiment	NOUN
fbem-10171	41	30	tendencies	tendency	NOUN
fbem-10171	41	31	.	.	PUNCT
fbem-10171	42	1	however	however	ADV
fbem-10171	42	2	,	,	PUNCT
fbem-10171	42	3	appropriate	appropriate	ADJ
fbem-10171	42	4	tools	tool	NOUN
fbem-10171	42	5	need	need	VERB
fbem-10171	42	6	to	to	PART
fbem-10171	42	7	be	be	AUX
fbem-10171	42	8	selected	select	VERB
fbem-10171	42	9	,	,	PUNCT
fbem-10171	42	10	tuned	tune	VERB
fbem-10171	42	11	and	and	CCONJ
fbem-10171	42	12	optimized	optimize	VERB
fbem-10171	42	13	according	accord	VERB
fbem-10171	42	14	to	to	ADP
fbem-10171	42	15	the	the	DET
fbem-10171	42	16	task	task	NOUN
fbem-10171	42	17	and	and	CCONJ
fbem-10171	42	18	dataset	dataset	NOUN
fbem-10171	42	19	.	.	PUNCT
fbem-10171	43	1	2.2	2.2	NUM
fbem-10171	43	2	.	.	PUNCT
fbem-10171	44	1	feature	feature	NOUN
fbem-10171	44	2	extraction	extraction	NOUN
fbem-10171	44	3	and	and	CCONJ
fbem-10171	44	4	representation	representation	NOUN
fbem-10171	44	5	methods	method	NOUN
fbem-10171	44	6	2.2.1	2.2.1	NUM
fbem-10171	44	7	.	.	PUNCT
fbem-10171	45	1	bag	bag	NOUN
fbem-10171	45	2	-	-	PUNCT
fbem-10171	45	3	of	of	ADP
fbem-10171	45	4	-	-	PUNCT
fbem-10171	45	5	words	word	NOUN
fbem-10171	45	6	model	model	NOUN
fbem-10171	45	7	and	and	CCONJ
fbem-10171	45	8	tf	tf	PROPN
fbem-10171	45	9	-	-	PUNCT
fbem-10171	45	10	idf	idf	PROPN
fbem-10171	45	11	feature	feature	NOUN
fbem-10171	45	12	extraction	extraction	NOUN
fbem-10171	45	13	and	and	CCONJ
fbem-10171	45	14	representation	representation	NOUN
fbem-10171	45	15	methods	method	NOUN
fbem-10171	45	16	are	be	AUX
fbem-10171	45	17	crucial	crucial	ADJ
fbem-10171	45	18	in	in	ADP
fbem-10171	45	19	sentiment	sentiment	NOUN
fbem-10171	45	20	analysis	analysis	NOUN
fbem-10171	45	21	,	,	PUNCT
fbem-10171	45	22	transforming	transform	VERB
fbem-10171	45	23	raw	raw	ADJ
fbem-10171	45	24	text	text	NOUN
fbem-10171	45	25	into	into	ADP
fbem-10171	45	26	computerprocessable	computerprocessable	ADJ
fbem-10171	45	27	numerical	numerical	PROPN
fbem-10171	45	28	features[23	features[23	PROPN
fbem-10171	45	29	]	]	PUNCT
fbem-10171	45	30	.	.	PUNCT
fbem-10171	46	1	bag	bag	NOUN
fbem-10171	46	2	-	-	PUNCT
fbem-10171	46	3	of	of	ADP
fbem-10171	46	4	-	-	PUNCT
fbem-10171	46	5	words	word	NOUN
fbem-10171	46	6	model	model	NOUN
fbem-10171	46	7	and	and	CCONJ
fbem-10171	46	8	tf	tf	PROPN
fbem-10171	46	9	-	-	PUNCT
fbem-10171	46	10	idf	idf	PROPN
fbem-10171	46	11	are	be	AUX
fbem-10171	46	12	commonly	commonly	ADV
fbem-10171	46	13	used	use	VERB
fbem-10171	46	14	text	text	NOUN
fbem-10171	46	15	feature	feature	NOUN
fbem-10171	46	16	extraction	extraction	NOUN
fbem-10171	46	17	and	and	CCONJ
fbem-10171	46	18	representation	representation	NOUN
fbem-10171	46	19	methods	method	NOUN
fbem-10171	46	20	for	for	ADP
fbem-10171	46	21	capturing	capture	VERB
fbem-10171	46	22	the	the	DET
fbem-10171	46	23	frequency	frequency	NOUN
fbem-10171	46	24	and	and	CCONJ
fbem-10171	46	25	importance	importance	NOUN
fbem-10171	46	26	of	of	ADP
fbem-10171	46	27	words	word	NOUN
fbem-10171	46	28	and	and	CCONJ
fbem-10171	46	29	applying	apply	VERB
fbem-10171	46	30	them	they	PRON
fbem-10171	46	31	to	to	PART
fbem-10171	46	32	sentiment	sentiment	VERB
fbem-10171	46	33	analysis	analysis	NOUN
fbem-10171	46	34	tasks[24	tasks[24	NOUN
fbem-10171	46	35	]	]	X
fbem-10171	46	36	.	.	PUNCT
fbem-10171	47	1	the	the	DET
fbem-10171	47	2	bag	bag	NOUN
fbem-10171	47	3	-	-	PUNCT
fbem-10171	47	4	of	of	ADP
fbem-10171	47	5	-	-	PUNCT
fbem-10171	47	6	words	word	NOUN
fbem-10171	47	7	model	model	NOUN
fbem-10171	47	8	treats	treat	NOUN
fbem-10171	47	9	text	text	NOUN
fbem-10171	47	10	as	as	ADP
fbem-10171	47	11	a	a	DET
fbem-10171	47	12	collection	collection	NOUN
fbem-10171	47	13	of	of	ADP
fbem-10171	47	14	words	word	NOUN
fbem-10171	47	15	,	,	PUNCT
fbem-10171	47	16	ignoring	ignore	VERB
fbem-10171	47	17	order	order	NOUN
fbem-10171	47	18	and	and	CCONJ
fbem-10171	47	19	grammatical	grammatical	ADJ
fbem-10171	47	20	structure	structure	NOUN
fbem-10171	47	21	and	and	CCONJ
fbem-10171	47	22	focusing	focus	VERB
fbem-10171	47	23	only	only	ADV
fbem-10171	47	24	on	on	ADP
fbem-10171	47	25	the	the	DET
fbem-10171	47	26	frequency	frequency	NOUN
fbem-10171	47	27	of	of	ADP
fbem-10171	47	28	word	word	NOUN
fbem-10171	47	29	occurrences[25	occurrences[25	X
fbem-10171	47	30	]	]	X
fbem-10171	47	31	.	.	PUNCT
fbem-10171	48	1	the	the	DET
fbem-10171	48	2	text	text	NOUN
fbem-10171	48	3	is	be	AUX
fbem-10171	48	4	converted	convert	VERB
fbem-10171	48	5	into	into	ADP
fbem-10171	48	6	a	a	DET
fbem-10171	48	7	numerical	numerical	ADJ
fbem-10171	48	8	representation	representation	NOUN
fbem-10171	48	9	by	by	ADP
fbem-10171	48	10	constructing	construct	VERB
fbem-10171	48	11	a	a	DET
fbem-10171	48	12	document	document	NOUN
fbem-10171	48	13	-	-	PUNCT
fbem-10171	48	14	word	word	NOUN
fbem-10171	48	15	matrix	matrix	NOUN
fbem-10171	48	16	.	.	PUNCT
fbem-10171	49	1	the	the	DET
fbem-10171	49	2	bag	bag	NOUN
fbem-10171	49	3	-	-	PUNCT
fbem-10171	49	4	of	of	ADP
fbem-10171	49	5	-	-	PUNCT
fbem-10171	49	6	words	word	NOUN
fbem-10171	49	7	model	model	NOUN
fbem-10171	49	8	assesses	assess	VERB
fbem-10171	49	9	the	the	DET
fbem-10171	49	10	importance	importance	NOUN
fbem-10171	49	11	of	of	ADP
fbem-10171	49	12	words	word	NOUN
fbem-10171	49	13	by	by	ADP
fbem-10171	49	14	assuming	assume	VERB
fbem-10171	49	15	that	that	SCONJ
fbem-10171	49	16	words	word	NOUN
fbem-10171	49	17	with	with	ADP
fbem-10171	49	18	higher	high	ADJ
fbem-10171	49	19	frequencies	frequency	NOUN
fbem-10171	49	20	are	be	AUX
fbem-10171	49	21	given	give	VERB
fbem-10171	49	22	more	more	ADJ
fbem-10171	49	23	weight	weight	NOUN
fbem-10171	49	24	in	in	ADP
fbem-10171	49	25	the	the	DET
fbem-10171	49	26	representation	representation	NOUN
fbem-10171	49	27	.	.	PUNCT
fbem-10171	50	1	tf	tf	ADJ
fbem-10171	50	2	-	-	PUNCT
fbem-10171	50	3	idf	idf	ADJ
fbem-10171	50	4	(	(	PUNCT
fbem-10171	50	5	word	word	NOUN
fbem-10171	50	6	frequency	frequency	NOUN
fbem-10171	50	7	-	-	PUNCT
fbem-10171	50	8	inverse	inverse	NOUN
fbem-10171	50	9	document	document	NOUN
fbem-10171	50	10	frequency	frequency	NOUN
fbem-10171	50	11	)	)	PUNCT
fbem-10171	50	12	is	be	AUX
fbem-10171	50	13	a	a	DET
fbem-10171	50	14	method	method	NOUN
fbem-10171	50	15	that	that	PRON
fbem-10171	50	16	integrates	integrate	VERB
fbem-10171	50	17	word	word	NOUN
fbem-10171	50	18	frequency	frequency	NOUN
fbem-10171	50	19	and	and	CCONJ
fbem-10171	50	20	document	document	NOUN
fbem-10171	50	21	frequency	frequency	NOUN
fbem-10171	50	22	to	to	PART
fbem-10171	50	23	evaluate	evaluate	VERB
fbem-10171	50	24	the	the	DET
fbem-10171	50	25	importance	importance	NOUN
fbem-10171	50	26	of	of	ADP
fbem-10171	50	27	words	word	NOUN
fbem-10171	50	28	in	in	ADP
fbem-10171	50	29	text[26	text[26	NOUN
fbem-10171	50	30	]	]	PUNCT
fbem-10171	50	31	.	.	PUNCT
fbem-10171	51	1	tf	tf	PROPN
fbem-10171	51	2	-	-	PUNCT
fbem-10171	51	3	idf	idf	PROPN
fbem-10171	51	4	filters	filter	VERB
fbem-10171	51	5	out	out	ADP
fbem-10171	51	6	common	common	ADJ
fbem-10171	51	7	words	word	NOUN
fbem-10171	51	8	and	and	CCONJ
fbem-10171	51	9	highlights	highlight	NOUN
fbem-10171	51	10	the	the	DET
fbem-10171	51	11	important	important	ADJ
fbem-10171	51	12	words	word	NOUN
fbem-10171	51	13	in	in	ADP
fbem-10171	51	14	a	a	DET
fbem-10171	51	15	given	give	VERB
fbem-10171	51	16	document	document	NOUN
fbem-10171	51	17	.	.	PUNCT
fbem-10171	52	1	its	its	PRON
fbem-10171	52	2	formula	formula	NOUN
fbem-10171	52	3	is	be	AUX
fbem-10171	52	4	defined	define	VERB
fbem-10171	52	5	as	as	SCONJ
fbem-10171	52	6	follows	follow	VERB
fbem-10171	52	7	:	:	PUNCT
fbem-10171	52	8			PROPN
fbem-10171	52	9			PROPN
fbem-10171	52	10	,	,	PUNCT
fbem-10171	52	11	,	,	PUNCT
fbem-10171	52	12	log	log	VERB
fbem-10171	52	13	:	:	PUNCT
fbem-10171	53	1	i	i	PRON
fbem-10171	53	2	i	i	PRON
fbem-10171	54	1	j	j	VERB
fbem-10171	55	1	i	i	PRON
fbem-10171	55	2	i	i	PRON
fbem-10171	56	1	j	j	VERB
fbem-10171	57	1	ij	ij	INTJ
fbem-10171	57	2	k	k	PROPN
fbem-10171	57	3	k	k	PROPN
fbem-10171	57	4	j	j	PROPN
fbem-10171	57	5	ij	ij	INTJ
fbem-10171	58	1	n	n	INTJ
fbem-10171	58	2	tf	tf	INTJ
fbem-10171	58	3	n	n	PROPN
fbem-10171	59	1	d	d	PROPN
fbem-10171	59	2	idf	idf	PROPN
fbem-10171	60	1	j	j	PROPN
fbem-10171	60	2	t	t	PROPN
fbem-10171	61	1	d	d	X
fbem-10171	61	2	tf	tf	PROPN
fbem-10171	62	1	idf	idf	PROPN
fbem-10171	62	2	f	f	PROPN
fbem-10171	62	3	idf	idf	PROPN
fbem-10171	62	4			PROPN
fbem-10171	63	1			PROPN
fbem-10171	63	2			PROPN
fbem-10171	63	3			NOUN
fbem-10171	63	4			NUM
fbem-10171	63	5			NOUN
fbem-10171	63	6			X
fbem-10171	63	7	(	(	PUNCT
fbem-10171	63	8	1	1	X
fbem-10171	63	9	)	)	PUNCT
fbem-10171	63	10	the	the	DET
fbem-10171	63	11	advantages	advantage	NOUN
fbem-10171	63	12	of	of	ADP
fbem-10171	63	13	the	the	DET
fbem-10171	63	14	bag	bag	NOUN
fbem-10171	63	15	-	-	PUNCT
fbem-10171	63	16	of	of	ADP
fbem-10171	63	17	-	-	PUNCT
fbem-10171	63	18	words	word	NOUN
fbem-10171	63	19	model	model	NOUN
fbem-10171	63	20	and	and	CCONJ
fbem-10171	63	21	tf	tf	PROPN
fbem-10171	63	22	-	-	PUNCT
fbem-10171	63	23	idf	idf	PROPN
fbem-10171	63	24	are	be	AUX
fbem-10171	63	25	their	their	PRON
fbem-10171	63	26	simplicity	simplicity	NOUN
fbem-10171	63	27	and	and	CCONJ
fbem-10171	63	28	ease	ease	NOUN
fbem-10171	63	29	of	of	ADP
fbem-10171	63	30	implementation	implementation	NOUN
fbem-10171	63	31	,	,	PUNCT
fbem-10171	63	32	and	and	CCONJ
fbem-10171	63	33	their	their	PRON
fbem-10171	63	34	ability	ability	NOUN
fbem-10171	63	35	to	to	PART
fbem-10171	63	36	effectively	effectively	ADV
fbem-10171	63	37	represent	represent	VERB
fbem-10171	63	38	lexical	lexical	ADJ
fbem-10171	63	39	information	information	NOUN
fbem-10171	63	40	of	of	ADP
fbem-10171	63	41	text	text	NOUN
fbem-10171	63	42	.	.	PUNCT
fbem-10171	64	1	they	they	PRON
fbem-10171	64	2	are	be	AUX
fbem-10171	64	3	suitable	suitable	ADJ
fbem-10171	64	4	for	for	ADP
fbem-10171	64	5	sentiment	sentiment	NOUN
fbem-10171	64	6	analysis	analysis	NOUN
fbem-10171	64	7	,	,	PUNCT
fbem-10171	64	8	revealing	reveal	VERB
fbem-10171	64	9	users	user	NOUN
fbem-10171	64	10	'	'	PART
fbem-10171	64	11	opinions	opinion	NOUN
fbem-10171	64	12	and	and	CCONJ
fbem-10171	64	13	emotional	emotional	ADJ
fbem-10171	64	14	tendencies	tendency	NOUN
fbem-10171	64	15	,	,	PUNCT
fbem-10171	64	16	capturing	capture	VERB
fbem-10171	64	17	the	the	DET
fbem-10171	64	18	frequency	frequency	NOUN
fbem-10171	64	19	and	and	CCONJ
fbem-10171	64	20	importance	importance	NOUN
fbem-10171	64	21	of	of	ADP
fbem-10171	64	22	keywords	keyword	NOUN
fbem-10171	64	23	.	.	PUNCT
fbem-10171	65	1	however	however	ADV
fbem-10171	65	2	,	,	PUNCT
fbem-10171	65	3	they	they	PRON
fbem-10171	65	4	also	also	ADV
fbem-10171	65	5	have	have	VERB
fbem-10171	65	6	limitations	limitation	NOUN
fbem-10171	65	7	,	,	PUNCT
fbem-10171	65	8	such	such	ADJ
fbem-10171	65	9	as	as	ADP
fbem-10171	65	10	the	the	DET
fbem-10171	65	11	inability	inability	NOUN
fbem-10171	65	12	to	to	PART
fbem-10171	65	13	capture	capture	VERB
fbem-10171	65	14	word	word	NOUN
fbem-10171	65	15	order	order	NOUN
fbem-10171	65	16	and	and	CCONJ
fbem-10171	65	17	contextual	contextual	ADJ
fbem-10171	65	18	information	information	NOUN
fbem-10171	65	19	,	,	PUNCT
fbem-10171	65	20	and	and	CCONJ
fbem-10171	65	21	ignoring	ignore	VERB
fbem-10171	65	22	semantic	semantic	ADJ
fbem-10171	65	23	relationships	relationship	NOUN
fbem-10171	65	24	.	.	PUNCT
fbem-10171	66	1	2.2.2	2.2.2	X
fbem-10171	66	2	.	.	PUNCT
fbem-10171	66	3	word	word	NOUN
fbem-10171	66	4	embedding	embed	VERB
fbem-10171	66	5	and	and	CCONJ
fbem-10171	66	6	word2vec	word2vec	ADJ
fbem-10171	66	7	word	word	NOUN
fbem-10171	66	8	embeddings	embedding	NOUN
fbem-10171	66	9	and	and	CCONJ
fbem-10171	66	10	word2vec	word2vec	PRON
fbem-10171	66	11	play	play	VERB
fbem-10171	66	12	an	an	DET
fbem-10171	66	13	important	important	ADJ
fbem-10171	66	14	role	role	NOUN
fbem-10171	66	15	in	in	ADP
fbem-10171	66	16	sentiment	sentiment	NOUN
fbem-10171	66	17	analysis	analysis	NOUN
fbem-10171	66	18	.	.	PUNCT
fbem-10171	67	1	word	word	NOUN
fbem-10171	67	2	embedding	embed	VERB
fbem-10171	67	3	converts	convert	NOUN
fbem-10171	67	4	text	text	NOUN
fbem-10171	67	5	into	into	ADP
fbem-10171	67	6	a	a	DET
fbem-10171	67	7	dense	dense	ADJ
fbem-10171	67	8	vector	vector	NOUN
fbem-10171	67	9	representation	representation	NOUN
fbem-10171	67	10	that	that	PRON
fbem-10171	67	11	better	well	ADJ
fbem-10171	67	12	captures	capture	VERB
fbem-10171	67	13	the	the	DET
fbem-10171	67	14	semantic	semantic	ADJ
fbem-10171	67	15	information	information	NOUN
fbem-10171	67	16	of	of	ADP
fbem-10171	67	17	words[27	words[27	PROPN
fbem-10171	67	18	]	]	PUNCT
fbem-10171	67	19	.	.	PUNCT
fbem-10171	68	1	this	this	DET
fbem-10171	68	2	continuous	continuous	ADJ
fbem-10171	68	3	vector	vector	NOUN
fbem-10171	68	4	representation	representation	NOUN
fbem-10171	68	5	provides	provide	VERB
fbem-10171	68	6	better	well	ADJ
fbem-10171	68	7	feature	feature	NOUN
fbem-10171	68	8	representation	representation	NOUN
fbem-10171	68	9	and	and	CCONJ
fbem-10171	68	10	semantic	semantic	ADJ
fbem-10171	68	11	similarity	similarity	NOUN
fbem-10171	68	12	computation	computation	NOUN
fbem-10171	68	13	,	,	PUNCT
fbem-10171	68	14	which	which	PRON
fbem-10171	68	15	helps	help	VERB
fbem-10171	68	16	to	to	PART
fbem-10171	68	17	identify	identify	VERB
fbem-10171	68	18	sentiment	sentiment	NOUN
fbem-10171	68	19	tendencies	tendency	NOUN
fbem-10171	68	20	and	and	CCONJ
fbem-10171	68	21	expressions	expression	NOUN
fbem-10171	68	22	.	.	PUNCT
fbem-10171	69	1	using	use	VERB
fbem-10171	69	2	the	the	DET
fbem-10171	69	3	pre	pre	ADJ
fbem-10171	69	4	-	-	ADJ
fbem-10171	69	5	trained	trained	ADJ
fbem-10171	69	6	word2vec	word2vec	PROPN
fbem-10171	69	7	model	model	NOUN
fbem-10171	69	8	can	can	AUX
fbem-10171	69	9	extract	extract	VERB
fbem-10171	69	10	the	the	DET
fbem-10171	69	11	contextual	contextual	ADJ
fbem-10171	69	12	relationships	relationship	NOUN
fbem-10171	69	13	of	of	ADP
fbem-10171	69	14	words	word	NOUN
fbem-10171	69	15	and	and	CCONJ
fbem-10171	69	16	improve	improve	VERB
fbem-10171	69	17	the	the	DET
fbem-10171	69	18	performance	performance	NOUN
fbem-10171	69	19	of	of	ADP
fbem-10171	69	20	sentiment	sentiment	NOUN
fbem-10171	69	21	analysis	analysis	NOUN
fbem-10171	69	22	models[28	models[28	NOUN
fbem-10171	69	23	]	]	PUNCT
fbem-10171	69	24	.	.	PUNCT
fbem-10171	70	1	word	word	NOUN
fbem-10171	70	2	embedding	embed	VERB
fbem-10171	70	3	maps	map	NOUN
fbem-10171	70	4	words	word	NOUN
fbem-10171	70	5	to	to	ADP
fbem-10171	70	6	a	a	DET
fbem-10171	70	7	continuous	continuous	ADJ
fbem-10171	70	8	vector	vector	NOUN
fbem-10171	70	9	space	space	NOUN
fbem-10171	70	10	and	and	CCONJ
fbem-10171	70	11	provides	provide	VERB
fbem-10171	70	12	rich	rich	ADJ
fbem-10171	70	13	feature	feature	NOUN
fbem-10171	70	14	representation	representation	NOUN
fbem-10171	70	15	by	by	ADP
fbem-10171	70	16	learning	learn	VERB
fbem-10171	70	17	to	to	PART
fbem-10171	70	18	capture	capture	VERB
fbem-10171	70	19	the	the	DET
fbem-10171	70	20	semantic	semantic	ADJ
fbem-10171	70	21	relationships	relationship	NOUN
fbem-10171	70	22	of	of	ADP
fbem-10171	70	23	words	word	NOUN
fbem-10171	70	24	,	,	PUNCT
fbem-10171	70	25	making	make	VERB
fbem-10171	70	26	semantically	semantically	ADV
fbem-10171	70	27	similar	similar	ADJ
fbem-10171	70	28	words	word	NOUN
fbem-10171	70	29	closer	close	ADV
fbem-10171	70	30	together	together	ADV
fbem-10171	70	31	in	in	ADP
fbem-10171	70	32	the	the	DET
fbem-10171	70	33	vector	vector	NOUN
fbem-10171	70	34	space	space	NOUN
fbem-10171	70	35	.	.	PUNCT
fbem-10171	71	1	it	it	PRON
fbem-10171	71	2	uses	use	VERB
fbem-10171	71	3	language	language	NOUN
fbem-10171	71	4	models	model	NOUN
fbem-10171	71	5	or	or	CCONJ
fbem-10171	71	6	neural	neural	ADJ
fbem-10171	71	7	networks	network	NOUN
fbem-10171	71	8	to	to	PART
fbem-10171	71	9	learn	learn	VERB
fbem-10171	71	10	word	word	NOUN
fbem-10171	71	11	embedding	embed	VERB
fbem-10171	71	12	models	model	NOUN
fbem-10171	71	13	on	on	ADP
fbem-10171	71	14	large	large	ADJ
fbem-10171	71	15	-	-	PUNCT
fbem-10171	71	16	scale	scale	NOUN
fbem-10171	71	17	text	text	NOUN
fbem-10171	71	18	data	datum	NOUN
fbem-10171	71	19	,	,	PUNCT
fbem-10171	71	20	making	make	VERB
fbem-10171	71	21	similarly	similarly	ADV
fbem-10171	71	22	meaningful	meaningful	ADJ
fbem-10171	71	23	words	word	NOUN
fbem-10171	71	24	closer	close	ADV
fbem-10171	71	25	together	together	ADV
fbem-10171	71	26	in	in	ADP
fbem-10171	71	27	vector	vector	NOUN
fbem-10171	71	28	space	space	NOUN
fbem-10171	71	29	.	.	PUNCT
fbem-10171	72	1	word2vec	word2vec	X
fbem-10171	72	2	is	be	AUX
fbem-10171	72	3	a	a	DET
fbem-10171	72	4	commonly	commonly	ADV
fbem-10171	72	5	used	use	VERB
fbem-10171	72	6	word	word	NOUN
fbem-10171	72	7	embedding	embed	VERB
fbem-10171	72	8	algorithm	algorithm	NOUN
fbem-10171	72	9	based	base	VERB
fbem-10171	72	10	on	on	ADP
fbem-10171	72	11	a	a	DET
fbem-10171	72	12	shallow	shallow	ADJ
fbem-10171	72	13	neural	neural	ADJ
fbem-10171	72	14	network	network	NOUN
fbem-10171	72	15	model	model	NOUN
fbem-10171	72	16	trained	train	VERB
fbem-10171	72	17	on	on	ADP
fbem-10171	72	18	largescale	largescale	NOUN
fbem-10171	72	19	text	text	NOUN
fbem-10171	72	20	data	datum	NOUN
fbem-10171	72	21	to	to	PART
fbem-10171	72	22	generate	generate	VERB
fbem-10171	72	23	high	high	ADJ
fbem-10171	72	24	-	-	PUNCT
fbem-10171	72	25	quality	quality	NOUN
fbem-10171	72	26	word	word	NOUN
fbem-10171	72	27	vectors	vector	NOUN
fbem-10171	72	28	by	by	ADP
fbem-10171	72	29	predicting	predict	VERB
fbem-10171	72	30	contextual	contextual	ADJ
fbem-10171	72	31	information	information	NOUN
fbem-10171	72	32	around	around	ADP
fbem-10171	72	33	words	word	NOUN
fbem-10171	72	34	.	.	PUNCT
fbem-10171	73	1	it	it	PRON
fbem-10171	73	2	consists	consist	VERB
fbem-10171	73	3	of	of	ADP
fbem-10171	73	4	a	a	DET
fbem-10171	73	5	continuous	continuous	ADJ
fbem-10171	73	6	bag	bag	NOUN
fbem-10171	73	7	-	-	PUNCT
fbem-10171	73	8	of	of	ADP
fbem-10171	73	9	-	-	PUNCT
fbem-10171	73	10	words	word	NOUN
fbem-10171	73	11	model	model	NOUN
fbem-10171	73	12	(	(	PUNCT
fbem-10171	73	13	cbow	cbow	VERB
fbem-10171	73	14	)	)	PUNCT
fbem-10171	73	15	and	and	CCONJ
fbem-10171	73	16	a	a	DET
fbem-10171	73	17	skip	skip	ADJ
fbem-10171	73	18	-	-	PUNCT
fbem-10171	73	19	gram	gram	NOUN
fbem-10171	73	20	model	model	NOUN
fbem-10171	73	21	,	,	PUNCT
fbem-10171	73	22	both	both	PRON
fbem-10171	73	23	of	of	ADP
fbem-10171	73	24	which	which	PRON
fbem-10171	73	25	generate	generate	VERB
fbem-10171	73	26	word	word	NOUN
fbem-10171	73	27	embedding	embed	VERB
fbem-10171	73	28	vectors	vector	NOUN
fbem-10171	73	29	that	that	PRON
fbem-10171	73	30	capture	capture	VERB
fbem-10171	73	31	the	the	DET
fbem-10171	73	32	semantic	semantic	ADJ
fbem-10171	73	33	features	feature	NOUN
fbem-10171	73	34	of	of	ADP
fbem-10171	73	35	words	word	NOUN
fbem-10171	73	36	.	.	PUNCT
fbem-10171	74	1	its	its	PRON
fbem-10171	74	2	network	network	NOUN
fbem-10171	74	3	model	model	NOUN
fbem-10171	74	4	diagram	diagram	NOUN
fbem-10171	74	5	is	be	AUX
fbem-10171	74	6	as	as	SCONJ
fbem-10171	74	7	follows	follow	VERB
fbem-10171	74	8	:	:	PUNCT
fbem-10171	74	9	figure	figure	NOUN
fbem-10171	74	10	2	2	NUM
fbem-10171	74	11	.	.	PUNCT
fbem-10171	74	12	cbow	cbow	VERB
fbem-10171	74	13	and	and	CCONJ
fbem-10171	74	14	skip	skip	VERB
fbem-10171	74	15	-	-	PUNCT
fbem-10171	74	16	gram	gram	NOUN
fbem-10171	74	17	models	model	NOUN
fbem-10171	74	18	word	word	NOUN
fbem-10171	74	19	embeddings	embedding	NOUN
fbem-10171	74	20	better	well	ADV
fbem-10171	74	21	capture	capture	VERB
fbem-10171	74	22	the	the	DET
fbem-10171	74	23	semantic	semantic	ADJ
fbem-10171	74	24	information	information	NOUN
fbem-10171	74	25	of	of	ADP
fbem-10171	74	26	words	word	NOUN
fbem-10171	74	27	and	and	CCONJ
fbem-10171	74	28	usually	usually	ADV
fbem-10171	74	29	perform	perform	VERB
fbem-10171	74	30	well	well	ADV
fbem-10171	74	31	in	in	ADP
fbem-10171	74	32	sentiment	sentiment	NOUN
fbem-10171	74	33	analysis	analysis	NOUN
fbem-10171	74	34	.	.	PUNCT
fbem-10171	75	1	word2vec	word2vec	X
fbem-10171	75	2	is	be	AUX
fbem-10171	75	3	an	an	DET
fbem-10171	75	4	important	important	ADJ
fbem-10171	75	5	feature	feature	NOUN
fbem-10171	75	6	extraction	extraction	NOUN
fbem-10171	75	7	and	and	CCONJ
fbem-10171	75	8	representation	representation	NOUN
fbem-10171	75	9	method	method	NOUN
fbem-10171	75	10	that	that	PRON
fbem-10171	75	11	captures	capture	VERB
fbem-10171	75	12	semantic	semantic	ADJ
fbem-10171	75	13	relations	relation	NOUN
fbem-10171	75	14	and	and	CCONJ
fbem-10171	75	15	contextual	contextual	ADJ
fbem-10171	75	16	information	information	NOUN
fbem-10171	75	17	of	of	ADP
fbem-10171	75	18	words	word	NOUN
fbem-10171	75	19	and	and	CCONJ
fbem-10171	75	20	provides	provide	VERB
fbem-10171	75	21	rich	rich	ADJ
fbem-10171	75	22	feature	feature	NOUN
fbem-10171	75	23	representation	representation	NOUN
fbem-10171	75	24	,	,	PUNCT
fbem-10171	75	25	thus	thus	ADV
fbem-10171	75	26	improving	improve	VERB
fbem-10171	75	27	the	the	DET
fbem-10171	75	28	accuracy	accuracy	NOUN
fbem-10171	75	29	and	and	CCONJ
fbem-10171	75	30	effectiveness	effectiveness	NOUN
fbem-10171	75	31	of	of	ADP
fbem-10171	75	32	sentiment	sentiment	NOUN
fbem-10171	75	33	analysis	analysis	NOUN
fbem-10171	75	34	.	.	PUNCT
fbem-10171	76	1	2.2.3	2.2.3	NUM
fbem-10171	76	2	.	.	PUNCT
fbem-10171	76	3	attention	attention	NOUN
