id	sid	tid	token	lemma	pos
sc-5860	1	1	eai	eai	PROPN
sc-5860	1	2	endorsed	endorse	VERB
sc-5860	1	3	transactions	transaction	NOUN
sc-5860	1	4	on	on	ADP
sc-5860	1	5	smart	smart	ADJ
sc-5860	1	6	cities	city	NOUN
sc-5860	1	7	research	research	NOUN
sc-5860	1	8	article	article	NOUN
sc-5860	1	9	multimodal	multimodal	NOUN
sc-5860	1	10	sentiment	sentiment	NOUN
sc-5860	1	11	analysis	analysis	NOUN
sc-5860	1	12	in	in	ADP
sc-5860	1	13	natural	natural	ADJ
sc-5860	1	14	disaster	disaster	NOUN
sc-5860	1	15	data	datum	NOUN
sc-5860	1	16	on	on	ADP
sc-5860	1	17	social	social	ADJ
sc-5860	1	18	media	medium	NOUN
sc-5860	1	19	sefa	sefa	NOUN
sc-5860	1	20	dursun1	dursun1	PROPN
sc-5860	1	21	and	and	CCONJ
sc-5860	1	22	süleyman	süleyman	ADJ
sc-5860	1	23	eken1,∗	eken1,∗	PROPN
sc-5860	1	24	1kocaeli	1kocaeli	NUM
sc-5860	1	25	university	university	NOUN
sc-5860	1	26	,	,	PUNCT
sc-5860	1	27	department	department	NOUN
sc-5860	1	28	of	of	ADP
sc-5860	1	29	information	information	NOUN
sc-5860	1	30	systems	system	NOUN
sc-5860	1	31	engineering	engineering	NOUN
sc-5860	1	32	,	,	PUNCT
sc-5860	1	33	i̇zmit	i̇zmit	NOUN
sc-5860	1	34	41001	41001	NUM
sc-5860	1	35	,	,	PUNCT
sc-5860	1	36	türkiye	türkiye	PROPN
sc-5860	1	37	abstract	abstract	ADJ
sc-5860	1	38	introduction	introduction	NOUN
sc-5860	1	39	:	:	PUNCT
sc-5860	1	40	with	with	ADP
sc-5860	1	41	the	the	DET
sc-5860	1	42	development	development	NOUN
sc-5860	1	43	of	of	ADP
sc-5860	1	44	the	the	DET
sc-5860	1	45	internet	internet	NOUN
sc-5860	1	46	,	,	PUNCT
sc-5860	1	47	users	user	NOUN
sc-5860	1	48	tend	tend	VERB
sc-5860	1	49	to	to	PART
sc-5860	1	50	express	express	VERB
sc-5860	1	51	their	their	PRON
sc-5860	1	52	opinions	opinion	NOUN
sc-5860	1	53	and	and	CCONJ
sc-5860	1	54	emotions	emotion	NOUN
sc-5860	1	55	through	through	ADP
sc-5860	1	56	text	text	NOUN
sc-5860	1	57	,	,	PUNCT
sc-5860	1	58	visual	visual	ADJ
sc-5860	1	59	and/or	and/or	CCONJ
sc-5860	1	60	audio	audio	ADJ
sc-5860	1	61	content	content	NOUN
sc-5860	1	62	.	.	PUNCT
sc-5860	2	1	this	this	PRON
sc-5860	2	2	has	have	AUX
sc-5860	2	3	increased	increase	VERB
sc-5860	2	4	the	the	DET
sc-5860	2	5	interest	interest	NOUN
sc-5860	2	6	in	in	ADP
sc-5860	2	7	multimodal	multimodal	NOUN
sc-5860	2	8	analysis	analysis	NOUN
sc-5860	2	9	methods	method	NOUN
sc-5860	2	10	.	.	PUNCT
sc-5860	3	1	objectives	objective	NOUN
sc-5860	3	2	:	:	PUNCT
sc-5860	3	3	this	this	DET
sc-5860	3	4	study	study	NOUN
sc-5860	3	5	addresses	address	VERB
sc-5860	3	6	multimodal	multimodal	ADJ
sc-5860	3	7	sentiment	sentiment	NOUN
sc-5860	3	8	analysis	analysis	NOUN
sc-5860	3	9	(	(	PUNCT
sc-5860	3	10	msa	msa	PROPN
sc-5860	3	11	)	)	PUNCT
sc-5860	3	12	on	on	ADP
sc-5860	3	13	tweets	tweet	NOUN
sc-5860	3	14	related	relate	VERB
sc-5860	3	15	to	to	ADP
sc-5860	3	16	natural	natural	ADJ
sc-5860	3	17	disasters	disaster	NOUN
sc-5860	3	18	by	by	ADP
sc-5860	3	19	combining	combine	VERB
sc-5860	3	20	textual	textual	ADJ
sc-5860	3	21	and	and	CCONJ
sc-5860	3	22	visual	visual	ADJ
sc-5860	3	23	embeddings	embedding	NOUN
sc-5860	3	24	.	.	PUNCT
sc-5860	4	1	methods	method	NOUN
sc-5860	4	2	:	:	PUNCT
sc-5860	4	3	the	the	DET
sc-5860	4	4	use	use	NOUN
sc-5860	4	5	of	of	ADP
sc-5860	4	6	textual	textual	ADJ
sc-5860	4	7	representations	representation	NOUN
sc-5860	4	8	together	together	ADV
sc-5860	4	9	with	with	ADP
sc-5860	4	10	the	the	DET
sc-5860	4	11	emotional	emotional	ADJ
sc-5860	4	12	expressions	expression	NOUN
sc-5860	4	13	of	of	ADP
sc-5860	4	14	the	the	DET
sc-5860	4	15	visual	visual	ADJ
sc-5860	4	16	content	content	NOUN
sc-5860	4	17	provides	provide	VERB
sc-5860	4	18	a	a	DET
sc-5860	4	19	more	more	ADV
sc-5860	4	20	comprehensive	comprehensive	ADJ
sc-5860	4	21	analysis	analysis	NOUN
sc-5860	4	22	.	.	PUNCT
sc-5860	5	1	to	to	PART
sc-5860	5	2	investigate	investigate	VERB
sc-5860	5	3	the	the	DET
sc-5860	5	4	impact	impact	NOUN
sc-5860	5	5	of	of	ADP
sc-5860	5	6	high	high	ADJ
sc-5860	5	7	-	-	PUNCT
sc-5860	5	8	level	level	NOUN
sc-5860	5	9	visual	visual	ADJ
sc-5860	5	10	and	and	CCONJ
sc-5860	5	11	texual	texual	ADJ
sc-5860	5	12	features	feature	NOUN
sc-5860	5	13	,	,	PUNCT
sc-5860	5	14	a	a	DET
sc-5860	5	15	three	three	NUM
sc-5860	5	16	-	-	PUNCT
sc-5860	5	17	layer	layer	NOUN
sc-5860	5	18	neural	neural	ADJ
sc-5860	5	19	network	network	NOUN
sc-5860	5	20	is	be	AUX
sc-5860	5	21	used	use	VERB
sc-5860	5	22	in	in	ADP
sc-5860	5	23	the	the	DET
sc-5860	5	24	study	study	NOUN
sc-5860	5	25	,	,	PUNCT
sc-5860	5	26	where	where	SCONJ
sc-5860	5	27	the	the	DET
sc-5860	5	28	first	first	ADJ
sc-5860	5	29	two	two	NUM
sc-5860	5	30	layers	layer	NOUN
sc-5860	5	31	collect	collect	VERB
sc-5860	5	32	features	feature	NOUN
sc-5860	5	33	from	from	ADP
sc-5860	5	34	different	different	ADJ
sc-5860	5	35	modalities	modality	NOUN
sc-5860	5	36	and	and	CCONJ
sc-5860	5	37	the	the	DET
sc-5860	5	38	third	third	ADJ
sc-5860	5	39	layer	layer	NOUN
sc-5860	5	40	is	be	AUX
sc-5860	5	41	used	use	VERB
sc-5860	5	42	to	to	PART
sc-5860	5	43	analyze	analyze	VERB
sc-5860	5	44	sentiments	sentiment	NOUN
sc-5860	5	45	.	.	PUNCT
sc-5860	6	1	results	result	NOUN
sc-5860	6	2	:	:	PUNCT
sc-5860	6	3	according	accord	VERB
sc-5860	6	4	to	to	ADP
sc-5860	6	5	experimental	experimental	ADJ
sc-5860	6	6	tests	test	NOUN
sc-5860	6	7	on	on	ADP
sc-5860	6	8	our	our	PRON
sc-5860	6	9	dataset	dataset	NOUN
sc-5860	6	10	,	,	PUNCT
sc-5860	6	11	the	the	DET
sc-5860	6	12	highest	high	ADJ
sc-5860	6	13	performance	performance	NOUN
sc-5860	6	14	values	value	NOUN
sc-5860	6	15	(	(	PUNCT
sc-5860	6	16	77	77	NUM
sc-5860	6	17	%	%	NOUN
sc-5860	6	18	accuracy	accuracy	NOUN
sc-5860	6	19	,	,	PUNCT
sc-5860	6	20	71	71	NUM
sc-5860	6	21	%	%	NOUN
sc-5860	6	22	f1	f1	NOUN
sc-5860	6	23	-	-	PUNCT
sc-5860	6	24	score	score	NOUN
sc-5860	6	25	)	)	PUNCT
sc-5860	6	26	are	be	AUX
sc-5860	6	27	achieved	achieve	VERB
sc-5860	6	28	by	by	ADP
sc-5860	6	29	using	use	VERB
sc-5860	6	30	the	the	DET
sc-5860	6	31	clip	clip	NOUN
sc-5860	6	32	model	model	NOUN
sc-5860	6	33	in	in	ADP
sc-5860	6	34	the	the	DET
sc-5860	6	35	image	image	NOUN
sc-5860	6	36	and	and	CCONJ
sc-5860	6	37	the	the	DET
sc-5860	6	38	roberta	roberta	PROPN
sc-5860	6	39	model	model	NOUN
sc-5860	6	40	in	in	ADP
sc-5860	6	41	the	the	DET
sc-5860	6	42	text	text	NOUN
sc-5860	6	43	.	.	PUNCT
sc-5860	7	1	conclusion	conclusion	NOUN
sc-5860	7	2	:	:	PUNCT
sc-5860	7	3	such	such	ADJ
sc-5860	7	4	analyzes	analyze	NOUN
sc-5860	7	5	can	can	AUX
sc-5860	7	6	be	be	AUX
sc-5860	7	7	used	use	VERB
sc-5860	7	8	in	in	ADP
sc-5860	7	9	different	different	ADJ
sc-5860	7	10	application	application	NOUN
sc-5860	7	11	areas	area	NOUN
sc-5860	7	12	such	such	ADJ
sc-5860	7	13	as	as	ADP
sc-5860	7	14	agencies	agency	NOUN
sc-5860	7	15	,	,	PUNCT
sc-5860	7	16	advertising	advertising	NOUN
sc-5860	7	17	,	,	PUNCT
sc-5860	7	18	social	social	ADJ
sc-5860	7	19	/	/	SYM
sc-5860	7	20	digital	digital	ADJ
sc-5860	7	21	media	medium	NOUN
sc-5860	7	22	content	content	NOUN
sc-5860	7	23	producers	producer	NOUN
sc-5860	7	24	,	,	PUNCT
sc-5860	7	25	humanitarian	humanitarian	ADJ
sc-5860	7	26	aid	aid	NOUN
sc-5860	7	27	organizations	organization	NOUN
sc-5860	7	28	and	and	CCONJ
sc-5860	7	29	can	can	AUX
sc-5860	7	30	provide	provide	VERB
sc-5860	7	31	important	important	ADJ
sc-5860	7	32	information	information	NOUN
sc-5860	7	33	in	in	ADP
sc-5860	7	34	terms	term	NOUN
sc-5860	7	35	of	of	ADP
sc-5860	7	36	social	social	ADJ
sc-5860	7	37	awareness	awareness	NOUN
sc-5860	7	38	.	.	PUNCT
sc-5860	8	1	received	receive	VERB
sc-5860	8	2	on	on	ADP
sc-5860	8	3	22	22	NUM
sc-5860	8	4	04	04	NUM
sc-5860	8	5	2024	2024	NUM
sc-5860	8	6	;	;	PUNCT
sc-5860	8	7	accepted	accept	VERB
sc-5860	8	8	on	on	ADP
sc-5860	8	9	03	03	NUM
sc-5860	8	10	11	11	NUM
sc-5860	8	11	2024	2024	NUM
sc-5860	8	12	;	;	PUNCT
sc-5860	8	13	published	publish	VERB
sc-5860	8	14	on	on	ADP
sc-5860	8	15	13	13	NUM
sc-5860	8	16	11	11	NUM
sc-5860	8	17	2024	2024	NUM