fbem-10171	76	4	mechanism	mechanism	NOUN
fbem-10171	76	5	and	and	CCONJ
fbem-10171	76	6	transformer	transformer	VERB
fbem-10171	76	7	the	the	DET
fbem-10171	76	8	attention	attention	NOUN
fbem-10171	76	9	mechanism	mechanism	NOUN
fbem-10171	76	10	is	be	AUX
fbem-10171	76	11	an	an	DET
fbem-10171	76	12	important	important	ADJ
fbem-10171	76	13	feature	feature	NOUN
fbem-10171	76	14	representation	representation	NOUN
fbem-10171	76	15	,	,	PUNCT
fbem-10171	76	16	similar	similar	ADJ
fbem-10171	76	17	to	to	ADP
fbem-10171	76	18	the	the	DET
fbem-10171	76	19	attention	attention	NOUN
fbem-10171	76	20	mechanism	mechanism	NOUN
fbem-10171	76	21	used	use	VERB
fbem-10171	76	22	by	by	ADP
fbem-10171	76	23	humans	human	NOUN
fbem-10171	76	24	in	in	ADP
fbem-10171	76	25	understanding	understand	VERB
fbem-10171	76	26	text[29	text[29	NOUN
fbem-10171	76	27	]	]	PUNCT
fbem-10171	76	28	.	.	PUNCT
fbem-10171	77	1	it	it	PRON
fbem-10171	77	2	solves	solve	VERB
fbem-10171	77	3	the	the	DET
fbem-10171	77	4	parallelism	parallelism	NOUN
fbem-10171	77	5	and	and	CCONJ
fbem-10171	77	6	long	long	ADJ
fbem-10171	77	7	sequence	sequence	NOUN
fbem-10171	77	8	dependency	dependency	NOUN
fbem-10171	77	9	problems	problem	NOUN
fbem-10171	77	10	in	in	ADP
fbem-10171	77	11	rnn	rnn	PROPN
fbem-10171	77	12	and	and	CCONJ
fbem-10171	77	13	lstm	lstm	NOUN
fbem-10171	77	14	networks	network	NOUN
fbem-10171	77	15	.	.	PUNCT
fbem-10171	78	1	the	the	DET
fbem-10171	78	2	attention	attention	NOUN
fbem-10171	78	3	mechanism	mechanism	NOUN
fbem-10171	78	4	learns	learn	VERB
fbem-10171	78	5	the	the	DET
fbem-10171	78	6	correlation	correlation	NOUN
fbem-10171	78	7	between	between	ADP
fbem-10171	78	8	different	different	ADJ
fbem-10171	78	9	parts	part	NOUN
fbem-10171	78	10	of	of	ADP
fbem-10171	78	11	a	a	DET
fbem-10171	78	12	text	text	NOUN
fbem-10171	78	13	and	and	CCONJ
fbem-10171	78	14	focuses	focus	VERB
fbem-10171	78	15	on	on	ADP
fbem-10171	78	16	important	important	ADJ
fbem-10171	78	17	pieces	piece	NOUN
fbem-10171	78	18	of	of	ADP
fbem-10171	78	19	information	information	NOUN
fbem-10171	78	20	,	,	PUNCT
fbem-10171	78	21	assigning	assign	VERB
fbem-10171	78	22	weights	weight	NOUN
fbem-10171	78	23	to	to	ADP
fbem-10171	78	24	different	different	ADJ
fbem-10171	78	25	parts	part	NOUN
fbem-10171	78	26	in	in	ADP
fbem-10171	78	27	a	a	DET
fbem-10171	78	28	specific	specific	ADJ
fbem-10171	78	29	task	task	NOUN
fbem-10171	78	30	.	.	PUNCT
fbem-10171	79	1	in	in	ADP
fbem-10171	79	2	text	text	NOUN
fbem-10171	79	3	sentiment	sentiment	NOUN
fbem-10171	79	4	analysis	analysis	NOUN
fbem-10171	79	5	,	,	PUNCT
fbem-10171	79	6	the	the	DET
fbem-10171	79	7	attention	attention	NOUN
fbem-10171	79	8	mechanism	mechanism	NOUN
fbem-10171	79	9	improves	improve	VERB
fbem-10171	79	10	model	model	NOUN
fbem-10171	79	11	performance	performance	NOUN
fbem-10171	79	12	by	by	ADP
fbem-10171	79	13	focusing	focus	VERB
fbem-10171	79	14	on	on	ADP
fbem-10171	79	15	important	important	ADJ
fbem-10171	79	16	contextual	contextual	ADJ
fbem-10171	79	17	information	information	NOUN
fbem-10171	79	18	to	to	PART
fbem-10171	79	19	better	well	ADV
fbem-10171	79	20	capture	capture	VERB
fbem-10171	79	21	the	the	DET
fbem-10171	79	22	semantic	semantic	ADJ
fbem-10171	79	23	and	and	CCONJ
fbem-10171	79	24	sentiment	sentiment	NOUN
fbem-10171	79	25	expressions	expression	NOUN
fbem-10171	79	26	of	of	ADP
fbem-10171	79	27	the	the	DET
fbem-10171	79	28	text	text	NOUN
fbem-10171	79	29	and	and	CCONJ
fbem-10171	79	30	identify	identify	VERB
fbem-10171	79	31	the	the	DET
fbem-10171	79	32	words	word	NOUN
fbem-10171	79	33	that	that	PRON
fbem-10171	79	34	contribute	contribute	VERB
fbem-10171	79	35	to	to	ADP
fbem-10171	79	36	sentiment	sentiment	NOUN
fbem-10171	79	37	classification	classification	NOUN
fbem-10171	79	38	.	.	PUNCT
fbem-10171	80	1	transformer[30	transformer[30	NOUN
fbem-10171	80	2	]	]	PUNCT
fbem-10171	80	3	model	model	NOUN
fbem-10171	80	4	is	be	AUX
fbem-10171	80	5	a	a	DET
fbem-10171	80	6	deep	deep	ADJ
fbem-10171	80	7	learning	learning	NOUN
fbem-10171	80	8	model	model	NOUN
fbem-10171	80	9	based	base	VERB
fbem-10171	80	10	on	on	ADP
fbem-10171	80	11	60	60	NUM
fbem-10171	80	12	the	the	DET
fbem-10171	80	13	attention	attention	NOUN
fbem-10171	80	14	mechanism	mechanism	NOUN
fbem-10171	80	15	,	,	PUNCT
fbem-10171	80	16	a	a	DET
fbem-10171	80	17	sequence	sequence	NOUN
fbem-10171	80	18	transformation	transformation	NOUN
fbem-10171	80	19	model	model	NOUN
fbem-10171	80	20	based	base	VERB
fbem-10171	80	21	entirely	entirely	ADV
fbem-10171	80	22	on	on	ADP
fbem-10171	80	23	attention	attention	NOUN
fbem-10171	80	24	that	that	PRON
fbem-10171	80	25	replaces	replace	VERB
fbem-10171	80	26	the	the	DET
fbem-10171	80	27	cyclic	cyclic	ADJ
fbem-10171	80	28	layer	layer	NOUN
fbem-10171	80	29	with	with	ADP
fbem-10171	80	30	a	a	DET
fbem-10171	80	31	multi	multi	ADJ
fbem-10171	80	32	-	-	ADJ
fbem-10171	80	33	headed	headed	ADJ
fbem-10171	80	34	self	self	NOUN
fbem-10171	80	35	-	-	PUNCT
fbem-10171	80	36	attentive	attentive	ADJ
fbem-10171	80	37	mechanism.transformer	mechanism.transformer	NUM
fbem-10171	80	38	model	model	NOUN
fbem-10171	80	39	uses	use	VERB
fbem-10171	80	40	multiple	multiple	ADJ
fbem-10171	80	41	self	self	NOUN
fbem-10171	80	42	-	-	PUNCT
fbem-10171	80	43	attentive	attentive	ADJ
fbem-10171	80	44	layers	layer	NOUN
fbem-10171	80	45	to	to	PART
fbem-10171	80	46	encode	encode	VERB
fbem-10171	80	47	the	the	DET
fbem-10171	80	48	input	input	NOUN
fbem-10171	80	49	sequence	sequence	NOUN
fbem-10171	80	50	by	by	ADP
fbem-10171	80	51	simultaneously	simultaneously	ADV
fbem-10171	80	52	considering	consider	VERB
fbem-10171	80	53	information	information	NOUN
fbem-10171	80	54	at	at	ADP
fbem-10171	80	55	different	different	ADJ
fbem-10171	80	56	locations	location	NOUN
fbem-10171	80	57	in	in	ADP
fbem-10171	80	58	the	the	DET
fbem-10171	80	59	sequence	sequence	NOUN
fbem-10171	80	60	,	,	PUNCT
fbem-10171	80	61	capturing	capture	VERB
fbem-10171	80	62	the	the	DET
fbem-10171	80	63	dependencies	dependency	NOUN
fbem-10171	80	64	between	between	ADP
fbem-10171	80	65	global	global	ADJ
fbem-10171	80	66	and	and	CCONJ
fbem-10171	80	67	local	local	ADJ
fbem-10171	80	68	,	,	PUNCT
fbem-10171	80	69	and	and	CCONJ
fbem-10171	80	70	improving	improve	VERB
fbem-10171	80	71	the	the	DET
fbem-10171	80	72	sentiment	sentiment	NOUN
fbem-10171	80	73	analysis	analysis	NOUN
fbem-10171	80	74	accuracy	accuracy	NOUN
fbem-10171	80	75	and	and	CCONJ
fbem-10171	80	76	effectiveness	effectiveness	NOUN
fbem-10171	80	77	.	.	PUNCT
fbem-10171	81	1	compared	compare	VERB
fbem-10171	81	2	with	with	ADP
fbem-10171	81	3	the	the	DET
fbem-10171	81	4	traditional	traditional	ADJ
fbem-10171	81	5	recurrent	recurrent	ADJ
fbem-10171	81	6	neural	neural	ADJ
fbem-10171	81	7	network	network	NOUN
fbem-10171	81	8	(	(	PUNCT
fbem-10171	81	9	rnn	rnn	PROPN
fbem-10171	81	10	)	)	PUNCT
fbem-10171	81	11	,	,	PUNCT
fbem-10171	81	12	the	the	DET
fbem-10171	81	13	transformer	transformer	NOUN
fbem-10171	81	14	model	model	NOUN
fbem-10171	81	15	has	have	AUX
fbem-10171	81	16	parallel	parallel	ADJ
fbem-10171	81	17	computing	computing	NOUN
fbem-10171	81	18	capability	capability	NOUN
fbem-10171	81	19	,	,	PUNCT
fbem-10171	81	20	improves	improve	VERB
fbem-10171	81	21	training	training	NOUN
fbem-10171	81	22	efficiency	efficiency	NOUN
fbem-10171	81	23	,	,	PUNCT
fbem-10171	81	24	and	and	CCONJ
fbem-10171	81	25	can	can	AUX
fbem-10171	81	26	handle	handle	VERB
fbem-10171	81	27	longer	long	ADJ
fbem-10171	81	28	text	text	VERB
fbem-10171	81	29	sequences	sequence	NOUN
fbem-10171	81	30	,	,	PUNCT
fbem-10171	81	31	performing	perform	VERB
fbem-10171	81	32	well	well	ADV
fbem-10171	81	33	in	in	ADP
fbem-10171	81	34	sentiment	sentiment	NOUN
fbem-10171	81	35	analysis	analysis	NOUN
fbem-10171	81	36	tasks	task	NOUN
fbem-10171	81	37	.	.	PUNCT
fbem-10171	82	1	in	in	ADP
fbem-10171	82	2	summary	summary	NOUN
fbem-10171	82	3	,	,	PUNCT
fbem-10171	82	4	feature	feature	NOUN
fbem-10171	82	5	extraction	extraction	NOUN
fbem-10171	82	6	and	and	CCONJ
fbem-10171	82	7	representation	representation	NOUN
fbem-10171	82	8	methods	method	NOUN
fbem-10171	82	9	play	play	VERB
fbem-10171	82	10	a	a	DET
fbem-10171	82	11	key	key	ADJ
fbem-10171	82	12	role	role	NOUN
fbem-10171	82	13	in	in	ADP
fbem-10171	82	14	natural	natural	ADJ
fbem-10171	82	15	language	language	NOUN
fbem-10171	82	16	processing	processing	NOUN
fbem-10171	82	17	sentiment	sentiment	NOUN
fbem-10171	82	18	analysis	analysis	NOUN
fbem-10171	82	19	.	.	PUNCT
fbem-10171	83	1	bag	bag	NOUN
fbem-10171	83	2	-	-	PUNCT
fbem-10171	83	3	of	of	ADP
fbem-10171	83	4	-	-	PUNCT
fbem-10171	83	5	words	word	NOUN
fbem-10171	83	6	model	model	NOUN
fbem-10171	83	7	and	and	CCONJ
fbem-10171	83	8	tf	tf	PROPN
fbem-10171	83	9	-	-	PUNCT
fbem-10171	83	10	idf	idf	PROPN
fbem-10171	83	11	are	be	AUX
fbem-10171	83	12	used	use	VERB
fbem-10171	83	13	to	to	PART
fbem-10171	83	14	capture	capture	VERB
fbem-10171	83	15	lexical	lexical	ADJ
fbem-10171	83	16	information	information	NOUN
fbem-10171	83	17	,	,	PUNCT
fbem-10171	83	18	word	word	NOUN
fbem-10171	83	19	embedding	embed	VERB
fbem-10171	83	20	and	and	CCONJ
fbem-10171	83	21	word2vec	word2vec	PRON
fbem-10171	83	22	are	be	AUX
fbem-10171	83	23	used	use	VERB
fbem-10171	83	24	to	to	PART
fbem-10171	83	25	represent	represent	VERB
fbem-10171	83	26	semantic	semantic	ADJ
fbem-10171	83	27	relationships	relationship	NOUN
fbem-10171	83	28	of	of	ADP
fbem-10171	83	29	words	word	NOUN
fbem-10171	83	30	,	,	PUNCT
fbem-10171	83	31	and	and	CCONJ
fbem-10171	83	32	attention	attention	NOUN
fbem-10171	83	33	mechanism	mechanism	NOUN
fbem-10171	83	34	and	and	CCONJ
fbem-10171	83	35	transformer	transformer	NOUN
fbem-10171	83	36	model	model	NOUN
fbem-10171	83	37	can	can	AUX
fbem-10171	83	38	focus	focus	VERB
fbem-10171	83	39	on	on	ADP
fbem-10171	83	40	important	important	ADJ
fbem-10171	83	41	text	text	NOUN
fbem-10171	83	42	fragments	fragment	NOUN
fbem-10171	83	43	and	and	CCONJ
fbem-10171	83	44	contextual	contextual	ADJ
fbem-10171	83	45	information	information	NOUN
fbem-10171	83	46	.	.	PUNCT
fbem-10171	84	1	the	the	DET
fbem-10171	84	2	combined	combined	ADJ
fbem-10171	84	3	application	application	NOUN
fbem-10171	84	4	of	of	ADP
fbem-10171	84	5	these	these	DET
fbem-10171	84	6	methods	method	NOUN
fbem-10171	84	7	can	can	AUX
fbem-10171	84	8	improve	improve	VERB
fbem-10171	84	9	the	the	DET
fbem-10171	84	10	accuracy	accuracy	NOUN
fbem-10171	84	11	and	and	CCONJ
fbem-10171	84	12	effectiveness	effectiveness	NOUN
fbem-10171	84	13	of	of	ADP
fbem-10171	84	14	sentiment	sentiment	NOUN
fbem-10171	84	15	analysis	analysis	NOUN
fbem-10171	84	16	and	and	CCONJ
fbem-10171	84	17	help	help	VERB
fbem-10171	84	18	analyze	analyze	VERB
fbem-10171	84	19	people	people	NOUN
fbem-10171	84	20	's	's	PART
fbem-10171	84	21	opinions	opinion	NOUN
fbem-10171	84	22	,	,	PUNCT
fbem-10171	84	23	attitudes	attitude	NOUN
fbem-10171	84	24	and	and	CCONJ
fbem-10171	84	25	emotions	emotion	NOUN
fbem-10171	84	26	,	,	PUNCT
fbem-10171	84	27	etc	etc	X
fbem-10171	84	28	.	.	X
fbem-10171	84	29	2.3	2.3	NUM
fbem-10171	84	30	.	.	PUNCT
fbem-10171	85	1	emotional	emotional	ADJ
fbem-10171	85	2	analysis	analysis	NOUN
fbem-10171	85	3	method	method	NOUN
fbem-10171	85	4	2.3.1	2.3.1	NUM
fbem-10171	85	5	.	.	PUNCT
fbem-10171	85	6	emotional	emotional	ADJ
fbem-10171	85	7	lexicon	lexicon	NOUN
fbem-10171	85	8	and	and	CCONJ
fbem-10171	85	9	rules	rule	NOUN
fbem-10171	85	10	-	-	PUNCT
fbem-10171	85	11	based	base	VERB
fbem-10171	85	12	approach	approach	NOUN
fbem-10171	85	13	the	the	DET
fbem-10171	85	14	sentiment	sentiment	NOUN
fbem-10171	85	15	lexicon	lexicon	NOUN
fbem-10171	85	16	and	and	CCONJ
fbem-10171	85	17	rule	rule	NOUN
fbem-10171	85	18	-	-	PUNCT
fbem-10171	85	19	based	base	VERB
fbem-10171	85	20	approach	approach	NOUN
fbem-10171	85	21	is	be	AUX
fbem-10171	85	22	a	a	DET
fbem-10171	85	23	commonly	commonly	ADV
fbem-10171	85	24	used	use	VERB
fbem-10171	85	25	sentiment	sentiment	NOUN
fbem-10171	85	26	classification	classification	NOUN
fbem-10171	85	27	technique	technique	NOUN
fbem-10171	85	28	.	.	PUNCT
fbem-10171	86	1	the	the	DET
fbem-10171	86	2	method	method	NOUN
fbem-10171	86	3	uses	use	VERB
fbem-10171	86	4	pre	pre	ADJ
fbem-10171	86	5	-	-	ADJ
fbem-10171	86	6	constructed	constructed	ADJ
fbem-10171	86	7	sentiment	sentiment	NOUN
fbem-10171	86	8	dictionaries	dictionary	NOUN
fbem-10171	86	9	and	and	CCONJ
fbem-10171	86	10	rules	rule	NOUN
fbem-10171	86	11	for	for	ADP
fbem-10171	86	12	sentiment	sentiment	NOUN
fbem-10171	86	13	classification	classification	NOUN
fbem-10171	86	14	,	,	PUNCT
fbem-10171	86	15	and	and	CCONJ
fbem-10171	86	16	the	the	DET
fbem-10171	86	17	overall	overall	ADJ
fbem-10171	86	18	sentiment	sentiment	NOUN
fbem-10171	86	19	polarity	polarity	NOUN
fbem-10171	86	20	is	be	AUX
fbem-10171	86	21	derived	derive	VERB
fbem-10171	86	22	by	by	ADP
fbem-10171	86	23	matching	match	VERB
fbem-10171	86	24	and	and	CCONJ
fbem-10171	86	25	weighting	weight	VERB
fbem-10171	86	26	the	the	DET
fbem-10171	86	27	sentiment	sentiment	NOUN
fbem-10171	86	28	words	word	NOUN
fbem-10171	86	29	in	in	ADP
fbem-10171	86	30	the	the	DET
fbem-10171	86	31	text[31	text[31	NOUN
fbem-10171	86	32	]	]	X
fbem-10171	86	33	.	.	PUNCT
fbem-10171	87	1	the	the	DET
fbem-10171	87	2	sentiment	sentiment	NOUN
fbem-10171	87	3	lexicon	lexicon	NOUN
fbem-10171	87	4	contains	contain	VERB
fbem-10171	87	5	vocabulary	vocabulary	NOUN
fbem-10171	87	6	and	and	CCONJ
fbem-10171	87	7	corresponding	corresponding	ADJ
fbem-10171	87	8	sentiment	sentiment	NOUN
fbem-10171	87	9	tendencies	tendency	NOUN
fbem-10171	87	10	,	,	PUNCT
fbem-10171	87	11	which	which	PRON
fbem-10171	87	12	can	can	AUX
fbem-10171	87	13	be	be	AUX
fbem-10171	87	14	constructed	construct	VERB
fbem-10171	87	15	manually	manually	ADV
fbem-10171	87	16	or	or	CCONJ
fbem-10171	87	17	obtained	obtain	VERB
fbem-10171	87	18	by	by	ADP
fbem-10171	87	19	automatic	automatic	ADJ
fbem-10171	87	20	mining	mining	NOUN
fbem-10171	87	21	and	and	CCONJ
fbem-10171	87	22	machine	machine	NOUN
fbem-10171	87	23	learning	learning	NOUN
fbem-10171	87	24	.	.	PUNCT
fbem-10171	88	1	the	the	DET
fbem-10171	88	2	process	process	NOUN
fbem-10171	88	3	of	of	ADP
fbem-10171	88	4	sentiment	sentiment	NOUN
fbem-10171	88	5	lexicon	lexicon	NOUN
fbem-10171	88	6	analysis	analysis	NOUN
fbem-10171	88	7	is	be	AUX
fbem-10171	88	8	as	as	SCONJ
fbem-10171	88	9	follows	follow	VERB
fbem-10171	88	10	:	:	PUNCT
fbem-10171	88	11	word	word	NOUN
fbem-10171	88	12	segmentation	segmentation	NOUN
fbem-10171	88	13	and	and	CCONJ
fbem-10171	88	14	lexical	lexical	ADJ
fbem-10171	88	15	annotation	annotation	NOUN
fbem-10171	88	16	segment	segment	NOUN
fbem-10171	88	17	the	the	DET
fbem-10171	88	18	text	text	NOUN
fbem-10171	88	19	into	into	ADP
fbem-10171	88	20	words	word	NOUN
fbem-10171	88	21	or	or	CCONJ
fbem-10171	88	22	phrases	phrase	NOUN
fbem-10171	88	23	and	and	CCONJ
fbem-10171	88	24	determine	determine	VERB
fbem-10171	88	25	the	the	DET
fbem-10171	88	26	lexical	lexical	ADJ
fbem-10171	88	27	nature	nature	NOUN
fbem-10171	88	28	of	of	ADP
fbem-10171	88	29	each	each	DET
fbem-10171	88	30	word	word	NOUN
fbem-10171	88	31	.	.	PUNCT
fbem-10171	89	1	by	by	ADP
fbem-10171	89	2	matching	match	VERB
fbem-10171	89	3	the	the	DET
fbem-10171	89	4	sentiment	sentiment	NOUN
fbem-10171	89	5	lexicon	lexicon	NOUN
fbem-10171	89	6	,	,	PUNCT
fbem-10171	89	7	the	the	DET
fbem-10171	89	8	sentiment	sentiment	NOUN
fbem-10171	89	9	words	word	NOUN
fbem-10171	89	10	present	present	ADJ
fbem-10171	89	11	in	in	ADP
fbem-10171	89	12	the	the	DET
fbem-10171	89	13	text	text	NOUN
fbem-10171	89	14	are	be	AUX
fbem-10171	89	15	identified	identify	VERB
fbem-10171	89	16	.	.	PUNCT
fbem-10171	90	1	the	the	DET
fbem-10171	90	2	sentiment	sentiment	NOUN
fbem-10171	90	3	score	score	NOUN
fbem-10171	90	4	of	of	ADP
fbem-10171	90	5	the	the	DET
fbem-10171	90	6	text	text	NOUN
fbem-10171	90	7	is	be	AUX
fbem-10171	90	8	calculated	calculate	VERB
fbem-10171	90	9	based	base	VERB
fbem-10171	90	10	on	on	ADP
fbem-10171	90	11	the	the	DET
fbem-10171	90	12	number	number	NOUN
fbem-10171	90	13	,	,	PUNCT
fbem-10171	90	14	position	position	NOUN
fbem-10171	90	15	and	and	CCONJ
fbem-10171	90	16	sentiment	sentiment	NOUN
fbem-10171	90	17	polarity	polarity	NOUN
fbem-10171	90	18	of	of	ADP
fbem-10171	90	19	the	the	DET
fbem-10171	90	20	sentiment	sentiment	NOUN
fbem-10171	90	21	words	word	NOUN
fbem-10171	90	22	.	.	PUNCT
fbem-10171	91	1	common	common	ADJ
fbem-10171	91	2	calculation	calculation	NOUN
fbem-10171	91	3	methods	method	NOUN
fbem-10171	91	4	include	include	VERB
fbem-10171	91	5	simple	simple	ADJ
fbem-10171	91	6	counting	counting	NOUN
fbem-10171	91	7	,	,	PUNCT
fbem-10171	91	8	weighted	weight	VERB
fbem-10171	91	9	counting	counting	NOUN
fbem-10171	91	10	and	and	CCONJ
fbem-10171	91	11	rule	rule	NOUN
fbem-10171	91	12	-	-	PUNCT
fbem-10171	91	13	based	base	VERB
fbem-10171	91	14	matching[32	matching[32	NOUN
fbem-10171	91	15	]	]	PUNCT
fbem-10171	91	16	.	.	PUNCT
fbem-10171	92	1	figure	figure	VERB
fbem-10171	92	2	3	3	NUM
fbem-10171	92	3	:	:	PUNCT
fbem-10171	92	4	flow	flow	VERB
fbem-10171	92	5	chart	chart	NOUN
fbem-10171	92	6	of	of	ADP
fbem-10171	92	7	sentiment	sentiment	NOUN
fbem-10171	92	8	analysis	analysis	NOUN
fbem-10171	92	9	method	method	NOUN
fbem-10171	92	10	based	base	VERB
fbem-10171	92	11	on	on	ADP
fbem-10171	92	12	sentiment	sentiment	NOUN
fbem-10171	92	13	dictionary	dictionary	ADJ
fbem-10171	92	14	rules	rule	NOUN
fbem-10171	92	15	can	can	AUX
fbem-10171	92	16	also	also	ADV
fbem-10171	92	17	be	be	AUX
fbem-10171	92	18	used	use	VERB
fbem-10171	92	19	to	to	PART
fbem-10171	92	20	handle	handle	VERB
fbem-10171	92	21	negation	negation	NOUN
fbem-10171	92	22	,	,	PUNCT
fbem-10171	92	23	degree	degree	NOUN
fbem-10171	92	24	adverbs	adverb	NOUN
fbem-10171	92	25	and	and	CCONJ
fbem-10171	92	26	context	context	NOUN
fbem-10171	92	27	dependencies	dependency	NOUN
fbem-10171	92	28	.	.	PUNCT
fbem-10171	93	1	negation	negation	NOUN
fbem-10171	93	2	words	word	NOUN
fbem-10171	93	3	can	can	AUX
fbem-10171	93	4	change	change	VERB
fbem-10171	93	5	the	the	DET
fbem-10171	93	6	sentiment	sentiment	NOUN
fbem-10171	93	7	polarity	polarity	NOUN
fbem-10171	93	8	of	of	ADP
fbem-10171	93	9	subsequent	subsequent	ADJ
fbem-10171	93	10	sentiment	sentiment	NOUN
fbem-10171	93	11	words	word	NOUN
fbem-10171	93	12	,	,	PUNCT
fbem-10171	93	13	and	and	CCONJ
fbem-10171	93	14	degree	degree	NOUN
fbem-10171	93	15	adverbs	adverb	NOUN
fbem-10171	93	16	can	can	AUX
fbem-10171	93	17	adjust	adjust	VERB
fbem-10171	93	18	the	the	DET
fbem-10171	93	19	weights	weight	NOUN
fbem-10171	93	20	of	of	ADP
fbem-10171	93	21	sentiment	sentiment	NOUN
fbem-10171	93	22	words	word	NOUN
fbem-10171	93	23	to	to	PART
fbem-10171	93	24	reflect	reflect	VERB
fbem-10171	93	25	sentiment	sentiment	NOUN
fbem-10171	93	26	intensity	intensity	NOUN
fbem-10171	93	27	.	.	PUNCT
fbem-10171	94	1	rule	rule	NOUN
fbem-10171	94	2	matching	matching	NOUN
fbem-10171	94	3	and	and	CCONJ
fbem-10171	94	4	sentiment	sentiment	NOUN
fbem-10171	94	5	adjustment	adjustment	NOUN
fbem-10171	94	6	can	can	AUX
fbem-10171	94	7	be	be	AUX
fbem-10171	94	8	performed	perform	VERB
fbem-10171	94	9	according	accord	VERB
fbem-10171	94	10	to	to	ADP
fbem-10171	94	11	the	the	DET
fbem-10171	94	12	context	context	NOUN
fbem-10171	94	13	dependency	dependency	NOUN
fbem-10171	94	14	of	of	ADP
fbem-10171	94	15	sentiment	sentiment	NOUN
fbem-10171	94	16	expressions	expression	NOUN
fbem-10171	94	17	.	.	PUNCT
fbem-10171	95	1	english	english	ADJ
fbem-10171	95	2	sentiment	sentiment	NOUN
fbem-10171	95	3	dictionaries	dictionary	NOUN
fbem-10171	95	4	are	be	AUX
fbem-10171	95	5	more	more	ADV
fbem-10171	95	6	maturely	maturely	ADV
fbem-10171	95	7	developed	develop	VERB
fbem-10171	95	8	,	,	PUNCT
fbem-10171	95	9	and	and	CCONJ
fbem-10171	95	10	the	the	DET
fbem-10171	95	11	common	common	ADJ
fbem-10171	95	12	ones	one	NOUN
fbem-10171	95	13	are	be	AUX
fbem-10171	95	14	sentiwordnet	sentiwordnet	NOUN
fbem-10171	95	15	and	and	CCONJ
fbem-10171	95	16	general	general	ADJ
fbem-10171	95	17	inquirer	inquirer	PROPN
fbem-10171	95	18	opinion	opinion	PROPN
fbem-10171	95	19	lexicon[33	lexicon[33	PROPN
fbem-10171	95	20	]	]	PUNCT
fbem-10171	95	21	.	.	PUNCT
fbem-10171	96	1	chinese	chinese	ADJ
fbem-10171	96	2	sentiment	sentiment	NOUN
fbem-10171	96	3	dictionaries	dictionary	NOUN
fbem-10171	96	4	include	include	VERB
fbem-10171	96	5	hownet	hownet	PROPN
fbem-10171	96	6	,	,	PUNCT
fbem-10171	96	7	a	a	DET
fbem-10171	96	8	zhiwang	zhiwang	PROPN
fbem-10171	96	9	dictionary	dictionary	PROPN
fbem-10171	96	10	,	,	PUNCT
fbem-10171	96	11	ntusd	ntusd	NOUN
fbem-10171	96	12	of	of	ADP
fbem-10171	96	13	national	national	PROPN
fbem-10171	96	14	taiwan	taiwan	PROPN
fbem-10171	96	15	university	university	PROPN
fbem-10171	96	16	and	and	CCONJ
fbem-10171	96	17	dalian	dalian	PROPN
fbem-10171	96	18	university	university	PROPN
fbem-10171	96	19	of	of	ADP
fbem-10171	96	20	technology	technology	NOUN
fbem-10171	96	21	's	's	PART
fbem-10171	96	22	chinese	chinese	ADJ
fbem-10171	96	23	sentiment	sentiment	NOUN
fbem-10171	96	24	vocabulary	vocabulary	NOUN
fbem-10171	96	25	ontology	ontology	NOUN
fbem-10171	96	26	library[1	library[1	PROPN
fbem-10171	96	27	]	]	X
fbem-10171	96	28	.	.	PUNCT
fbem-10171	97	1	the	the	DET
fbem-10171	97	2	method	method	NOUN
fbem-10171	97	3	based	base	VERB
fbem-10171	97	4	on	on	ADP
fbem-10171	97	5	sentiment	sentiment	NOUN
fbem-10171	97	6	lexicon	lexicon	NOUN
fbem-10171	97	7	and	and	CCONJ
fbem-10171	97	8	rules	rule	NOUN
fbem-10171	97	9	has	have	VERB
fbem-10171	97	10	the	the	DET
fbem-10171	97	11	advantages	advantage	NOUN
fbem-10171	97	12	of	of	ADP
fbem-10171	97	13	simplicity	simplicity	NOUN
fbem-10171	97	14	,	,	PUNCT
fbem-10171	97	15	intuitiveness	intuitiveness	NOUN
fbem-10171	97	16	and	and	CCONJ
fbem-10171	97	17	interpretability	interpretability	NOUN
fbem-10171	97	18	.	.	PUNCT
fbem-10171	98	1	it	it	PRON
fbem-10171	98	2	does	do	AUX
fbem-10171	98	3	not	not	PART
fbem-10171	98	4	require	require	VERB
fbem-10171	98	5	a	a	DET
fbem-10171	98	6	large	large	ADJ
fbem-10171	98	7	amount	amount	NOUN
fbem-10171	98	8	of	of	ADP
fbem-10171	98	9	training	training	NOUN
fbem-10171	98	10	data	datum	NOUN
fbem-10171	98	11	and	and	CCONJ
fbem-10171	98	12	is	be	AUX
fbem-10171	98	13	computationally	computationally	ADV
fbem-10171	98	14	efficient	efficient	ADJ
fbem-10171	98	15	.	.	PUNCT
fbem-10171	99	1	however	however	ADV
fbem-10171	99	2	,	,	PUNCT
fbem-10171	99	3	the	the	DET
fbem-10171	99	4	method	method	NOUN
fbem-10171	99	5	also	also	ADV
fbem-10171	99	6	has	have	VERB
fbem-10171	99	7	limitations	limitation	NOUN
fbem-10171	99	8	.	.	PUNCT
fbem-10171	100	1	the	the	DET
fbem-10171	100	2	quality	quality	NOUN
fbem-10171	100	3	and	and	CCONJ
fbem-10171	100	4	coverage	coverage	NOUN
fbem-10171	100	5	of	of	ADP
fbem-10171	100	6	sentiment	sentiment	NOUN
fbem-10171	100	7	dictionaries	dictionary	NOUN
fbem-10171	100	8	directly	directly	ADV
fbem-10171	100	9	affect	affect	VERB
fbem-10171	100	10	the	the	DET
fbem-10171	100	11	classification	classification	NOUN
fbem-10171	100	12	accuracy	accuracy	NOUN
fbem-10171	100	13	,	,	PUNCT
fbem-10171	100	14	and	and	CCONJ
fbem-10171	100	15	the	the	DET
fbem-10171	100	16	lack	lack	NOUN
fbem-10171	100	17	of	of	ADP
fbem-10171	100	18	domain	domain	NOUN
fbem-10171	100	19	-	-	PUNCT
fbem-10171	100	20	specific	specific	ADJ
fbem-10171	100	21	or	or	CCONJ
fbem-10171	100	22	new	new	ADJ
fbem-10171	100	23	vocabulary	vocabulary	ADJ
fbem-10171	100	24	sentiment	sentiment	NOUN
fbem-10171	100	25	information	information	NOUN
fbem-10171	100	26	may	may	AUX
fbem-10171	100	27	lead	lead	VERB
fbem-10171	100	28	to	to	ADP
fbem-10171	100	29	classification	classification	NOUN
fbem-10171	100	30	errors[34	errors[34	NOUN
fbem-10171	100	31	]	]	PUNCT
fbem-10171	100	32	.	.	PUNCT
fbem-10171	101	1	in	in	ADP
fbem-10171	101	2	addition	addition	NOUN
fbem-10171	101	3	,	,	PUNCT
fbem-10171	101	4	the	the	DET
fbem-10171	101	5	rule	rule	NOUN
fbem-10171	101	6	-	-	PUNCT
fbem-10171	101	7	based	base	VERB
fbem-10171	101	8	approach	approach	NOUN
fbem-10171	101	9	has	have	VERB
fbem-10171	101	10	difficulty	difficulty	NOUN
fbem-10171	101	11	capturing	capture	VERB
fbem-10171	101	12	complex	complex	ADJ
fbem-10171	101	13	semantic	semantic	ADJ
fbem-10171	101	14	and	and	CCONJ
fbem-10171	101	15	contextual	contextual	ADJ
fbem-10171	101	16	relationships	relationship	NOUN
fbem-10171	101	17	.	.	PUNCT
fbem-10171	102	1	2.3.2	2.3.2	X
fbem-10171	102	2	.	.	PUNCT
fbem-10171	102	3	traditional	traditional	ADJ
fbem-10171	102	4	machine	machine	NOUN
fbem-10171	102	5	learning	learning	NOUN
fbem-10171	102	6	based	base	VERB
fbem-10171	102	7	approach	approach	NOUN
fbem-10171	102	8	in	in	ADP
fbem-10171	102	9	sentiment	sentiment	NOUN
fbem-10171	102	10	analysis	analysis	NOUN
fbem-10171	102	11	,	,	PUNCT
fbem-10171	102	12	traditional	traditional	ADJ
fbem-10171	102	13	machine	machine	NOUN
fbem-10171	102	14	learning	learning	NOUN
fbem-10171	102	15	-	-	PUNCT
fbem-10171	102	16	based	base	VERB
fbem-10171	102	17	approaches	approach	NOUN
fbem-10171	102	18	use	use	VERB
fbem-10171	102	19	labeled	label	VERB
fbem-10171	102	20	sentiment	sentiment	NOUN
fbem-10171	102	21	datasets	dataset	NOUN
fbem-10171	102	22	for	for	ADP
fbem-10171	102	23	model	model	NOUN
fbem-10171	102	24	training	training	NOUN
fbem-10171	102	25	and	and	CCONJ
fbem-10171	102	26	prediction	prediction	NOUN
fbem-10171	102	27	.	.	PUNCT
fbem-10171	103	1	the	the	DET
fbem-10171	103	2	approach	approach	NOUN
fbem-10171	103	3	converts	convert	VERB
fbem-10171	103	4	text	text	NOUN
fbem-10171	103	5	features	feature	NOUN
fbem-10171	103	6	into	into	ADP
fbem-10171	103	7	numerical	numerical	ADJ
fbem-10171	103	8	features	feature	NOUN
fbem-10171	103	9	and	and	CCONJ
fbem-10171	103	10	performs	perform	VERB
fbem-10171	103	11	sentiment	sentiment	NOUN
fbem-10171	103	12	classification	classification	NOUN
fbem-10171	103	13	using	use	VERB
fbem-10171	103	14	classification	classification	NOUN
fbem-10171	103	15	algorithms	algorithm	NOUN
fbem-10171	103	16	,	,	PUNCT
fbem-10171	103	17	such	such	ADJ
fbem-10171	103	18	as	as	ADP
fbem-10171	103	19	plain	plain	ADJ
fbem-10171	103	20	bayes[35	bayes[35	NOUN
fbem-10171	103	21	]	]	PUNCT
fbem-10171	103	22	,	,	PUNCT
fbem-10171	103	23	support	support	VERB
fbem-10171	103	24	vector	vector	NOUN
fbem-10171	103	25	machines[36	machines[36	NOUN
fbem-10171	103	26	]	]	PUNCT
fbem-10171	103	27	,	,	PUNCT
fbem-10171	103	28	decision	decision	NOUN
fbem-10171	103	29	trees[37	trees[37	NOUN
fbem-10171	103	30	]	]	PUNCT
fbem-10171	103	31	and	and	CCONJ
fbem-10171	103	32	random	random	ADJ
fbem-10171	103	33	forests[38	forests[38	NOUN
fbem-10171	103	34	]	]	PUNCT
fbem-10171	103	35	.	.	PUNCT
fbem-10171	104	1	feature	feature	NOUN
fbem-10171	104	2	selection	selection	NOUN
fbem-10171	104	3	and	and	CCONJ
fbem-10171	104	4	model	model	NOUN
fbem-10171	104	5	tuning	tuning	NOUN
fbem-10171	104	6	are	be	AUX
fbem-10171	104	7	key	key	ADJ
fbem-10171	104	8	steps	step	NOUN
fbem-10171	104	9	,	,	PUNCT
fbem-10171	104	10	where	where	SCONJ
fbem-10171	104	11	important	important	ADJ
fbem-10171	104	12	features	feature	NOUN
fbem-10171	104	13	can	can	AUX
fbem-10171	104	14	be	be	AUX
fbem-10171	104	15	selected	select	VERB
fbem-10171	104	16	by	by	ADP
fbem-10171	104	17	information	information	NOUN
fbem-10171	104	18	gain	gain	NOUN
fbem-10171	104	19	and	and	CCONJ
fbem-10171	104	20	mutual	mutual	ADJ
fbem-10171	104	21	information	information	NOUN
fbem-10171	104	22	,	,	PUNCT
fbem-10171	104	23	and	and	CCONJ
fbem-10171	104	24	parameters	parameter	NOUN
fbem-10171	104	25	can	can	AUX
fbem-10171	104	26	be	be	AUX
fbem-10171	104	27	tuned	tune	VERB
fbem-10171	104	28	using	use	VERB
fbem-10171	104	29	cross	cross	NOUN
fbem-10171	104	30	-	-	ADJ
fbem-10171	104	31	validation	validation	ADJ
fbem-10171	104	32	,	,	PUNCT
fbem-10171	104	33	grid	grid	NOUN
fbem-10171	104	34	search	search	NOUN
fbem-10171	104	35	,	,	PUNCT
fbem-10171	104	36	etc	etc	X
fbem-10171	104	37	.	.	X
fbem-10171	104	38	sentiment	sentiment	NOUN
fbem-10171	104	39	analysis	analysis	NOUN
fbem-10171	104	40	methods	method	NOUN
fbem-10171	104	41	based	base	VERB
fbem-10171	104	42	on	on	ADP
fbem-10171	104	43	traditional	traditional	ADJ
fbem-10171	104	44	machine	machine	NOUN
fbem-10171	104	45	learning	learning	NOUN
fbem-10171	104	46	have	have	AUX
fbem-10171	104	47	the	the	DET
fbem-10171	104	48	advantages	advantage	NOUN
fbem-10171	104	49	of	of	ADP
fbem-10171	104	50	being	be	AUX
fbem-10171	104	51	explanatory	explanatory	ADJ
fbem-10171	104	52	and	and	CCONJ
fbem-10171	104	53	suitable	suitable	ADJ
fbem-10171	104	54	for	for	ADP
fbem-10171	104	55	small	small	ADJ
fbem-10171	104	56	-	-	PUNCT
fbem-10171	104	57	scale	scale	NOUN
fbem-10171	104	58	datasets	dataset	NOUN
fbem-10171	104	59	.	.	PUNCT
fbem-10171	105	1	they	they	PRON
fbem-10171	105	2	can	can	AUX
fbem-10171	105	3	provide	provide	VERB
fbem-10171	105	4	explanatory	explanatory	ADJ
fbem-10171	105	5	relationships	relationship	NOUN
fbem-10171	105	6	between	between	ADP
fbem-10171	105	7	features	feature	NOUN
fbem-10171	105	8	and	and	CCONJ
fbem-10171	105	9	classification	classification	NOUN
fbem-10171	105	10	results	result	NOUN
fbem-10171	105	11	,	,	PUNCT
fbem-10171	105	12	and	and	CCONJ
fbem-10171	105	13	can	can	AUX
fbem-10171	105	14	incorporate	incorporate	VERB
fbem-10171	105	15	domain	domain	NOUN
fbem-10171	105	16	knowledge	knowledge	NOUN
fbem-10171	105	17	for	for	ADP
fbem-10171	105	18	feature	feature	NOUN
fbem-10171	105	19	engineering	engineering	NOUN
fbem-10171	105	20	.	.	PUNCT
fbem-10171	106	1	however	however	ADV
fbem-10171	106	2	,	,	PUNCT
fbem-10171	106	3	the	the	DET
fbem-10171	106	4	performance	performance	NOUN
fbem-10171	106	5	of	of	ADP
fbem-10171	106	6	traditional	traditional	ADJ
fbem-10171	106	7	machine	machine	NOUN
fbem-10171	106	8	learning	learning	NOUN
fbem-10171	106	9	methods	method	NOUN
fbem-10171	106	10	may	may	AUX
fbem-10171	106	11	be	be	AUX
fbem-10171	106	12	limited	limit	VERB
fbem-10171	106	13	when	when	SCONJ
fbem-10171	106	14	dealing	deal	VERB
fbem-10171	106	15	with	with	ADP
fbem-10171	106	16	large	large	ADJ
fbem-10171	106	17	-	-	PUNCT
fbem-10171	106	18	scale	scale	NOUN
fbem-10171	106	19	datasets	dataset	NOUN
fbem-10171	106	20	and	and	CCONJ
fbem-10171	106	21	complex	complex	ADJ
fbem-10171	106	22	contexts	contexts	NOUN
fbem-10171	106	23	.	.	PUNCT
fbem-10171	107	1	in	in	ADP
fbem-10171	107	2	addition	addition	NOUN
fbem-10171	107	3	,	,	PUNCT
fbem-10171	107	4	feature	feature	NOUN
fbem-10171	107	5	engineering	engineering	NOUN
fbem-10171	107	6	requires	require	VERB
fbem-10171	107	7	manual	manual	ADJ
fbem-10171	107	8	involvement	involvement	NOUN
fbem-10171	107	9	and	and	CCONJ
fbem-10171	107	10	high	high	ADJ
fbem-10171	107	11	requirements	requirement	NOUN
fbem-10171	107	12	for	for	ADP
fbem-10171	107	13	domain	domain	NOUN
fbem-10171	107	14	knowledge	knowledge	NOUN
fbem-10171	107	15	.	.	PUNCT
fbem-10171	108	1	2.3.3	2.3.3	NUM
fbem-10171	108	2	.	.	X
fbem-10171	108	3	deep	deep	ADJ
fbem-10171	108	4	learning	learning	NOUN
fbem-10171	108	5	based	base	VERB
fbem-10171	108	6	sentiment	sentiment	NOUN
fbem-10171	108	7	analysis	analysis	NOUN
fbem-10171	108	8	method	method	NOUN
fbem-10171	108	9	deep	deep	ADJ
fbem-10171	108	10	learning	learning	NOUN
fbem-10171	108	11	-	-	PUNCT
fbem-10171	108	12	based	base	VERB
fbem-10171	108	13	sentiment	sentiment	NOUN
fbem-10171	108	14	analysis	analysis	NOUN
fbem-10171	108	15	methods	method	NOUN
fbem-10171	108	16	use	use	VERB
fbem-10171	108	17	multilayer	multilayer	ADJ
fbem-10171	108	18	neural	neural	ADJ
fbem-10171	108	19	networks	network	NOUN
fbem-10171	108	20	to	to	PART
fbem-10171	108	21	learn	learn	VERB
fbem-10171	108	22	abstract	abstract	ADJ
fbem-10171	108	23	feature	feature	NOUN
fbem-10171	108	24	representations	representation	NOUN
fbem-10171	108	25	.	.	PUNCT
fbem-10171	109	1	trained	train	VERB
fbem-10171	109	2	with	with	ADP
fbem-10171	109	3	large	large	ADJ
fbem-10171	109	4	-	-	PUNCT
fbem-10171	109	5	scale	scale	NOUN
fbem-10171	109	6	datasets	dataset	NOUN
fbem-10171	109	7	,	,	PUNCT
fbem-10171	109	8	these	these	DET
fbem-10171	109	9	methods	method	NOUN
fbem-10171	109	10	are	be	AUX
fbem-10171	109	11	able	able	ADJ
fbem-10171	109	12	to	to	PART
fbem-10171	109	13	automatically	automatically	ADV
fbem-10171	109	14	extract	extract	VERB
fbem-10171	109	15	key	key	ADJ
fbem-10171	109	16	semantic	semantic	ADJ
fbem-10171	109	17	features	feature	NOUN
fbem-10171	109	18	of	of	ADP
fbem-10171	109	19	text	text	NOUN
fbem-10171	109	20	and	and	CCONJ
fbem-10171	109	21	implement	implement	VERB
fbem-10171	109	22	sentiment	sentiment	NOUN
fbem-10171	109	23	classification	classification	NOUN
fbem-10171	109	24	tasks	task	NOUN
fbem-10171	109	25	.	.	PUNCT
fbem-10171	110	1	at	at	ADP
fbem-10171	110	2	the	the	DET
fbem-10171	110	3	core	core	NOUN
fbem-10171	110	4	of	of	ADP
fbem-10171	110	5	deep	deep	ADJ
fbem-10171	110	6	learning	learning	NOUN
fbem-10171	110	7	methods	method	NOUN
fbem-10171	110	8	are	be	AUX
fbem-10171	110	9	neural	neural	ADJ
fbem-10171	110	10	network	network	NOUN
fbem-10171	110	11	models	model	NOUN
fbem-10171	110	12	,	,	PUNCT
fbem-10171	110	13	as	as	SCONJ
fbem-10171	110	14	shown	show	VERB
fbem-10171	110	15	in	in	ADP
fbem-10171	110	16	figure	figure	NOUN
fbem-10171	110	17	4	4	NUM
fbem-10171	110	18	.	.	PUNCT
fbem-10171	111	1	in	in	ADP
fbem-10171	111	2	sentiment	sentiment	NOUN
fbem-10171	111	3	analysis	analysis	NOUN
fbem-10171	111	4	,	,	PUNCT
fbem-10171	111	5	commonly	commonly	ADV
fbem-10171	111	6	used	use	VERB
fbem-10171	111	7	neural	neural	ADJ
fbem-10171	111	8	network	network	NOUN
fbem-10171	111	9	models	model	NOUN
fbem-10171	111	10	include	include	VERB
fbem-10171	111	11	convolutional	convolutional	ADJ
fbem-10171	111	12	neural	neural	ADJ
fbem-10171	111	13	network	network	NOUN
fbem-10171	111	14	(	(	PUNCT
fbem-10171	111	15	cnn)[39	cnn)[39	NOUN
fbem-10171	111	16	]	]	PUNCT
fbem-10171	111	17	,	,	PUNCT
fbem-10171	111	18	recurrent	recurrent	ADJ
fbem-10171	111	19	neural	neural	ADJ
fbem-10171	111	20	network	network	NOUN
fbem-10171	111	21	(	(	PUNCT
fbem-10171	111	22	rnn	rnn	PROPN
fbem-10171	111	23	)	)	PUNCT
fbem-10171	111	24	and	and	CCONJ
fbem-10171	111	25	long	long	ADJ
fbem-10171	111	26	and	and	CCONJ
fbem-10171	111	27	short	short	ADJ
fbem-10171	111	28	term	term	NOUN
fbem-10171	111	29	memory	memory	NOUN
fbem-10171	111	30	network	network	NOUN
fbem-10171	111	31	(	(	PUNCT
fbem-10171	111	32	lstm)[40	lstm)[40	PROPN
fbem-10171	111	33	]	]	PUNCT
fbem-10171	111	34	,	,	PUNCT
fbem-10171	111	35	transformer[30	transformer[30	NOUN
fbem-10171	111	36	]	]	X
fbem-10171	111	37	,	,	PUNCT
fbem-10171	111	38	etc	etc	X
fbem-10171	111	39	.	.	X
fbem-10171	111	40	figure	figure	VERB
fbem-10171	111	41	4	4	NUM
fbem-10171	111	42	.	.	PUNCT
fbem-10171	111	43	neural	neural	ADJ
fbem-10171	111	44	network	network	PROPN
fbem-10171	111	45	model	model	PROPN
fbem-10171	111	46	cnn	cnn	PROPN
fbem-10171	111	47	(	(	PUNCT
fbem-10171	111	48	convolutional	convolutional	ADJ
fbem-10171	111	49	neural	neural	ADJ
fbem-10171	111	50	network	network	NOUN
fbem-10171	111	51	)	)	PUNCT
fbem-10171	111	52	suffers	suffer	VERB
fbem-10171	111	53	from	from	ADP
fbem-10171	111	54	the	the	DET
fbem-10171	111	55	lack	lack	NOUN
fbem-10171	111	56	of	of	ADP
fbem-10171	111	57	sequence	sequence	NOUN
fbem-10171	111	58	modeling	modeling	NOUN
fbem-10171	111	59	capability	capability	NOUN
fbem-10171	111	60	in	in	ADP
fbem-10171	111	61	sentiment	sentiment	NOUN
fbem-10171	111	62	analysis	analysis	NOUN
fbem-10171	111	63	61	61	NUM
fbem-10171	111	64	and	and	CCONJ
fbem-10171	111	65	has	have	AUX
fbem-10171	111	66	limited	limit	VERB
fbem-10171	111	67	ability	ability	NOUN
fbem-10171	111	68	to	to	PART
fbem-10171	111	69	model	model	VERB
fbem-10171	111	70	text	text	NOUN
fbem-10171	111	71	sequences	sequence	NOUN
fbem-10171	111	72	due	due	ADP
fbem-10171	111	73	to	to	ADP
fbem-10171	111	74	the	the	DET
fbem-10171	111	75	limitation	limitation	NOUN
fbem-10171	111	76	of	of	ADP
fbem-10171	111	77	local	local	ADJ
fbem-10171	111	78	perceptual	perceptual	ADJ
fbem-10171	111	79	fields	field	NOUN
fbem-10171	111	80	and	and	CCONJ
fbem-10171	111	81	weight	weight	NOUN
fbem-10171	111	82	sharing	sharing	NOUN
fbem-10171	111	83	to	to	PART
fbem-10171	111	84	capture	capture	VERB
fbem-10171	111	85	long	long	ADJ
fbem-10171	111	86	-	-	PUNCT
fbem-10171	111	87	range	range	NOUN
fbem-10171	111	88	dependencies	dependency	NOUN
fbem-10171	111	89	.	.	PUNCT
fbem-10171	112	1	figure	figure	NOUN
fbem-10171	112	2	5	5	NUM
fbem-10171	112	3	.	.	PUNCT
fbem-10171	112	4	convolutional	convolutional	ADJ
fbem-10171	112	5	neural	neural	ADJ
fbem-10171	112	6	networks	network	NOUN
fbem-10171	112	7	on	on	ADP
fbem-10171	112	8	the	the	DET
fbem-10171	112	9	other	other	ADJ
fbem-10171	112	10	hand	hand	NOUN
fbem-10171	112	11	,	,	PUNCT
fbem-10171	112	12	rnn	rnn	VERB
fbem-10171	112	13	(	(	PUNCT
fbem-10171	112	14	recurrent	recurrent	ADJ
fbem-10171	112	15	neural	neural	ADJ
fbem-10171	112	16	network	network	NOUN
fbem-10171	112	17	)	)	PUNCT
fbem-10171	112	18	is	be	AUX
fbem-10171	112	19	able	able	ADJ
fbem-10171	112	20	to	to	PART
fbem-10171	112	21	retain	retain	VERB
fbem-10171	112	22	contextual	contextual	ADJ
fbem-10171	112	23	information	information	NOUN
fbem-10171	112	24	and	and	CCONJ
fbem-10171	112	25	model	model	NOUN
fbem-10171	112	26	sequence	sequence	NOUN
fbem-10171	112	27	dependencies	dependency	NOUN
fbem-10171	112	28	,	,	PUNCT
fbem-10171	112	29	but	but	CCONJ
fbem-10171	112	30	is	be	AUX
fbem-10171	112	31	prone	prone	ADJ
fbem-10171	112	32	to	to	ADP
fbem-10171	112	33	gradient	gradient	VERB
fbem-10171	112	34	disappearance	disappearance	NOUN
fbem-10171	112	35	and	and	CCONJ
fbem-10171	112	36	gradient	gradient	ADJ
fbem-10171	112	37	explosion	explosion	NOUN
fbem-10171	112	38	problems	problem	NOUN
fbem-10171	112	39	,	,	PUNCT
fbem-10171	112	40	and	and	CCONJ
fbem-10171	112	41	has	have	VERB
fbem-10171	112	42	limited	limit	VERB
fbem-10171	112	43	memory	memory	NOUN
fbem-10171	112	44	capacity	capacity	NOUN
fbem-10171	112	45	when	when	SCONJ
fbem-10171	112	46	dealing	deal	VERB
fbem-10171	112	47	with	with	ADP
fbem-10171	112	48	long	long	ADJ
fbem-10171	112	49	sequences	sequence	NOUN
fbem-10171	112	50	,	,	PUNCT
fbem-10171	112	51	which	which	PRON
fbem-10171	112	52	affects	affect	VERB
fbem-10171	112	53	the	the	DET
fbem-10171	112	54	accuracy	accuracy	NOUN
fbem-10171	112	55	of	of	ADP
fbem-10171	112	56	sentiment	sentiment	NOUN
fbem-10171	112	57	classification[41	classification[41	PROPN
fbem-10171	112	58	]	]	PUNCT
fbem-10171	112	59	.	.	PUNCT
fbem-10171	113	1	figure	figure	NOUN
fbem-10171	113	2	6	6	NUM
fbem-10171	113	3	.	.	PUNCT
fbem-10171	114	1	recurrent	recurrent	ADJ
fbem-10171	114	2	neural	neural	ADJ
fbem-10171	114	3	networks	network	NOUN
fbem-10171	114	4	to	to	PART
fbem-10171	114	5	overcome	overcome	VERB
fbem-10171	114	6	these	these	DET
fbem-10171	114	7	problems	problem	NOUN
fbem-10171	114	8	,	,	PUNCT
fbem-10171	114	9	lstm	lstm	NOUN
fbem-10171	114	10	(	(	PUNCT
fbem-10171	114	11	long	long	ADJ
fbem-10171	114	12	short	short	ADJ
fbem-10171	114	13	-	-	PUNCT
fbem-10171	114	14	term	term	NOUN
fbem-10171	114	15	memory	memory	NOUN
fbem-10171	114	16	network	network	NOUN
fbem-10171	114	17	)	)	PUNCT
fbem-10171	114	18	introduces	introduce	VERB
fbem-10171	114	19	a	a	DET
fbem-10171	114	20	gating	gate	VERB
fbem-10171	114	21	mechanism	mechanism	NOUN
fbem-10171	114	22	and	and	CCONJ
fbem-10171	114	23	memory	memory	NOUN
fbem-10171	114	24	units	unit	NOUN
fbem-10171	114	25	to	to	PART
fbem-10171	114	26	solve	solve	VERB
fbem-10171	114	27	the	the	DET
fbem-10171	114	28	gradient	gradient	ADJ
fbem-10171	114	29	problem	problem	NOUN
fbem-10171	114	30	and	and	CCONJ
fbem-10171	114	31	long	long	ADJ
fbem-10171	114	32	-	-	PUNCT
fbem-10171	114	33	term	term	NOUN
fbem-10171	114	34	dependency	dependency	NOUN
fbem-10171	114	35	problem	problem	NOUN
fbem-10171	114	36	and	and	CCONJ
fbem-10171	114	37	improve	improve	VERB
fbem-10171	114	38	the	the	DET
fbem-10171	114	39	performance	performance	NOUN
fbem-10171	114	40	of	of	ADP
fbem-10171	114	41	sentiment	sentiment	NOUN
fbem-10171	114	42	analysis	analysis	NOUN
fbem-10171	114	43	.	.	PUNCT
fbem-10171	115	1	figure	figure	NOUN
fbem-10171	115	2	7	7	NUM
fbem-10171	115	3	.	.	PUNCT
fbem-10171	115	4	long	long	ADJ
fbem-10171	115	5	short	short	ADJ
fbem-10171	115	6	-	-	PUNCT
fbem-10171	115	7	term	term	NOUN
fbem-10171	115	8	memory	memory	NOUN
fbem-10171	115	9	networks	network	NOUN
fbem-10171	115	10	however	however	ADV
fbem-10171	115	11	,	,	PUNCT
fbem-10171	115	12	the	the	DET
fbem-10171	115	13	computational	computational	ADJ
fbem-10171	115	14	complexity	complexity	NOUN
fbem-10171	115	15	of	of	ADP
fbem-10171	115	16	lstm	lstm	NOUN
fbem-10171	115	17	is	be	AUX
fbem-10171	115	18	high	high	ADJ
fbem-10171	115	19	and	and	CCONJ
fbem-10171	115	20	still	still	ADV
fbem-10171	115	21	has	have	VERB
fbem-10171	115	22	limitations	limitation	NOUN
fbem-10171	115	23	for	for	ADP
fbem-10171	115	24	very	very	ADV
fbem-10171	115	25	long	long	ADJ
fbem-10171	115	26	text	text	NOUN
fbem-10171	115	27	sequences	sequence	NOUN
fbem-10171	115	28	.	.	PUNCT
fbem-10171	116	1	when	when	SCONJ
fbem-10171	116	2	lstm	lstm	NOUN
fbem-10171	116	3	is	be	AUX
fbem-10171	116	4	used	use	VERB
fbem-10171	116	5	for	for	ADP
fbem-10171	116	6	text	text	NOUN
fbem-10171	116	7	sequences	sequence	NOUN
fbem-10171	116	8	of	of	ADP
fbem-10171	116	9	more	more	ADJ
fbem-10171	116	10	than	than	ADP
fbem-10171	116	11	200	200	NUM
fbem-10171	116	12	words	word	NOUN
fbem-10171	116	13	,	,	PUNCT
fbem-10171	116	14	it	it	PRON
fbem-10171	116	15	still	still	ADV
fbem-10171	116	16	has	have	VERB
fbem-10171	116	17	more	more	ADV
fbem-10171	116	18	serious	serious	ADJ
fbem-10171	116	19	long	long	ADJ
fbem-10171	116	20	-	-	PUNCT
fbem-10171	116	21	term	term	NOUN
fbem-10171	116	22	since	since	SCONJ
fbem-10171	116	23	problem	problem	NOUN
fbem-10171	116	24	and	and	CCONJ
fbem-10171	116	25	generates	generate	VERB
fbem-10171	116	26	higher	high	ADJ
fbem-10171	116	27	errors	error	NOUN
fbem-10171	116	28	.	.	PUNCT
fbem-10171	117	1	transformer	transformer	NOUN
fbem-10171	117	2	model	model	NOUN
fbem-10171	117	3	is	be	AUX
fbem-10171	117	4	an	an	DET
fbem-10171	117	5	emerging	emerge	VERB
fbem-10171	117	6	model	model	NOUN
fbem-10171	117	7	after	after	ADP
fbem-10171	117	8	rnn	rnn	NOUN
fbem-10171	117	9	and	and	CCONJ
fbem-10171	117	10	ltsm	ltsm	PROPN
fbem-10171	117	11	,	,	PUNCT
fbem-10171	117	12	which	which	PRON
fbem-10171	117	13	is	be	AUX