sc-5860	8	18	keywords	keyword	NOUN
sc-5860	8	19	:	:	PUNCT
sc-5860	8	20	multimodal	multimodal	ADJ
sc-5860	8	21	learning	learning	NOUN
sc-5860	8	22	,	,	PUNCT
sc-5860	8	23	sentiment	sentiment	NOUN
sc-5860	8	24	analysis	analysis	NOUN
sc-5860	8	25	,	,	PUNCT
sc-5860	8	26	natural	natural	ADJ
sc-5860	8	27	disaster	disaster	NOUN
sc-5860	8	28	,	,	PUNCT
sc-5860	8	29	natural	natural	ADJ
sc-5860	8	30	language	language	NOUN
sc-5860	8	31	processing	processing	NOUN
sc-5860	8	32	,	,	PUNCT
sc-5860	8	33	image	image	NOUN
sc-5860	8	34	processing	processing	NOUN
sc-5860	8	35	copyright	copyright	NOUN
sc-5860	8	36	©	©	ADP
sc-5860	8	37	2024	2024	NUM
sc-5860	8	38	s.	s.	PROPN
sc-5860	8	39	dursun	dursun	PROPN
sc-5860	8	40	and	and	CCONJ
sc-5860	8	41	s.	s.	PROPN
sc-5860	8	42	eken	eken	PROPN
sc-5860	8	43	,	,	PUNCT
sc-5860	8	44	licensed	license	VERB
sc-5860	8	45	to	to	ADP
sc-5860	8	46	eai	eai	PROPN
sc-5860	8	47	.	.	PUNCT
sc-5860	9	1	this	this	PRON
sc-5860	9	2	is	be	AUX
sc-5860	9	3	an	an	DET
sc-5860	9	4	open	open	ADJ
sc-5860	9	5	access	access	NOUN
sc-5860	9	6	article	article	NOUN
sc-5860	9	7	distributed	distribute	VERB
sc-5860	9	8	under	under	ADP
sc-5860	9	9	the	the	DET
sc-5860	9	10	terms	term	NOUN
sc-5860	9	11	of	of	ADP
sc-5860	9	12	the	the	DET
sc-5860	9	13	cc	cc	PROPN
sc-5860	9	14	by	by	ADP
sc-5860	9	15	-	-	PUNCT
sc-5860	9	16	nc	nc	PROPN
sc-5860	9	17	-	-	PUNCT
sc-5860	9	18	sa	sa	PROPN
sc-5860	9	19	4.0	4.0	NUM
sc-5860	9	20	,	,	PUNCT
sc-5860	9	21	which	which	PRON
sc-5860	9	22	permits	permit	VERB
sc-5860	9	23	copying	copy	VERB
sc-5860	9	24	,	,	PUNCT
sc-5860	9	25	redistributing	redistributing	NOUN
sc-5860	9	26	,	,	PUNCT
sc-5860	9	27	remixing	remixing	NOUN
sc-5860	9	28	,	,	PUNCT
sc-5860	9	29	transformation	transformation	NOUN
sc-5860	9	30	,	,	PUNCT
sc-5860	9	31	and	and	CCONJ
sc-5860	9	32	building	build	VERB
sc-5860	9	33	upon	upon	SCONJ
sc-5860	9	34	the	the	DET
sc-5860	9	35	material	material	NOUN
sc-5860	9	36	in	in	ADP
sc-5860	9	37	any	any	DET
sc-5860	9	38	medium	medium	NOUN
sc-5860	9	39	so	so	ADV
sc-5860	9	40	long	long	ADV
sc-5860	9	41	as	as	SCONJ
sc-5860	9	42	the	the	DET
sc-5860	9	43	original	original	ADJ
sc-5860	9	44	work	work	NOUN
sc-5860	9	45	is	be	AUX
sc-5860	9	46	properly	properly	ADV
sc-5860	9	47	cited	cite	VERB
sc-5860	9	48	.	.	PUNCT
sc-5860	10	1	doi:10.4108	doi:10.4108	VERB
sc-5860	10	2	/	/	SYM
sc-5860	10	3	eetsc.5860	eetsc.5860	PROPN
sc-5860	10	4	1	1	NUM
sc-5860	10	5	.	.	PUNCT
sc-5860	10	6	introduction	introduction	NOUN
sc-5860	10	7	with	with	ADP
sc-5860	10	8	the	the	DET
sc-5860	10	9	advent	advent	NOUN
sc-5860	10	10	of	of	ADP
sc-5860	10	11	the	the	DET
sc-5860	10	12	internet	internet	NOUN
sc-5860	10	13	,	,	PUNCT
sc-5860	10	14	social	social	ADJ
sc-5860	10	15	media	medium	NOUN
sc-5860	10	16	platforms	platform	NOUN
sc-5860	10	17	in	in	ADP
sc-5860	10	18	particular	particular	ADJ
sc-5860	10	19	have	have	AUX
sc-5860	10	20	become	become	VERB
sc-5860	10	21	multimodal	multimodal	ADJ
sc-5860	10	22	,	,	PUNCT
sc-5860	10	23	with	with	ADP
sc-5860	10	24	content	content	NOUN
sc-5860	10	25	containing	contain	VERB
sc-5860	10	26	text	text	NOUN
sc-5860	10	27	,	,	PUNCT
sc-5860	10	28	audio	audio	NOUN
sc-5860	10	29	,	,	PUNCT
sc-5860	10	30	images	image	NOUN
sc-5860	10	31	,	,	PUNCT
sc-5860	10	32	and	and	CCONJ
sc-5860	10	33	videos	video	NOUN
sc-5860	10	34	to	to	PART
sc-5860	10	35	evoke	evoke	VERB
sc-5860	10	36	different	different	ADJ
sc-5860	10	37	emotions	emotion	NOUN
sc-5860	10	38	of	of	ADP
sc-5860	10	39	a	a	DET
sc-5860	10	40	user	user	NOUN
sc-5860	10	41	.	.	PUNCT
sc-5860	11	1	the	the	DET
sc-5860	11	2	increasing	increase	VERB
sc-5860	11	3	popularity	popularity	NOUN
sc-5860	11	4	of	of	ADP
sc-5860	11	5	social	social	ADJ
sc-5860	11	6	networks	network	NOUN
sc-5860	11	7	and	and	CCONJ
sc-5860	11	8	the	the	DET
sc-5860	11	9	tendency	tendency	NOUN
sc-5860	11	10	of	of	ADP
sc-5860	11	11	users	user	NOUN
sc-5860	11	12	to	to	PART
sc-5860	11	13	share	share	VERB
sc-5860	11	14	their	their	PRON
sc-5860	11	15	emotions	emotion	NOUN
sc-5860	11	16	,	,	PUNCT
sc-5860	11	17	expressions	expression	NOUN
sc-5860	11	18	,	,	PUNCT
sc-5860	11	19	and	and	CCONJ
sc-5860	11	20	thoughts	thought	NOUN
sc-5860	11	21	in	in	ADP
sc-5860	11	22	text	text	NOUN
sc-5860	11	23	,	,	PUNCT
sc-5860	11	24	visual	visual	ADJ
sc-5860	11	25	,	,	PUNCT
sc-5860	11	26	and	and	CCONJ
sc-5860	11	27	audio	audio	NOUN
sc-5860	11	28	content	content	NOUN
sc-5860	11	29	have	have	AUX
sc-5860	11	30	created	create	VERB
sc-5860	11	31	new	new	ADJ
sc-5860	11	32	opportunities	opportunity	NOUN
sc-5860	11	33	and	and	CCONJ
sc-5860	11	34	challenges	challenge	NOUN
sc-5860	11	35	in	in	ADP
sc-5860	11	36	sentiment	sentiment	NOUN
sc-5860	11	37	analysis	analysis	NOUN
sc-5860	11	38	[	[	X
sc-5860	11	39	1	1	NUM
sc-5860	11	40	]	]	PUNCT
sc-5860	11	41	.	.	PUNCT
sc-5860	12	1	while	while	SCONJ
sc-5860	12	2	sentiment	sentiment	NOUN
sc-5860	12	3	analysis	analysis	NOUN
sc-5860	12	4	from	from	ADP
sc-5860	12	5	texts	text	NOUN
sc-5860	12	6	has	have	AUX
sc-5860	12	7	been	be	AUX
sc-5860	12	8	widely	widely	ADV
sc-5860	12	9	researched	research	VERB
sc-5860	12	10	in	in	ADP
sc-5860	12	11	the	the	DET
sc-5860	12	12	literature	literature	NOUN
sc-5860	12	13	[	[	X
sc-5860	12	14	2–5	2–5	X
sc-5860	12	15	]	]	PUNCT
sc-5860	12	16	,	,	PUNCT
sc-5860	12	17	sentiment	sentiment	NOUN
sc-5860	12	18	analysis	analysis	NOUN
sc-5860	12	19	from	from	ADP
sc-5860	12	20	images	image	NOUN
sc-5860	12	21	and	and	CCONJ
sc-5860	12	22	videos	video	NOUN
sc-5860	12	23	is	be	AUX
sc-5860	12	24	relatively	relatively	ADV
sc-5860	12	25	new	new	ADJ
sc-5860	12	26	.	.	PUNCT
sc-5860	13	1	within	within	ADP
sc-5860	13	2	the	the	DET
sc-5860	13	3	scope	scope	NOUN
sc-5860	13	4	of	of	ADP
sc-5860	13	5	this	this	DET
sc-5860	13	6	study	study	NOUN
sc-5860	13	7	,	,	PUNCT
sc-5860	13	8	we	we	PRON
sc-5860	13	9	focused	focus	VERB
sc-5860	13	10	on	on	ADP
sc-5860	13	11	the	the	DET
sc-5860	13	12	joint	joint	ADJ
sc-5860	13	13	analysis	analysis	NOUN
sc-5860	13	14	of	of	ADP
sc-5860	13	15	visual	visual	ADJ
sc-5860	13	16	and	and	CCONJ
sc-5860	13	17	textual	textual	ADJ
sc-5860	13	18	content	content	NOUN
sc-5860	13	19	in	in	ADP
sc-5860	13	20	a	a	DET
sc-5860	13	21	socially	socially	ADV
sc-5860	13	22	important	important	ADJ
sc-5860	13	23	area	area	NOUN
sc-5860	13	24	and	and	CCONJ
sc-5860	13	25	examined	examine	VERB
sc-5860	13	26	the	the	DET
sc-5860	13	27	effect	effect	NOUN
sc-5860	13	28	of	of	ADP
sc-5860	13	29	different	different	ADJ
sc-5860	13	30	visual	visual	ADJ
sc-5860	13	31	features	feature	VERB
sc-5860	13	32	together	together	ADV
sc-5860	13	33	with	with	ADP
sc-5860	13	34	contextual	contextual	ADJ
sc-5860	13	35	text	text	NOUN
sc-5860	13	36	representations	representation	NOUN
sc-5860	13	37	for	for	ADP
sc-5860	13	38	multimodal	multimodal	NOUN
sc-5860	13	39	tweet	tweet	NOUN
sc-5860	13	40	sentiment	sentiment	NOUN
sc-5860	13	41	classification	classification	NOUN
sc-5860	13	42	.	.	PUNCT
sc-5860	14	1	∗corresponding	∗corresponde	VERB
sc-5860	14	2	author	author	NOUN
sc-5860	14	3	.	.	PUNCT
sc-5860	15	1	email	email	NOUN
sc-5860	15	2	:	:	PUNCT
sc-5860	15	3	suleyman.eken@kocaeli.edu.tr	suleyman.eken@kocaeli.edu.tr	PROPN
sc-5860	15	4	based	base	VERB
sc-5860	15	5	on	on	ADP
sc-5860	15	6	the	the	DET
sc-5860	15	7	approach	approach	NOUN
sc-5860	15	8	of	of	ADP
sc-5860	15	9	“	"	PUNCT
sc-5860	15	10	a	a	DET
sc-5860	15	11	picture	picture	NOUN
sc-5860	15	12	is	be	AUX
sc-5860	15	13	worth	worth	ADJ
sc-5860	15	14	a	a	DET
sc-5860	15	15	thousand	thousand	NUM
sc-5860	15	16	words	word	NOUN
sc-5860	15	17	”	"	PUNCT
sc-5860	15	18	,	,	PUNCT
sc-5860	15	19	visuals	visual	NOUN
sc-5860	15	20	are	be	AUX
sc-5860	15	21	an	an	DET
sc-5860	15	22	effective	effective	ADJ
sc-5860	15	23	tool	tool	NOUN
sc-5860	15	24	to	to	PART
sc-5860	15	25	convey	convey	VERB
sc-5860	15	26	not	not	PART
sc-5860	15	27	only	only	ADV
sc-5860	15	28	facts	fact	NOUN
sc-5860	15	29	but	but	CCONJ
sc-5860	15	30	also	also	ADV
sc-5860	15	31	clues	clue	VERB
sc-5860	15	32	about	about	ADP
sc-5860	15	33	feelings	feeling	NOUN
sc-5860	15	34	and	and	CCONJ
sc-5860	15	35	emotions	emotion	NOUN
sc-5860	15	36	.	.	PUNCT
sc-5860	16	1	such	such	ADJ
sc-5860	16	2	clues	clue	NOUN
sc-5860	16	3	representing	represent	VERB
sc-5860	16	4	emotions	emotion	NOUN
sc-5860	16	5	and	and	CCONJ
sc-5860	16	6	thoughts	thought	NOUN
sc-5860	16	7	can	can	AUX
sc-5860	16	8	trigger	trigger	VERB
sc-5860	16	9	similar	similar	ADJ
sc-5860	16	10	emotions	emotion	NOUN
sc-5860	16	11	in	in	ADP
sc-5860	16	12	the	the	DET
sc-5860	16	13	observer	observer	NOUN
sc-5860	16	14	and	and	CCONJ
sc-5860	16	15	help	help	AUX
sc-5860	16	16	understand	understand	VERB
sc-5860	16	17	visual	visual	ADJ
sc-5860	16	18	content	content	NOUN
sc-5860	16	19	beyond	beyond	ADP
sc-5860	16	20	textual	textual	ADJ
sc-5860	16	21	concepts	concept	NOUN
sc-5860	16	22	in	in	ADP
sc-5860	16	23	different	different	ADJ
sc-5860	16	24	application	application	NOUN
sc-5860	16	25	areas	area	NOUN
sc-5860	16	26	such	such	ADJ
sc-5860	16	27	as	as	ADP
sc-5860	16	28	education	education	NOUN
sc-5860	16	29	,	,	PUNCT
sc-5860	16	30	entertainment	entertainment	NOUN
sc-5860	16	31	,	,	PUNCT
sc-5860	16	32	advertising	advertising	NOUN
sc-5860	16	33	,	,	PUNCT
sc-5860	16	34	journalism	journalism	NOUN
sc-5860	16	35	,	,	PUNCT
sc-5860	16	36	and	and	CCONJ
sc-5860	16	37	smart	smart	ADJ
sc-5860	16	38	cities	city	NOUN
sc-5860	16	39	[	[	X
sc-5860	16	40	6	6	NUM
sc-5860	16	41	]	]	PUNCT
sc-5860	16	42	.	.	PUNCT
sc-5860	17	1	however	however	ADV
sc-5860	17	2	,	,	PUNCT
sc-5860	17	3	it	it	PRON
sc-5860	17	4	is	be	AUX
sc-5860	17	5	not	not	PART
sc-5860	17	6	entirely	entirely	ADV
sc-5860	17	7	clear	clear	ADJ
sc-5860	17	8	how	how	SCONJ
sc-5860	17	9	such	such	ADJ
sc-5860	17	10	emotional	emotional	ADJ
sc-5860	17	11	cues	cue	NOUN
sc-5860	17	12	can	can	AUX
sc-5860	17	13	be	be	AUX
sc-5860	17	14	evoked	evoke	VERB
sc-5860	17	15	by	by	ADP
sc-5860	17	16	visual	visual	ADJ
sc-5860	17	17	content	content	NOUN
sc-5860	17	18	and	and	CCONJ
sc-5860	17	19	,	,	PUNCT
sc-5860	17	20	more	more	ADV
sc-5860	17	21	importantly	importantly	ADV
sc-5860	17	22	,	,	PUNCT
sc-5860	17	23	how	how	SCONJ
sc-5860	17	24	emotions	emotion	NOUN
sc-5860	17	25	derived	derive	VERB
sc-5860	17	26	from	from	ADP
sc-5860	17	27	a	a	DET
sc-5860	17	28	scene	scene	NOUN
sc-5860	17	29	can	can	AUX
sc-5860	17	30	be	be	AUX
sc-5860	17	31	expressed	express	VERB
sc-5860	17	32	by	by	ADP
sc-5860	17	33	an	an	DET
sc-5860	17	34	automatic	automatic	ADJ
sc-5860	17	35	algorithm	algorithm	NOUN
sc-5860	17	36	.	.	PUNCT
sc-5860	18	1	in	in	ADP
sc-5860	18	2	this	this	DET
sc-5860	18	3	context	context	NOUN
sc-5860	18	4	,	,	PUNCT
sc-5860	18	5	the	the	DET
sc-5860	18	6	proposed	propose	VERB
sc-5860	18	7	system	system	NOUN
sc-5860	18	8	and	and	CCONJ
sc-5860	18	9	the	the	DET
sc-5860	18	10	analyses	analysis	NOUN
sc-5860	18	11	performed	perform	VERB
sc-5860	18	12	;	;	PUNCT
sc-5860	18	13	it	it	PRON
sc-5860	18	14	is	be	AUX
sc-5860	18	15	thought	think	VERB
sc-5860	18	16	to	to	PART
sc-5860	18	17	contribute	contribute	VERB
sc-5860	18	18	to	to	ADP
sc-5860	18	19	different	different	ADJ
sc-5860	18	20	stakeholders	stakeholder	NOUN
sc-5860	18	21	such	such	ADJ
sc-5860	18	22	as	as	ADP
sc-5860	18	23	news	news	NOUN
sc-5860	18	24	publishers	publisher	NOUN
sc-5860	18	25	,	,	PUNCT
sc-5860	18	26	agencies	agency	NOUN
sc-5860	18	27	,	,	PUNCT
sc-5860	18	28	social	social	ADJ
sc-5860	18	29	/	/	SYM
sc-5860	18	30	digital	digital	ADJ
sc-5860	18	31	media	medium	NOUN
sc-5860	18	32	content	content	NOUN
sc-5860	18	33	producers	producer	NOUN
sc-5860	18	34	,	,	PUNCT
sc-5860	18	35	humanitarian	humanitarian	ADJ
sc-5860	18	36	organizations	organization	NOUN
sc-5860	18	37	,	,	PUNCT
sc-5860	18	38	and	and	CCONJ
sc-5860	18	39	the	the	DET
sc-5860	18	40	general	general	ADJ
sc-5860	18	41	public	public	NOUN
sc-5860	18	42	[	[	X
sc-5860	18	43	7	7	NUM
sc-5860	18	44	]	]	PUNCT
sc-5860	18	45	.	.	PUNCT
sc-5860	19	1	in	in	ADP
sc-5860	19	2	recent	recent	ADJ
sc-5860	19	3	years	year	NOUN
sc-5860	19	4	,	,	PUNCT
sc-5860	19	5	a	a	DET
sc-5860	19	6	number	number	NOUN
sc-5860	19	7	of	of	ADP
sc-5860	19	8	major	major	ADJ
sc-5860	19	9	natural	natural	ADJ
sc-5860	19	10	disasters	disaster	NOUN
sc-5860	19	11	such	such	ADJ
sc-5860	19	12	as	as	ADP
sc-5860	19	13	earthquakes	earthquake	NOUN
sc-5860	19	14	,	,	PUNCT
sc-5860	19	15	hurricanes	hurricane	NOUN
sc-5860	19	16	,	,	PUNCT
sc-5860	19	17	forest	forest	NOUN
sc-5860	19	18	fires	fire	NOUN
sc-5860	19	19	,	,	PUNCT
sc-5860	19	20	and	and	CCONJ
sc-5860	19	21	floods	flood	NOUN
sc-5860	19	22	have	have	AUX
sc-5860	19	23	been	be	AUX
sc-5860	19	24	experienced	experience	VERB
sc-5860	19	25	in	in	ADP
sc-5860	19	26	different	different	ADJ
sc-5860	19	27	parts	part	NOUN
sc-5860	19	28	of	of	ADP
sc-5860	19	29	the	the	DET
sc-5860	19	30	world	world	NOUN
sc-5860	19	31	,	,	PUNCT
sc-5860	19	32	1	1	NUM
sc-5860	19	33	eai	eai	NOUN
sc-5860	19	34	endorsed	endorse	VERB
sc-5860	19	35	transactions	transaction	NOUN
sc-5860	19	36	on	on	ADP
sc-5860	19	37	smart	smart	ADJ
sc-5860	19	38	cities	city	NOUN
sc-5860	19	39	|	|	ADV
sc-5860	19	40	volume	volume	VERB
sc-5860	19	41	7	7	NUM
sc-5860	19	42	|	|	NOUN
sc-5860	19	43	issue	issue	VERB
sc-5860	19	44	4	4	NUM
sc-5860	19	45	|	|	CCONJ
sc-5860	19	46	2023	2023	NUM
sc-5860	20	1	|	|	ADV
sc-5860	20	2	https://creativecommons.org/licenses/by-nc-sa/4.0/	https://creativecommons.org/licenses/by-nc-sa/4.0/	PROPN
sc-5860	20	3	mailto:<suleyman.eken@kocaeli.edu.tr	mailto:<suleyman.eken@kocaeli.edu.tr	PROPN
sc-5860	20	4	>	>	X
sc-5860	20	5	s.	s.	PROPN
sc-5860	20	6	dursun	dursun	PROPN
sc-5860	20	7	and	and	CCONJ
sc-5860	20	8	s.	s.	PROPN
sc-5860	20	9	eken	eken	VERB
sc-5860	20	10	as	as	ADV
sc-5860	20	11	well	well	ADV
sc-5860	20	12	as	as	ADP
sc-5860	20	13	in	in	ADP
sc-5860	20	14	our	our	PRON
sc-5860	20	15	country	country	NOUN
sc-5860	20	16	.	.	PUNCT
sc-5860	21	1	in	in	ADP
sc-5860	21	2	these	these	DET
sc-5860	21	3	difficult	difficult	ADJ
sc-5860	21	4	times	time	NOUN
sc-5860	21	5	,	,	PUNCT
sc-5860	21	6	social	social	ADJ
sc-5860	21	7	media	medium	NOUN
sc-5860	21	8	plays	play	VERB
sc-5860	21	9	a	a	DET
sc-5860	21	10	key	key	ADJ
sc-5860	21	11	role	role	NOUN
sc-5860	21	12	in	in	ADP
sc-5860	21	13	disseminating	disseminate	VERB
sc-5860	21	14	information	information	NOUN
sc-5860	21	15	about	about	ADP
sc-5860	21	16	the	the	DET
sc-5860	21	17	damages	damage	NOUN
sc-5860	21	18	incurred	incur	VERB
sc-5860	21	19	in	in	ADP
sc-5860	21	20	these	these	DET
sc-5860	21	21	disasters	disaster	NOUN
sc-5860	21	22	[	[	X
sc-5860	21	23	8	8	NUM
sc-5860	21	24	]	]	PUNCT
sc-5860	21	25	.	.	PUNCT
sc-5860	22	1	within	within	ADP
sc-5860	22	2	the	the	DET
sc-5860	22	3	scope	scope	NOUN
sc-5860	22	4	of	of	ADP
sc-5860	22	5	this	this	DET
sc-5860	22	6	study	study	NOUN
sc-5860	22	7	,	,	PUNCT
sc-5860	22	8	multi	multi	ADJ
sc-5860	22	9	-	-	ADJ
sc-5860	22	10	modal	modal	ADJ
sc-5860	22	11	sentiment	sentiment	NOUN
sc-5860	22	12	analysis	analysis	NOUN
sc-5860	22	13	of	of	ADP
sc-5860	22	14	disaster	disaster	NOUN
sc-5860	22	15	images	image	NOUN
sc-5860	22	16	shared	share	VERB
sc-5860	22	17	on	on	ADP
sc-5860	22	18	social	social	ADJ
sc-5860	22	19	media	medium	NOUN
sc-5860	22	20	,	,	PUNCT
sc-5860	22	21	which	which	PRON
sc-5860	22	22	is	be	AUX
sc-5860	22	23	considered	consider	VERB
sc-5860	22	24	to	to	PART
sc-5860	22	25	be	be	AUX
sc-5860	22	26	an	an	DET
sc-5860	22	27	important	important	ADJ
sc-5860	22	28	area	area	NOUN
sc-5860	22	29	in	in	ADP
sc-5860	22	30	terms	term	NOUN
sc-5860	22	31	of	of	ADP
sc-5860	22	32	social	social	ADJ
sc-5860	22	33	sensitivity	sensitivity	NOUN
sc-5860	22	34	,	,	PUNCT
sc-5860	22	35	and	and	CCONJ
sc-5860	22	36	the	the	DET
sc-5860	22	37	sentiments	sentiment	NOUN