fbem-10171	117	14	widely	widely	ADV
fbem-10171	117	15	used	use	VERB
fbem-10171	117	16	in	in	ADP
fbem-10171	117	17	text	text	NOUN
fbem-10171	117	18	sentiment	sentiment	NOUN
fbem-10171	117	19	analysis	analysis	NOUN
fbem-10171	117	20	.	.	PUNCT
fbem-10171	118	1	based	base	VERB
fbem-10171	118	2	on	on	ADP
fbem-10171	118	3	the	the	DET
fbem-10171	118	4	self	self	NOUN
fbem-10171	118	5	-	-	PUNCT
fbem-10171	118	6	attention	attention	NOUN
fbem-10171	118	7	mechanism	mechanism	NOUN
fbem-10171	118	8	and	and	CCONJ
fbem-10171	118	9	the	the	DET
fbem-10171	118	10	introduction	introduction	NOUN
fbem-10171	118	11	of	of	ADP
fbem-10171	118	12	parallel	parallel	ADJ
fbem-10171	118	13	computing	computing	NOUN
fbem-10171	118	14	capability	capability	NOUN
fbem-10171	118	15	,	,	PUNCT
fbem-10171	118	16	the	the	DET
fbem-10171	118	17	processing	processing	NOUN
fbem-10171	118	18	and	and	CCONJ
fbem-10171	118	19	understanding	understanding	NOUN
fbem-10171	118	20	of	of	ADP
fbem-10171	118	21	text	text	NOUN
fbem-10171	118	22	sequences	sequence	NOUN
fbem-10171	118	23	are	be	AUX
fbem-10171	118	24	more	more	ADV
fbem-10171	118	25	efficient	efficient	ADJ
fbem-10171	118	26	,	,	PUNCT
fbem-10171	118	27	while	while	SCONJ
fbem-10171	118	28	the	the	DET
fbem-10171	118	29	attention	attention	NOUN
fbem-10171	118	30	mechanism	mechanism	NOUN
fbem-10171	118	31	can	can	AUX
fbem-10171	118	32	solve	solve	VERB
fbem-10171	118	33	the	the	DET
fbem-10171	118	34	problem	problem	NOUN
fbem-10171	118	35	of	of	ADP
fbem-10171	118	36	long	long	ADJ
fbem-10171	118	37	-	-	PUNCT
fbem-10171	118	38	distance	distance	NOUN
fbem-10171	118	39	dependence	dependence	NOUN
fbem-10171	118	40	of	of	ADP
fbem-10171	118	41	sequences	sequence	NOUN
fbem-10171	118	42	.	.	PUNCT
fbem-10171	119	1	the	the	DET
fbem-10171	119	2	essence	essence	NOUN
fbem-10171	119	3	of	of	ADP
fbem-10171	119	4	the	the	DET
fbem-10171	119	5	attention	attention	NOUN
fbem-10171	119	6	mechanism	mechanism	NOUN
fbem-10171	119	7	is	be	AUX
fbem-10171	119	8	to	to	PART
fbem-10171	119	9	selectively	selectively	ADV
fbem-10171	119	10	filter	filter	VERB
fbem-10171	119	11	a	a	DET
fbem-10171	119	12	small	small	ADJ
fbem-10171	119	13	amount	amount	NOUN
fbem-10171	119	14	of	of	ADP
fbem-10171	119	15	important	important	ADJ
fbem-10171	119	16	information	information	NOUN
fbem-10171	119	17	from	from	ADP
fbem-10171	119	18	a	a	DET
fbem-10171	119	19	large	large	ADJ
fbem-10171	119	20	amount	amount	NOUN
fbem-10171	119	21	of	of	ADP
fbem-10171	119	22	information	information	NOUN
fbem-10171	119	23	and	and	CCONJ
fbem-10171	119	24	focus	focus	VERB
fbem-10171	119	25	on	on	ADP
fbem-10171	119	26	this	this	DET
fbem-10171	119	27	important	important	ADJ
fbem-10171	119	28	information	information	NOUN
fbem-10171	119	29	,	,	PUNCT
fbem-10171	119	30	ignoring	ignore	VERB
fbem-10171	119	31	the	the	DET
fbem-10171	119	32	unimportant	unimportant	ADJ
fbem-10171	119	33	information[42	information[42	NOUN
fbem-10171	119	34	]	]	PUNCT
fbem-10171	119	35	.	.	PUNCT
fbem-10171	120	1	by	by	ADP
fbem-10171	120	2	weighting	weight	VERB
fbem-10171	120	3	the	the	DET
fbem-10171	120	4	weights	weight	NOUN
fbem-10171	120	5	at	at	ADP
fbem-10171	120	6	different	different	ADJ
fbem-10171	120	7	positions	position	NOUN
fbem-10171	120	8	in	in	ADP
fbem-10171	120	9	the	the	DET
fbem-10171	120	10	input	input	NOUN
fbem-10171	120	11	sequence	sequence	NOUN
fbem-10171	120	12	,	,	PUNCT
fbem-10171	120	13	it	it	PRON
fbem-10171	120	14	enables	enable	VERB
fbem-10171	120	15	the	the	DET
fbem-10171	120	16	model	model	NOUN
fbem-10171	120	17	to	to	PART
fbem-10171	120	18	focus	focus	VERB
fbem-10171	120	19	more	more	ADV
fbem-10171	120	20	on	on	ADP
fbem-10171	120	21	the	the	DET
fbem-10171	120	22	important	important	ADJ
fbem-10171	120	23	parts	part	NOUN
fbem-10171	120	24	when	when	SCONJ
fbem-10171	120	25	processing	process	VERB
fbem-10171	120	26	the	the	DET
fbem-10171	120	27	sequence	sequence	NOUN
fbem-10171	120	28	data	datum	NOUN
fbem-10171	120	29	.	.	PUNCT
fbem-10171	121	1	its	its	PRON
fbem-10171	121	2	structure	structure	NOUN
fbem-10171	121	3	diagram	diagram	NOUN
fbem-10171	121	4	is	be	AUX
fbem-10171	121	5	shown	show	VERB
fbem-10171	121	6	in	in	ADP
fbem-10171	121	7	figure	figure	NOUN
fbem-10171	121	8	8	8	NUM
fbem-10171	121	9	.	.	PUNCT
fbem-10171	122	1	figure	figure	NOUN
fbem-10171	122	2	8	8	NUM
fbem-10171	122	3	.	.	PUNCT
fbem-10171	122	4	attention	attention	NOUN
fbem-10171	122	5	mechanism	mechanism	NOUN
fbem-10171	122	6	module	module	NOUN
fbem-10171	122	7	in	in	ADP
fbem-10171	122	8	the	the	DET
fbem-10171	122	9	transformer	transformer	NOUN
fbem-10171	122	10	model	model	NOUN
fbem-10171	122	11	,	,	PUNCT
fbem-10171	122	12	the	the	DET
fbem-10171	122	13	attention	attention	NOUN
fbem-10171	122	14	mechanism	mechanism	NOUN
fbem-10171	122	15	is	be	AUX
fbem-10171	122	16	applied	apply	VERB
fbem-10171	122	17	to	to	ADP
fbem-10171	122	18	the	the	DET
fbem-10171	122	19	self	self	NOUN
fbem-10171	122	20	-	-	PUNCT
fbem-10171	122	21	attention	attention	NOUN
fbem-10171	122	22	mechanism	mechanism	NOUN
fbem-10171	122	23	(	(	PUNCT
fbem-10171	122	24	self	self	NOUN
fbem-10171	122	25	-	-	PUNCT
fbem-10171	122	26	attention)[43	attention)[43	NUM
fbem-10171	122	27	]	]	PUNCT
fbem-10171	122	28	.	.	PUNCT
fbem-10171	123	1	the	the	DET
fbem-10171	123	2	self	self	NOUN
fbem-10171	123	3	-	-	PUNCT
fbem-10171	123	4	attention	attention	NOUN
fbem-10171	123	5	mechanism	mechanism	NOUN
fbem-10171	123	6	calculates	calculate	VERB
fbem-10171	123	7	the	the	DET
fbem-10171	123	8	interactions	interaction	NOUN
fbem-10171	123	9	of	of	ADP
fbem-10171	123	10	each	each	DET
fbem-10171	123	11	position	position	NOUN
fbem-10171	123	12	in	in	ADP
fbem-10171	123	13	the	the	DET
fbem-10171	123	14	sequence	sequence	NOUN
fbem-10171	123	15	with	with	ADP
fbem-10171	123	16	other	other	ADJ
fbem-10171	123	17	positions	position	NOUN
fbem-10171	123	18	and	and	CCONJ
fbem-10171	123	19	assigns	assign	VERB
fbem-10171	123	20	a	a	DET
fbem-10171	123	21	weight	weight	NOUN
fbem-10171	123	22	to	to	ADP
fbem-10171	123	23	each	each	DET
fbem-10171	123	24	position	position	NOUN
fbem-10171	123	25	based	base	VERB
fbem-10171	123	26	on	on	ADP
fbem-10171	123	27	the	the	DET
fbem-10171	123	28	similarity	similarity	NOUN
fbem-10171	123	29	.	.	PUNCT
fbem-10171	124	1	by	by	ADP
fbem-10171	124	2	weighting	weight	VERB
fbem-10171	124	3	and	and	CCONJ
fbem-10171	124	4	summing	sum	VERB
fbem-10171	124	5	the	the	DET
fbem-10171	124	6	attention	attention	NOUN
fbem-10171	124	7	weights	weight	NOUN
fbem-10171	124	8	of	of	ADP
fbem-10171	124	9	each	each	DET
fbem-10171	124	10	position	position	NOUN
fbem-10171	124	11	,	,	PUNCT
fbem-10171	124	12	the	the	DET
fbem-10171	124	13	contextual	contextual	ADJ
fbem-10171	124	14	representation	representation	NOUN
fbem-10171	124	15	of	of	ADP
fbem-10171	124	16	each	each	DET
fbem-10171	124	17	position	position	NOUN
fbem-10171	124	18	and	and	CCONJ
fbem-10171	124	19	the	the	DET
fbem-10171	124	20	degree	degree	NOUN
fbem-10171	124	21	of	of	ADP
fbem-10171	124	22	contribution	contribution	NOUN
fbem-10171	124	23	can	can	AUX
fbem-10171	124	24	be	be	AUX
fbem-10171	124	25	obtained	obtain	VERB
fbem-10171	124	26	.	.	PUNCT
fbem-10171	125	1	the	the	DET
fbem-10171	125	2	advantage	advantage	NOUN
fbem-10171	125	3	of	of	ADP
fbem-10171	125	4	this	this	DET
fbem-10171	125	5	attention	attention	NOUN
fbem-10171	125	6	mechanism	mechanism	NOUN
fbem-10171	125	7	is	be	AUX
fbem-10171	125	8	that	that	SCONJ
fbem-10171	125	9	it	it	PRON
fbem-10171	125	10	can	can	AUX
fbem-10171	125	11	fuse	fuse	VERB
fbem-10171	125	12	information	information	NOUN
fbem-10171	125	13	from	from	ADP
fbem-10171	125	14	different	different	ADJ
fbem-10171	125	15	positions	position	NOUN
fbem-10171	125	16	in	in	ADP
fbem-10171	125	17	the	the	DET
fbem-10171	125	18	sequence	sequence	NOUN
fbem-10171	125	19	to	to	PART
fbem-10171	125	20	establish	establish	VERB
fbem-10171	125	21	global	global	ADJ
fbem-10171	125	22	semantic	semantic	ADJ
fbem-10171	125	23	associations	association	NOUN
fbem-10171	125	24	and	and	CCONJ
fbem-10171	125	25	thus	thus	ADV
fbem-10171	125	26	better	well	ADV
fbem-10171	125	27	understand	understand	VERB
fbem-10171	125	28	the	the	DET
fbem-10171	125	29	semantics	semantic	NOUN
fbem-10171	125	30	and	and	CCONJ
fbem-10171	125	31	structure	structure	NOUN
fbem-10171	125	32	of	of	ADP
fbem-10171	125	33	the	the	DET
fbem-10171	125	34	whole	whole	ADJ
fbem-10171	125	35	sequence	sequence	NOUN
fbem-10171	125	36	.	.	PUNCT
fbem-10171	126	1	figure	figure	NOUN
fbem-10171	126	2	9	9	NUM
fbem-10171	126	3	.	.	PUNCT
fbem-10171	126	4	self	self	NOUN
fbem-10171	126	5	-	-	PUNCT
fbem-10171	126	6	attention	attention	NOUN
fbem-10171	126	7	mechanism	mechanism	NOUN
fbem-10171	126	8	module	module	NOUN
fbem-10171	126	9	the	the	DET
fbem-10171	126	10	difference	difference	NOUN
fbem-10171	126	11	between	between	ADP
fbem-10171	126	12	the	the	DET
fbem-10171	126	13	self	self	NOUN
fbem-10171	126	14	-	-	PUNCT
fbem-10171	126	15	attention	attention	NOUN
fbem-10171	126	16	mechanism	mechanism	NOUN
fbem-10171	126	17	and	and	CCONJ
fbem-10171	126	18	the	the	DET
fbem-10171	126	19	attention	attention	NOUN
fbem-10171	126	20	mechanism	mechanism	NOUN
fbem-10171	126	21	lies	lie	VERB
fbem-10171	126	22	in	in	ADP
fbem-10171	126	23	the	the	DET
fbem-10171	126	24	fact	fact	NOUN
fbem-10171	126	25	that	that	SCONJ
fbem-10171	126	26	q	q	X
fbem-10171	126	27	,	,	PUNCT
fbem-10171	126	28	k	k	NOUN
fbem-10171	126	29	,	,	PUNCT
fbem-10171	126	30	and	and	CCONJ
fbem-10171	126	31	v	v	NOUN
fbem-10171	126	32	of	of	ADP
fbem-10171	126	33	the	the	DET
fbem-10171	126	34	self	self	NOUN
fbem-10171	126	35	-	-	PUNCT
fbem-10171	126	36	attention	attention	NOUN
fbem-10171	126	37	mechanism	mechanism	NOUN
fbem-10171	126	38	are	be	AUX
fbem-10171	126	39	homologous	homologous	ADJ
fbem-10171	126	40	,	,	PUNCT
fbem-10171	126	41	i.e.	i.e.	X
fbem-10171	126	42	,	,	PUNCT
fbem-10171	126	43	they	they	PRON
fbem-10171	126	44	all	all	PRON
fbem-10171	126	45	come	come	VERB
fbem-10171	126	46	from	from	ADP
fbem-10171	126	47	the	the	DET
fbem-10171	126	48	same	same	ADJ
fbem-10171	126	49	word	word	NOUN
fbem-10171	126	50	vector	vector	NOUN
fbem-10171	126	51	.	.	PUNCT
fbem-10171	127	1	multi	multi	ADJ
fbem-10171	127	2	-	-	ADJ
fbem-10171	127	3	head	head	ADJ
fbem-10171	127	4	self	self	NOUN
fbem-10171	127	5	attention[44	attention[44	PROPN
fbem-10171	127	6	]	]	PUNCT
fbem-10171	127	7	is	be	AUX
fbem-10171	127	8	an	an	DET
fbem-10171	127	9	extended	extended	ADJ
fbem-10171	127	10	self	self	NOUN
fbem-10171	127	11	-	-	PUNCT
fbem-10171	127	12	attentive	attentive	ADJ
fbem-10171	127	13	mechanism	mechanism	NOUN
fbem-10171	127	14	model	model	NOUN
fbem-10171	127	15	in	in	ADP
fbem-10171	127	16	transformer	transformer	NOUN
fbem-10171	127	17	,	,	PUNCT
fbem-10171	127	18	i.e.	i.e.	X
fbem-10171	127	19	,	,	PUNCT
fbem-10171	127	20	the	the	DET
fbem-10171	127	21	superposition	superposition	NOUN
fbem-10171	127	22	of	of	ADP
fbem-10171	127	23	62	62	NUM
fbem-10171	127	24	multiple	multiple	ADJ
fbem-10171	127	25	multi	multi	ADJ
fbem-10171	127	26	-	-	ADJ
fbem-10171	127	27	head	head	ADJ
fbem-10171	127	28	self	self	NOUN
fbem-10171	127	29	-	-	PUNCT
fbem-10171	127	30	attentive	attentive	ADJ
fbem-10171	127	31	mechanisms	mechanism	NOUN
fbem-10171	127	32	constitutes	constitute	VERB
fbem-10171	127	33	the	the	DET
fbem-10171	127	34	encoder	encoder	NOUN
fbem-10171	127	35	-	-	PUNCT
fbem-10171	127	36	decoder	decoder	NOUN
fbem-10171	127	37	structure	structure	NOUN
fbem-10171	127	38	in	in	ADP
fbem-10171	127	39	transformer	transformer	NOUN
fbem-10171	127	40	.	.	PUNCT
fbem-10171	128	1	it	it	PRON
fbem-10171	128	2	slices	slice	VERB
fbem-10171	128	3	the	the	DET
fbem-10171	128	4	input	input	NOUN
fbem-10171	128	5	into	into	ADP
fbem-10171	128	6	eight	eight	NUM
fbem-10171	128	7	independent	independent	ADJ
fbem-10171	128	8	attention	attention	NOUN
fbem-10171	128	9	heads	head	NOUN
fbem-10171	128	10	to	to	PART
fbem-10171	128	11	compute	compute	VERB
fbem-10171	128	12	the	the	DET
fbem-10171	128	13	selfattention	selfattention	NOUN
fbem-10171	128	14	in	in	ADP
fbem-10171	128	15	parallel	parallel	NOUN
fbem-10171	128	16	,	,	PUNCT
fbem-10171	128	17	each	each	PRON
fbem-10171	128	18	with	with	ADP
fbem-10171	128	19	different	different	ADJ
fbem-10171	128	20	k	k	PROPN
fbem-10171	128	21	,	,	PUNCT
fbem-10171	128	22	v	v	NOUN
fbem-10171	128	23	,	,	PUNCT
fbem-10171	128	24	and	and	CCONJ
fbem-10171	128	25	q	q	X
fbem-10171	128	26	,	,	PUNCT
fbem-10171	128	27	which	which	PRON
fbem-10171	128	28	is	be	AUX
fbem-10171	128	29	equivalent	equivalent	ADJ
fbem-10171	128	30	to	to	ADP
fbem-10171	128	31	eight	eight	NUM
fbem-10171	128	32	linear	linear	ADJ
fbem-10171	128	33	transformations	transformation	NOUN
fbem-10171	128	34	of	of	ADP
fbem-10171	128	35	the	the	DET
fbem-10171	128	36	original	original	ADJ
fbem-10171	128	37	self	self	NOUN
fbem-10171	128	38	-	-	PUNCT
fbem-10171	128	39	attention	attention	NOUN
fbem-10171	128	40	,	,	PUNCT
fbem-10171	128	41	so	so	SCONJ
fbem-10171	128	42	that	that	SCONJ
fbem-10171	128	43	different	different	ADJ
fbem-10171	128	44	representation	representation	NOUN
fbem-10171	128	45	spaces	space	NOUN
fbem-10171	128	46	can	can	AUX
fbem-10171	128	47	be	be	AUX
fbem-10171	128	48	learned	learn	VERB
fbem-10171	128	49	.	.	PUNCT
fbem-10171	129	1	transformer[30	transformer[30	NOUN
fbem-10171	129	2	]	]	PUNCT
fbem-10171	129	3	,	,	PUNCT
fbem-10171	129	4	proposed	propose	VERB
fbem-10171	129	5	by	by	ADP
fbem-10171	129	6	vaswani	vaswani	PROPN
fbem-10171	129	7	et	et	PROPN
fbem-10171	129	8	al	al	PROPN
fbem-10171	129	9	.	.	PROPN
fbem-10171	129	10	is	be	AUX
fbem-10171	129	11	a	a	DET
fbem-10171	129	12	neural	neural	ADJ
fbem-10171	129	13	network	network	NOUN
fbem-10171	129	14	integrated	integrate	VERB
fbem-10171	129	15	with	with	ADP
fbem-10171	129	16	a	a	DET
fbem-10171	129	17	multi	multi	ADJ
fbem-10171	129	18	-	-	ADJ
fbem-10171	129	19	headed	headed	ADJ
fbem-10171	129	20	self	self	NOUN
fbem-10171	129	21	-	-	PUNCT
fbem-10171	129	22	attentive	attentive	ADJ
fbem-10171	129	23	mechanism	mechanism	NOUN
fbem-10171	129	24	.	.	PUNCT
fbem-10171	130	1	the	the	DET
fbem-10171	130	2	model	model	NOUN
fbem-10171	130	3	adopts	adopt	VERB
fbem-10171	130	4	an	an	DET
fbem-10171	130	5	encoder	encoder	NOUN
fbem-10171	130	6	-	-	PUNCT
fbem-10171	130	7	decoder	decoder	NOUN
fbem-10171	130	8	structure	structure	NOUN
fbem-10171	130	9	,	,	PUNCT
fbem-10171	130	10	which	which	PRON
fbem-10171	130	11	solves	solve	VERB
fbem-10171	130	12	the	the	DET
fbem-10171	130	13	problems	problem	NOUN
fbem-10171	130	14	such	such	ADJ
fbem-10171	130	15	as	as	ADP
fbem-10171	130	16	inability	inability	NOUN
fbem-10171	130	17	to	to	PART
fbem-10171	130	18	compute	compute	VERB
fbem-10171	130	19	in	in	ADP
fbem-10171	130	20	parallel	parallel	NOUN
fbem-10171	130	21	with	with	ADP
fbem-10171	130	22	long	long	ADJ
fbem-10171	130	23	distance	distance	NOUN
fbem-10171	130	24	dependence	dependence	NOUN
fbem-10171	130	25	in	in	ADP
fbem-10171	130	26	rnn	rnn	NOUN
fbem-10171	130	27	and	and	CCONJ
fbem-10171	130	28	ltsm	ltsm	NOUN
fbem-10171	130	29	.	.	PUNCT
fbem-10171	131	1	in	in	ADP
fbem-10171	131	2	sentiment	sentiment	NOUN
fbem-10171	131	3	analysis	analysis	NOUN
fbem-10171	131	4	tasks	task	NOUN
fbem-10171	131	5	,	,	PUNCT
fbem-10171	131	6	the	the	DET
fbem-10171	131	7	transformer	transformer	NOUN
fbem-10171	131	8	model	model	NOUN
fbem-10171	131	9	possesses	possess	VERB
fbem-10171	131	10	an	an	DET
fbem-10171	131	11	extremely	extremely	ADV
fbem-10171	131	12	powerful	powerful	ADJ
fbem-10171	131	13	context	context	NOUN
fbem-10171	131	14	modeling	modeling	NOUN
fbem-10171	131	15	capability[45	capability[45	NOUN
fbem-10171	131	16	]	]	X
fbem-10171	131	17	.	.	PUNCT
fbem-10171	132	1	it	it	PRON
fbem-10171	132	2	can	can	AUX
fbem-10171	132	3	encode	encode	VERB
fbem-10171	132	4	and	and	CCONJ
fbem-10171	132	5	decode	decode	VERB
fbem-10171	132	6	text	text	NOUN
fbem-10171	132	7	sequences	sequence	NOUN
fbem-10171	132	8	,	,	PUNCT
fbem-10171	132	9	capture	capture	VERB
fbem-10171	132	10	semantic	semantic	ADJ
fbem-10171	132	11	information	information	NOUN
fbem-10171	132	12	between	between	ADP
fbem-10171	132	13	keywords	keyword	NOUN
fbem-10171	132	14	and	and	CCONJ
fbem-10171	132	15	contexts	context	NOUN
fbem-10171	132	16	,	,	PUNCT
fbem-10171	132	17	and	and	CCONJ
fbem-10171	132	18	effectively	effectively	ADV
fbem-10171	132	19	construct	construct	VERB
fbem-10171	132	20	dependencies	dependency	NOUN
fbem-10171	132	21	in	in	ADP
fbem-10171	132	22	sequences	sequence	NOUN
fbem-10171	132	23	.	.	PUNCT
fbem-10171	133	1	the	the	DET
fbem-10171	133	2	overlay	overlay	NOUN
fbem-10171	133	3	computation	computation	NOUN
fbem-10171	133	4	by	by	ADP
fbem-10171	133	5	multi	multi	ADJ
fbem-10171	133	6	-	-	ADJ
fbem-10171	133	7	headed	headed	ADJ
fbem-10171	133	8	self	self	NOUN
fbem-10171	133	9	-	-	PUNCT
fbem-10171	133	10	attentive	attentive	ADJ
fbem-10171	133	11	mechanism	mechanism	NOUN
fbem-10171	133	12	enables	enable	VERB
fbem-10171	133	13	the	the	DET
fbem-10171	133	14	model	model	NOUN
fbem-10171	133	15	to	to	PART
fbem-10171	133	16	better	well	ADV
fbem-10171	133	17	understand	understand	VERB
fbem-10171	133	18	the	the	DET
fbem-10171	133	19	semantic	semantic	ADJ
fbem-10171	133	20	and	and	CCONJ
fbem-10171	133	21	sentiment	sentiment	NOUN
fbem-10171	133	22	expressions	expression	NOUN
fbem-10171	133	23	in	in	ADP
fbem-10171	133	24	sentences	sentence	NOUN
fbem-10171	133	25	and	and	CCONJ
fbem-10171	133	26	extract	extract	VERB
fbem-10171	133	27	the	the	DET
fbem-10171	133	28	features	feature	NOUN
fbem-10171	133	29	that	that	PRON
fbem-10171	133	30	contribute	contribute	VERB
fbem-10171	133	31	to	to	ADP
fbem-10171	133	32	sentiment	sentiment	NOUN
fbem-10171	133	33	classification	classification	NOUN
fbem-10171	133	34	.	.	PUNCT
fbem-10171	134	1	meanwhile	meanwhile	ADV
fbem-10171	134	2	transformer	transformer	NOUN
fbem-10171	134	3	introduces	introduce	NOUN
fbem-10171	134	4	operations	operation	NOUN
fbem-10171	134	5	such	such	ADJ
fbem-10171	134	6	as	as	ADP
fbem-10171	134	7	residual	residual	ADJ
fbem-10171	134	8	module	module	NOUN
fbem-10171	134	9	and	and	CCONJ
fbem-10171	134	10	batch	batch	NOUN
fbem-10171	134	11	normalization	normalization	NOUN
fbem-10171	134	12	,	,	PUNCT
fbem-10171	134	13	which	which	PRON
fbem-10171	134	14	effectively	effectively	ADV
fbem-10171	134	15	avoid	avoid	VERB
fbem-10171	134	16	the	the	DET
fbem-10171	134	17	problems	problem	NOUN
fbem-10171	134	18	of	of	ADP
fbem-10171	134	19	gradient	gradient	ADJ
fbem-10171	134	20	disappearance	disappearance	NOUN
fbem-10171	134	21	and	and	CCONJ
fbem-10171	134	22	gradient	gradient	NOUN
fbem-10171	134	23	explosion	explosion	NOUN
fbem-10171	134	24	during	during	ADP
fbem-10171	134	25	network	network	NOUN
fbem-10171	134	26	training	training	NOUN
fbem-10171	134	27	,	,	PUNCT
fbem-10171	134	28	while	while	SCONJ
fbem-10171	134	29	improving	improve	VERB
fbem-10171	134	30	the	the	DET
fbem-10171	134	31	training	training	NOUN
fbem-10171	134	32	efficiency	efficiency	NOUN
fbem-10171	134	33	.	.	PUNCT
fbem-10171	135	1	its	its	PRON
fbem-10171	135	2	schematic	schematic	ADJ
fbem-10171	135	3	diagram	diagram	NOUN
fbem-10171	135	4	is	be	AUX
fbem-10171	135	5	shown	show	VERB
fbem-10171	135	6	in	in	ADP
fbem-10171	135	7	figure	figure	NOUN
fbem-10171	135	8	10	10	NUM
fbem-10171	135	9	.	.	PUNCT
fbem-10171	136	1	figure	figure	NOUN
fbem-10171	136	2	10	10	NUM
fbem-10171	136	3	.	.	PUNCT
fbem-10171	137	1	transformer	transformer	NOUN
fbem-10171	137	2	module	module	NOUN
fbem-10171	137	3	however	however	ADV
fbem-10171	137	4	,	,	PUNCT
fbem-10171	137	5	deep	deep	ADJ
fbem-10171	137	6	learning	learning	NOUN
fbem-10171	137	7	-	-	PUNCT
fbem-10171	137	8	based	base	VERB
fbem-10171	137	9	sentiment	sentiment	NOUN
fbem-10171	137	10	analysis	analysis	NOUN
fbem-10171	137	11	methods	method	NOUN
fbem-10171	137	12	also	also	ADV
fbem-10171	137	13	have	have	VERB
fbem-10171	137	14	some	some	DET
fbem-10171	137	15	challenges	challenge	NOUN
fbem-10171	137	16	.	.	PUNCT
fbem-10171	138	1	first	first	ADV
fbem-10171	138	2	,	,	PUNCT
fbem-10171	138	3	deep	deep	ADJ
fbem-10171	138	4	learning	learning	NOUN
fbem-10171	138	5	methods	method	NOUN
fbem-10171	138	6	have	have	VERB
fbem-10171	138	7	a	a	DET
fbem-10171	138	8	high	high	ADJ
fbem-10171	138	9	demand	demand	NOUN
fbem-10171	138	10	for	for	ADP
fbem-10171	138	11	data	datum	NOUN
fbem-10171	138	12	and	and	CCONJ
fbem-10171	138	13	require	require	VERB
fbem-10171	138	14	a	a	DET
fbem-10171	138	15	large	large	ADJ
fbem-10171	138	16	amount	amount	NOUN
fbem-10171	138	17	of	of	ADP
fbem-10171	138	18	labeled	label	VERB
fbem-10171	138	19	data	datum	NOUN
fbem-10171	138	20	for	for	ADP
fbem-10171	138	21	model	model	NOUN
fbem-10171	138	22	training	training	NOUN
fbem-10171	138	23	.	.	PUNCT
fbem-10171	139	1	second	second	ADJ
fbem-10171	139	2	,	,	PUNCT
fbem-10171	139	3	deep	deep	ADJ
fbem-10171	139	4	learning	learning	NOUN
fbem-10171	139	5	models	model	NOUN
fbem-10171	139	6	tend	tend	VERB
fbem-10171	139	7	to	to	PART
fbem-10171	139	8	be	be	AUX
fbem-10171	139	9	more	more	ADV
fbem-10171	139	10	complex	complex	ADJ
fbem-10171	139	11	and	and	CCONJ
fbem-10171	139	12	require	require	VERB
fbem-10171	139	13	more	more	ADV