sc-5860	22	38	in	in	ADP
sc-5860	22	39	the	the	DET
sc-5860	22	40	texts	text	NOUN
sc-5860	22	41	shared	share	VERB
sc-5860	22	42	with	with	ADP
sc-5860	22	43	the	the	DET
sc-5860	22	44	images	image	NOUN
sc-5860	22	45	,	,	PUNCT
sc-5860	22	46	was	be	AUX
sc-5860	22	47	carried	carry	VERB
sc-5860	22	48	out	out	ADP
sc-5860	22	49	with	with	ADP
sc-5860	22	50	a	a	DET
sc-5860	22	51	threelayer	threelayer	NOUN
sc-5860	22	52	deep	deep	ADJ
sc-5860	22	53	learning	learning	NOUN
sc-5860	22	54	model	model	NOUN
sc-5860	22	55	.	.	PUNCT
sc-5860	23	1	the	the	DET
sc-5860	23	2	data	datum	NOUN
sc-5860	23	3	set	set	VERB
sc-5860	23	4	collected	collect	VERB
sc-5860	23	5	from	from	ADP
sc-5860	23	6	social	social	ADJ
sc-5860	23	7	media	medium	NOUN
sc-5860	23	8	within	within	ADP
sc-5860	23	9	the	the	DET
sc-5860	23	10	scope	scope	NOUN
sc-5860	23	11	of	of	ADP
sc-5860	23	12	multimodality	multimodality	NOUN
sc-5860	23	13	(	(	PUNCT
sc-5860	23	14	image	image	NOUN
sc-5860	23	15	and	and	CCONJ
sc-5860	23	16	text	text	NOUN
sc-5860	23	17	)	)	PUNCT
sc-5860	23	18	will	will	AUX
sc-5860	23	19	provide	provide	VERB
sc-5860	23	20	a	a	DET
sc-5860	23	21	benchmark	benchmark	NOUN
sc-5860	23	22	for	for	ADP
sc-5860	23	23	future	future	ADJ
sc-5860	23	24	research	research	NOUN
sc-5860	23	25	in	in	ADP
sc-5860	23	26	this	this	DET
sc-5860	23	27	field	field	NOUN
sc-5860	23	28	,	,	PUNCT
sc-5860	23	29	and	and	CCONJ
sc-5860	23	30	the	the	DET
sc-5860	23	31	presented	present	VERB
sc-5860	23	32	analysis	analysis	NOUN
sc-5860	23	33	model	model	NOUN
sc-5860	23	34	can	can	AUX
sc-5860	23	35	be	be	AUX
sc-5860	23	36	used	use	VERB
sc-5860	23	37	in	in	ADP
sc-5860	23	38	other	other	ADJ
sc-5860	23	39	fields	field	NOUN
sc-5860	23	40	through	through	ADP
sc-5860	23	41	transfer	transfer	NOUN
sc-5860	23	42	learning	learning	NOUN
sc-5860	23	43	.	.	PUNCT
sc-5860	24	1	the	the	DET
sc-5860	24	2	remaining	remain	VERB
sc-5860	24	3	part	part	NOUN
sc-5860	24	4	of	of	ADP
sc-5860	24	5	the	the	DET
sc-5860	24	6	paper	paper	NOUN
sc-5860	24	7	is	be	AUX
sc-5860	24	8	organized	organize	VERB
sc-5860	24	9	as	as	SCONJ
sc-5860	24	10	follows	follow	VERB
sc-5860	24	11	:	:	PUNCT
sc-5860	24	12	in	in	ADP
sc-5860	24	13	the	the	DET
sc-5860	24	14	2nd	2nd	ADJ
sc-5860	24	15	section	section	NOUN
sc-5860	24	16	,	,	PUNCT
sc-5860	24	17	the	the	DET
sc-5860	24	18	relevant	relevant	ADJ
sc-5860	24	19	studies	study	NOUN
sc-5860	24	20	are	be	AUX
sc-5860	24	21	mentioned	mention	VERB
sc-5860	24	22	,	,	PUNCT
sc-5860	24	23	in	in	ADP
sc-5860	24	24	the	the	DET
sc-5860	24	25	3rd	3rd	ADJ
sc-5860	24	26	section	section	NOUN
sc-5860	24	27	,	,	PUNCT
sc-5860	24	28	the	the	DET
sc-5860	24	29	visual	visual	ADJ
sc-5860	24	30	and	and	CCONJ
sc-5860	24	31	textual	textual	ADJ
sc-5860	24	32	features	feature	NOUN
sc-5860	24	33	are	be	AUX
sc-5860	24	34	explained	explain	VERB
sc-5860	24	35	.	.	PUNCT
sc-5860	25	1	in	in	ADP
sc-5860	25	2	the	the	DET
sc-5860	25	3	4th	4th	ADJ
sc-5860	25	4	section	section	NOUN
sc-5860	25	5	,	,	PUNCT
sc-5860	25	6	the	the	DET
sc-5860	25	7	performed	perform	VERB
sc-5860	25	8	tests	test	NOUN
sc-5860	25	9	are	be	AUX
sc-5860	25	10	given	give	VERB
sc-5860	25	11	,	,	PUNCT
sc-5860	25	12	and	and	CCONJ
sc-5860	25	13	conclusion	conclusion	NOUN
sc-5860	25	14	and	and	CCONJ
sc-5860	25	15	future	future	ADJ
sc-5860	25	16	works	work	NOUN
sc-5860	25	17	are	be	AUX
sc-5860	25	18	mentioned	mention	VERB
sc-5860	25	19	in	in	ADP
sc-5860	25	20	the	the	DET
sc-5860	25	21	last	last	ADJ
sc-5860	25	22	section	section	NOUN
sc-5860	25	23	.	.	PUNCT
sc-5860	26	1	2	2	X
sc-5860	26	2	.	.	NUM
sc-5860	26	3	related	relate	VERB
sc-5860	26	4	works	work	NOUN
sc-5860	26	5	in	in	ADP
sc-5860	26	6	recent	recent	ADJ
sc-5860	26	7	years	year	NOUN
sc-5860	26	8	,	,	PUNCT
sc-5860	26	9	multimodal	multimodal	ADJ
sc-5860	26	10	learning	learning	NOUN
sc-5860	26	11	has	have	AUX
sc-5860	26	12	trended	trend	VERB
sc-5860	26	13	upward	upward	ADV
sc-5860	26	14	in	in	ADP
sc-5860	26	15	ai	ai	ADJ
sc-5860	26	16	applications	application	NOUN
sc-5860	26	17	as	as	SCONJ
sc-5860	26	18	researchers	researcher	NOUN
sc-5860	26	19	integrate	integrate	VERB
sc-5860	26	20	data	datum	NOUN
sc-5860	26	21	in	in	ADP
sc-5860	26	22	different	different	ADJ
sc-5860	26	23	modes	mode	NOUN
sc-5860	26	24	/	/	SYM
sc-5860	26	25	types	type	NOUN
sc-5860	26	26	into	into	ADP
sc-5860	26	27	modeling	modeling	NOUN
sc-5860	26	28	,	,	PUNCT
sc-5860	26	29	such	such	ADJ
sc-5860	26	30	as	as	ADP
sc-5860	26	31	text	text	NOUN
sc-5860	26	32	,	,	PUNCT
sc-5860	26	33	images	image	NOUN
sc-5860	26	34	,	,	PUNCT
sc-5860	26	35	speech	speech	NOUN
sc-5860	26	36	,	,	PUNCT
sc-5860	26	37	etc	etc	X
sc-5860	27	1	[	[	X
sc-5860	27	2	9	9	NUM
sc-5860	27	3	]	]	PUNCT
sc-5860	27	4	.	.	PUNCT
sc-5860	28	1	to	to	PART
sc-5860	28	2	achieve	achieve	VERB
sc-5860	28	3	the	the	DET
sc-5860	28	4	best	good	ADJ
sc-5860	28	5	results	result	NOUN
sc-5860	28	6	.	.	PUNCT
sc-5860	29	1	matrix	matrix	NOUN
sc-5860	29	2	factorization	factorization	NOUN
sc-5860	29	3	methods	method	NOUN
sc-5860	29	4	are	be	AUX
sc-5860	29	5	among	among	ADP
sc-5860	29	6	the	the	DET
sc-5860	29	7	preferred	preferred	ADJ
sc-5860	29	8	methods	method	NOUN
sc-5860	29	9	for	for	ADP
sc-5860	29	10	multimodal	multimodal	NOUN
sc-5860	29	11	data	datum	NOUN
sc-5860	29	12	setup	setup	NOUN
sc-5860	29	13	.	.	PUNCT
sc-5860	30	1	some	some	PRON
sc-5860	30	2	of	of	ADP
sc-5860	30	3	the	the	DET
sc-5860	30	4	common	common	ADJ
sc-5860	30	5	methods	method	NOUN
sc-5860	30	6	that	that	PRON
sc-5860	30	7	have	have	AUX
sc-5860	30	8	been	be	AUX
sc-5860	30	9	shown	show	VERB
sc-5860	30	10	to	to	PART
sc-5860	30	11	be	be	AUX
sc-5860	30	12	useful	useful	ADJ
sc-5860	30	13	in	in	ADP
sc-5860	30	14	learning	learn	VERB
sc-5860	30	15	representations	representation	NOUN
sc-5860	30	16	of	of	ADP
sc-5860	30	17	entities	entity	NOUN
sc-5860	30	18	from	from	ADP
sc-5860	30	19	a	a	DET
sc-5860	30	20	collection	collection	NOUN
sc-5860	30	21	of	of	ADP
sc-5860	30	22	matrices	matrix	NOUN
sc-5860	30	23	are	be	AUX
sc-5860	30	24	collective	collective	ADJ
sc-5860	30	25	matrix	matrix	NOUN
sc-5860	30	26	factorization	factorization	NOUN
sc-5860	30	27	,	,	PUNCT
sc-5860	30	28	data	datum	NOUN
sc-5860	30	29	fusion	fusion	NOUN
sc-5860	30	30	with	with	ADP
sc-5860	30	31	matrix	matrix	NOUN
sc-5860	30	32	factorization	factorization	NOUN
sc-5860	30	33	,	,	PUNCT
sc-5860	30	34	and	and	CCONJ
sc-5860	30	35	deep	deep	ADJ
sc-5860	30	36	collective	collective	ADJ
sc-5860	30	37	matrix	matrix	NOUN
sc-5860	30	38	factorization	factorization	NOUN
sc-5860	30	39	.	.	PUNCT
sc-5860	31	1	these	these	DET
sc-5860	31	2	methods	method	NOUN
sc-5860	31	3	learn	learn	VERB
sc-5860	31	4	entityspecific	entityspecific	ADJ
sc-5860	31	5	representations	representation	NOUN
sc-5860	31	6	for	for	ADP
sc-5860	31	7	entities	entity	NOUN
sc-5860	31	8	across	across	ADP
sc-5860	31	9	rows	row	NOUN
sc-5860	31	10	and	and	CCONJ
sc-5860	31	11	columns	column	NOUN
sc-5860	31	12	of	of	ADP
sc-5860	31	13	matrices	matrix	NOUN
sc-5860	31	14	and	and	CCONJ
sc-5860	31	15	use	use	VERB
sc-5860	31	16	them	they	PRON
sc-5860	31	17	to	to	PART
sc-5860	31	18	reconstruct	reconstruct	VERB
sc-5860	31	19	the	the	DET
sc-5860	31	20	matrices	matrix	NOUN
sc-5860	31	21	.	.	PUNCT
sc-5860	32	1	however	however	ADV
sc-5860	32	2	,	,	PUNCT
sc-5860	32	3	these	these	DET
sc-5860	32	4	methods	method	NOUN
sc-5860	32	5	expect	expect	VERB
sc-5860	32	6	the	the	DET
sc-5860	32	7	input	input	NOUN
sc-5860	32	8	to	to	PART
sc-5860	32	9	be	be	AUX
sc-5860	32	10	a	a	DET
sc-5860	32	11	collection	collection	NOUN
sc-5860	32	12	of	of	ADP
sc-5860	32	13	matrices	matrix	NOUN
sc-5860	32	14	and	and	CCONJ
sc-5860	32	15	tensors	tensor	NOUN
sc-5860	32	16	are	be	AUX
sc-5860	32	17	not	not	PART
sc-5860	32	18	supported	support	VERB
sc-5860	32	19	.	.	PUNCT
sc-5860	33	1	jayagopal	jayagopal	PROPN
sc-5860	33	2	et	et	PROPN