fbem-10171	139	14	computational	computational	ADJ
fbem-10171	139	15	resources	resource	NOUN
fbem-10171	139	16	for	for	ADP
fbem-10171	139	17	training	training	NOUN
fbem-10171	139	18	and	and	CCONJ
fbem-10171	139	19	inference	inference	NOUN
fbem-10171	139	20	,	,	PUNCT
fbem-10171	139	21	and	and	CCONJ
fbem-10171	139	22	also	also	ADV
fbem-10171	139	23	lack	lack	VERB
fbem-10171	139	24	interpretability	interpretability	NOUN
fbem-10171	139	25	to	to	PART
fbem-10171	139	26	explain	explain	VERB
fbem-10171	139	27	the	the	DET
fbem-10171	139	28	prediction	prediction	NOUN
fbem-10171	139	29	results	result	NOUN
fbem-10171	139	30	of	of	ADP
fbem-10171	139	31	the	the	DET
fbem-10171	139	32	models	model	NOUN
fbem-10171	139	33	.	.	PUNCT
fbem-10171	140	1	in	in	ADP
fbem-10171	140	2	summary	summary	NOUN
fbem-10171	140	3	,	,	PUNCT
fbem-10171	140	4	sentiment	sentiment	NOUN
fbem-10171	140	5	lexicon	lexicon	NOUN
fbem-10171	140	6	and	and	CCONJ
fbem-10171	140	7	rule	rule	NOUN
fbem-10171	140	8	-	-	PUNCT
fbem-10171	140	9	based	base	VERB
fbem-10171	140	10	methods	method	NOUN
fbem-10171	140	11	are	be	AUX
fbem-10171	140	12	simple	simple	ADJ
fbem-10171	140	13	and	and	CCONJ
fbem-10171	140	14	intuitive	intuitive	ADJ
fbem-10171	140	15	,	,	PUNCT
fbem-10171	140	16	and	and	CCONJ
fbem-10171	140	17	are	be	AUX
fbem-10171	140	18	suitable	suitable	ADJ
fbem-10171	140	19	for	for	ADP
fbem-10171	140	20	small	small	ADJ
fbem-10171	140	21	-	-	PUNCT
fbem-10171	140	22	scale	scale	NOUN
fbem-10171	140	23	data	datum	NOUN
fbem-10171	140	24	and	and	CCONJ
fbem-10171	140	25	simple	simple	ADJ
fbem-10171	140	26	sentiment	sentiment	NOUN
fbem-10171	140	27	analysis	analysis	NOUN
fbem-10171	140	28	tasks	task	NOUN
fbem-10171	140	29	.	.	PUNCT
fbem-10171	141	1	traditional	traditional	ADJ
fbem-10171	141	2	machine	machine	NOUN
fbem-10171	141	3	learning	learning	NOUN
fbem-10171	141	4	methods	method	NOUN
fbem-10171	141	5	have	have	VERB
fbem-10171	141	6	good	good	ADJ
fbem-10171	141	7	explanations	explanation	NOUN
fbem-10171	141	8	and	and	CCONJ
fbem-10171	141	9	interpretability	interpretability	NOUN
fbem-10171	141	10	and	and	CCONJ
fbem-10171	141	11	perform	perform	VERB
fbem-10171	141	12	better	well	ADV
fbem-10171	141	13	on	on	ADP
fbem-10171	141	14	small	small	ADJ
fbem-10171	141	15	data	data	NOUN
fbem-10171	141	16	sets	set	NOUN
fbem-10171	141	17	.	.	PUNCT
fbem-10171	142	1	deep	deep	ADJ
fbem-10171	142	2	learning	learning	NOUN
fbem-10171	142	3	methods	method	NOUN
fbem-10171	142	4	have	have	AUX
fbem-10171	142	5	achieved	achieve	VERB
fbem-10171	142	6	remarkable	remarkable	ADJ
fbem-10171	142	7	results	result	NOUN
fbem-10171	142	8	on	on	ADP
fbem-10171	142	9	large	large	ADJ
fbem-10171	142	10	-	-	PUNCT
fbem-10171	142	11	scale	scale	NOUN
fbem-10171	142	12	data	datum	NOUN
fbem-10171	142	13	and	and	CCONJ
fbem-10171	142	14	complex	complex	ADJ
fbem-10171	142	15	contexts	context	NOUN
fbem-10171	142	16	by	by	ADP
fbem-10171	142	17	learning	learn	VERB
fbem-10171	142	18	abstract	abstract	ADJ
fbem-10171	142	19	semantic	semantic	ADJ
fbem-10171	142	20	feature	feature	NOUN
fbem-10171	142	21	representations	representation	NOUN
fbem-10171	142	22	.	.	PUNCT
fbem-10171	143	1	future	future	ADJ
fbem-10171	143	2	research	research	NOUN
fbem-10171	143	3	can	can	AUX
fbem-10171	143	4	explore	explore	VERB
fbem-10171	143	5	the	the	DET
fbem-10171	143	6	fusion	fusion	NOUN
fbem-10171	143	7	of	of	ADP
fbem-10171	143	8	different	different	ADJ
fbem-10171	143	9	methods	method	NOUN
fbem-10171	143	10	to	to	PART
fbem-10171	143	11	improve	improve	VERB
fbem-10171	143	12	the	the	DET
fbem-10171	143	13	accuracy	accuracy	NOUN
fbem-10171	143	14	and	and	CCONJ
fbem-10171	143	15	robustness	robustness	NOUN
fbem-10171	143	16	of	of	ADP
fbem-10171	143	17	sentiment	sentiment	NOUN
fbem-10171	143	18	classification	classification	NOUN
fbem-10171	143	19	and	and	CCONJ
fbem-10171	143	20	combine	combine	VERB
fbem-10171	143	21	domain	domain	NOUN
fbem-10171	143	22	knowledge	knowledge	NOUN
fbem-10171	143	23	and	and	CCONJ
fbem-10171	143	24	contextual	contextual	ADJ
fbem-10171	143	25	information	information	NOUN
fbem-10171	143	26	to	to	PART
fbem-10171	143	27	achieve	achieve	VERB
fbem-10171	143	28	more	more	ADV
fbem-10171	143	29	refined	refined	ADJ
fbem-10171	143	30	and	and	CCONJ
fbem-10171	143	31	personalized	personalized	ADJ
fbem-10171	143	32	sentiment	sentiment	NOUN
fbem-10171	143	33	analysis	analysis	NOUN
fbem-10171	143	34	applications	application	NOUN
fbem-10171	143	35	.	.	PUNCT
fbem-10171	144	1	3	3	X
fbem-10171	144	2	.	.	X
fbem-10171	144	3	text	text	NOUN
fbem-10171	144	4	sentiment	sentiment	NOUN
fbem-10171	144	5	analysis	analysis	NOUN
fbem-10171	144	6	application	application	NOUN
fbem-10171	144	7	areas	area	VERB
fbem-10171	144	8	the	the	DET
fbem-10171	144	9	large	large	ADJ
fbem-10171	144	10	amount	amount	NOUN
fbem-10171	144	11	of	of	ADP
fbem-10171	144	12	user	user	NOUN
fbem-10171	144	13	-	-	PUNCT
fbem-10171	144	14	generated	generate	VERB
fbem-10171	144	15	text	text	NOUN
fbem-10171	144	16	data	datum	NOUN
fbem-10171	144	17	on	on	ADP
fbem-10171	144	18	social	social	ADJ
fbem-10171	144	19	media	medium	NOUN
fbem-10171	144	20	platforms	platform	NOUN
fbem-10171	144	21	contains	contain	VERB
fbem-10171	144	22	rich	rich	ADJ
fbem-10171	144	23	sentiment	sentiment	NOUN
fbem-10171	144	24	information	information	NOUN
fbem-10171	144	25	,	,	PUNCT
fbem-10171	144	26	and	and	CCONJ
fbem-10171	144	27	sentiment	sentiment	NOUN
fbem-10171	144	28	analysis	analysis	NOUN
fbem-10171	144	29	can	can	AUX
fbem-10171	144	30	be	be	AUX
fbem-10171	144	31	used	use	VERB
fbem-10171	144	32	to	to	PART
fbem-10171	144	33	analyze	analyze	VERB
fbem-10171	144	34	users	user	NOUN
fbem-10171	144	35	'	'	PART
fbem-10171	144	36	sentiment	sentiment	NOUN
fbem-10171	144	37	tendencies	tendency	NOUN
fbem-10171	144	38	towards	towards	ADP
fbem-10171	144	39	specific	specific	ADJ
fbem-10171	144	40	events	event	NOUN
fbem-10171	144	41	,	,	PUNCT
fbem-10171	144	42	products	product	NOUN
fbem-10171	144	43	or	or	CCONJ
fbem-10171	144	44	topics	topic	NOUN
fbem-10171	144	45	,	,	PUNCT
fbem-10171	144	46	thus	thus	ADV
fbem-10171	144	47	providing	provide	VERB
fbem-10171	144	48	insights	insight	NOUN
fbem-10171	144	49	about	about	ADP
fbem-10171	144	50	users	user	NOUN
fbem-10171	144	51	'	'	PART
fbem-10171	144	52	emotional	emotional	ADJ
fbem-10171	144	53	attitudes	attitude	NOUN
fbem-10171	144	54	.	.	PUNCT
fbem-10171	145	1	for	for	ADP
fbem-10171	145	2	example	example	NOUN
fbem-10171	145	3	,	,	PUNCT
fbem-10171	145	4	in	in	ADP
fbem-10171	145	5	microblog	microblog	ADJ
fbem-10171	145	6	sentiment	sentiment	NOUN
fbem-10171	145	7	analysis	analysis	NOUN
fbem-10171	145	8	,	,	PUNCT
fbem-10171	145	9	based	base	VERB
fbem-10171	145	10	on	on	ADP
fbem-10171	145	11	sentiment	sentiment	NOUN
fbem-10171	145	12	lexicon	lexicon	NOUN
fbem-10171	145	13	,	,	PUNCT
fbem-10171	145	14	yuqing	yuqe	VERB
fbem-10171	145	15	li	li	PROPN
fbem-10171	145	16	et	et	PROPN
fbem-10171	145	17	al	al	PROPN
fbem-10171	145	18	.	.	PROPN
fbem-10171	145	19	proposed	propose	VERB
fbem-10171	145	20	a	a	DET
fbem-10171	145	21	bilingual	bilingual	ADJ
fbem-10171	145	22	lexicon	lexicon	ADJ
fbem-10171	145	23	sentiment	sentiment	NOUN
fbem-10171	145	24	analysis	analysis	NOUN
fbem-10171	145	25	method	method	NOUN
fbem-10171	145	26	based	base	VERB
fbem-10171	145	27	on	on	ADP
fbem-10171	145	28	knn	knn	PROPN
fbem-10171	145	29	algorithm[46	algorithm[46	PROPN
fbem-10171	145	30	]	]	PUNCT
fbem-10171	145	31	,	,	PUNCT
fbem-10171	145	32	and	and	CCONJ
fbem-10171	145	33	yanyan	yanyan	ADJ
fbem-10171	145	34	zhao	zhao	PROPN
fbem-10171	145	35	used	use	VERB
fbem-10171	145	36	the	the	DET
fbem-10171	145	37	huge	huge	ADJ
fbem-10171	145	38	data	datum	NOUN
fbem-10171	145	39	volume	volume	NOUN
fbem-10171	145	40	of	of	ADP
fbem-10171	145	41	microblogs	microblog	NOUN
fbem-10171	145	42	to	to	PART
fbem-10171	145	43	build	build	VERB
fbem-10171	145	44	a	a	DET
fbem-10171	145	45	mega	mega	ADJ
fbem-10171	145	46	sentiment	sentiment	NOUN
fbem-10171	145	47	lexicon[47	lexicon[47	PROPN
fbem-10171	145	48	]	]	PUNCT
fbem-10171	145	49	,	,	PUNCT
fbem-10171	145	50	which	which	PRON
fbem-10171	145	51	has	have	AUX
fbem-10171	145	52	significantly	significantly	ADV
fbem-10171	145	53	improved	improve	VERB
fbem-10171	145	54	the	the	DET
fbem-10171	145	55	effect	effect	NOUN
fbem-10171	145	56	of	of	ADP
fbem-10171	145	57	sentiment	sentiment	NOUN
fbem-10171	145	58	classification	classification	NOUN
fbem-10171	145	59	on	on	ADP
fbem-10171	145	60	microblogs	microblog	NOUN
fbem-10171	145	61	.	.	PUNCT
fbem-10171	146	1	sentiment	sentiment	NOUN
fbem-10171	146	2	analysis	analysis	NOUN
fbem-10171	146	3	can	can	AUX
fbem-10171	146	4	also	also	ADV
fbem-10171	146	5	help	help	VERB
fbem-10171	146	6	companies	company	NOUN
fbem-10171	146	7	understand	understand	VERB
fbem-10171	146	8	consumers	consumer	NOUN
fbem-10171	146	9	'	'	PART
fbem-10171	146	10	emotional	emotional	ADJ
fbem-10171	146	11	attitudes	attitude	NOUN
fbem-10171	146	12	and	and	CCONJ
fbem-10171	146	13	emotional	emotional	ADJ
fbem-10171	146	14	needs	need	NOUN
fbem-10171	146	15	towards	towards	ADP
fbem-10171	146	16	brands	brand	NOUN
fbem-10171	146	17	,	,	PUNCT
fbem-10171	146	18	so	so	SCONJ
fbem-10171	146	19	that	that	SCONJ
fbem-10171	146	20	they	they	PRON
fbem-10171	146	21	can	can	AUX
fbem-10171	146	22	develop	develop	VERB
fbem-10171	146	23	more	more	ADV
fbem-10171	146	24	targeted	target	VERB
fbem-10171	146	25	marketing	marketing	NOUN
fbem-10171	146	26	strategies	strategy	NOUN
fbem-10171	146	27	and	and	CCONJ
fbem-10171	146	28	brand	brand	NOUN
fbem-10171	146	29	management	management	NOUN
fbem-10171	146	30	strategies	strategy	NOUN
fbem-10171	146	31	.	.	PUNCT
fbem-10171	147	1	for	for	ADP
fbem-10171	147	2	example	example	NOUN
fbem-10171	147	3	,	,	PUNCT
fbem-10171	147	4	asghar[48	asghar[48	PROPN
fbem-10171	147	5	]	]	X
fbem-10171	147	6	et	et	PROPN
fbem-10171	147	7	al	al	PROPN
fbem-10171	147	8	.	.	PROPN
fbem-10171	147	9	produce	produce	VERB
fbem-10171	147	10	sentiment	sentiment	NOUN
fbem-10171	147	11	dictionaries	dictionary	NOUN
fbem-10171	147	12	integrating	integrate	VERB
fbem-10171	147	13	emoticons	emoticon	NOUN
fbem-10171	147	14	and	and	CCONJ
fbem-10171	147	15	domain	domain	NOUN
fbem-10171	147	16	terms	term	NOUN
fbem-10171	147	17	for	for	ADP
fbem-10171	147	18	hotel	hotel	NOUN
fbem-10171	147	19	review	review	NOUN
fbem-10171	147	20	data	datum	NOUN
fbem-10171	147	21	,	,	PUNCT
fbem-10171	147	22	and	and	CCONJ
fbem-10171	147	23	obtain	obtain	VERB
fbem-10171	147	24	hotel	hotel	NOUN
fbem-10171	147	25	review	review	NOUN
fbem-10171	147	26	sentiment	sentiment	NOUN
fbem-10171	147	27	for	for	ADP
fbem-10171	147	28	improving	improve	VERB
fbem-10171	147	29	hotel	hotel	NOUN
fbem-10171	147	30	facilities	facility	NOUN
fbem-10171	147	31	to	to	PART
fbem-10171	147	32	meet	meet	VERB
fbem-10171	147	33	people	people	NOUN
fbem-10171	147	34	's	's	PART
fbem-10171	147	35	needs	need	NOUN
fbem-10171	147	36	.	.	PUNCT
fbem-10171	148	1	min	min	PROPN
fbem-10171	148	2	peng[49	peng[49	PROPN
fbem-10171	148	3	]	]	X
fbem-10171	148	4	et	et	PROPN
fbem-10171	148	5	al	al	PROPN
fbem-10171	148	6	.	.	PUNCT
fbem-10171	148	7	use	use	VERB
fbem-10171	148	8	lda	lda	PROPN
fbem-10171	148	9	topic	topic	NOUN
fbem-10171	148	10	model	model	NOUN
fbem-10171	148	11	to	to	PART
fbem-10171	148	12	analyze	analyze	VERB
fbem-10171	148	13	potential	potential	ADJ
fbem-10171	148	14	attributes	attribute	NOUN
fbem-10171	148	15	of	of	ADP
fbem-10171	148	16	goods	good	NOUN
fbem-10171	148	17	and	and	CCONJ
fbem-10171	148	18	calculate	calculate	NOUN
fbem-10171	148	19	users	user	NOUN
fbem-10171	148	20	'	'	PART
fbem-10171	148	21	emotional	emotional	ADJ
fbem-10171	148	22	tendency	tendency	NOUN
fbem-10171	148	23	and	and	CCONJ
fbem-10171	148	24	user	user	NOUN
fbem-10171	148	25	similarity	similarity	NOUN
fbem-10171	148	26	to	to	ADP
fbem-10171	148	27	the	the	DET
fbem-10171	148	28	attributes	attribute	NOUN
fbem-10171	148	29	of	of	ADP
fbem-10171	148	30	goods	good	NOUN
fbem-10171	148	31	,	,	PUNCT
fbem-10171	148	32	which	which	PRON
fbem-10171	148	33	enables	enable	VERB
fbem-10171	148	34	merchants	merchant	NOUN
fbem-10171	148	35	to	to	PART
fbem-10171	148	36	better	well	ADV
fbem-10171	148	37	design	design	VERB
fbem-10171	148	38	more	more	ADV
fbem-10171	148	39	perfect	perfect	ADJ
fbem-10171	148	40	goods	good	NOUN
fbem-10171	148	41	according	accord	VERB
fbem-10171	148	42	to	to	ADP
fbem-10171	148	43	customers	customer	NOUN
fbem-10171	148	44	'	'	PART
fbem-10171	148	45	emotional	emotional	ADJ
fbem-10171	148	46	preferences	preference	NOUN
fbem-10171	148	47	.	.	PUNCT
fbem-10171	149	1	jinmeng	jinmeng	PROPN
fbem-10171	149	2	li	li	PROPN
fbem-10171	149	3	combines	combine	VERB
fbem-10171	149	4	sentiment	sentiment	NOUN
fbem-10171	149	5	analysis	analysis	NOUN
fbem-10171	149	6	algorithms	algorithm	NOUN
fbem-10171	149	7	with	with	ADP
fbem-10171	149	8	eyetracking	eyetracke	VERB
fbem-10171	149	9	devices[50	devices[50	NOUN
fbem-10171	149	10	]	]	PUNCT
fbem-10171	149	11	,	,	PUNCT
fbem-10171	149	12	where	where	SCONJ
fbem-10171	149	13	users	user	NOUN
fbem-10171	149	14	wear	wear	VERB
fbem-10171	149	15	eye	eye	NOUN
fbem-10171	149	16	-	-	PUNCT
fbem-10171	149	17	tracking	track	VERB
fbem-10171	149	18	devices	device	NOUN
fbem-10171	149	19	to	to	PART
fbem-10171	149	20	derive	derive	VERB
fbem-10171	149	21	the	the	DET
fbem-10171	149	22	area	area	NOUN
fbem-10171	149	23	of	of	ADP
fbem-10171	149	24	gaze	gaze	NOUN
fbem-10171	149	25	on	on	ADP
fbem-10171	149	26	the	the	DET
fbem-10171	149	27	text	text	NOUN
fbem-10171	149	28	,	,	PUNCT
fbem-10171	149	29	and	and	CCONJ
fbem-10171	149	30	finally	finally	ADV
fbem-10171	149	31	the	the	DET
fbem-10171	149	32	sentiment	sentiment	NOUN
fbem-10171	149	33	value	value	NOUN
fbem-10171	149	34	of	of	ADP
fbem-10171	149	35	the	the	DET
fbem-10171	149	36	user	user	NOUN
fbem-10171	149	37	's	's	PART
fbem-10171	149	38	area	area	NOUN
fbem-10171	149	39	of	of	ADP
fbem-10171	149	40	interest	interest	NOUN
fbem-10171	149	41	is	be	AUX
fbem-10171	149	42	derived	derive	VERB
fbem-10171	149	43	from	from	ADP
fbem-10171	149	44	the	the	DET
fbem-10171	149	45	algorithmic	algorithmic	ADJ
fbem-10171	149	46	model	model	NOUN
fbem-10171	149	47	to	to	PART
fbem-10171	149	48	determine	determine	VERB
fbem-10171	149	49	the	the	DET
fbem-10171	149	50	user	user	NOUN
fbem-10171	149	51	's	's	PART
fbem-10171	149	52	preference	preference	NOUN
fbem-10171	149	53	for	for	ADP
fbem-10171	149	54	the	the	DET
fbem-10171	149	55	content	content	NOUN
fbem-10171	149	56	.	.	PUNCT
fbem-10171	150	1	4	4	X
fbem-10171	150	2	.	.	X
fbem-10171	150	3	conclusion	conclusion	NOUN
fbem-10171	150	4	this	this	DET
fbem-10171	150	5	review	review	NOUN
fbem-10171	150	6	provides	provide	VERB
fbem-10171	150	7	a	a	DET
fbem-10171	150	8	comprehensive	comprehensive	ADJ
fbem-10171	150	9	overview	overview	NOUN
fbem-10171	150	10	of	of	ADP
fbem-10171	150	11	the	the	DET
fbem-10171	150	12	application	application	NOUN
fbem-10171	150	13	of	of	ADP
fbem-10171	150	14	natural	natural	ADJ
fbem-10171	150	15	language	language	NOUN
fbem-10171	150	16	processing	processing	NOUN
fbem-10171	150	17	in	in	ADP
fbem-10171	150	18	text	text	NOUN
fbem-10171	150	19	sentiment	sentiment	NOUN
fbem-10171	150	20	analysis	analysis	NOUN
fbem-10171	150	21	.	.	PUNCT
fbem-10171	151	1	sentiment	sentiment	NOUN
fbem-10171	151	2	features	feature	NOUN
fbem-10171	151	3	can	can	AUX
fbem-10171	151	4	be	be	AUX
fbem-10171	151	5	effectively	effectively	ADV
fbem-10171	151	6	extracted	extract	VERB
fbem-10171	151	7	from	from	ADP
fbem-10171	151	8	text	text	NOUN
fbem-10171	151	9	by	by	ADP
fbem-10171	151	10	text	text	NOUN
fbem-10171	151	11	pre	pre	ADJ
fbem-10171	151	12	-	-	ADJ
fbem-10171	151	13	processing	processing	ADJ
fbem-10171	151	14	techniques	technique	NOUN
fbem-10171	151	15	and	and	CCONJ
fbem-10171	151	16	feature	feature	NOUN
fbem-10171	151	17	extraction	extraction	NOUN
fbem-10171	151	18	methods	method	NOUN
fbem-10171	151	19	.	.	PUNCT
fbem-10171	152	1	different	different	ADJ
fbem-10171	152	2	sentiment	sentiment	NOUN
fbem-10171	152	3	classification	classification	NOUN
fbem-10171	152	4	algorithms	algorithm	NOUN
fbem-10171	152	5	and	and	CCONJ
fbem-10171	152	6	sentiment	sentiment	NOUN
fbem-10171	152	7	intensity	intensity	NOUN
fbem-10171	152	8	analysis	analysis	NOUN
fbem-10171	152	9	methods	method	NOUN
fbem-10171	152	10	can	can	AUX
fbem-10171	152	11	help	help	VERB
fbem-10171	152	12	to	to	PART
fbem-10171	152	13	understand	understand	VERB
fbem-10171	152	14	the	the	DET
fbem-10171	152	15	sentiment	sentiment	NOUN
fbem-10171	152	16	tendency	tendency	NOUN
fbem-10171	152	17	and	and	CCONJ
fbem-10171	152	18	intensity	intensity	NOUN
fbem-10171	152	19	in	in	ADP
fbem-10171	152	20	texts	text	NOUN
fbem-10171	152	21	.	.	PUNCT
fbem-10171	153	1	moreover	moreover	ADV
fbem-10171	153	2	,	,	PUNCT
fbem-10171	153	3	multilingual	multilingual	ADJ
fbem-10171	153	4	sentiment	sentiment	NOUN
fbem-10171	153	5	analysis	analysis	NOUN
fbem-10171	153	6	and	and	CCONJ
fbem-10171	153	7	case	case	NOUN
fbem-10171	153	8	studies	study	NOUN
fbem-10171	153	9	in	in	ADP
fbem-10171	153	10	different	different	ADJ
fbem-10171	153	11	application	application	NOUN
fbem-10171	153	12	domains	domain	NOUN
fbem-10171	153	13	demonstrate	demonstrate	VERB
fbem-10171	153	14	the	the	DET
fbem-10171	153	15	wide	wide	ADJ
fbem-10171	153	16	application	application	NOUN
fbem-10171	153	17	of	of	ADP
fbem-10171	153	18	sentiment	sentiment	NOUN
fbem-10171	153	19	analysis	analysis	NOUN
fbem-10171	153	20	.	.	PUNCT
fbem-10171	154	1	however	however	ADV
fbem-10171	154	2	,	,	PUNCT
fbem-10171	154	3	there	there	PRON
fbem-10171	154	4	are	be	VERB
fbem-10171	154	5	still	still	ADV
fbem-10171	154	6	challenges	challenge	NOUN
fbem-10171	154	7	such	such	ADJ
fbem-10171	154	8	as	as	ADP
fbem-10171	154	9	data	datum	NOUN
fbem-10171	154	10	scarcity	scarcity	NOUN
fbem-10171	154	11	,	,	PUNCT
fbem-10171	154	12	model	model	NOUN
fbem-10171	154	13	generalization	generalization	NOUN
fbem-10171	154	14	capability	capability	NOUN
fbem-10171	154	15	and	and	CCONJ
fbem-10171	154	16	interpretability	interpretability	NOUN
fbem-10171	154	17	that	that	PRON
fbem-10171	154	18	need	need	VERB
fbem-10171	154	19	to	to	PART
fbem-10171	154	20	be	be	AUX
fbem-10171	154	21	addressed	address	VERB
fbem-10171	154	22	.	.	PUNCT
fbem-10171	155	1	future	future	ADJ
fbem-10171	155	2	research	research	NOUN
fbem-10171	155	3	can	can	AUX
fbem-10171	155	4	combine	combine	VERB
fbem-10171	155	5	multimodal	multimodal	ADJ
fbem-10171	155	6	data	datum	NOUN
fbem-10171	155	7	and	and	CCONJ
fbem-10171	155	8	emerging	emerge	VERB
fbem-10171	155	9	technologies	technology	NOUN
fbem-10171	155	10	to	to	PART
fbem-10171	155	11	improve	improve	VERB
fbem-10171	155	12	the	the	DET
fbem-10171	155	13	accuracy	accuracy	NOUN
fbem-10171	155	14	and	and	CCONJ
fbem-10171	155	15	efficiency	efficiency	NOUN
fbem-10171	155	16	of	of	ADP
fbem-10171	155	17	sentiment	sentiment	NOUN
fbem-10171	155	18	analysis	analysis	NOUN
fbem-10171	155	19	and	and	CCONJ
fbem-10171	155	20	provide	provide	VERB
fbem-10171	155	21	more	more	ADV
fbem-10171	155	22	valuable	valuable	ADJ
fbem-10171	155	23	information	information	NOUN
fbem-10171	155	24	for	for	ADP
fbem-10171	155	25	decision	decision	NOUN
fbem-10171	155	26	making	making	NOUN
fbem-10171	155	27	in	in	ADP
fbem-10171	155	28	various	various	ADJ
fbem-10171	155	29	industries	industry	NOUN
fbem-10171	155	30	.	.	PUNCT
fbem-10171	156	1	acknowledgment	acknowledgment	NOUN
fbem-10171	156	2	fundamental	fundamental	ADJ
fbem-10171	156	3	project	project	NOUN
fbem-10171	156	4	:	:	PUNCT
fbem-10171	156	5	sichuan	sichuan	PROPN
fbem-10171	156	6	university	university	PROPN
fbem-10171	156	7	of	of	ADP
fbem-10171	156	8	science	science	PROPN
fbem-10171	156	9	&	&	CCONJ
fbem-10171	156	10	engineering	engineering	PROPN
fbem-10171	156	11	graduate	graduate	NOUN
fbem-10171	156	12	student	student	NOUN
fbem-10171	156	13	innovation	innovation	NOUN
fbem-10171	156	14	fund	fund	PROPN
fbem-10171	156	15	(	(	PUNCT
fbem-10171	156	16	y2022141	y2022141	PROPN
fbem-10171	156	17	)	)	PUNCT
fbem-10171	156	18	,	,	PUNCT
fbem-10171	156	19	design	design	NOUN
fbem-10171	156	20	and	and	CCONJ
fbem-10171	156	21	application	application	NOUN
fbem-10171	156	22	of	of	ADP
fbem-10171	156	23	text	text	NOUN
fbem-10171	156	24	sentiment	sentiment	NOUN
fbem-10171	156	25	analysis	analysis	NOUN
fbem-10171	156	26	model	model	NOUN
fbem-10171	156	27	based	base	VERB
fbem-10171	156	28	on	on	ADP
fbem-10171	156	29	eye	eye	NOUN
fbem-10171	156	30	tracking	track	VERB
fbem-10171	156	31	core+	core+	ADP
fbem-10171	156	32	eye	eye	NOUN
fbem-10171	156	33	movement	movement	NOUN
fbem-10171	156	34	instrument	instrument	NOUN
fbem-10171	156	35	.	.	PUNCT
fbem-10171	157	1	63	63	NUM
fbem-10171	157	2	references	reference	NOUN
fbem-10171	157	3	[	[	X
fbem-10171	157	4	1	1	NUM
fbem-10171	157	5	]	]	PUNCT
fbem-10171	157	6	zhong	zhong	PROPN
fbem-10171	157	7	jiawa	jiawa	PROPN
fbem-10171	157	8	,	,	PUNCT
fbem-10171	157	9	liu	liu	PROPN
fbem-10171	157	10	wei	wei	PROPN
fbem-10171	157	11	,	,	PUNCT
fbem-10171	157	12	wang	wang	PROPN
fbem-10171	157	13	sili	sili	PROPN
fbem-10171	157	14	et	et	PROPN