sc-5860	33	3	al	al	PROPN
sc-5860	33	4	.	.	PUNCT
sc-5860	34	1	[	[	X
sc-5860	34	2	10	10	NUM
sc-5860	34	3	]	]	PUNCT
sc-5860	34	4	overcome	overcome	VERB
sc-5860	34	5	this	this	DET
sc-5860	34	6	limitation	limitation	NOUN
sc-5860	34	7	in	in	ADP
sc-5860	34	8	their	their	PRON
sc-5860	34	9	proposed	propose	VERB
sc-5860	34	10	model	model	NOUN
sc-5860	34	11	by	by	ADP
sc-5860	34	12	combining	combine	VERB
sc-5860	34	13	the	the	DET
sc-5860	34	14	reconstruction	reconstruction	NOUN
sc-5860	34	15	capabilities	capability	NOUN
sc-5860	34	16	of	of	ADP
sc-5860	34	17	matrix	matrix	NOUN
sc-5860	34	18	factorization	factorization	NOUN
sc-5860	34	19	and	and	CCONJ
sc-5860	34	20	the	the	DET
sc-5860	34	21	ability	ability	NOUN
sc-5860	34	22	of	of	ADP
sc-5860	34	23	convolutional	convolutional	ADJ
sc-5860	34	24	autoencoders	autoencoder	NOUN
sc-5860	34	25	to	to	PART
sc-5860	34	26	operate	operate	VERB
sc-5860	34	27	on	on	ADP
sc-5860	34	28	tensor	tensor	NOUN
sc-5860	34	29	inputs	input	NOUN
sc-5860	34	30	.	.	PUNCT
sc-5860	35	1	thus	thus	ADV
sc-5860	35	2	,	,	PUNCT
sc-5860	35	3	with	with	ADP
sc-5860	35	4	the	the	DET
sc-5860	35	5	proposed	propose	VERB
sc-5860	35	6	architecture	architecture	NOUN
sc-5860	35	7	,	,	PUNCT
sc-5860	35	8	it	it	PRON
sc-5860	35	9	can	can	AUX
sc-5860	35	10	jointly	jointly	ADV
sc-5860	35	11	learn	learn	VERB
sc-5860	35	12	representations	representation	NOUN
sc-5860	35	13	for	for	ADP
sc-5860	35	14	entities	entity	NOUN
sc-5860	35	15	based	base	VERB
sc-5860	35	16	on	on	ADP
sc-5860	35	17	inputs	input	NOUN
sc-5860	35	18	(	(	PUNCT
sc-5860	35	19	matrices	matrix	NOUN
sc-5860	35	20	or	or	CCONJ
sc-5860	35	21	tensors	tensor	NOUN
sc-5860	35	22	)	)	PUNCT
sc-5860	35	23	of	of	ADP
sc-5860	35	24	any	any	DET
sc-5860	35	25	size	size	NOUN
sc-5860	35	26	.	.	PUNCT
sc-5860	36	1	joint	joint	ADJ
sc-5860	36	2	representation	representation	NOUN
sc-5860	36	3	learning	learning	NOUN
sc-5860	36	4	in	in	ADP
sc-5860	36	5	the	the	DET
sc-5860	36	6	multimodal	multimodal	NOUN
sc-5860	36	7	domain	domain	NOUN
sc-5860	36	8	occurs	occur	VERB
sc-5860	36	9	in	in	ADP
sc-5860	36	10	different	different	ADJ
sc-5860	36	11	ways	way	NOUN
sc-5860	36	12	.	.	PUNCT
sc-5860	37	1	the	the	DET
sc-5860	37	2	first	first	ADJ
sc-5860	37	3	group	group	NOUN
sc-5860	37	4	of	of	ADP
sc-5860	37	5	works	work	NOUN
sc-5860	37	6	optimizes	optimize	VERB
sc-5860	37	7	an	an	DET
sc-5860	37	8	image	image	NOUN
sc-5860	37	9	encoder	encoder	NOUN
sc-5860	37	10	and	and	CCONJ
sc-5860	37	11	a	a	DET
sc-5860	37	12	language	language	NOUN
sc-5860	37	13	decoder	decoder	NOUN
sc-5860	37	14	for	for	ADP
sc-5860	37	15	the	the	DET
sc-5860	37	16	image	image	NOUN
sc-5860	37	17	captioning	captioning	NOUN
sc-5860	37	18	task	task	NOUN
sc-5860	37	19	and	and	CCONJ
sc-5860	37	20	transfers	transfer	NOUN
sc-5860	37	21	the	the	DET
sc-5860	37	22	learned	learn	VERB
sc-5860	37	23	visual	visual	ADJ
sc-5860	37	24	representations	representation	NOUN
sc-5860	37	25	to	to	ADP
sc-5860	37	26	subsequent	subsequent	ADJ
sc-5860	37	27	applications	application	NOUN
sc-5860	37	28	.	.	PUNCT
sc-5860	38	1	the	the	DET
sc-5860	38	2	second	second	ADJ
sc-5860	38	3	group	group	NOUN
sc-5860	38	4	jointly	jointly	ADV
sc-5860	38	5	learns	learn	VERB
sc-5860	38	6	multimodal	multimodal	ADJ
sc-5860	38	7	pretext	pretext	NOUN
sc-5860	38	8	tasks	task	NOUN
sc-5860	38	9	,	,	PUNCT
sc-5860	38	10	such	such	ADJ
sc-5860	38	11	as	as	ADP
sc-5860	38	12	reconstructing	reconstruct	VERB
sc-5860	38	13	masked	mask	VERB
sc-5860	38	14	image	image	NOUN
sc-5860	38	15	regions	region	NOUN
sc-5860	38	16	and	and	CCONJ
sc-5860	38	17	language	language	NOUN
sc-5860	38	18	symbols	symbol	NOUN
sc-5860	38	19	/	/	SYM
sc-5860	38	20	tokens	token	NOUN
sc-5860	38	21	,	,	PUNCT
sc-5860	38	22	as	as	ADV
sc-5860	38	23	well	well	ADV
sc-5860	38	24	as	as	ADP
sc-5860	38	25	directly	directly	ADV
sc-5860	38	26	estimating	estimate	VERB
sc-5860	38	27	the	the	DET
sc-5860	38	28	alignment	alignment	NOUN
sc-5860	38	29	between	between	ADP
sc-5860	38	30	image	image	NOUN
sc-5860	38	31	and	and	CCONJ
sc-5860	38	32	text	text	NOUN
sc-5860	38	33	.	.	PUNCT
sc-5860	39	1	the	the	DET
sc-5860	39	2	cross	cross	ADJ
sc-5860	39	3	-	-	ADJ
sc-5860	39	4	modal	modal	ADJ
sc-5860	39	5	attention	attention	NOUN
sc-5860	39	6	modules	module	NOUN
sc-5860	39	7	that	that	PRON
sc-5860	39	8	arise	arise	VERB
sc-5860	39	9	in	in	ADP
sc-5860	39	10	these	these	DET
sc-5860	39	11	methods	method	NOUN
sc-5860	39	12	cause	cause	VERB
sc-5860	39	13	them	they	PRON
sc-5860	39	14	to	to	PART
sc-5860	39	15	be	be	AUX
sc-5860	39	16	less	less	ADV
sc-5860	39	17	efficient	efficient	ADJ
sc-5860	39	18	in	in	ADP
sc-5860	39	19	practical	practical	ADJ
sc-5860	39	20	retrieval	retrieval	NOUN
sc-5860	39	21	systems	system	NOUN
sc-5860	39	22	.	.	PUNCT
sc-5860	40	1	the	the	DET
sc-5860	40	2	third	third	ADJ
sc-5860	40	3	group	group	NOUN
sc-5860	40	4	,	,	PUNCT
sc-5860	40	5	closer	close	ADV
sc-5860	40	6	to	to	ADP
sc-5860	40	7	the	the	DET
sc-5860	40	8	contrast	contrast	NOUN
sc-5860	40	9	methodology	methodology	NOUN
sc-5860	40	10	in	in	ADP
sc-5860	40	11	visual	visual	ADJ
sc-5860	40	12	representation	representation	NOUN
sc-5860	40	13	learning	learning	NOUN
sc-5860	40	14	,	,	PUNCT
sc-5860	40	15	uses	use	VERB
sc-5860	40	16	dual	dual	ADJ
sc-5860	40	17	encoder	encoder	NOUN
sc-5860	40	18	architecture	architecture	NOUN
sc-5860	40	19	to	to	PART
sc-5860	40	20	directly	directly	ADV
sc-5860	40	21	map	map	VERB
sc-5860	40	22	image	image	NOUN
sc-5860	40	23	and	and	CCONJ
sc-5860	40	24	text	text	NOUN
sc-5860	40	25	data	datum	NOUN
sc-5860	40	26	into	into	ADP
sc-5860	40	27	a	a	DET
sc-5860	40	28	common	common	ADJ
sc-5860	40	29	embedding	embed	VERB
sc-5860	40	30	space	space	NOUN
sc-5860	40	31	.	.	PUNCT
sc-5860	41	1	here	here	ADV
sc-5860	41	2	,	,	PUNCT
sc-5860	41	3	the	the	DET
sc-5860	41	4	agreement	agreement	NOUN
sc-5860	41	5	between	between	ADP
sc-5860	41	6	paired	pair	VERB
sc-5860	41	7	samples	sample	NOUN
sc-5860	41	8	is	be	AUX
sc-5860	41	9	maximized	maximize	VERB
sc-5860	41	10	while	while	SCONJ
sc-5860	41	11	the	the	DET
sc-5860	41	12	agreement	agreement	NOUN
sc-5860	41	13	between	between	ADP
sc-5860	41	14	unpaired	unpaired	ADJ
sc-5860	41	15	samples	sample	NOUN
sc-5860	41	16	is	be	AUX
sc-5860	41	17	minimized	minimize	VERB
sc-5860	41	18	.	.	PUNCT
sc-5860	42	1	however	however	ADV
sc-5860	42	2	,	,	PUNCT
sc-5860	42	3	recent	recent	ADJ
sc-5860	42	4	studies	study	NOUN
sc-5860	42	5	focus	focus	VERB
sc-5860	42	6	on	on	ADP
sc-5860	42	7	data	datum	NOUN
sc-5860	42	8	scale	scale	NOUN
sc-5860	42	9	and	and	CCONJ
sc-5860	42	10	model	model	NOUN
sc-5860	42	11	architectures	architecture	NOUN
sc-5860	42	12	,	,	PUNCT
sc-5860	42	13	using	use	VERB
sc-5860	42	14	state	state	NOUN
sc-5860	42	15	-	-	PUNCT
sc-5860	42	16	of	of	ADP
sc-5860	42	17	-	-	PUNCT
sc-5860	42	18	the	the	DET
sc-5860	42	19	-	-	PUNCT
sc-5860	42	20	art	art	NOUN
sc-5860	42	21	multimodal	multimodal	NOUN
sc-5860	42	22	contrastive	contrastive	ADJ
sc-5860	42	23	learning	learning	NOUN
sc-5860	42	24	frameworks	framework	NOUN
sc-5860	42	25	such	such	ADJ
sc-5860	42	26	as	as	ADP
sc-5860	42	27	convirt	convirt	PROPN
sc-5860	42	28	[	[	X
sc-5860	42	29	11	11	NUM
sc-5860	42	30	]	]	PUNCT
sc-5860	42	31	and	and	CCONJ
sc-5860	42	32	clip	clip	VERB
sc-5860	43	1	[	[	X
sc-5860	43	2	12	12	NUM
sc-5860	43	3	]	]	PUNCT
sc-5860	43	4	,	,	PUNCT
sc-5860	43	5	which	which	PRON
sc-5860	43	6	relax	relax	VERB
sc-5860	43	7	the	the	DET
sc-5860	43	8	driving	drive	VERB
sc-5860	43	9	force	force	NOUN
sc-5860	43	10	between	between	ADP
sc-5860	43	11	negative	negative	ADJ
sc-5860	43	12	pairs	pair	NOUN
sc-5860	43	13	(	(	PUNCT
sc-5860	43	14	relaxation	relaxation	NOUN
sc-5860	43	15	of	of	ADP
sc-5860	43	16	contrastiveness	contrastiveness	NOUN
sc-5860	43	17	)	)	PUNCT
sc-5860	43	18	in	in	ADP
sc-5860	43	19	pretraining	pretraine	VERB
sc-5860	43	20	,	,	PUNCT
sc-5860	43	21	are	be	AUX
sc-5860	43	22	used	use	VERB
sc-5860	43	23	[	[	PUNCT
sc-5860	43	24	13	13	NUM
sc-5860	43	25	]	]	PUNCT
sc-5860	43	26	.	.	PUNCT
sc-5860	44	1	sentiment	sentiment	NOUN
sc-5860	44	2	analysis	analysis	NOUN
sc-5860	44	3	studies	study	NOUN
sc-5860	44	4	conducted	conduct	VERB
sc-5860	44	5	on	on	ADP
sc-5860	44	6	social	social	ADJ
sc-5860	44	7	media	medium	NOUN
sc-5860	44	8	data	datum	NOUN
sc-5860	44	9	refer	refer	VERB
sc-5860	44	10	to	to	ADP
sc-5860	44	11	a	a	DET
sc-5860	44	12	field	field	NOUN
sc-5860	44	13	that	that	PRON
sc-5860	44	14	aims	aim	VERB
sc-5860	44	15	to	to	PART
sc-5860	44	16	determine	determine	VERB
sc-5860	44	17	the	the	DET
sc-5860	44	18	emotional	emotional	ADJ
sc-5860	44	19	state	state	NOUN
sc-5860	44	20	of	of	ADP
sc-5860	44	21	text	text	NOUN
sc-5860	44	22	,	,	PUNCT
sc-5860	44	23	visuals	visual	NOUN
sc-5860	44	24	and	and	CCONJ
sc-5860	44	25	other	other	ADJ
sc-5860	44	26	content	content	NOUN
sc-5860	44	27	shared	share	VERB
sc-5860	44	28	on	on	ADP
sc-5860	44	29	social	social	ADJ
sc-5860	44	30	media	medium	NOUN
sc-5860	44	31	platforms	platform	NOUN
sc-5860	44	32	.	.	PUNCT
sc-5860	45	1	this	this	DET
sc-5860	45	2	type	type	NOUN
sc-5860	45	3	of	of	ADP
sc-5860	45	4	analysis	analysis	NOUN
sc-5860	45	5	can	can	AUX
sc-5860	45	6	serve	serve	VERB
sc-5860	45	7	many	many	ADJ
sc-5860	45	8	different	different	ADJ
sc-5860	45	9	purposes	purpose	NOUN
sc-5860	45	10	.	.	PUNCT
sc-5860	46	1	for	for	ADP
sc-5860	46	2	example	example	NOUN
sc-5860	46	3	,	,	PUNCT
sc-5860	46	4	within	within	ADP
sc-5860	46	5	marketing	marketing	NOUN
sc-5860	46	6	and	and	CCONJ
sc-5860	46	7	advertising	advertising	NOUN
sc-5860	46	8	it	it	PRON
sc-5860	46	9	can	can	AUX
sc-5860	46	10	be	be	AUX
sc-5860	46	11	used	use	VERB
sc-5860	46	12	to	to	PART
sc-5860	46	13	understand	understand	VERB
sc-5860	46	14	people	people	NOUN
sc-5860	46	15	’s	’s	PART
sc-5860	46	16	emotional	emotional	ADJ
sc-5860	46	17	reactions	reaction	NOUN
sc-5860	46	18	to	to	ADP
sc-5860	46	19	products	product	NOUN
sc-5860	46	20	,	,	PUNCT
sc-5860	46	21	services	service	NOUN
sc-5860	46	22	or	or	CCONJ
sc-5860	46	23	brands	brand	NOUN
sc-5860	46	24	,	,	PUNCT
sc-5860	46	25	to	to	ADP
sc-5860	46	26	better	well	ADJ
sc-5860	46	27	target	target	VERB
sc-5860	46	28	marketing	marketing	NOUN
sc-5860	46	29	strategies	strategy	NOUN
sc-5860	46	30	and	and	CCONJ
sc-5860	46	31	increase	increase	VERB
sc-5860	46	32	customer	customer	NOUN
sc-5860	46	33	satisfaction	satisfaction	NOUN
sc-5860	46	34	.	.	PUNCT
sc-5860	47	1	again	again	ADV
sc-5860	47	2	,	,	PUNCT
sc-5860	47	3	in	in	ADP
sc-5860	47	4	the	the	DET
sc-5860	47	5	field	field	NOUN
sc-5860	47	6	of	of	ADP
sc-5860	47	7	social	social	ADJ
sc-5860	47	8	monitoring	monitoring	NOUN
sc-5860	47	9	,	,	PUNCT
sc-5860	47	10	it	it	PRON
sc-5860	47	11	is	be	AUX
sc-5860	47	12	possible	possible	ADJ
sc-5860	47	13	to	to	PART
sc-5860	47	14	monitor	monitor	VERB
sc-5860	47	15	the	the	DET
sc-5860	47	16	general	general	ADJ
sc-5860	47	17	mood	mood	NOUN
sc-5860	47	18	about	about	ADP
sc-5860	47	19	a	a	DET
sc-5860	47	20	particular	particular	ADJ
sc-5860	47	21	event	event	NOUN
sc-5860	47	22	or	or	CCONJ
sc-5860	47	23	issue	issue	NOUN
sc-5860	47	24	on	on	ADP
sc-5860	47	25	social	social	ADJ
sc-5860	47	26	media	medium	NOUN
sc-5860	47	27	and	and	CCONJ
sc-5860	47	28	use	use	VERB
sc-5860	47	29	this	this	DET
sc-5860	47	30	information	information	NOUN
sc-5860	47	31	as	as	ADP
sc-5860	47	32	an	an	DET
sc-5860	47	33	alternative	alternative	NOUN
sc-5860	47	34	to	to	ADP
sc-5860	47	35	public	public	ADJ
sc-5860	47	36	opinion	opinion	NOUN
sc-5860	47	37	surveys	survey	NOUN
sc-5860	47	38	.	.	PUNCT
sc-5860	48	1	studies	study	NOUN
sc-5860	48	2	in	in	ADP
sc-5860	48	3	this	this	DET
sc-5860	48	4	field	field	NOUN
sc-5860	48	5	are	be	AUX
sc-5860	48	6	becoming	become	VERB
sc-5860	48	7	increasingly	increasingly	ADV
sc-5860	48	8	important	important	ADJ
sc-5860	48	9	with	with	ADP
sc-5860	48	10	the	the	DET
sc-5860	48	11	growth	growth	NOUN
sc-5860	48	12	of	of	ADP
sc-5860	48	13	social	social	ADJ
sc-5860	48	14	media	medium	NOUN
sc-5860	48	15	platforms	platform	NOUN
sc-5860	49	1	[	[	X
sc-5860	49	2	14	14	NUM
sc-5860	49	3	]	]	PUNCT
sc-5860	49	4	.	.	PUNCT
sc-5860	50	1	sentiment	sentiment	NOUN
sc-5860	50	2	analysis	analysis	NOUN
sc-5860	50	3	works	work	NOUN
sc-5860	50	4	often	often	ADV
sc-5860	50	5	use	use	VERB
sc-5860	50	6	natural	natural	ADJ
sc-5860	50	7	language	language	NOUN
sc-5860	50	8	processing	processing	NOUN
sc-5860	50	9	(	(	PUNCT
sc-5860	50	10	nlp	nlp	NOUN
sc-5860	50	11	)	)	PUNCT
sc-5860	50	12	techniques	technique	NOUN
sc-5860	50	13	.	.	PUNCT
sc-5860	51	1	these	these	DET
sc-5860	51	2	techniques	technique	NOUN
sc-5860	51	3	process	process	NOUN
sc-5860	51	4	text	text	NOUN
sc-5860	51	5	data	datum	NOUN
sc-5860	51	6	,	,	PUNCT
sc-5860	51	7	trying	try	VERB
sc-5860	51	8	to	to	PART
sc-5860	51	9	detect	detect	VERB
sc-5860	51	10	specific	specific	ADJ
sc-5860	51	11	sentiments	sentiment	NOUN
sc-5860	51	12	or	or	CCONJ
sc-5860	51	13	sentiment	sentiment	NOUN
sc-5860	51	14	intensity	intensity	NOUN
sc-5860	51	15	.	.	PUNCT
sc-5860	52	1	sentiment	sentiment	NOUN
sc-5860	52	2	analysis	analysis	NOUN
sc-5860	52	3	has	have	AUX
sc-5860	52	4	been	be	AUX
sc-5860	52	5	extensively	extensively	ADV
sc-5860	52	6	explored	explore	VERB
sc-5860	52	7	for	for	ADP
sc-5860	52	8	textual	textual	ADJ
sc-5860	52	9	social	social	ADJ
sc-5860	52	10	media	medium	NOUN
sc-5860	52	11	data	datum	NOUN
sc-5860	52	12	,	,	PUNCT
sc-5860	52	13	with	with	ADP
sc-5860	52	14	previous	previous	ADJ
sc-5860	52	15	dictionary	dictionary	ADJ
sc-5860	52	16	-	-	PUNCT
sc-5860	52	17	based	base	VERB
sc-5860	52	18	approaches	approach	NOUN
sc-5860	52	19	evolving	evolve	VERB
sc-5860	52	20	into	into	ADP
sc-5860	52	21	statistical	statistical	ADJ
sc-5860	52	22	and	and	CCONJ
sc-5860	52	23	machine	machine	NOUN
sc-5860	52	24	learning	learning	NOUN
sc-5860	52	25	-	-	PUNCT
sc-5860	52	26	based	base	VERB
sc-5860	52	27	classification	classification	NOUN
sc-5860	52	28	over	over	ADP
sc-5860	52	29	the	the	DET
sc-5860	52	30	last	last	ADJ
sc-5860	52	31	decade	decade	NOUN
sc-5860	52	32	.	.	PUNCT
sc-5860	53	1	sentistrength	sentistrength	NOUN
sc-5860	54	1	[	[	X
sc-5860	54	2	15	15	NUM
sc-5860	54	3	]	]	X
sc-5860	54	4	is	be	AUX
sc-5860	54	5	a	a	DET
sc-5860	54	6	well	well	ADV
sc-5860	54	7	-	-	PUNCT
sc-5860	54	8	known	know	VERB
sc-5860	54	9	dictionary	dictionary	NOUN
sc-5860	54	10	-	-	PUNCT
sc-5860	54	11	based	base	VERB
sc-5860	54	12	approach	approach	NOUN
sc-5860	54	13	for	for	ADP
sc-5860	54	14	short	short	ADJ
sc-5860	54	15	texts	text	NOUN
sc-5860	54	16	created	create	VERB
sc-5860	54	17	using	use	VERB
sc-5860	54	18	words	word	NOUN
sc-5860	54	19	and	and	CCONJ
sc-5860	54	20	expressions	expression	NOUN
sc-5860	54	21	commonly	commonly	ADV
sc-5860	54	22	used	use	VERB
sc-5860	54	23	on	on	ADP
sc-5860	54	24	social	social	ADJ
sc-5860	54	25	media	medium	NOUN
sc-5860	54	26	.	.	PUNCT
sc-5860	55	1	later	later	ADV
sc-5860	55	2	,	,	PUNCT
sc-5860	55	3	senticircles	senticircle	VERB
sc-5860	55	4	[	[	PUNCT
sc-5860	55	5	16	16	NUM
sc-5860	55	6	]	]	PUNCT
sc-5860	55	7	is	be	AUX
sc-5860	55	8	developed	develop	VERB
sc-5860	55	9	for	for	ADP
sc-5860	55	10	twitter	twitter	NOUN
sc-5860	55	11	sentiment	sentiment	NOUN
sc-5860	55	12	analysis	analysis	NOUN
sc-5860	55	13	by	by	ADP
sc-5860	55	14	taking	take	VERB
sc-5860	55	15	into	into	ADP
sc-5860	55	16	account	account	NOUN
sc-5860	55	17	the	the	DET
sc-5860	55	18	co	co	NOUN
sc-5860	55	19	-	-	NOUN
sc-5860	55	20	occurrence	occurrence	NOUN
sc-5860	55	21	of	of	ADP
sc-5860	55	22	words	word	NOUN
sc-5860	55	23	in	in	ADP
sc-5860	55	24	tweets	tweet	NOUN
sc-5860	55	25	in	in	ADP
sc-5860	55	26	different	different	ADJ
sc-5860	55	27	contexts	contexts	NOUN
sc-5860	55	28	.	.	PUNCT
sc-5860	56	1	with	with	ADP