fbem-10171	157	15	al	al	PROPN
fbem-10171	157	16	.	.	PROPN
fbem-10171	157	17	review	review	NOUN
fbem-10171	157	18	of	of	ADP
fbem-10171	157	19	methods	method	NOUN
fbem-10171	157	20	and	and	CCONJ
fbem-10171	157	21	applications	application	NOUN
fbem-10171	157	22	of	of	ADP
fbem-10171	157	23	text	text	NOUN
fbem-10171	157	24	sentiment	sentiment	NOUN
fbem-10171	157	25	analysis[j	analysis[j	PROPN
fbem-10171	157	26	]	]	PUNCT
fbem-10171	157	27	.	.	PUNCT
fbem-10171	158	1	data	datum	NOUN
fbem-10171	158	2	analysis	analysis	NOUN
fbem-10171	158	3	and	and	CCONJ
fbem-10171	158	4	knowledge	knowledge	NOUN
fbem-10171	158	5	discovery,2021,5(06):1	discovery,2021,5(06):1	PROPN
fbem-10171	158	6	-	-	PUNCT
fbem-10171	158	7	13	13	NUM
fbem-10171	158	8	.	.	PUNCT
fbem-10171	159	1	[	[	X
fbem-10171	159	2	2	2	X
fbem-10171	159	3	]	]	X
fbem-10171	159	4	wang	wang	PROPN
fbem-10171	159	5	yj	yj	PROPN
fbem-10171	159	6	,	,	PUNCT
fbem-10171	159	7	zhu	zhu	PROPN
fbem-10171	159	8	jq	jq	PROPN
fbem-10171	159	9	,	,	PUNCT
fbem-10171	159	10	wang	wang	PROPN
fbem-10171	159	11	zm	zm	PROPN
fbem-10171	159	12	et	et	PROPN
fbem-10171	159	13	al	al	PROPN
fbem-10171	159	14	.	.	PROPN
fbem-10171	159	15	review	review	NOUN
fbem-10171	159	16	of	of	ADP
fbem-10171	159	17	applications	application	NOUN
fbem-10171	159	18	of	of	ADP
fbem-10171	159	19	natural	natural	ADJ
fbem-10171	159	20	language	language	NOUN
fbem-10171	159	21	processing	processing	NOUN
fbem-10171	159	22	in	in	ADP
fbem-10171	159	23	text	text	NOUN
fbem-10171	159	24	sentiment	sentiment	NOUN
fbem-10171	159	25	analysis	analysis	NOUN
fbem-10171	159	26	[	[	X
fbem-10171	159	27	j	j	X
fbem-10171	159	28	/	/	SYM
fbem-10171	159	29	ol	ol	PROPN
fbem-10171	159	30	]	]	PUNCT
fbem-10171	159	31	.	.	PUNCT
fbem-10171	160	1	computer	computer	NOUN
fbem-10171	160	2	applications:1	applications:1	NOUN
fbem-10171	160	3	-	-	SYM
fbem-10171	160	4	12	12	NUM
fbem-10171	161	1	[	[	X
fbem-10171	161	2	2023	2023	NUM
fbem-10171	161	3	-	-	SYM
fbem-10171	161	4	07	07	NUM
fbem-10171	161	5	-	-	PUNCT
fbem-10171	161	6	03	03	NUM
fbem-10171	161	7	]	]	PUNCT
fbem-10171	161	8	.	.	PUNCT
fbem-10171	162	1	[	[	X
fbem-10171	162	2	3	3	NUM
fbem-10171	162	3	]	]	X
fbem-10171	162	4	zhao	zhao	PROPN
fbem-10171	162	5	jingsheng	jingsheng	PROPN
fbem-10171	162	6	,	,	PUNCT
fbem-10171	162	7	song	song	NOUN
fbem-10171	162	8	mengxue	mengxue	NOUN
fbem-10171	162	9	,	,	PUNCT
fbem-10171	162	10	gao	gao	PROPN
fbem-10171	162	11	xiang	xiang	PROPN
fbem-10171	162	12	,	,	PUNCT
fbem-10171	162	13	et	et	PROPN
fbem-10171	162	14	al	al	PROPN
fbem-10171	162	15	.	.	PUNCT
fbem-10171	163	1	a	a	DET
fbem-10171	163	2	study	study	NOUN
fbem-10171	163	3	of	of	ADP
fbem-10171	163	4	text	text	NOUN
fbem-10171	163	5	representation	representation	NOUN
fbem-10171	163	6	in	in	ADP
fbem-10171	163	7	natural	natural	ADJ
fbem-10171	163	8	language	language	NOUN
fbem-10171	163	9	processing[j	processing[j	NOUN
fbem-10171	163	10	]	]	PUNCT
fbem-10171	163	11	.	.	PUNCT
fbem-10171	164	1	journal	journal	PROPN
fbem-10171	164	2	of	of	ADP
fbem-10171	164	3	software	software	NOUN
fbem-10171	164	4	,	,	PUNCT
fbem-10171	164	5	2021	2021	NUM
fbem-10171	164	6	,	,	PUNCT
fbem-10171	164	7	33(1	33(1	NUM
fbem-10171	164	8	):	):	PUNCT
fbem-10171	164	9	102	102	NUM
fbem-10171	164	10	-	-	SYM
fbem-10171	164	11	128	128	NUM
fbem-10171	164	12	.	.	PUNCT
fbem-10171	165	1	[	[	X
fbem-10171	165	2	4	4	NUM
fbem-10171	165	3	]	]	X
fbem-10171	165	4	li	li	PROPN
fbem-10171	165	5	yachao	yachao	PROPN
fbem-10171	165	6	,	,	PUNCT
fbem-10171	165	7	xiong	xiong	PROPN
fbem-10171	165	8	deyi	deyi	PROPN
fbem-10171	165	9	,	,	PUNCT
fbem-10171	165	10	zhang	zhang	PROPN
fbem-10171	165	11	min	min	PROPN
fbem-10171	165	12	.	.	PROPN
fbem-10171	166	1	a	a	DET
fbem-10171	166	2	review	review	NOUN
fbem-10171	166	3	of	of	ADP
fbem-10171	166	4	neural	neural	ADJ
fbem-10171	166	5	machine	machine	NOUN
fbem-10171	166	6	translation[j	translation[j	PROPN
fbem-10171	166	7	]	]	PUNCT
fbem-10171	166	8	.	.	PUNCT
fbem-10171	167	1	journal	journal	PROPN
fbem-10171	167	2	of	of	ADP
fbem-10171	167	3	computer	computer	NOUN
fbem-10171	167	4	science	science	NOUN
fbem-10171	167	5	,	,	PUNCT
fbem-10171	167	6	2018	2018	NUM
fbem-10171	167	7	,	,	PUNCT
fbem-10171	167	8	41(12	41(12	NUM
fbem-10171	167	9	):	):	PUNCT
fbem-10171	167	10	2734	2734	NUM
fbem-10171	167	11	-	-	SYM
fbem-10171	167	12	2755	2755	NUM
fbem-10171	167	13	.	.	PUNCT
fbem-10171	168	1	[	[	X
fbem-10171	168	2	5	5	NUM
fbem-10171	168	3	]	]	PUNCT
fbem-10171	168	4	ao	ao	PROPN
fbem-10171	168	5	sheng	sheng	PROPN
fbem-10171	168	6	,	,	PUNCT
fbem-10171	168	7	xu	xu	PROPN
fbem-10171	168	8	lan	lan	PROPN
fbem-10171	168	9	,	,	PUNCT
fbem-10171	168	10	ao	ao	PROPN
fbem-10171	168	11	qingwen.application	qingwen.application	NUM
fbem-10171	168	12	of	of	ADP
fbem-10171	168	13	nlp	nlp	ADJ
fbem-10171	168	14	chinese	chinese	ADJ
fbem-10171	168	15	word	word	NOUN
fbem-10171	168	16	separation	separation	NOUN
fbem-10171	168	17	technology	technology	NOUN
fbem-10171	168	18	in	in	ADP
fbem-10171	168	19	bridge	bridge	NOUN
fbem-10171	168	20	report	report	NOUN
fbem-10171	168	21	data	datum	NOUN
fbem-10171	168	22	processing[j	processing[j	NOUN
fbem-10171	168	23	]	]	PUNCT
fbem-10171	168	24	.	.	PUNCT
fbem-10171	169	1	traffic	traffic	NOUN
fbem-10171	169	2	world,2020(17):3	world,2020(17):3	PROPN
fbem-10171	169	3	-	-	PUNCT
fbem-10171	169	4	5.doi:10.16248	5.doi:10.16248	NUM
fbem-10171	169	5	/	/	SYM
fbem-10171	169	6	j.cnki.113723	j.cnki.113723	NOUN
fbem-10171	169	7	/	/	SYM
fbem-10171	169	8	u.2020.17.001	u.2020.17.001	ADJ
fbem-10171	169	9	.	.	PUNCT
fbem-10171	170	1	[	[	X
fbem-10171	170	2	6	6	NUM
fbem-10171	170	3	]	]	PUNCT
fbem-10171	170	4	wang	wang	PROPN
fbem-10171	170	5	yy	yy	PROPN
fbem-10171	170	6	,	,	PUNCT
fbem-10171	170	7	dong	dong	PROPN
fbem-10171	170	8	gw	gw	PROPN
fbem-10171	170	9	.	.	PUNCT
fbem-10171	170	10	research	research	NOUN
fbem-10171	170	11	on	on	ADP
fbem-10171	170	12	the	the	DET
fbem-10171	170	13	problems	problem	NOUN
fbem-10171	170	14	and	and	CCONJ
fbem-10171	170	15	countermeasures	countermeasure	NOUN
fbem-10171	170	16	of	of	ADP
fbem-10171	170	17	netroots	netroot	NOUN
fbem-10171	170	18	libraries	library	NOUN
fbem-10171	170	19	based	base	VERB
fbem-10171	170	20	on	on	ADP
fbem-10171	170	21	jieba	jieba	PROPN
fbem-10171	170	22	subtext[j	subtext[j	PROPN
fbem-10171	170	23	]	]	PUNCT
fbem-10171	170	24	.	.	PUNCT
fbem-10171	171	1	jiangsu	jiangsu	PROPN
fbem-10171	171	2	science	science	PROPN
fbem-10171	171	3	and	and	CCONJ
fbem-10171	171	4	technology	technology	NOUN
fbem-10171	171	5	information,2023,40(06):3538	information,2023,40(06):3538	NOUN
fbem-10171	171	6	.	.	PUNCT
fbem-10171	172	1	[	[	X
fbem-10171	172	2	7	7	X
fbem-10171	172	3	]	]	X
fbem-10171	172	4	yin	yin	PROPN
fbem-10171	172	5	r	r	PROPN
fbem-10171	172	6	,	,	PUNCT
fbem-10171	172	7	wang	wang	PROPN
fbem-10171	172	8	q	q	PROPN
fbem-10171	172	9	,	,	PUNCT
fbem-10171	172	10	li	li	PROPN
fbem-10171	172	11	p	p	PROPN
fbem-10171	172	12	,	,	PUNCT
fbem-10171	172	13	et	et	PROPN
fbem-10171	172	14	al	al	PROPN
fbem-10171	172	15	.	.	PUNCT
fbem-10171	172	16	multi	multi	ADJ
fbem-10171	172	17	-	-	ADJ
fbem-10171	172	18	granularity	granularity	ADJ
fbem-10171	172	19	chinese	chinese	ADJ
fbem-10171	172	20	word	word	NOUN
fbem-10171	173	1	embedding[c]//proceedings	embedding[c]//proceeding	NOUN
fbem-10171	173	2	of	of	ADP
fbem-10171	173	3	the	the	DET
fbem-10171	173	4	2016	2016	NUM
fbem-10171	173	5	conference	conference	NOUN
fbem-10171	173	6	on	on	ADP
fbem-10171	173	7	empirical	empirical	ADJ
fbem-10171	173	8	methods	method	NOUN
fbem-10171	173	9	in	in	ADP
fbem-10171	173	10	natural	natural	ADJ
fbem-10171	173	11	language	language	NOUN
fbem-10171	173	12	processing	processing	NOUN
fbem-10171	173	13	.	.	PUNCT
fbem-10171	174	1	2016	2016	NUM
fbem-10171	174	2	:	:	PUNCT
fbem-10171	174	3	981986	981986	NUM
fbem-10171	174	4	.	.	PUNCT
fbem-10171	175	1	[	[	X
fbem-10171	175	2	8	8	NUM
fbem-10171	175	3	]	]	X
fbem-10171	175	4	zhao	zhao	PROPN
fbem-10171	175	5	pengfei	pengfei	PROPN
fbem-10171	175	6	,	,	PUNCT
fbem-10171	175	7	zhao	zhao	PROPN
fbem-10171	175	8	chunjiang	chunjiang	PROPN
fbem-10171	175	9	,	,	PUNCT
fbem-10171	175	10	wu	wu	PROPN
fbem-10171	175	11	huarui	huarui	PROPN
fbem-10171	175	12	et	et	PROPN
fbem-10171	175	13	al	al	PROPN
fbem-10171	175	14	.	.	PUNCT
fbem-10171	176	1	bert	bert	PROPN
fbem-10171	176	2	-	-	PUNCT
fbem-10171	176	3	based	base	VERB
fbem-10171	176	4	multi	multi	ADJ
fbem-10171	176	5	-	-	ADJ
fbem-10171	176	6	feature	feature	ADJ
fbem-10171	176	7	fusion	fusion	NOUN
fbem-10171	176	8	for	for	ADP
fbem-10171	176	9	named	name	VERB
fbem-10171	176	10	entity	entity	NOUN
fbem-10171	176	11	recognition	recognition	NOUN
fbem-10171	176	12	in	in	ADP
fbem-10171	176	13	agriculture[j	agriculture[j	NOUN
fbem-10171	176	14	]	]	PUNCT
fbem-10171	176	15	.	.	PUNCT
fbem-10171	177	1	journal	journal	PROPN
fbem-10171	177	2	of	of	ADP
fbem-10171	177	3	agricultural	agricultural	ADJ
fbem-10171	177	4	engineering	engineering	NOUN
fbem-10171	177	5	,	,	PUNCT
fbem-10171	177	6	2022	2022	NUM
fbem-10171	177	7	,	,	PUNCT
fbem-10171	177	8	38(03	38(03	NUM
fbem-10171	177	9	):	):	PUNCT
fbem-10171	177	10	112	112	NUM
fbem-10171	177	11	-	-	SYM
fbem-10171	177	12	118	118	NUM
fbem-10171	177	13	.	.	PUNCT
fbem-10171	178	1	[	[	X
fbem-10171	178	2	9	9	NUM
fbem-10171	178	3	]	]	X
fbem-10171	178	4	chen	chen	PROPN
fbem-10171	178	5	zhiyan	zhiyan	PROPN
fbem-10171	178	6	,	,	PUNCT
fbem-10171	178	7	li	li	PROPN
fbem-10171	178	8	xiaojie	xiaojie	PROPN
fbem-10171	178	9	,	,	PUNCT
fbem-10171	178	10	zhu	zhu	PROPN
fbem-10171	178	11	shuhua	shuhua	PROPN
fbem-10171	178	12	et	et	PROPN
fbem-10171	178	13	al	al	PROPN
fbem-10171	178	14	.	.	PUNCT
fbem-10171	179	1	a	a	DET
fbem-10171	179	2	two	two	NUM
fbem-10171	179	3	-	-	PUNCT
fbem-10171	179	4	way	way	NOUN
fbem-10171	179	5	maximal	maximal	ADJ
fbem-10171	179	6	matching	matching	NOUN
fbem-10171	179	7	word	word	NOUN
fbem-10171	179	8	splitting	splitting	NOUN
fbem-10171	179	9	method	method	NOUN
fbem-10171	179	10	based	base	VERB
fbem-10171	179	11	on	on	ADP
fbem-10171	179	12	hash	hash	NOUN
fbem-10171	179	13	structured	structure	VERB
fbem-10171	179	14	dictionary[j	dictionary[j	NOUN
fbem-10171	179	15	]	]	X
fbem-10171	179	16	.	.	PUNCT
fbem-10171	180	1	computer	computer	NOUN
fbem-10171	180	2	science,2015,42(s2):49	science,2015,42(s2):49	PROPN
fbem-10171	180	3	-	-	NUM
fbem-10171	180	4	54	54	NUM
fbem-10171	180	5	.	.	PUNCT
fbem-10171	181	1	[	[	X
fbem-10171	181	2	10	10	NUM
fbem-10171	181	3	]	]	X
fbem-10171	181	4	li	li	PROPN
fbem-10171	181	5	ship	ship	NOUN
fbem-10171	181	6	.	.	PUNCT
fbem-10171	182	1	statistics	statistic	NOUN
fbem-10171	182	2	in	in	ADP
fbem-10171	182	3	chinese	chinese	ADJ
fbem-10171	182	4	word	word	NOUN
fbem-10171	182	5	splitting[j	splitting[j	PROPN
fbem-10171	182	6	]	]	PUNCT
fbem-10171	182	7	.	.	PUNCT
fbem-10171	183	1	china	china	PROPN
fbem-10171	183	2	statistics	statistics	PROPN
fbem-10171	183	3	,	,	PUNCT
fbem-10171	183	4	2020(10):34	2020(10):34	NUM
fbem-10171	183	5	-	-	SYM
fbem-10171	183	6	35	35	NUM
fbem-10171	183	7	.	.	PUNCT
fbem-10171	184	1	[	[	X
fbem-10171	184	2	11	11	NUM
fbem-10171	184	3	]	]	X
fbem-10171	184	4	liu	liu	PROPN
fbem-10171	184	5	,	,	PUNCT
fbem-10171	184	6	x.y.y	x.y.y	PROPN
fbem-10171	184	7	.	.	PROPN
fbem-10171	184	8	,	,	PUNCT
fbem-10171	184	9	sheng	sheng	PROPN
fbem-10171	184	10	,	,	PUNCT
fbem-10171	184	11	y.h	y.h	PROPN
fbem-10171	184	12	.	.	PROPN
fbem-10171	184	13	,	,	PUNCT
fbem-10171	184	14	qin	qin	PROPN
fbem-10171	184	15	,	,	PUNCT
fbem-10171	184	16	j.r	j.r	PROPN
fbem-10171	184	17	.	.	PROPN
fbem-10171	184	18	,	,	PUNCT
fbem-10171	184	19	et	et	PROPN
fbem-10171	184	20	al	al	PROPN
fbem-10171	184	21	.	.	PUNCT
fbem-10171	185	1	a	a	DET
fbem-10171	185	2	semantic	semantic	ADJ
fbem-10171	185	3	matching	matching	NOUN
fbem-10171	185	4	method	method	NOUN
fbem-10171	185	5	for	for	ADP
fbem-10171	185	6	spatio	spatio	PROPN
fbem-10171	185	7	-	-	PUNCT
fbem-10171	185	8	temporal	temporal	ADJ
fbem-10171	185	9	trajectories	trajectory	NOUN
fbem-10171	185	10	based	base	VERB
fbem-10171	185	11	on	on	ADP
fbem-10171	185	12	hidden	hidden	ADJ
fbem-10171	185	13	markov	markov	NOUN
fbem-10171	185	14	model[j	model[j	PROPN
fbem-10171	185	15	]	]	PUNCT
fbem-10171	185	16	.	.	PUNCT
fbem-10171	186	1	geography	geography	NOUN
fbem-10171	186	2	and	and	CCONJ
fbem-10171	186	3	geographic	geographic	ADJ
fbem-10171	186	4	information	information	NOUN
fbem-10171	186	5	science,2023,39(03):1	science,2023,39(03):1	PROPN
fbem-10171	186	6	-	-	PUNCT
fbem-10171	186	7	6	6	NUM
fbem-10171	186	8	.	.	PUNCT
fbem-10171	187	1	[	[	X
fbem-10171	187	2	12	12	NUM
fbem-10171	187	3	]	]	X
fbem-10171	187	4	liu	liu	PROPN
fbem-10171	187	5	,	,	PUNCT
fbem-10171	187	6	yunzhong	yunzhong	PROPN
fbem-10171	187	7	,	,	PUNCT
fbem-10171	187	8	lin	lin	PROPN
fbem-10171	187	9	,	,	PUNCT
fbem-10171	187	10	yaping	yaping	PROPN
fbem-10171	187	11	,	,	PUNCT
fbem-10171	187	12	chen	chen	PROPN
fbem-10171	187	13	,	,	PUNCT
fbem-10171	187	14	zhiping	zhipe	VERB
fbem-10171	187	15	.	.	PUNCT
fbem-10171	188	1	hidden	hide	VERB
fbem-10171	188	2	markov	markov	NOUN
fbem-10171	188	3	model	model	NOUN
fbem-10171	188	4	-	-	PUNCT
fbem-10171	188	5	based	base	VERB
fbem-10171	188	6	text	text	NOUN
fbem-10171	188	7	information	information	NOUN
fbem-10171	188	8	extraction[j	extraction[j	PROPN
fbem-10171	188	9	]	]	PUNCT
fbem-10171	188	10	.	.	PUNCT
fbem-10171	189	1	journal	journal	PROPN
fbem-10171	189	2	of	of	ADP
fbem-10171	189	3	system	system	NOUN
fbem-10171	189	4	simulation	simulation	NOUN
fbem-10171	189	5	,	,	PUNCT
fbem-10171	189	6	2004	2004	NUM
fbem-10171	189	7	,	,	PUNCT
fbem-10171	189	8	16(3	16(3	NUM
fbem-10171	189	9	):	):	PUNCT
fbem-10171	189	10	507	507	NUM
fbem-10171	189	11	-	-	SYM
fbem-10171	189	12	510	510	NUM
fbem-10171	189	13	.	.	PUNCT
fbem-10171	190	1	[	[	X
fbem-10171	190	2	13	13	NUM
fbem-10171	190	3	]	]	SYM
fbem-10171	190	4	li	li	PROPN
fbem-10171	190	5	ronglu	ronglu	PROPN
fbem-10171	190	6	,	,	PUNCT
fbem-10171	190	7	wang	wang	PROPN
fbem-10171	190	8	jianhui	jianhui	PROPN
fbem-10171	190	9	,	,	PUNCT
fbem-10171	190	10	chen	chen	PROPN
fbem-10171	190	11	xiaoyun	xiaoyun	PROPN
fbem-10171	190	12	,	,	PUNCT
fbem-10171	190	13	et	et	PROPN
fbem-10171	190	14	al	al	PROPN
fbem-10171	190	15	.	.	PUNCT
fbem-10171	191	1	chinese	chinese	ADJ
fbem-10171	191	2	text	text	NOUN
fbem-10171	191	3	classification	classification	NOUN
fbem-10171	191	4	using	use	VERB
fbem-10171	191	5	maximum	maximum	ADJ
fbem-10171	191	6	entropy	entropy	NOUN
fbem-10171	191	7	model[d	model[d	PROPN
fbem-10171	191	8	]	]	PUNCT
fbem-10171	191	9	.	.	PUNCT
fbem-10171	192	1	,	,	PUNCT
fbem-10171	192	2	2005	2005	NUM
fbem-10171	192	3	.	.	PUNCT
fbem-10171	193	1	[	[	X
fbem-10171	193	2	14	14	NUM
fbem-10171	193	3	]	]	X
fbem-10171	193	4	zhou	zhou	PROPN
fbem-10171	193	5	j.s	j.s	PROPN
fbem-10171	193	6	.	.	PROPN
fbem-10171	193	7	,	,	PUNCT
fbem-10171	193	8	dai	dai	PROPN
fbem-10171	193	9	x.y	x.y	PROPN
fbem-10171	193	10	.	.	PROPN
fbem-10171	193	11	,	,	PUNCT
fbem-10171	193	12	yin	yin	PROPN
fbem-10171	193	13	cunyan	cunyan	PROPN
fbem-10171	193	14	,	,	PUNCT
fbem-10171	193	15	et	et	PROPN
fbem-10171	193	16	al	al	PROPN
fbem-10171	193	17	.	.	PROPN
fbem-10171	193	18	automatic	automatic	ADJ
fbem-10171	193	19	recognition	recognition	NOUN
fbem-10171	193	20	of	of	ADP
fbem-10171	193	21	chinese	chinese	ADJ
fbem-10171	193	22	organization	organization	NOUN
fbem-10171	193	23	names	name	NOUN
fbem-10171	193	24	based	base	VERB
fbem-10171	193	25	on	on	ADP
fbem-10171	193	26	cascaded	cascade	VERB
fbem-10171	193	27	conditional	conditional	ADJ
fbem-10171	193	28	random	random	ADJ
fbem-10171	193	29	field	field	NOUN
fbem-10171	193	30	model[j	model[j	PROPN
fbem-10171	193	31	]	]	PUNCT
fbem-10171	193	32	.	.	PUNCT
fbem-10171	194	1	journal	journal	PROPN
fbem-10171	194	2	of	of	ADP
fbem-10171	194	3	electronics	electronic	NOUN
fbem-10171	194	4	,	,	PUNCT
fbem-10171	194	5	2006	2006	NUM
fbem-10171	194	6	,	,	PUNCT
fbem-10171	194	7	34(5	34(5	NUM
fbem-10171	194	8	):	):	PUNCT
fbem-10171	194	9	804	804	NUM
fbem-10171	194	10	.	.	PUNCT
fbem-10171	195	1	[	[	X
fbem-10171	195	2	15	15	NUM
fbem-10171	195	3	]	]	X
fbem-10171	195	4	huang	huang	PROPN
fbem-10171	195	5	bo	bo	PROPN
fbem-10171	195	6	,	,	PUNCT
fbem-10171	195	7	liu	liu	PROPN
fbem-10171	195	8	chuancai	chuancai	PROPN
fbem-10171	195	9	.	.	PUNCT
fbem-10171	196	1	automatic	automatic	ADJ
fbem-10171	196	2	chinese	chinese	ADJ
fbem-10171	196	3	text	text	NOUN
fbem-10171	196	4	summarization	summarization	NOUN
fbem-10171	196	5	based	base	VERB
fbem-10171	196	6	on	on	ADP
fbem-10171	196	7	weighted	weight	VERB
fbem-10171	196	8	textrank[j	textrank[j	NUM
fbem-10171	196	9	]	]	PUNCT
fbem-10171	196	10	.	.	PUNCT
fbem-10171	197	1	application	application	NOUN
fbem-10171	197	2	research	research	NOUN
fbem-10171	197	3	of	of	ADP
fbem-10171	197	4	computers	computer	NOUN
fbem-10171	197	5	/	/	SYM
fbem-10171	197	6	jisuanji	jisuanji	PROPN
fbem-10171	197	7	yingyong	yingyong	PROPN
fbem-10171	197	8	yanjiu	yanjiu	PROPN
fbem-10171	197	9	,	,	PUNCT
fbem-10171	197	10	2020	2020	NUM
fbem-10171	197	11	,	,	PUNCT
fbem-10171	197	12	37(2	37(2	NOUN
fbem-10171	197	13	)	)	PUNCT
fbem-10171	197	14	.	.	PUNCT
fbem-10171	198	1	[	[	X
fbem-10171	198	2	16	16	NUM
fbem-10171	198	3	]	]	X
fbem-10171	198	4	zhang	zhang	PROPN
fbem-10171	198	5	,	,	PUNCT
fbem-10171	198	6	liu	liu	PROPN
fbem-10171	198	7	,	,	PUNCT
fbem-10171	198	8	wang	wang	PROPN
fbem-10171	198	9	,	,	PUNCT
fbem-10171	198	10	liwei	liwei	PROPN
fbem-10171	198	11	,	,	PUNCT
fbem-10171	198	12	huang	huang	PROPN
fbem-10171	198	13	,	,	PUNCT
fbem-10171	198	14	bo	bo	PROPN
fbem-10171	198	15	,	,	PUNCT
fbem-10171	198	16	et	et	PROPN
fbem-10171	198	17	al	al	PROPN
fbem-10171	198	18	.	.	PROPN
fbem-10171	198	19	sentiment	sentiment	PROPN
fbem-10171	198	20	classification	classification	NOUN
fbem-10171	198	21	model	model	NOUN
fbem-10171	198	22	and	and	CCONJ
fbem-10171	198	23	experimental	experimental	ADJ
fbem-10171	198	24	study	study	NOUN
fbem-10171	198	25	of	of	ADP
fbem-10171	198	26	microblog	microblog	ADJ
fbem-10171	198	27	comments	comment	NOUN
fbem-10171	198	28	based	base	VERB
fbem-10171	198	29	on	on	ADP
fbem-10171	198	30	word	word	NOUN
fbem-10171	198	31	vectors	vector	NOUN
fbem-10171	198	32	with	with	ADP
fbem-10171	198	33	multi	multi	ADJ
fbem-10171	198	34	-	-	ADJ
fbem-10171	198	35	scale	scale	ADJ
fbem-10171	198	36	convolutional	convolutional	ADJ
fbem-10171	198	37	neural	neural	ADJ
fbem-10171	198	38	network[j	network[j	NOUN
fbem-10171	198	39	]	]	PUNCT
fbem-10171	198	40	.	.	PUNCT
fbem-10171	199	1	library	library	NOUN
fbem-10171	199	2	intelligence	intelligence	NOUN
fbem-10171	199	3	work	work	NOUN
fbem-10171	199	4	,	,	PUNCT
fbem-10171	199	5	2019	2019	NUM
fbem-10171	199	6	,	,	PUNCT
fbem-10171	199	7	63(18	63(18	NOUN
fbem-10171	199	8	):	):	PUNCT
fbem-10171	199	9	99	99	NUM
fbem-10171	199	10	.	.	PUNCT
fbem-10171	200	1	[	[	X
fbem-10171	200	2	17	17	NUM
fbem-10171	200	3	]	]	X
fbem-10171	200	4	gong	gong	PROPN
fbem-10171	200	5	,	,	PUNCT
fbem-10171	200	6	shuang	shuang	PROPN
fbem-10171	200	7	-	-	PUNCT
fbem-10171	200	8	shuang	shuang	PROPN
fbem-10171	200	9	,	,	PUNCT
fbem-10171	200	10	chen	chen	PROPN
fbem-10171	200	11	,	,	PUNCT
fbem-10171	200	12	yu	yu	PROPN
fbem-10171	200	13	-	-	PUNCT
fbem-10171	200	14	feng	feng	PROPN
fbem-10171	200	15	,	,	PUNCT
fbem-10171	200	16	xu	xu	PROPN
fbem-10171	200	17	,	,	PUNCT
fbem-10171	200	18	jin	jin	NOUN
fbem-10171	200	19	-	-	PUNCT
fbem-10171	200	20	an	an	PROPN
fbem-10171	200	21	,	,	PUNCT
fbem-10171	200	22	et	et	PROPN
fbem-10171	200	23	al	al	PROPN
fbem-10171	200	24	.	.	PUNCT
fbem-10171	201	1	a	a	DET
fbem-10171	201	2	multi	multi	ADJ
fbem-10171	201	3	-	-	ADJ
fbem-10171	201	4	word	word	ADJ
fbem-10171	201	5	expression	expression	NOUN
fbem-10171	201	6	extraction	extraction	NOUN
fbem-10171	201	7	method	method	NOUN
fbem-10171	201	8	for	for	ADP
fbem-10171	201	9	chinese	chinese	PROPN
fbem-10171	201	10	based	base	VERB
fbem-10171	201	11	on	on	ADP
fbem-10171	201	12	web	web	NOUN
fbem-10171	201	13	text[j	text[j	PROPN
fbem-10171	201	14	]	]	PUNCT
fbem-10171	201	15	.	.	PUNCT
fbem-10171	202	1	journal	journal	PROPN
fbem-10171	202	2	of	of	ADP
fbem-10171	202	3	shandong	shandong	PROPN
fbem-10171	202	4	university	university	PROPN
fbem-10171	202	5	(	(	PUNCT
fbem-10171	202	6	science	science	NOUN
fbem-10171	202	7	edition	edition	PROPN
fbem-10171	202	8	)	)	PUNCT
fbem-10171	202	9	,	,	PUNCT
fbem-10171	202	10	2018	2018	NUM
fbem-10171	202	11	,	,	PUNCT
fbem-10171	202	12	53(9	53(9	NUM
fbem-10171	202	13	):	):	PUNCT
fbem-10171	202	14	40	40	NUM
fbem-10171	202	15	-	-	SYM
fbem-10171	202	16	48	48	NUM
fbem-10171	202	17	.	.	PUNCT
fbem-10171	203	1	[	[	X
fbem-10171	203	2	18	18	NUM
fbem-10171	203	3	]	]	X
fbem-10171	203	4	zheng	zheng	PROPN
fbem-10171	203	5	zheng	zheng	PROPN
fbem-10171	203	6	,	,	PUNCT
fbem-10171	203	7	wang	wang	PROPN
fbem-10171	203	8	huadan	huadan	PROPN
fbem-10171	203	9	,	,	PUNCT
fbem-10171	203	10	hao	hao	PROPN
fbem-10171	203	11	lihua	lihua	PROPN
fbem-10171	203	12	.	.	PUNCT
fbem-10171	204	1	a	a	DET
fbem-10171	204	2	study	study	NOUN
fbem-10171	204	3	of	of	ADP
fbem-10171	204	4	the	the	DET
fbem-10171	204	5	u.s	u.s	PROPN
fbem-10171	204	6	.	.	PROPN
fbem-10171	204	7	military	military	ADJ
fbem-10171	204	8	language	language	PROPN
fbem-10171	204	9	dictionary	dictionary	NOUN
fbem-10171	204	10	from	from	ADP
fbem-10171	204	11	a	a	DET
fbem-10171	204	12	terminological	terminological	ADJ
fbem-10171	204	13	perspective[j	perspective[j	NOUN
fbem-10171	204	14	]	]	PUNCT
fbem-10171	204	15	.	.	PUNCT
fbem-10171	205	1	china	china	PROPN
fbem-10171	205	2	science	science	PROPN
fbem-10171	205	3	and	and	CCONJ
fbem-10171	205	4	technology	technology	NOUN
fbem-10171	205	5	terminology	terminology	NOUN
fbem-10171	205	6	,	,	PUNCT
fbem-10171	205	7	2021	2021	NUM
fbem-10171	205	8	,	,	PUNCT
fbem-10171	205	9	23(03	23(03	NUM
fbem-10171	205	10	):	):	PUNCT
fbem-10171	205	11	33	33	NUM
fbem-10171	205	12	-	-	SYM
fbem-10171	205	13	41	41	NUM
fbem-10171	205	14	.	.	PUNCT
fbem-10171	206	1	[	[	X
fbem-10171	206	2	19	19	NUM
fbem-10171	206	3	]	]	X
fbem-10171	206	4	zhou	zhou	PROPN
fbem-10171	206	5	zhi	zhi	PROPN
fbem-10171	206	6	,	,	PUNCT
fbem-10171	206	7	wang	wang	PROPN
fbem-10171	206	8	chunying	chunying	PROPN
fbem-10171	206	9	,	,	PUNCT
fbem-10171	206	10	zhu	zhu	PROPN
fbem-10171	206	11	jia	jia	PROPN
fbem-10171	206	12	-	-	PUNCT
fbem-10171	206	13	li	li	PROPN
fbem-10171	206	14	.	.	PROPN
fbem-10171	206	15	research	research	NOUN
fbem-10171	206	16	on	on	ADP
fbem-10171	206	17	the	the	DET
fbem-10171	206	18	construction	construction	NOUN
fbem-10171	206	19	of	of	ADP
fbem-10171	206	20	an	an	DET
fbem-10171	206	21	emotion	emotion	NOUN
fbem-10171	206	22	lexicon	lexicon	NOUN
fbem-10171	206	23	in	in	ADP
fbem-10171	206	24	book	book	NOUN
fbem-10171	206	25	domain	domain	NOUN
fbem-10171	206	26	based	base	VERB
fbem-10171	206	27	on	on	ADP
fbem-10171	206	28	ultra	ultra	ADJ
fbem-10171	206	29	-	-	ADJ
fbem-10171	206	30	short	short	ADJ
fbem-10171	206	31	reviews[j	reviews[j	PROPN
fbem-10171	206	32	]	]	PUNCT
fbem-10171	206	33	.	.	PUNCT
fbem-10171	207	1	intelligence	intelligence	NOUN
fbem-10171	207	2	theory	theory	NOUN
fbem-10171	207	3	and	and	CCONJ
fbem-10171	207	4	practice	practice	NOUN
fbem-10171	207	5	,	,	PUNCT
fbem-10171	207	6	2021	2021	NUM
fbem-10171	207	7	,	,	PUNCT
fbem-10171	207	8	44	44	NUM
fbem-10171	207	9	(	(	PUNCT
fbem-10171	207	10	09	09	NUM
fbem-10171	207	11	):	):	PUNCT
fbem-10171	207	12	183	183	NUM
fbem-10171	207	13	-	-	SYM
fbem-10171	207	14	189	189	NUM
fbem-10171	207	15	+	+	NOUN
fbem-10171	207	16	197.doi:10.16353	197.doi:10.16353	NUM
fbem-10171	207	17	/	/	SYM
fbem-10171	207	18	j.cnki.10007490.2021.09.026	j.cnki.10007490.2021.09.026	NOUN
fbem-10171	207	19	.	.	PUNCT
fbem-10171	208	1	[	[	X
fbem-10171	208	2	20	20	NUM
fbem-10171	208	3	]	]	X
fbem-10171	208	4	liu	liu	PROPN
fbem-10171	208	5	p	p	PROPN
fbem-10171	208	6	,	,	PUNCT
fbem-10171	208	7	wang	wang	PROPN
fbem-10171	208	8	w	w	PROPN
fbem-10171	208	9	,	,	PUNCT
fbem-10171	208	10	qiu	qiu	PROPN
fbem-10171	208	11	l	l	PROPN
fbem-10171	208	12	,	,	PUNCT
fbem-10171	208	13	et	et	PROPN
fbem-10171	208	14	al	al	PROPN
fbem-10171	208	15	.	.	PROPN
fbem-10171	208	16	cdcpp	cdcpp	PROPN
fbem-10171	208	17	:	:	PUNCT
fbem-10171	208	18	(	(	PUNCT
fbem-10171	208	19	cdcpp	cdcpp	NOUN
fbem-10171	208	20	:	:	PUNCT
fbem-10171	208	21	cross	cross	ADJ
fbem-10171	208	22	-	-	ADJ
fbem-10171	208	23	domain	domain	ADJ
fbem-10171	208	24	chinese	chinese	ADJ
fbem-10171	208	25	punctuation	punctuation	NOUN
fbem-10171	208	26	prediction)[c]//proceedings	prediction)[c]//proceeding	NOUN
fbem-10171	208	27	of	of	ADP
fbem-10171	208	28	the	the	DET
fbem-10171	208	29	19th	19th	ADJ
fbem-10171	208	30	chinese	chinese	ADJ
fbem-10171	208	31	national	national	ADJ
fbem-10171	208	32	conference	conference	NOUN
fbem-10171	208	33	on	on	ADP
fbem-10171	208	34	computational	computational	ADJ
fbem-10171	208	35	linguistics	linguistic	NOUN
fbem-10171	208	36	.	.	PUNCT
fbem-10171	209	1	2020	2020	NUM
fbem-10171	209	2	:	:	PUNCT
fbem-10171	210	1	593	593	NUM
fbem-10171	210	2	-	-	SYM
fbem-10171	210	3	603	603	NUM
fbem-10171	210	4	.	.	PUNCT
fbem-10171	211	1	[	[	X
fbem-10171	211	2	21	21	NUM
fbem-10171	211	3	]	]	X
fbem-10171	211	4	yi	yi	PROPN
fbem-10171	211	5	,	,	PUNCT
fbem-10171	211	6	shun	shun	NOUN
fbem-10171	211	7	-	-	PUNCT
fbem-10171	211	8	ming	ming	PROPN
fbem-10171	211	9	,	,	PUNCT
fbem-10171	211	10	yi	yi	PROPN
fbem-10171	211	11	,	,	PUNCT
fbem-10171	211	12	hao	hao	PROPN
fbem-10171	211	13	,	,	PUNCT
fbem-10171	211	14	zhou	zhou	PROPN
fbem-10171	211	15	,	,	PUNCT
fbem-10171	211	16	guo	guo	PROPN
fbem-10171	211	17	-	-	PUNCT
fbem-10171	211	18	dong	dong	PROPN
fbem-10171	211	19	.	.	PUNCT
fbem-10171	212	1	research	research	NOUN
fbem-10171	212	2	on	on	ADP
fbem-10171	212	3	twitter	twitter	NOUN
fbem-10171	212	4	sentiment	sentiment	NOUN
fbem-10171	212	5	classification	classification	NOUN
fbem-10171	212	6	method	method	NOUN
fbem-10171	212	7	using	use	VERB
fbem-10171	212	8	sentiment	sentiment	NOUN
fbem-10171	212	9	feature	feature	NOUN
fbem-10171	212	10	vector[j	vector[j	NOUN
fbem-10171	212	11	]	]	PUNCT
fbem-10171	212	12	.	.	PUNCT
fbem-10171	213	1	small	small	ADJ
fbem-10171	213	2	microcomputer	microcomputer	NOUN
fbem-10171	213	3	systems	system	NOUN
fbem-10171	213	4	,	,	PUNCT
fbem-10171	213	5	2016	2016	NUM
fbem-10171	213	6	,	,	PUNCT
fbem-10171	213	7	37(11	37(11	NUM
fbem-10171	213	8	):	):	PUNCT
fbem-10171	213	9	2454	2454	NUM
fbem-10171	213	10	-	-	SYM
fbem-10171	213	11	2458	2458	NUM
fbem-10171	213	12	.	.	PUNCT
fbem-10171	214	1	[	[	X
fbem-10171	214	2	22	22	NUM
fbem-10171	214	3	]	]	X
fbem-10171	214	4	zhang	zhang	PROPN
fbem-10171	214	5	huaping	huaping	PROPN
fbem-10171	214	6	,	,	PUNCT
fbem-10171	214	7	liu	liu	PROPN
fbem-10171	214	8	qun	qun	PROPN
fbem-10171	214	9	.	.	PUNCT
fbem-10171	215	1	a	a	DET
fbem-10171	215	2	coarse	coarse	ADJ
fbem-10171	215	3	classification	classification	NOUN
fbem-10171	215	4	model	model	NOUN
fbem-10171	215	5	for	for	ADP
fbem-10171	215	6	chinese	chinese	ADJ
fbem-10171	215	7	words	word	NOUN
fbem-10171	215	8	based	base	VERB
fbem-10171	215	9	on	on	ADP
fbem-10171	215	10	the	the	DET
fbem-10171	215	11	n	n	ADV
fbem-10171	215	12	shortest	short	ADJ
fbem-10171	215	13	path	path	NOUN
fbem-10171	215	14	method[j	method[j	NOUN
fbem-10171	215	15	]	]	PUNCT
fbem-10171	215	16	.	.	PUNCT
fbem-10171	216	1	chinese	chinese	ADJ
fbem-10171	216	2	journal	journal	PROPN
fbem-10171	216	3	of	of	ADP
fbem-10171	216	4	information	information	NOUN
fbem-10171	216	5	,	,	PUNCT
fbem-10171	216	6	2002	2002	NUM
fbem-10171	216	7	,	,	PUNCT
fbem-10171	216	8	16(5	16(5	NUM
fbem-10171	216	9	):	):	PUNCT
fbem-10171	216	10	1	1	NUM
fbem-10171	216	11	-	-	SYM
fbem-10171	216	12	7	7	NUM
fbem-10171	216	13	.	.	PUNCT
fbem-10171	217	1	[	[	X
fbem-10171	217	2	23	23	NUM
fbem-10171	217	3	]	]	X
fbem-10171	217	4	tang	tang	X
fbem-10171	217	5	huifeng	huifeng	PROPN
fbem-10171	217	6	,	,	PUNCT
fbem-10171	217	7	tan	tan	PROPN
fbem-10171	217	8	songbo	songbo	PROPN
fbem-10171	217	9	,	,	PUNCT
fbem-10171	217	10	cheng	cheng	PROPN
fbem-10171	217	11	xueqi	xueqi	PROPN
fbem-10171	217	12	.	.	PUNCT
fbem-10171	218	1	a	a	DET
fbem-10171	218	2	comparative	comparative	ADJ
fbem-10171	218	3	study	study	NOUN
fbem-10171	218	4	of	of	ADP
fbem-10171	218	5	chinese	chinese	ADJ
fbem-10171	218	6	sentiment	sentiment	NOUN
fbem-10171	218	7	classification	classification	NOUN
fbem-10171	218	8	techniques	technique	NOUN
fbem-10171	218	9	based	base	VERB
fbem-10171	218	10	on	on	ADP
fbem-10171	218	11	supervised	supervised	ADJ
fbem-10171	218	12	learning	learn	VERB
fbem-10171	218	13	[	[	X
fbem-10171	218	14	d	d	X
fbem-10171	218	15	]	]	X
fbem-10171	218	16	.	.	PUNCT
fbem-10171	218	17	,	,	PUNCT
fbem-10171	218	18	2007	2007	NUM
fbem-10171	218	19	.	.	PUNCT
fbem-10171	219	1	[	[	X
fbem-10171	219	2	24	24	NUM
fbem-10171	219	3	]	]	X
fbem-10171	219	4	lu	lu	PROPN
fbem-10171	219	5	yh	yh	PROPN
fbem-10171	219	6	,	,	PUNCT
fbem-10171	219	7	li	li	PROPN
fbem-10171	219	8	yf	yf	PROPN
fbem-10171	219	9	.	.	PUNCT
fbem-10171	220	1	an	an	DET
fbem-10171	220	2	improved	improved	ADJ
fbem-10171	220	3	tf	tf	PROPN
fbem-10171	220	4	-	-	PUNCT
fbem-10171	220	5	idf	idf	PROPN
fbem-10171	220	6	algorithm	algorithm	NOUN
fbem-10171	220	7	for	for	ADP
fbem-10171	220	8	computing	compute	VERB
fbem-10171	220	9	the	the	DET
fbem-10171	220	10	weights	weight	NOUN
fbem-10171	220	11	of	of	ADP
fbem-10171	220	12	textual	textual	ADJ
fbem-10171	220	13	feature	feature	NOUN
fbem-10171	220	14	items[j	items[j	PROPN
fbem-10171	220	15	]	]	PUNCT
fbem-10171	220	16	.	.	PUNCT
fbem-10171	221	1	library	library	NOUN
fbem-10171	221	2	intelligence	intelligence	NOUN
fbem-10171	221	3	work	work	NOUN
fbem-10171	221	4	,	,	PUNCT
fbem-10171	221	5	2013	2013	NUM
fbem-10171	221	6	,	,	PUNCT
fbem-10171	221	7	57(03	57(03	PROPN
fbem-10171	221	8	):	):	PUNCT
fbem-10171	221	9	90	90	NUM
fbem-10171	221	10	.	.	PUNCT
fbem-10171	222	1	[	[	X
fbem-10171	222	2	25	25	NUM
fbem-10171	222	3	]	]	X
fbem-10171	222	4	han	han	PROPN
fbem-10171	222	5	,	,	PUNCT
fbem-10171	222	6	tong	tong	PROPN
fbem-10171	222	7	-	-	PUNCT
fbem-10171	222	8	hui	hui	PROPN
fbem-10171	222	9	,	,	PUNCT
fbem-10171	222	10	yang	yang	PROPN
fbem-10171	222	11	,	,	PUNCT
fbem-10171	222	12	dong	dong	PROPN
fbem-10171	222	13	-	-	PUNCT
fbem-10171	222	14	qiang	qiang	PROPN
fbem-10171	222	15	,	,	PUNCT
fbem-10171	222	16	ma	ma	PROPN
fbem-10171	222	17	,	,	PUNCT
fbem-10171	222	18	hong	hong	PROPN
fbem-10171	222	19	-	-	PUNCT
fbem-10171	222	20	wei	wei	PROPN
fbem-10171	222	21	.	.	PUNCT
fbem-10171	223	1	a	a	DET
fbem-10171	223	2	method	method	NOUN
fbem-10171	223	3	for	for	ADP
fbem-10171	223	4	constructing	construct	VERB
fbem-10171	223	5	text	text	NOUN
fbem-10171	223	6	feature	feature	NOUN
fbem-10171	223	7	representations	representation	NOUN
fbem-10171	223	8	using	use	VERB
fbem-10171	223	9	sentiment	sentiment	NOUN
fbem-10171	223	10	word	word	NOUN
fbem-10171	223	11	statistics[j	statistics[j	PROPN
fbem-10171	223	12	]	]	PUNCT
fbem-10171	223	13	.	.	PUNCT
fbem-10171	224	1	application	application	NOUN
fbem-10171	224	2	research	research	NOUN
fbem-10171	224	3	of	of	ADP
fbem-10171	224	4	computers	computer	NOUN
fbem-10171	224	5	/	/	SYM
fbem-10171	224	6	jisuanji	jisuanji	PROPN
fbem-10171	224	7	yingyong	yingyong	PROPN
fbem-10171	224	8	yanjiu	yanjiu	PROPN
fbem-10171	224	9	,	,	PUNCT
fbem-10171	224	10	2019	2019	NUM
fbem-10171	224	11	,	,	PUNCT
fbem-10171	224	12	36(7	36(7	NUM
fbem-10171	224	13	)	)	PUNCT
fbem-10171	224	14	.	.	PUNCT
fbem-10171	225	1	[	[	X
fbem-10171	225	2	26	26	NUM
fbem-10171	225	3	]	]	X
fbem-10171	225	4	ramos	ramos	PROPN
fbem-10171	225	5	j.	j.	PROPN
fbem-10171	225	6	using	use	VERB
fbem-10171	225	7	tf	tf	PROPN
fbem-10171	225	8	-	-	PUNCT
fbem-10171	225	9	idf	idf	NOUN
fbem-10171	225	10	to	to	PART
fbem-10171	225	11	determine	determine	VERB
fbem-10171	225	12	word	word	NOUN
fbem-10171	225	13	relevance	relevance	NOUN
fbem-10171	225	14	in	in	ADP
fbem-10171	225	15	document	document	NOUN
fbem-10171	225	16	queries[c]//proceedings	queries[c]//proceeding	NOUN
fbem-10171	225	17	of	of	ADP
fbem-10171	225	18	the	the	DET
fbem-10171	225	19	first	first	ADJ
fbem-10171	225	20	instructional	instructional	ADJ
fbem-10171	225	21	conference	conference	NOUN
fbem-10171	225	22	on	on	ADP
fbem-10171	225	23	machine	machine	NOUN
fbem-10171	225	24	learning	learning	NOUN
fbem-10171	225	25	.	.	PUNCT
fbem-10171	226	1	2003	2003	NUM
fbem-10171	226	2	,	,	PUNCT
fbem-10171	226	3	242(1	242(1	NUM
fbem-10171	226	4	):	):	PUNCT
fbem-10171	226	5	29	29	NUM
fbem-10171	226	6	-	-	SYM
fbem-10171	226	7	48	48	NUM
fbem-10171	226	8	.	.	PUNCT
fbem-10171	227	1	[	[	X
fbem-10171	227	2	27	27	NUM
fbem-10171	227	3	]	]	X
fbem-10171	227	4	xue	xue	PROPN
fbem-10171	227	5	yanfei	yanfei	PROPN
fbem-10171	227	6	,	,	PUNCT
fbem-10171	227	7	mao	mao	X
fbem-10171	228	1	qirong	qirong	PROPN
fbem-10171	228	2	,	,	PUNCT
fbem-10171	228	3	zhang	zhang	PROPN
fbem-10171	228	4	jianming	jianming	PROPN
fbem-10171	228	5	.	.	PUNCT
fbem-10171	229	1	a	a	DET
fbem-10171	229	2	deep	deep	ADJ
fbem-10171	229	3	recommendation	recommendation	NOUN
fbem-10171	229	4	model	model	NOUN
fbem-10171	229	5	incorporating	incorporate	VERB
fbem-10171	229	6	contextual	contextual	ADJ
fbem-10171	229	7	information	information	NOUN
fbem-10171	230	1	[	[	X
fbem-10171	230	2	j	j	X
fbem-10171	230	3	]	]	X
fbem-10171	230	4	.	.	PUNCT
fbem-10171	231	1	application	application	NOUN
fbem-10171	231	2	research	research	NOUN
fbem-10171	231	3	of	of	ADP
fbem-10171	231	4	computers	computer	NOUN
fbem-10171	231	5	/	/	SYM
fbem-10171	231	6	jisuanji	jisuanji	PROPN
fbem-10171	231	7	yingyong	yingyong	PROPN
fbem-10171	231	8	yanjiu	yanjiu	PROPN
fbem-10171	231	9	,	,	PUNCT
fbem-10171	231	10	2021	2021	NUM
fbem-10171	231	11	,	,	PUNCT
fbem-10171	231	12	38(4	38(4	NUM
fbem-10171	231	13	)	)	PUNCT
fbem-10171	231	14	.	.	PUNCT
fbem-10171	232	1	[	[	X
fbem-10171	232	2	28	28	NUM
fbem-10171	232	3	]	]	X
fbem-10171	232	4	rong	rong	PROPN
fbem-10171	232	5	x.	x.	PROPN
fbem-10171	232	6	word2vec	word2vec	PROPN
fbem-10171	232	7	parameter	parameter	NOUN
fbem-10171	232	8	learning	learn	VERB
fbem-10171	232	9	explained[j	explained[j	PROPN
fbem-10171	232	10	]	]	PUNCT
fbem-10171	232	11	.	.	PUNCT
fbem-10171	233	1	arxiv	arxiv	PROPN
fbem-10171	233	2	preprint	preprint	VERB
fbem-10171	233	3	arxiv:1411.2738	arxiv:1411.2738	NOUN
fbem-10171	233	4	,	,	PUNCT
fbem-10171	233	5	2014	2014	NUM
fbem-10171	233	6	.	.	PUNCT
fbem-10171	234	1	[	[	X
fbem-10171	234	2	29	29	NUM
fbem-10171	234	3	]	]	X
fbem-10171	234	4	shi	shi	PROPN
fbem-10171	234	5	l	l	PROPN
fbem-10171	234	6	,	,	PUNCT
fbem-10171	234	7	wang	wang	PROPN
fbem-10171	234	8	y	y	PROPN
fbem-10171	234	9	,	,	PUNCT
fbem-10171	234	10	cheng	cheng	PROPN
fbem-10171	234	11	y	y	PROPN
fbem-10171	234	12	,	,	PUNCT
fbem-10171	234	13	et	et	PROPN
fbem-10171	234	14	al	al	PROPN
fbem-10171	234	15	.	.	PUNCT
fbem-10171	235	1	a	a	DET
fbem-10171	235	2	review	review	NOUN
fbem-10171	235	3	of	of	ADP
fbem-10171	235	4	attention	attention	NOUN
fbem-10171	235	5	mechanisms	mechanism	NOUN
fbem-10171	235	6	in	in	ADP
fbem-10171	235	7	natural	natural	ADJ
fbem-10171	235	8	language	language	NOUN
fbem-10171	235	9	processing[j	processing[j	NOUN
fbem-10171	235	10	]	]	PUNCT
fbem-10171	235	11	.	.	PUNCT
fbem-10171	236	1	data	datum	NOUN
fbem-10171	236	2	analysis	analysis	NOUN
fbem-10171	236	3	and	and	CCONJ
fbem-10171	236	4	knowledge	knowledge	NOUN
fbem-10171	236	5	discovery	discovery	NOUN
fbem-10171	236	6	,	,	PUNCT
fbem-10171	236	7	2020	2020	NUM
fbem-10171	236	8	,	,	PUNCT
fbem-10171	236	9	4(5	4(5	NUM
fbem-10171	236	10	):	):	PUNCT
fbem-10171	236	11	1	1	NUM
fbem-10171	236	12	-	-	SYM
fbem-10171	236	13	14	14	NUM
fbem-10171	236	14	.	.	PUNCT
fbem-10171	237	1	ren	ren	PROPN
fbem-10171	237	2	huan	huan	PROPN
fbem-10171	237	3	,	,	PUNCT
fbem-10171	237	4	wang	wang	PROPN
fbem-10171	237	5	xuguang	xuguang	PROPN
fbem-10171	237	6	.	.	PUNCT
fbem-10171	238	1	a	a	DET
fbem-10171	238	2	review	review	NOUN
fbem-10171	238	3	of	of	ADP
fbem-10171	238	4	attention	attention	NOUN
fbem-10171	238	5	mechanisms[j	mechanisms[j	PROPN
fbem-10171	238	6	]	]	PUNCT
fbem-10171	238	7	.	.	PUNCT
fbem-10171	239	1	computer	computer	NOUN
fbem-10171	239	2	applications	application	NOUN
fbem-10171	239	3	,	,	PUNCT
fbem-10171	239	4	2021	2021	NUM
fbem-10171	239	5	,	,	PUNCT
fbem-10171	239	6	41(s01	41(s01	NUM
fbem-10171	239	7	):	):	PUNCT
fbem-10171	239	8	1	1	NUM
fbem-10171	239	9	-	-	SYM
fbem-10171	239	10	6	6	NUM
fbem-10171	239	11	.	.	PUNCT
fbem-10171	240	1	[	[	X
fbem-10171	240	2	30	30	NUM
fbem-10171	240	3	]	]	X
fbem-10171	240	4	vaswani	vaswani	NOUN
fbem-10171	240	5	a	a	PRON
fbem-10171	240	6	,	,	PUNCT
fbem-10171	240	7	shazeer	shazeer	NOUN
fbem-10171	240	8	n	n	SYM
fbem-10171	240	9	,	,	PUNCT
fbem-10171	240	10	parmar	parmar	PROPN
fbem-10171	240	11	n	n	CCONJ
fbem-10171	240	12	,	,	PUNCT
fbem-10171	240	13	et	et	PROPN
fbem-10171	240	14	al	al	PROPN
fbem-10171	240	15	.	.	PUNCT
fbem-10171	240	16	attention	attention	NOUN
fbem-10171	240	17	is	be	AUX
fbem-10171	240	18	all	all	PRON
fbem-10171	240	19	you	you	PRON
fbem-10171	240	20	need[j	need[j	VERB
fbem-10171	240	21	]	]	PUNCT
fbem-10171	240	22	.	.	PUNCT
fbem-10171	241	1	advances	advance	NOUN
fbem-10171	241	2	in	in	ADP
fbem-10171	241	3	neural	neural	ADJ
fbem-10171	241	4	information	information	NOUN
fbem-10171	241	5	processing	processing	NOUN
fbem-10171	241	6	systems	system	NOUN
fbem-10171	241	7	,	,	PUNCT
fbem-10171	241	8	2017	2017	NUM
fbem-10171	241	9	,	,	PUNCT
fbem-10171	241	10	30	30	NUM
fbem-10171	241	11	.	.	PUNCT
fbem-10171	242	1	[	[	X
fbem-10171	242	2	31	31	NUM
fbem-10171	242	3	]	]	PUNCT
fbem-10171	242	4	chen	chen	PROPN
fbem-10171	242	5	l	l	PROPN
fbem-10171	242	6	c	c	PROPN
fbem-10171	242	7	,	,	PUNCT
fbem-10171	242	8	lee	lee	PROPN
fbem-10171	242	9	c	c	PROPN
fbem-10171	242	10	m	m	PROPN
fbem-10171	242	11	,	,	PUNCT
fbem-10171	242	12	chen	chen	PROPN
fbem-10171	242	13	m	m	PROPN
fbem-10171	242	14	y.	y.	PROPN
fbem-10171	242	15	exploration	exploration	NOUN
fbem-10171	242	16	of	of	ADP
fbem-10171	242	17	social	social	ADJ
fbem-10171	242	18	media	medium	NOUN
fbem-10171	242	19	for	for	ADP
fbem-10171	242	20	sentiment	sentiment	NOUN
fbem-10171	242	21	analysis	analysis	NOUN
fbem-10171	242	22	using	use	VERB
fbem-10171	242	23	deep	deep	ADJ
fbem-10171	242	24	learning[j	learning[j	NOUN
fbem-10171	242	25	]	]	PUNCT
fbem-10171	242	26	.	.	PUNCT
fbem-10171	243	1	soft	soft	ADJ
fbem-10171	243	2	computing	computing	NOUN
fbem-10171	243	3	,	,	PUNCT
fbem-10171	243	4	2020	2020	NUM
fbem-10171	243	5	,	,	PUNCT
fbem-10171	243	6	24	24	NUM
fbem-10171	243	7	:	:	PUNCT
fbem-10171	243	8	8187	8187	NUM
fbem-10171	243	9	-	-	SYM
fbem-10171	243	10	8197	8197	NUM
fbem-10171	243	11	.	.	PUNCT
fbem-10171	244	1	pan	pan	PROPN
fbem-10171	244	2	d	d	PROPN
fbem-10171	244	3	,	,	PUNCT
fbem-10171	244	4	yuan	yuan	PROPN
fbem-10171	244	5	j	j	PROPN
fbem-10171	244	6	,	,	PUNCT
fbem-10171	244	7	li	li	PROPN
fbem-10171	244	8	l	l	PROPN
fbem-10171	244	9	,	,	PUNCT
fbem-10171	244	10	et	et	PROPN
fbem-10171	244	11	al	al	PROPN
fbem-10171	244	12	.	.	PUNCT
fbem-10171	245	1	deep	deep	ADJ
fbem-10171	245	2	neural	neural	ADJ
fbem-10171	245	3	network	network	NOUN
fbem-10171	245	4	-	-	PUNCT
fbem-10171	245	5	based	base	VERB
fbem-10171	245	6	classification	classification	NOUN
fbem-10171	245	7	model	model	NOUN
fbem-10171	245	8	for	for	ADP
fbem-10171	245	9	sentiment	sentiment	NOUN
fbem-10171	245	10	analysis	analysis	NOUN
fbem-10171	245	11	[	[	X
fbem-10171	245	12	c]//2019	c]//2019	NUM
fbem-10171	245	13	6th	6th	ADJ
fbem-10171	245	14	international	international	ADJ
fbem-10171	245	15	conference	conference	NOUN
fbem-10171	245	16	on	on	ADP
fbem-10171	245	17	behavioral	behavioral	ADJ
fbem-10171	245	18	,	,	PUNCT
fbem-10171	245	19	economic	economic	ADJ
fbem-10171	245	20	and	and	CCONJ
fbem-10171	245	21	socio	socio	NOUN
fbem-10171	245	22	-	-	PUNCT
fbem-10171	245	23	cultural	cultural	ADJ
fbem-10171	245	24	computing	computing	NOUN
fbem-10171	245	25	(	(	PUNCT
fbem-10171	245	26	besc	besc	PROPN
fbem-10171	245	27	)	)	PUNCT
fbem-10171	245	28	.	.	PUNCT
fbem-10171	246	1	ieee	ieee	PROPN
fbem-10171	246	2	,	,	PUNCT
fbem-10171	246	3	2019	2019	NUM
fbem-10171	246	4	:	:	PUNCT
fbem-10171	246	5	1	1	NUM
fbem-10171	246	6	-	-	SYM
fbem-10171	246	7	4	4	NUM
fbem-10171	246	8	.	.	PUNCT
fbem-10171	247	1	[	[	X
fbem-10171	247	2	32	32	NUM
fbem-10171	247	3	]	]	X
fbem-10171	247	4	zhang	zhang	PROPN
fbem-10171	247	5	zq	zq	PROPN
fbem-10171	247	6	,	,	PUNCT
fbem-10171	247	7	ye	ye	PROPN
fbem-10171	247	8	q	q	PROPN
fbem-10171	247	9	,	,	PUNCT
fbem-10171	247	10	li	li	PROPN
fbem-10171	247	11	yj	yj	PROPN
fbem-10171	247	12	.	.	PROPN
fbem-10171	247	13	a	a	DET
fbem-10171	247	14	review	review	NOUN
fbem-10171	247	15	of	of	ADP
fbem-10171	247	16	sentiment	sentiment	NOUN
fbem-10171	247	17	analysis	analysis	NOUN
fbem-10171	247	18	of	of	ADP
fbem-10171	247	19	internet	internet	NOUN
fbem-10171	247	20	product	product	NOUN
fbem-10171	247	21	reviews[j	reviews[j	PROPN
fbem-10171	247	22	]	]	PUNCT
fbem-10171	247	23	.	.	PUNCT
fbem-10171	248	1	journal	journal	PROPN
fbem-10171	248	2	of	of	ADP
fbem-10171	248	3	management	management	NOUN
fbem-10171	248	4	science	science	NOUN
fbem-10171	248	5	,	,	PUNCT
fbem-10171	248	6	2010	2010	NUM
fbem-10171	248	7	,	,	PUNCT
fbem-10171	248	8	13(006	13(006	NUM
fbem-10171	248	9	):	):	PUNCT
fbem-10171	248	10	84	84	NUM
fbem-10171	248	11	-	-	SYM
fbem-10171	248	12	96	96	NUM
fbem-10171	248	13	.	.	PUNCT
fbem-10171	249	1	[	[	X
fbem-10171	249	2	33	33	NUM
fbem-10171	249	3	]	]	X
fbem-10171	249	4	deng	deng	PROPN
fbem-10171	249	5	s	s	PROPN
fbem-10171	249	6	,	,	PUNCT
fbem-10171	249	7	sinha	sinha	VERB
fbem-10171	249	8	a	a	DET
fbem-10171	249	9	p	p	X
fbem-10171	249	10	,	,	PUNCT
fbem-10171	249	11	zhao	zhao	PROPN
fbem-10171	249	12	h.	h.	PROPN
fbem-10171	249	13	adapting	adapt	VERB
fbem-10171	249	14	sentiment	sentiment	NOUN
fbem-10171	249	15	lexicons	lexicon	NOUN
fbem-10171	249	16	to	to	ADP
fbem-10171	249	17	domain	domain	VERB
fbem-10171	249	18	-	-	PUNCT
fbem-10171	249	19	specific	specific	ADJ
fbem-10171	249	20	social	social	ADJ
fbem-10171	249	21	media	medium	NOUN
fbem-10171	249	22	texts[j	texts[j	NOUN
fbem-10171	249	23	]	]	X
fbem-10171	249	24	.	.	PUNCT
fbem-10171	250	1	decision	decision	NOUN
fbem-10171	250	2	support	support	NOUN
fbem-10171	250	3	systems	system	NOUN
fbem-10171	250	4	,	,	PUNCT
fbem-10171	250	5	2017	2017	NUM
fbem-10171	250	6	,	,	PUNCT
fbem-10171	250	7	94	94	NUM
fbem-10171	250	8	:	:	SYM
fbem-10171	250	9	65	65	NUM
fbem-10171	250	10	-	-	SYM
fbem-10171	250	11	76	76	NUM
fbem-10171	250	12	.	.	PUNCT
fbem-10171	251	1	[	[	X