sc-5860	56	2	the	the	DET
sc-5860	56	3	proliferation	proliferation	NOUN
sc-5860	56	4	of	of	ADP
sc-5860	56	5	deep	deep	ADJ
sc-5860	56	6	learning	learning	NOUN
sc-5860	56	7	,	,	PUNCT
sc-5860	56	8	sequential	sequential	ADJ
sc-5860	56	9	models	model	NOUN
sc-5860	56	10	such	such	ADJ
sc-5860	56	11	as	as	ADP
sc-5860	56	12	convolutional	convolutional	ADJ
sc-5860	56	13	neural	neural	ADJ
sc-5860	56	14	networks	network	NOUN
sc-5860	56	15	(	(	PUNCT
sc-5860	56	16	cnns	cnns	PROPN
sc-5860	56	17	)	)	PUNCT
sc-5860	56	18	and	and	CCONJ
sc-5860	56	19	long	long	ADJ
sc-5860	56	20	short	short	ADJ
sc-5860	56	21	-	-	PUNCT
sc-5860	56	22	term	term	NOUN
sc-5860	56	23	memory	memory	NOUN
sc-5860	56	24	(	(	PUNCT
sc-5860	56	25	lstm	lstm	NOUN
sc-5860	56	26	)	)	PUNCT
sc-5860	56	27	networks	network	NOUN
sc-5860	56	28	have	have	AUX
sc-5860	56	29	been	be	AUX
sc-5860	56	30	successfully	successfully	ADV
sc-5860	56	31	used	use	VERB
sc-5860	56	32	for	for	ADP
sc-5860	56	33	tweet	tweet	NOUN
sc-5860	56	34	sentiment	sentiment	NOUN
sc-5860	56	35	classification	classification	NOUN
sc-5860	56	36	.	.	PUNCT
sc-5860	57	1	with	with	ADP
sc-5860	57	2	the	the	DET
sc-5860	57	3	increase	increase	NOUN
sc-5860	57	4	in	in	ADP
sc-5860	57	5	image	image	NOUN
sc-5860	57	6	and	and	CCONJ
sc-5860	57	7	video	video	NOUN
sc-5860	57	8	data	datum	NOUN
sc-5860	57	9	on	on	ADP
sc-5860	57	10	social	social	ADJ
sc-5860	57	11	media	medium	NOUN
sc-5860	57	12	sites	site	NOUN
sc-5860	57	13	such	such	ADJ
sc-5860	57	14	as	as	ADP
sc-5860	57	15	instagram	instagram	PROPN
sc-5860	57	16	,	,	PUNCT
sc-5860	57	17	flickr	flickr	PROPN
sc-5860	57	18	and	and	CCONJ
sc-5860	57	19	twitter	twitter	NOUN
sc-5860	57	20	,	,	PUNCT
sc-5860	57	21	visual	visual	ADJ
sc-5860	57	22	sentiment	sentiment	NOUN
sc-5860	57	23	analysis	analysis	NOUN
sc-5860	57	24	has	have	AUX
sc-5860	57	25	recently	recently	ADV
sc-5860	57	26	begun	begin	VERB
sc-5860	57	27	to	to	PART
sc-5860	57	28	attract	attract	VERB
sc-5860	57	29	great	great	ADJ
sc-5860	57	30	attention	attention	NOUN
sc-5860	57	31	.	.	PUNCT
sc-5860	58	1	techniques	technique	NOUN
sc-5860	58	2	in	in	ADP
sc-5860	58	3	this	this	DET
sc-5860	58	4	scope	scope	NOUN
sc-5860	58	5	can	can	AUX
sc-5860	58	6	be	be	AUX
sc-5860	58	7	generally	generally	ADV
sc-5860	58	8	divided	divide	VERB
sc-5860	58	9	into	into	ADP
sc-5860	58	10	two	two	NUM
sc-5860	58	11	as	as	ADP
sc-5860	58	12	midlevel	midlevel	NOUN
sc-5860	58	13	and	and	CCONJ
sc-5860	58	14	deep	deep	ADJ
sc-5860	58	15	learning	learning	NOUN
sc-5860	58	16	representations	representation	NOUN
sc-5860	58	17	.	.	PUNCT
sc-5860	59	1	using	use	VERB
sc-5860	59	2	the	the	DET
sc-5860	59	3	power	power	NOUN
sc-5860	59	4	of	of	ADP
sc-5860	59	5	pre	pre	ADJ
sc-5860	59	6	-	-	ADJ
sc-5860	59	7	trained	train	VERB
sc-5860	59	8	cnns	cnn	NOUN
sc-5860	59	9	,	,	PUNCT
sc-5860	59	10	you	you	PRON
sc-5860	59	11	et	et	VERB
sc-5860	59	12	al	al	PROPN
sc-5860	59	13	.	.	PUNCT
sc-5860	60	1	[	[	X
sc-5860	60	2	17	17	NUM
sc-5860	60	3	]	]	X
sc-5860	60	4	fine	fine	ADV
sc-5860	60	5	-	-	PUNCT
sc-5860	60	6	tuned	tune	VERB
sc-5860	60	7	networks	network	NOUN
sc-5860	60	8	for	for	ADP
sc-5860	60	9	binary	binary	ADJ
sc-5860	60	10	sentiment	sentiment	NOUN
sc-5860	60	11	classification	classification	NOUN
sc-5860	60	12	on	on	ADP
sc-5860	60	13	flickr	flickr	PROPN
sc-5860	60	14	and	and	CCONJ
sc-5860	60	15	twitter	twitter	NOUN
sc-5860	60	16	image	image	NOUN
sc-5860	60	17	datasets	dataset	NOUN
sc-5860	60	18	.	.	PUNCT
sc-5860	61	1	they	they	PRON
sc-5860	61	2	later	later	ADV
sc-5860	61	3	extended	extend	VERB
sc-5860	61	4	the	the	DET
sc-5860	61	5	visual	visual	ADJ
sc-5860	61	6	sensitivity	sensitivity	NOUN
sc-5860	61	7	problem	problem	NOUN
sc-5860	61	8	to	to	ADP
sc-5860	61	9	sentimental	sentimental	ADJ
sc-5860	61	10	image	image	NOUN
sc-5860	61	11	content	content	NOUN
sc-5860	61	12	analysis	analysis	NOUN
sc-5860	61	13	to	to	PART
sc-5860	61	14	predict	predict	VERB
sc-5860	61	15	sentiments	sentiment	NOUN
sc-5860	61	16	such	such	ADJ
sc-5860	61	17	as	as	ADP
sc-5860	61	18	amusement	amusement	NOUN
sc-5860	61	19	,	,	PUNCT
sc-5860	61	20	anger	anger	NOUN
sc-5860	61	21	,	,	PUNCT
sc-5860	61	22	awe	awe	PROPN
sc-5860	61	23	,	,	PUNCT
sc-5860	61	24	fear	fear	NOUN
sc-5860	61	25	,	,	PUNCT
sc-5860	61	26	sadness	sadness	NOUN
sc-5860	61	27	,	,	PUNCT
sc-5860	61	28	and	and	CCONJ
sc-5860	61	29	excitement	excitement	NOUN
sc-5860	61	30	[	[	X
sc-5860	61	31	18	18	NUM
sc-5860	61	32	]	]	PUNCT
sc-5860	61	33	.	.	PUNCT
sc-5860	62	1	recently	recently	ADV
sc-5860	62	2	,	,	PUNCT
sc-5860	62	3	jiang	jiang	PROPN
sc-5860	62	4	et	et	PROPN
sc-5860	62	5	al	al	PROPN
sc-5860	62	6	.	.	PUNCT
sc-5860	63	1	[	[	X
sc-5860	63	2	19	19	NUM
sc-5860	63	3	]	]	PUNCT
sc-5860	63	4	proposed	propose	VERB
sc-5860	63	5	another	another	DET
sc-5860	63	6	attention	attention	NOUN
sc-5860	63	7	mechanism	mechanism	NOUN
sc-5860	63	8	in	in	ADP
sc-5860	63	9	which	which	PRON
sc-5860	63	10	they	they	PRON
sc-5860	63	11	use	use	VERB
sc-5860	63	12	both	both	DET
sc-5860	63	13	cross	cross	ADJ
sc-5860	63	14	-	-	ADJ
sc-5860	63	15	modal	modal	ADJ
sc-5860	63	16	attention	attention	NOUN
sc-5860	63	17	fusion	fusion	NOUN
sc-5860	63	18	and	and	CCONJ
sc-5860	63	19	2	2	NUM
sc-5860	63	20	eai	eai	NOUN
sc-5860	63	21	endorsed	endorse	VERB
sc-5860	63	22	transactions	transaction	NOUN
sc-5860	63	23	on	on	ADP
sc-5860	63	24	smart	smart	ADJ
sc-5860	63	25	cities	city	NOUN
sc-5860	63	26	|	|	ADV
sc-5860	63	27	volume	volume	VERB
sc-5860	63	28	7	7	NUM
sc-5860	63	29	|	|	NOUN
sc-5860	63	30	issue	issue	VERB
sc-5860	63	31	4	4	NUM
sc-5860	63	32	|	|	CCONJ
sc-5860	63	33	2023	2023	NUM
sc-5860	63	34	|	|	NOUN
sc-5860	63	35	multimodal	multimodal	ADJ
sc-5860	63	36	sentiment	sentiment	NOUN
sc-5860	63	37	analysis	analysis	NOUN
sc-5860	63	38	in	in	ADP
sc-5860	63	39	natural	natural	ADJ
sc-5860	63	40	disaster	disaster	NOUN
sc-5860	63	41	data	datum	NOUN
sc-5860	63	42	on	on	ADP
sc-5860	63	43	social	social	ADJ
sc-5860	63	44	media	medium	NOUN
sc-5860	63	45	modality	modality	NOUN
sc-5860	63	46	-	-	PUNCT
sc-5860	63	47	specific	specific	ADJ
sc-5860	63	48	cnn	cnn	ADJ
sc-5860	63	49	-	-	PUNCT
sc-5860	63	50	transitive	transitive	ADJ
sc-5860	63	51	feature	feature	NOUN
sc-5860	63	52	extraction	extraction	NOUN
sc-5860	63	53	to	to	PART
sc-5860	63	54	learn	learn	VERB
sc-5860	63	55	a	a	DET
sc-5860	63	56	better	well	ADJ
sc-5860	63	57	representation	representation	NOUN
sc-5860	63	58	.	.	PUNCT
sc-5860	64	1	they	they	PRON
sc-5860	64	2	used	use	VERB
sc-5860	64	3	imagenet	imagenet	NOUN
sc-5860	65	1	[	[	X
sc-5860	65	2	20	20	NUM
sc-5860	65	3	]	]	X
sc-5860	65	4	pre	pre	ADJ
sc-5860	65	5	-	-	ADJ
sc-5860	65	6	trained	train	VERB
sc-5860	65	7	resnet	resnet	NOUN
sc-5860	65	8	[	[	X
sc-5860	65	9	21	21	NUM
sc-5860	65	10	]	]	PUNCT
sc-5860	65	11	for	for	ADP
sc-5860	65	12	visual	visual	ADJ
sc-5860	65	13	features	feature	NOUN
sc-5860	65	14	,	,	PUNCT
sc-5860	65	15	and	and	CCONJ
sc-5860	65	16	glove	glove	NOUN
sc-5860	66	1	[	[	X
sc-5860	66	2	22	22	NUM
sc-5860	66	3	]	]	PUNCT
sc-5860	66	4	and	and	CCONJ
sc-5860	66	5	bert	bert	X
sc-5860	67	1	[	[	X
sc-5860	67	2	23	23	NUM
sc-5860	67	3	]	]	PUNCT
sc-5860	67	4	for	for	ADP
sc-5860	67	5	textual	textual	ADJ
sc-5860	67	6	features	feature	NOUN
sc-5860	67	7	to	to	PART
sc-5860	67	8	achieve	achieve	VERB
sc-5860	67	9	the	the	DET
sc-5860	67	10	best	good	ADJ
sc-5860	67	11	results	result	NOUN
sc-5860	67	12	on	on	ADP
sc-5860	67	13	the	the	DET
sc-5860	67	14	mvsa	mvsa	NOUN
sc-5860	67	15	dataset	dataset	NOUN
sc-5860	67	16	.	.	PUNCT
sc-5860	68	1	bica	bica	PROPN
sc-5860	68	2	et	et	PROPN
sc-5860	68	3	al	al	PROPN
sc-5860	68	4	.	.	PUNCT
sc-5860	69	1	[	[	X
sc-5860	69	2	24	24	NUM
sc-5860	69	3	]	]	PUNCT
sc-5860	69	4	combined	combine	VERB
sc-5860	69	5	textual	textual	ADJ
sc-5860	69	6	and	and	CCONJ
sc-5860	69	7	visual	visual	ADJ
sc-5860	69	8	content	content	NOUN