fbem-10171	251	2	34	34	NUM
fbem-10171	251	3	]	]	X
fbem-10171	251	4	chen	chen	PROPN
fbem-10171	251	5	yan	yan	PROPN
fbem-10171	251	6	-	-	PUNCT
fbem-10171	251	7	fang	fang	PROPN
fbem-10171	251	8	,	,	PUNCT
fbem-10171	251	9	li	li	PROPN
fbem-10171	251	10	zhi	zhi	PROPN
fbem-10171	251	11	-	-	PUNCT
fbem-10171	251	12	yu	yu	PROPN
fbem-10171	251	13	,	,	PUNCT
fbem-10171	251	14	liang	liang	PROPN
fbem-10171	251	15	zhuan	zhuan	PROPN
fbem-10171	251	16	,	,	PUNCT
fbem-10171	251	17	et	et	PROPN
fbem-10171	251	18	al	al	PROPN
fbem-10171	251	19	.	.	PUNCT
fbem-10171	252	1	a	a	DET
fbem-10171	252	2	review	review	NOUN
fbem-10171	252	3	of	of	ADP
fbem-10171	252	4	online	online	ADJ
fbem-10171	252	5	social	social	ADJ
fbem-10171	252	6	network	network	NOUN
fbem-10171	252	7	rumor	rumor	NOUN
fbem-10171	252	8	detection[j	detection[j	PROPN
fbem-10171	252	9	]	]	PUNCT
fbem-10171	252	10	.	.	PUNCT
fbem-10171	253	1	journal	journal	PROPN
fbem-10171	253	2	of	of	ADP
fbem-10171	253	3	computer	computer	NOUN
fbem-10171	253	4	science	science	NOUN
fbem-10171	253	5	,	,	PUNCT
fbem-10171	253	6	2018	2018	NUM
fbem-10171	253	7	,	,	PUNCT
fbem-10171	253	8	41(7	41(7	NUM
fbem-10171	253	9	):	):	PUNCT
fbem-10171	253	10	1648	1648	NUM
fbem-10171	253	11	-	-	SYM
fbem-10171	253	12	1676	1676	NUM
fbem-10171	253	13	.	.	PUNCT
fbem-10171	254	1	[	[	X
fbem-10171	254	2	35	35	NUM
fbem-10171	254	3	]	]	X
fbem-10171	254	4	li	li	PROPN
fbem-10171	254	5	jingmei	jingmei	PROPN
fbem-10171	254	6	,	,	PUNCT
fbem-10171	254	7	sun	sun	PROPN
fbem-10171	254	8	lihua	lihua	PROPN
fbem-10171	254	9	,	,	PUNCT
fbem-10171	254	10	zhang	zhang	PROPN
fbem-10171	254	11	qiaorong	qiaorong	PROPN
fbem-10171	254	12	,	,	PUNCT
fbem-10171	254	13	et	et	PROPN
fbem-10171	254	14	al	al	PROPN
fbem-10171	254	15	.	.	PUNCT
fbem-10171	255	1	a	a	DET
fbem-10171	255	2	plain	plain	ADJ
fbem-10171	255	3	bayesian	bayesian	NOUN
fbem-10171	255	4	classifier	classifier	NOUN
fbem-10171	255	5	for	for	ADP
fbem-10171	255	6	text	text	NOUN
fbem-10171	255	7	processing[j	processing[j	NOUN
fbem-10171	255	8	]	]	PUNCT
fbem-10171	255	9	.	.	PUNCT
fbem-10171	256	1	journal	journal	PROPN
fbem-10171	256	2	of	of	ADP
fbem-10171	256	3	harbin	harbin	PROPN
fbem-10171	256	4	engineering	engineering	PROPN
fbem-10171	256	5	university	university	PROPN
fbem-10171	256	6	,	,	PUNCT
fbem-10171	256	7	2003	2003	NUM
fbem-10171	256	8	,	,	PUNCT
fbem-10171	256	9	24(1	24(1	NUM
fbem-10171	256	10	):	):	PUNCT
fbem-10171	256	11	71	71	NUM
fbem-10171	256	12	-	-	SYM
fbem-10171	256	13	74	74	NUM
fbem-10171	256	14	.	.	PUNCT
fbem-10171	256	15	64	64	NUM
fbem-10171	257	1	[	[	SYM
fbem-10171	257	2	36	36	NUM
fbem-10171	257	3	]	]	X
fbem-10171	257	4	xiao	xiao	PROPN
fbem-10171	257	5	z	z	PROPN
fbem-10171	257	6	,	,	PUNCT
fbem-10171	257	7	liu	liu	PROPN
fbem-10171	257	8	f	f	PROPN
fbem-10171	257	9	,	,	PUNCT
fbem-10171	257	10	li	li	PROPN
fbem-10171	257	11	b.	b.	PROPN
fbem-10171	257	12	a	a	DET
fbem-10171	257	13	semantic	semantic	ADJ
fbem-10171	257	14	distance	distance	NOUN
fbem-10171	257	15	-	-	PUNCT
fbem-10171	257	16	based	base	VERB
fbem-10171	257	17	svm	svm	ADJ
fbem-10171	257	18	sentiment	sentiment	NOUN
fbem-10171	257	19	classification	classification	NOUN
fbem-10171	257	20	method	method	NOUN
fbem-10171	257	21	for	for	ADP
fbem-10171	257	22	web	web	NOUN
fbem-10171	257	23	reviews[j	reviews[j	PROPN
fbem-10171	257	24	]	]	X
fbem-10171	257	25	.	.	PUNCT
fbem-10171	258	1	computer	computer	NOUN
fbem-10171	258	2	science	science	NOUN
fbem-10171	258	3	,	,	PUNCT
fbem-10171	258	4	2014	2014	NUM
fbem-10171	258	5	,	,	PUNCT
fbem-10171	258	6	41(9	41(9	NUM
fbem-10171	258	7	):	):	PUNCT
fbem-10171	258	8	248	248	NUM
fbem-10171	258	9	-	-	SYM
fbem-10171	258	10	252	252	NUM
fbem-10171	258	11	.	.	PUNCT
fbem-10171	259	1	[	[	X
fbem-10171	259	2	37	37	NUM
fbem-10171	259	3	]	]	X
fbem-10171	259	4	liu	liu	PROPN
fbem-10171	259	5	gang	gang	PROPN
fbem-10171	259	6	,	,	PUNCT
fbem-10171	259	7	zhang	zhang	PROPN
fbem-10171	259	8	weishi	weishi	PROPN
fbem-10171	259	9	.	.	PUNCT
fbem-10171	260	1	sentiment	sentiment	NOUN
fbem-10171	260	2	analysis	analysis	NOUN
fbem-10171	260	3	of	of	ADP
fbem-10171	260	4	internet	internet	NOUN
fbem-10171	260	5	users	user	NOUN
fbem-10171	260	6	'	'	PART
fbem-10171	260	7	evaluation	evaluation	NOUN
fbem-10171	260	8	based	base	VERB
fbem-10171	260	9	on	on	ADP
fbem-10171	260	10	decision	decision	NOUN
fbem-10171	260	11	tree[j	tree[j	NOUN
fbem-10171	260	12	]	]	PUNCT
fbem-10171	260	13	.	.	PUNCT
fbem-10171	261	1	modern	modern	ADJ
fbem-10171	261	2	computer	computer	NOUN
fbem-10171	261	3	:	:	PUNCT
fbem-10171	261	4	midterm	midterm	PROPN
fbem-10171	261	5	journal	journal	PROPN
fbem-10171	261	6	,	,	PUNCT
fbem-10171	261	7	2017	2017	NUM
fbem-10171	261	8	(	(	PUNCT
fbem-10171	261	9	11	11	NUM
fbem-10171	261	10	):	):	SYM
fbem-10171	261	11	15	15	NUM
fbem-10171	261	12	-	-	SYM
fbem-10171	261	13	19	19	NUM
fbem-10171	261	14	.	.	PUNCT
fbem-10171	262	1	[	[	X
fbem-10171	262	2	38	38	NUM
fbem-10171	262	3	]	]	PUNCT
fbem-10171	262	4	neethu	neethu	NOUN
fbem-10171	262	5	m	m	PROPN
fbem-10171	262	6	s	s	PROPN
fbem-10171	262	7	,	,	PUNCT
fbem-10171	262	8	rajasree	rajasree	ADJ
fbem-10171	262	9	r.	r.	PROPN
fbem-10171	262	10	sentiment	sentiment	PROPN
fbem-10171	262	11	analysis	analysis	NOUN
fbem-10171	262	12	in	in	ADP
fbem-10171	262	13	twitter	twitter	NOUN
fbem-10171	262	14	using	use	VERB
fbem-10171	262	15	machine	machine	NOUN
fbem-10171	262	16	learning	learn	VERB
fbem-10171	262	17	techniques[c]//2013	techniques[c]//2013	PROPN
fbem-10171	262	18	fourth	fourth	ADJ
fbem-10171	262	19	international	international	ADJ
fbem-10171	262	20	conference	conference	NOUN
fbem-10171	262	21	on	on	ADP
fbem-10171	262	22	computing	computing	NOUN
fbem-10171	262	23	,	,	PUNCT
fbem-10171	262	24	communications	communication	NOUN
fbem-10171	262	25	and	and	CCONJ
fbem-10171	262	26	networking	networking	NOUN
fbem-10171	262	27	technologies	technology	NOUN
fbem-10171	262	28	(	(	PUNCT
fbem-10171	262	29	icccnt	icccnt	ADJ
fbem-10171	262	30	)	)	PUNCT
fbem-10171	262	31	.	.	PUNCT
fbem-10171	263	1	ieee	ieee	PROPN
fbem-10171	263	2	,	,	PUNCT
fbem-10171	263	3	2013	2013	NUM
fbem-10171	263	4	:	:	PUNCT
fbem-10171	263	5	1	1	NUM
fbem-10171	263	6	-	-	SYM
fbem-10171	263	7	5	5	NUM
fbem-10171	263	8	.	.	PUNCT
fbem-10171	264	1	[	[	X
fbem-10171	264	2	39	39	NUM
fbem-10171	264	3	]	]	PUNCT
fbem-10171	264	4	o'shea	o'shea	PROPN
fbem-10171	264	5	k	k	PROPN
fbem-10171	264	6	,	,	PUNCT
fbem-10171	264	7	nash	nash	PROPN
fbem-10171	264	8	r.	r.	PROPN
fbem-10171	264	9	an	an	DET
fbem-10171	264	10	introduction	introduction	NOUN
fbem-10171	264	11	to	to	ADP
fbem-10171	264	12	convolutional	convolutional	ADJ
fbem-10171	264	13	neural	neural	ADJ
fbem-10171	264	14	networks[j	networks[j	NOUN
fbem-10171	264	15	]	]	PUNCT
fbem-10171	264	16	.	.	PUNCT
fbem-10171	265	1	arxiv	arxiv	PROPN
fbem-10171	265	2	preprint	preprint	VERB
fbem-10171	265	3	arxiv:1511.08458	arxiv:1511.08458	PROPN
fbem-10171	265	4	,	,	PUNCT
fbem-10171	265	5	2015	2015	NUM
fbem-10171	265	6	.	.	PUNCT
fbem-10171	266	1	[	[	X
fbem-10171	266	2	40	40	NUM
fbem-10171	266	3	]	]	X
fbem-10171	266	4	sherstinsky	sherstinsky	ADJ
fbem-10171	266	5	a.	a.	NOUN
fbem-10171	266	6	fundamentals	fundamental	NOUN
fbem-10171	266	7	of	of	ADP
fbem-10171	266	8	recurrent	recurrent	ADJ
fbem-10171	266	9	neural	neural	ADJ
fbem-10171	266	10	network	network	NOUN
fbem-10171	266	11	(	(	PUNCT
fbem-10171	266	12	rnn	rnn	PROPN
fbem-10171	266	13	)	)	PUNCT
fbem-10171	266	14	and	and	CCONJ
fbem-10171	266	15	long	long	ADJ
fbem-10171	266	16	short	short	ADJ
fbem-10171	266	17	-	-	PUNCT
fbem-10171	266	18	term	term	NOUN
fbem-10171	266	19	memory	memory	NOUN
fbem-10171	266	20	(	(	PUNCT
fbem-10171	266	21	lstm	lstm	NOUN
fbem-10171	266	22	)	)	PUNCT
fbem-10171	266	23	network[j	network[j	PROPN
fbem-10171	266	24	]	]	PUNCT
fbem-10171	266	25	.	.	PUNCT
fbem-10171	267	1	physica	physica	PROPN
fbem-10171	267	2	d	d	NOUN
fbem-10171	267	3	:	:	PUNCT
fbem-10171	267	4	nonlinear	nonlinear	ADJ
fbem-10171	267	5	phenomena	phenomenon	NOUN
fbem-10171	267	6	,	,	PUNCT
fbem-10171	267	7	2020	2020	NUM
fbem-10171	267	8	,	,	PUNCT
fbem-10171	267	9	404	404	NUM
fbem-10171	267	10	:	:	PUNCT
fbem-10171	267	11	132306	132306	NUM
fbem-10171	267	12	.	.	PUNCT
fbem-10171	268	1	[	[	X
fbem-10171	268	2	41	41	NUM
fbem-10171	268	3	]	]	X
fbem-10171	268	4	zhao	zhao	PROPN
fbem-10171	268	5	yy	yy	PROPN
fbem-10171	268	6	,	,	PUNCT
fbem-10171	268	7	wang	wang	PROPN
fbem-10171	268	8	zy	zy	PROPN
fbem-10171	268	9	,	,	PUNCT
fbem-10171	268	10	wang	wang	PROPN
fbem-10171	268	11	p	p	PROPN
fbem-10171	268	12	,	,	PUNCT
fbem-10171	268	13	et	et	PROPN
fbem-10171	268	14	al	al	PROPN
fbem-10171	268	15	.	.	PUNCT
fbem-10171	269	1	a	a	DET
fbem-10171	269	2	review	review	NOUN
fbem-10171	269	3	of	of	ADP
fbem-10171	269	4	task	task	NOUN
fbem-10171	269	5	-	-	PUNCT
fbem-10171	269	6	based	base	VERB
fbem-10171	269	7	dialogue	dialogue	NOUN
fbem-10171	269	8	systems[j	systems[j	PROPN
fbem-10171	269	9	]	]	PUNCT
fbem-10171	269	10	.	.	PUNCT
fbem-10171	270	1	journal	journal	PROPN
fbem-10171	270	2	of	of	ADP
fbem-10171	270	3	computer	computer	NOUN
fbem-10171	270	4	science	science	NOUN
fbem-10171	270	5	,	,	PUNCT
fbem-10171	270	6	2020	2020	NUM
fbem-10171	270	7	,	,	PUNCT
fbem-10171	270	8	43(10	43(10	NUM
fbem-10171	270	9	):	):	PUNCT
fbem-10171	270	10	1862	1862	NUM
fbem-10171	270	11	-	-	SYM
fbem-10171	270	12	1896	1896	NUM
fbem-10171	270	13	.	.	PUNCT
fbem-10171	271	1	[	[	X
fbem-10171	271	2	42	42	NUM
fbem-10171	271	3	]	]	PUNCT
fbem-10171	271	4	shan	shan	PROPN
fbem-10171	271	5	,	,	PUNCT
fbem-10171	271	6	w.f	w.f	PROPN
fbem-10171	271	7	.	.	PROPN
fbem-10171	271	8	,	,	PUNCT
fbem-10171	271	9	li	li	PROPN
fbem-10171	271	10	,	,	PUNCT
fbem-10171	271	11	c.y	c.y	PROPN
fbem-10171	271	12	.	.	PROPN
fbem-10171	271	13	,	,	PUNCT
fbem-10171	271	14	chen	chen	PROPN
fbem-10171	271	15	,	,	PUNCT
fbem-10171	271	16	j.	j.	PROPN
fbem-10171	271	17	,	,	PUNCT
fbem-10171	271	18	et	et	PROPN
fbem-10171	271	19	al	al	PROPN
fbem-10171	271	20	.	.	PUNCT
fbem-10171	271	21	application	application	NOUN
fbem-10171	271	22	of	of	ADP
fbem-10171	271	23	convolutional	convolutional	ADJ
fbem-10171	271	24	neural	neural	ADJ
fbem-10171	271	25	network	network	NOUN
fbem-10171	271	26	and	and	CCONJ
fbem-10171	271	27	self	self	NOUN
fbem-10171	271	28	-	-	PUNCT
fbem-10171	271	29	attentive	attentive	ADJ
fbem-10171	271	30	mechanism	mechanism	NOUN
fbem-10171	271	31	to	to	PART
fbem-10171	271	32	identify	identify	VERB
fbem-10171	271	33	geomagnetic	geomagnetic	ADJ
fbem-10171	271	34	field	field	NOUN
fbem-10171	271	35	disturbance	disturbance	NOUN
fbem-10171	271	36	events[j	events[j	PROPN
fbem-10171	271	37	]	]	PUNCT
fbem-10171	271	38	.	.	PUNCT
fbem-10171	272	1	seismic	seismic	ADJ
fbem-10171	272	2	geomagnetic	geomagnetic	ADJ
fbem-10171	272	3	observation	observation	NOUN
fbem-10171	272	4	and	and	CCONJ
fbem-10171	272	5	research	research	NOUN
fbem-10171	272	6	,	,	PUNCT
fbem-10171	272	7	2022	2022	NUM
fbem-10171	272	8	,	,	PUNCT
fbem-10171	272	9	43(5	43(5	NUM
fbem-10171	272	10	):	):	PUNCT
fbem-10171	272	11	49	49	NUM
fbem-10171	272	12	-	-	SYM
fbem-10171	272	13	63	63	NUM
fbem-10171	272	14	.	.	PUNCT
fbem-10171	273	1	[	[	X
fbem-10171	273	2	43	43	NUM
fbem-10171	273	3	]	]	X
fbem-10171	273	4	shaw	shaw	PROPN
fbem-10171	273	5	p	p	PROPN
fbem-10171	273	6	,	,	PUNCT
fbem-10171	273	7	uszkoreit	uszkoreit	PROPN
fbem-10171	273	8	j	j	PROPN
fbem-10171	273	9	,	,	PUNCT
fbem-10171	273	10	vaswani	vaswani	ADJ
fbem-10171	273	11	a.	a.	NOUN
fbem-10171	273	12	self	self	NOUN
fbem-10171	273	13	-	-	PUNCT
fbem-10171	273	14	attention	attention	NOUN
fbem-10171	273	15	with	with	ADP
fbem-10171	273	16	relative	relative	ADJ
fbem-10171	273	17	position	position	NOUN
fbem-10171	273	18	representations[j	representations[j	PROPN
fbem-10171	273	19	]	]	PUNCT
fbem-10171	273	20	.	.	PUNCT
fbem-10171	274	1	arxiv	arxiv	PROPN
fbem-10171	274	2	preprint	preprint	PROPN
fbem-10171	274	3	arxiv:1803.02155	arxiv:1803.02155	PROPN
fbem-10171	274	4	,	,	PUNCT
fbem-10171	274	5	2018	2018	NUM
fbem-10171	274	6	.	.	PUNCT
fbem-10171	275	1	[	[	X
fbem-10171	275	2	44	44	NUM
fbem-10171	275	3	]	]	X
fbem-10171	275	4	voita	voita	PROPN
fbem-10171	275	5	e	e	PROPN
fbem-10171	275	6	,	,	PUNCT
fbem-10171	275	7	talbot	talbot	PROPN
fbem-10171	275	8	d	d	PROPN
fbem-10171	275	9	,	,	PUNCT
fbem-10171	275	10	moiseev	moiseev	PROPN
fbem-10171	275	11	f	f	NOUN
fbem-10171	275	12	,	,	PUNCT
fbem-10171	275	13	et	et	PROPN
fbem-10171	275	14	al	al	PROPN
fbem-10171	275	15	.	.	PUNCT
fbem-10171	276	1	analyzing	analyze	VERB
fbem-10171	276	2	multi	multi	ADJ
fbem-10171	276	3	-	-	ADJ
fbem-10171	276	4	head	head	ADJ
fbem-10171	276	5	selfattention	selfattention	NOUN
fbem-10171	276	6	:	:	PUNCT
fbem-10171	276	7	specialized	specialized	ADJ
fbem-10171	276	8	heads	head	NOUN
fbem-10171	276	9	do	do	VERB
fbem-10171	276	10	the	the	DET
fbem-10171	276	11	heavy	heavy	ADJ
fbem-10171	276	12	lifting	lifting	NOUN
fbem-10171	276	13	,	,	PUNCT
fbem-10171	276	14	the	the	DET
fbem-10171	276	15	rest	rest	NOUN
fbem-10171	276	16	can	can	AUX
fbem-10171	276	17	be	be	AUX
fbem-10171	276	18	pruned[j	pruned[j	ADV
fbem-10171	276	19	]	]	PUNCT
fbem-10171	276	20	.	.	PUNCT
fbem-10171	277	1	arxiv	arxiv	PROPN
fbem-10171	277	2	preprint	preprint	VERB
fbem-10171	277	3	arxiv:1905.09418	arxiv:1905.09418	PROPN
fbem-10171	277	4	,	,	PUNCT
fbem-10171	277	5	2019	2019	NUM
fbem-10171	277	6	.	.	PUNCT
fbem-10171	278	1	[	[	X
fbem-10171	278	2	45	45	NUM
fbem-10171	278	3	]	]	X
fbem-10171	278	4	chen	chen	PROPN
fbem-10171	278	5	chen	chen	PROPN
fbem-10171	278	6	,	,	PUNCT
fbem-10171	278	7	xia	xia	PROPN
fbem-10171	278	8	-	-	PUNCT
fbem-10171	278	9	bing	bing	VERB
fbem-10171	278	10	zhou	zhou	PROPN
fbem-10171	278	11	,	,	PUNCT
fbem-10171	278	12	zhong	zhong	PROPN
fbem-10171	278	13	-	-	PUNCT
fbem-10171	278	14	qing	qing	PROPN
fbem-10171	278	15	wang	wang	PROPN
fbem-10171	278	16	,	,	PUNCT
fbem-10171	278	17	et	et	PROPN
fbem-10171	278	18	al	al	PROPN
fbem-10171	278	19	.	.	PROPN
fbem-10171	278	20	conversational	conversational	ADJ
fbem-10171	278	21	sentiment	sentiment	NOUN
fbem-10171	278	22	classification	classification	NOUN
fbem-10171	278	23	based	base	VERB
fbem-10171	278	24	on	on	ADP
fbem-10171	278	25	multi	multi	ADJ
fbem-10171	278	26	-	-	ADJ
fbem-10171	278	27	party	party	ADJ
fbem-10171	278	28	attention	attention	NOUN
fbem-10171	278	29	modeling[j	modeling[j	NOUN
fbem-10171	278	30	]	]	PUNCT
fbem-10171	278	31	.	.	PUNCT
fbem-10171	279	1	chinese	chinese	ADJ
fbem-10171	279	2	journal	journal	PROPN
fbem-10171	279	3	of	of	ADP
fbem-10171	279	4	informatics	informatics	PROPN
fbem-10171	279	5	,	,	PUNCT
fbem-10171	279	6	2022	2022	NUM
fbem-10171	279	7	,	,	PUNCT
fbem-10171	279	8	36(12	36(12	NUM
fbem-10171	279	9	):	):	PUNCT
fbem-10171	279	10	173	173	NUM
fbem-10171	279	11	-	-	SYM
fbem-10171	279	12	181	181	NUM
fbem-10171	279	13	.	.	PUNCT
fbem-10171	280	1	[	[	X
fbem-10171	280	2	46	46	NUM
fbem-10171	280	3	]	]	X
fbem-10171	280	4	li	li	PROPN
fbem-10171	280	5	yuqing	yuqing	PROPN
fbem-10171	280	6	,	,	PUNCT
fbem-10171	280	7	li	li	PROPN
fbem-10171	280	8	xin	xin	PROPN
fbem-10171	280	9	,	,	PUNCT
fbem-10171	280	10	han	han	PROPN
fbem-10171	280	11	xu	xu	PROPN
fbem-10171	280	12	et	et	PROPN
fbem-10171	280	13	al	al	PROPN
fbem-10171	280	14	.	.	PUNCT
fbem-10171	281	1	a	a	DET
fbem-10171	281	2	bilingual	bilingual	ADJ
fbem-10171	281	3	lexicon	lexicon	NOUN
fbem-10171	281	4	-	-	PUNCT
fbem-10171	281	5	based	base	VERB
fbem-10171	281	6	sentiment	sentiment	NOUN
fbem-10171	281	7	analysis	analysis	NOUN
fbem-10171	281	8	method	method	NOUN
fbem-10171	281	9	for	for	ADP
fbem-10171	281	10	micro	micro	ADJ
fbem-10171	281	11	-	-	ADJ
fbem-10171	281	12	bodied	bodied	ADJ
fbem-10171	281	13	categories[j	categories[j	NOUN
fbem-10171	281	14	]	]	PUNCT
fbem-10171	281	15	.	.	PUNCT
fbem-10171	282	1	journal	journal	PROPN
fbem-10171	282	2	of	of	ADP
fbem-10171	282	3	electronics,2016,44(09):2068	electronics,2016,44(09):2068	NOUN
fbem-10171	282	4	-	-	PUNCT
fbem-10171	282	5	2073	2073	NUM
fbem-10171	282	6	.	.	PUNCT
fbem-10171	283	1	[	[	X
fbem-10171	283	2	47	47	NUM
fbem-10171	283	3	]	]	X
fbem-10171	283	4	zhao	zhao	PROPN
fbem-10171	283	5	yan	yan	PROPN
fbem-10171	283	6	-	-	PUNCT
fbem-10171	283	7	yan	yan	PROPN
fbem-10171	283	8	,	,	PUNCT
fbem-10171	283	9	qin	qin	PROPN
fbem-10171	283	10	bing	bing	PROPN
fbem-10171	283	11	,	,	PUNCT
fbem-10171	283	12	shi	shi	PROPN
fbem-10171	283	13	qiu	qiu	PROPN
fbem-10171	283	14	-	-	PUNCT
fbem-10171	283	15	hui	hui	PROPN
fbem-10171	283	16	et	et	PROPN
fbem-10171	283	17	al	al	PROPN
fbem-10171	283	18	.	.	PUNCT
fbem-10171	284	1	construction	construction	NOUN
fbem-10171	284	2	of	of	ADP
fbem-10171	284	3	large	large	ADJ
fbem-10171	284	4	-	-	PUNCT
fbem-10171	284	5	scale	scale	NOUN
fbem-10171	284	6	sentiment	sentiment	NOUN
fbem-10171	284	7	lexicon	lexicon	NOUN
fbem-10171	284	8	and	and	CCONJ
fbem-10171	284	9	its	its	PRON
fbem-10171	284	10	application	application	NOUN
fbem-10171	284	11	in	in	ADP
fbem-10171	284	12	sentiment	sentiment	NOUN
fbem-10171	284	13	classification[j	classification[j	PROPN
fbem-10171	284	14	]	]	PUNCT
fbem-10171	284	15	.	.	PUNCT
fbem-10171	285	1	journal	journal	PROPN
fbem-10171	285	2	of	of	ADP
fbem-10171	285	3	chinese	chinese	ADJ
fbem-10171	285	4	information	information	NOUN
fbem-10171	285	5	,	,	PUNCT
fbem-10171	285	6	2017	2017	NUM
fbem-10171	285	7	,	,	PUNCT
fbem-10171	285	8	31(02	31(02	NUM
fbem-10171	285	9	):	):	PUNCT
fbem-10171	285	10	187	187	NUM
fbem-10171	285	11	-	-	SYM
fbem-10171	285	12	193	193	NUM
fbem-10171	285	13	.	.	PUNCT
fbem-10171	286	1	[	[	X
fbem-10171	286	2	48	48	NUM
fbem-10171	286	3	]	]	X
fbem-10171	286	4	asghar	asghar	PROPN
fbem-10171	286	5	m	m	PROPN
fbem-10171	286	6	z	z	PROPN
fbem-10171	286	7	,	,	PUNCT
fbem-10171	286	8	khan	khan	PROPN
fbem-10171	286	9	a	a	PROPN
fbem-10171	286	10	,	,	PUNCT
fbem-10171	286	11	ahmad	ahmad	PROPN
fbem-10171	286	12	s	s	PROPN
fbem-10171	286	13	,	,	PUNCT
fbem-10171	286	14	et	et	PROPN
fbem-10171	286	15	al	al	PROPN
fbem-10171	286	16	.	.	PROPN
fbem-10171	286	17	lexicon	lexicon	NOUN
fbem-10171	286	18	-	-	PUNCT
fbem-10171	286	19	enhanced	enhance	VERB
fbem-10171	286	20	sentiment	sentiment	NOUN
fbem-10171	286	21	analysis	analysis	NOUN
fbem-10171	286	22	framework	framework	NOUN
fbem-10171	286	23	using	use	VERB
fbem-10171	286	24	rule	rule	NOUN
fbem-10171	286	25	-	-	PUNCT
fbem-10171	286	26	based	base	VERB
fbem-10171	286	27	classification	classification	NOUN
fbem-10171	286	28	scheme[j	scheme[j	NOUN
fbem-10171	286	29	]	]	PUNCT
fbem-10171	286	30	.	.	PUNCT
fbem-10171	287	1	plos	plos	PROPN
fbem-10171	287	2	one	one	NUM
fbem-10171	287	3	,	,	PUNCT
fbem-10171	287	4	2017	2017	NUM
fbem-10171	287	5	,	,	PUNCT
fbem-10171	287	6	12(2	12(2	NUM
fbem-10171	287	7	):	):	PUNCT
fbem-10171	287	8	e0171649	e0171649	ADJ
fbem-10171	287	9	.	.	PUNCT
fbem-10171	288	1	[	[	X
fbem-10171	288	2	49	49	NUM
fbem-10171	288	3	]	]	X
fbem-10171	288	4	peng	peng	PROPN
fbem-10171	288	5	m	m	PROPN
fbem-10171	288	6	,	,	PUNCT
fbem-10171	288	7	xi	xi	PROPN
fbem-10171	288	8	j	j	PROPN
fbem-10171	288	9	,	,	PUNCT
fbem-10171	288	10	dai	dai	PROPN
fbem-10171	288	11	xy	xy	PROPN
fbem-10171	288	12	,	,	PUNCT
fbem-10171	288	13	et	et	PROPN
fbem-10171	288	14	al	al	PROPN
fbem-10171	288	15	.	.	PROPN
fbem-10171	288	16	collaborative	collaborative	ADJ
fbem-10171	288	17	filtering	filtering	NOUN
fbem-10171	288	18	recommendation	recommendation	NOUN
fbem-10171	288	19	algorithm	algorithm	NOUN
fbem-10171	288	20	based	base	VERB
fbem-10171	288	21	on	on	ADP
fbem-10171	288	22	sentiment	sentiment	NOUN
fbem-10171	288	23	analysis	analysis	NOUN
fbem-10171	288	24	and	and	CCONJ
fbem-10171	288	25	lda	lda	VERB
fbem-10171	288	26	topic	topic	NOUN
fbem-10171	288	27	model[j	model[j	PROPN
fbem-10171	288	28	]	]	PUNCT
fbem-10171	288	29	.	.	PUNCT
fbem-10171	289	1	chinese	chinese	ADJ
fbem-10171	289	2	journal	journal	PROPN
fbem-10171	289	3	of	of	ADP
fbem-10171	289	4	information	information	NOUN
fbem-10171	289	5	,	,	PUNCT
fbem-10171	289	6	2017	2017	NUM
fbem-10171	289	7	,	,	PUNCT
fbem-10171	289	8	31(02):194	31(02):194	NUM
fbem-10171	289	9	-	-	SYM
fbem-10171	289	10	203	203	NUM
fbem-10171	289	11	.	.	PUNCT
fbem-10171	290	1	[	[	X
fbem-10171	290	2	50	50	NUM
fbem-10171	290	3	]	]	X
fbem-10171	290	4	li	li	PROPN
fbem-10171	290	5	jinmeng	jinmeng	PROPN
fbem-10171	290	6	.	.	PUNCT
fbem-10171	291	1	research	research	NOUN
fbem-10171	291	2	on	on	ADP
fbem-10171	291	3	natural	natural	ADJ
fbem-10171	291	4	language	language	NOUN
fbem-10171	291	5	emotion	emotion	NOUN
fbem-10171	291	6	analysis	analysis	NOUN
fbem-10171	291	7	based	base	VERB
fbem-10171	291	8	on	on	ADP
fbem-10171	291	9	eye	eye	NOUN
fbem-10171	291	10	-	-	PUNCT
fbem-10171	291	11	tracking	tracking	NOUN
fbem-10171	292	1	[	[	X
fbem-10171	292	2	d	d	X
fbem-10171	292	3	]	]	X
fbem-10171	292	4	.	.	PUNCT
fbem-10171	293	1	beijing	beijing	PROPN
fbem-10171	293	2	university	university	PROPN
fbem-10171	293	3	of	of	ADP
fbem-10171	293	4	posts	post	NOUN
fbem-10171	293	5	and	and	CCONJ
fbem-10171	293	6	telecommunications	telecommunication	NOUN
fbem-10171	293	7	,	,	PUNCT
fbem-10171	293	8	2021	2021	NUM
fbem-10171	293	9	.	.	PUNCT
fbem-10171	294	1	doi:10.26969	doi:10.26969	ADJ
fbem-10171	294	2	/	/	SYM
fbem-10171	294	3	d.cnki.gbydu.2021.001688	d.cnki.gbydu.2021.001688	NOUN
fbem-10171	294	4	.	.	PUNCT