sc-5860	69	9	in	in	ADP
sc-5860	69	10	their	their	PRON
sc-5860	69	11	study	study	NOUN
sc-5860	69	12	,	,	PUNCT
sc-5860	69	13	examined	examine	VERB
sc-5860	69	14	geotagged	geotagge	VERB
sc-5860	69	15	images	image	NOUN
sc-5860	69	16	on	on	ADP
sc-5860	69	17	social	social	ADJ
sc-5860	69	18	media	medium	NOUN
sc-5860	69	19	published	publish	VERB
sc-5860	69	20	during	during	ADP
sc-5860	69	21	two	two	NUM
sc-5860	69	22	major	major	ADJ
sc-5860	69	23	earthquakes	earthquake	NOUN
sc-5860	69	24	in	in	ADP
sc-5860	69	25	nepal	nepal	NOUN
sc-5860	69	26	in	in	ADP
sc-5860	69	27	april	april	PROPN
sc-5860	69	28	-	-	PUNCT
sc-5860	69	29	may	may	PROPN
sc-5860	69	30	2015	2015	NUM
sc-5860	69	31	,	,	PUNCT
sc-5860	69	32	and	and	CCONJ
sc-5860	69	33	focused	focus	VERB
sc-5860	69	34	on	on	ADP
sc-5860	69	35	identifying	identify	VERB
sc-5860	69	36	the	the	DET
sc-5860	69	37	damages	damage	NOUN
sc-5860	69	38	associated	associate	VERB
sc-5860	69	39	with	with	ADP
sc-5860	69	40	them	they	PRON
sc-5860	69	41	.	.	PUNCT
sc-5860	70	1	sit	sit	VERB
sc-5860	70	2	et	et	PROPN
sc-5860	70	3	al	al	PROPN
sc-5860	70	4	.	.	PUNCT
sc-5860	71	1	[	[	X
sc-5860	71	2	25	25	NUM
sc-5860	71	3	]	]	PUNCT
sc-5860	71	4	analyzed	analyze	VERB
sc-5860	71	5	tweets	tweet	NOUN
sc-5860	71	6	to	to	PART
sc-5860	71	7	identify	identify	VERB
sc-5860	71	8	and	and	CCONJ
sc-5860	71	9	categorize	categorize	VERB
sc-5860	71	10	fine	fine	ADV
sc-5860	71	11	-	-	PUNCT
sc-5860	71	12	grained	grain	VERB
sc-5860	71	13	details	detail	NOUN
sc-5860	71	14	about	about	ADP
sc-5860	71	15	a	a	DET
sc-5860	71	16	disaster	disaster	NOUN
sc-5860	71	17	,	,	PUNCT
sc-5860	71	18	such	such	ADJ
sc-5860	71	19	as	as	ADP
sc-5860	71	20	affected	affected	ADJ
sc-5860	71	21	individuals	individual	NOUN
sc-5860	71	22	,	,	PUNCT
sc-5860	71	23	damaged	damaged	ADJ
sc-5860	71	24	infrastructure	infrastructure	NOUN
sc-5860	71	25	,	,	PUNCT
sc-5860	71	26	and	and	CCONJ
sc-5860	71	27	interrupted	interrupted	ADJ
sc-5860	71	28	services	service	NOUN
sc-5860	71	29	,	,	PUNCT
sc-5860	71	30	as	as	ADV
sc-5860	71	31	well	well	ADV
sc-5860	71	32	as	as	ADP
sc-5860	71	33	to	to	PART
sc-5860	71	34	distinguish	distinguish	VERB
sc-5860	71	35	domains	domain	NOUN
sc-5860	71	36	and	and	CCONJ
sc-5860	71	37	time	time	NOUN
sc-5860	71	38	periods	period	NOUN
sc-5860	71	39	and	and	CCONJ
sc-5860	71	40	the	the	DET
sc-5860	71	41	relative	relative	ADJ
sc-5860	71	42	importance	importance	NOUN
sc-5860	71	43	of	of	ADP
sc-5860	71	44	each	each	DET
sc-5860	71	45	category	category	NOUN
sc-5860	71	46	of	of	ADP
sc-5860	71	47	disasterrelated	disasterrelate	VERB
sc-5860	71	48	information	information	NOUN
sc-5860	71	49	in	in	ADP
sc-5860	71	50	space	space	NOUN
sc-5860	71	51	and	and	CCONJ
sc-5860	71	52	time	time	NOUN
sc-5860	71	53	.	.	PUNCT
sc-5860	72	1	hassan	hassan	PROPN
sc-5860	72	2	et	et	PROPN
sc-5860	72	3	al	al	PROPN
sc-5860	72	4	.	.	PUNCT
sc-5860	73	1	[	[	X
sc-5860	73	2	26	26	NUM
sc-5860	73	3	]	]	PUNCT
sc-5860	73	4	proposed	propose	VERB
sc-5860	73	5	a	a	DET
sc-5860	73	6	deep	deep	ADJ
sc-5860	73	7	visual	visual	ADJ
sc-5860	73	8	emotion	emotion	NOUN
sc-5860	73	9	analyzer	analyzer	NOUN
sc-5860	73	10	that	that	PRON
sc-5860	73	11	covers	cover	VERB
sc-5860	73	12	different	different	ADJ
sc-5860	73	13	aspects	aspect	NOUN
sc-5860	73	14	of	of	ADP
sc-5860	73	15	visual	visual	ADJ
sc-5860	73	16	sentiment	sentiment	NOUN
sc-5860	73	17	analysis	analysis	NOUN
sc-5860	73	18	,	,	PUNCT
sc-5860	73	19	starting	start	VERB
sc-5860	73	20	from	from	ADP
sc-5860	73	21	data	data	PROPN
sc-5860	73	22	collection	collection	NOUN
sc-5860	73	23	,	,	PUNCT
sc-5860	73	24	annotation	annotation	NOUN
sc-5860	73	25	,	,	PUNCT
sc-5860	73	26	model	model	NOUN
sc-5860	73	27	selection	selection	NOUN
sc-5860	73	28	,	,	PUNCT
sc-5860	73	29	application	application	NOUN
sc-5860	73	30	and	and	CCONJ
sc-5860	73	31	evaluations	evaluation	NOUN
sc-5860	73	32	for	for	ADP
sc-5860	73	33	disaster	disaster	NOUN
sc-5860	73	34	-	-	PUNCT
sc-5860	73	35	related	relate	VERB
sc-5860	73	36	images	image	NOUN
sc-5860	73	37	.	.	PUNCT
sc-5860	74	1	competitions	competition	NOUN
sc-5860	74	2	focusing	focus	VERB
sc-5860	74	3	on	on	ADP
sc-5860	74	4	visual	visual	ADJ
sc-5860	74	5	sentiment	sentiment	NOUN
sc-5860	74	6	analysis	analysis	NOUN
sc-5860	74	7	on	on	ADP
sc-5860	74	8	natural	natural	ADJ
sc-5860	74	9	disasters	disaster	NOUN
sc-5860	74	10	are	be	AUX
sc-5860	74	11	also	also	ADV
sc-5860	74	12	organized	organize	VERB
sc-5860	74	13	[	[	X
sc-5860	74	14	27	27	NUM
sc-5860	74	15	]	]	PUNCT
sc-5860	74	16	.	.	PUNCT
sc-5860	75	1	niu	niu	PROPN
sc-5860	75	2	et	et	PROPN
sc-5860	75	3	al	al	PROPN
sc-5860	75	4	.	.	PUNCT
sc-5860	76	1	[	[	X
sc-5860	76	2	28	28	NUM
sc-5860	76	3	]	]	PUNCT
sc-5860	76	4	introduce	introduce	VERB
sc-5860	76	5	a	a	DET
sc-5860	76	6	multi	multi	ADJ
sc-5860	76	7	-	-	ADJ
sc-5860	76	8	view	view	ADJ
sc-5860	76	9	sentiment	sentiment	NOUN
sc-5860	76	10	analysis	analysis	NOUN
sc-5860	76	11	dataset	dataset	NOUN
sc-5860	76	12	(	(	PUNCT
sc-5860	76	13	mvsa	mvsa	NOUN
sc-5860	76	14	)	)	PUNCT
sc-5860	76	15	including	include	VERB
sc-5860	76	16	a	a	DET
sc-5860	76	17	set	set	NOUN
sc-5860	76	18	of	of	ADP
sc-5860	76	19	image	image	NOUN
sc-5860	76	20	-	-	PUNCT
sc-5860	76	21	text	text	NOUN
sc-5860	76	22	pairs	pair	NOUN
sc-5860	76	23	with	with	ADP
sc-5860	76	24	manual	manual	ADJ
sc-5860	76	25	annotations	annotation	NOUN
sc-5860	76	26	collected	collect	VERB
sc-5860	76	27	from	from	ADP
sc-5860	76	28	twitter	twitter	NOUN
sc-5860	76	29	.	.	PUNCT
sc-5860	77	1	mvsa	mvsa	NOUN
sc-5860	77	2	can	can	AUX
sc-5860	77	3	be	be	AUX
sc-5860	77	4	utilized	utilize	VERB
sc-5860	77	5	as	as	ADP
sc-5860	77	6	a	a	DET
sc-5860	77	7	valuable	valuable	ADJ
sc-5860	77	8	benchmark	benchmark	NOUN
sc-5860	77	9	for	for	ADP
sc-5860	77	10	both	both	PRON
sc-5860	77	11	single	single	ADJ
sc-5860	77	12	-	-	PUNCT
sc-5860	77	13	view	view	NOUN
sc-5860	77	14	and	and	CCONJ
sc-5860	77	15	multi	multi	ADJ
sc-5860	77	16	-	-	ADJ
sc-5860	77	17	view	view	ADJ
sc-5860	77	18	sentiment	sentiment	NOUN
sc-5860	77	19	analysis	analysis	NOUN
sc-5860	77	20	.	.	PUNCT
sc-5860	78	1	in	in	ADP
sc-5860	78	2	this	this	DET
sc-5860	78	3	study	study	NOUN
sc-5860	78	4	,	,	PUNCT
sc-5860	78	5	multimodal	multimodal	ADJ
sc-5860	78	6	sentiment	sentiment	NOUN
sc-5860	78	7	classification	classification	NOUN
sc-5860	78	8	from	from	ADP
sc-5860	78	9	natural	natural	ADJ
sc-5860	78	10	disaster	disaster	NOUN
sc-5860	78	11	data	datum	NOUN
sc-5860	78	12	on	on	ADP
sc-5860	78	13	social	social	ADJ
sc-5860	78	14	media	medium	NOUN
sc-5860	78	15	is	be	AUX
sc-5860	78	16	discussed	discuss	VERB
sc-5860	78	17	by	by	ADP
sc-5860	78	18	combining	combine	VERB
sc-5860	78	19	text	text	NOUN
sc-5860	78	20	and	and	CCONJ
sc-5860	78	21	visual	visual	ADJ
sc-5860	78	22	content	content	NOUN
sc-5860	78	23	.	.	PUNCT
sc-5860	79	1	the	the	DET
sc-5860	79	2	use	use	NOUN
sc-5860	79	3	of	of	ADP
sc-5860	79	4	textual	textual	ADJ
sc-5860	79	5	representations	representation	NOUN
sc-5860	79	6	together	together	ADV
sc-5860	79	7	with	with	ADP
sc-5860	79	8	the	the	DET
sc-5860	79	9	sentimental	sentimental	ADJ
sc-5860	79	10	expressions	expression	NOUN
sc-5860	79	11	of	of	ADP
sc-5860	79	12	the	the	DET
sc-5860	79	13	visual	visual	ADJ
sc-5860	79	14	content	content	NOUN
sc-5860	79	15	provides	provide	VERB
sc-5860	79	16	a	a	DET
sc-5860	79	17	more	more	ADV
sc-5860	79	18	comprehensive	comprehensive	ADJ
sc-5860	79	19	analysis	analysis	NOUN
sc-5860	79	20	.	.	PUNCT
sc-5860	80	1	to	to	PART
sc-5860	80	2	investigate	investigate	VERB
sc-5860	80	3	the	the	DET
sc-5860	80	4	impact	impact	NOUN
sc-5860	80	5	of	of	ADP
sc-5860	80	6	high	high	ADJ
sc-5860	80	7	-	-	PUNCT
sc-5860	80	8	level	level	NOUN
sc-5860	80	9	visual	visual	ADJ
sc-5860	80	10	features	feature	NOUN
sc-5860	80	11	,	,	PUNCT
sc-5860	80	12	a	a	DET
sc-5860	80	13	three	three	NUM
sc-5860	80	14	-	-	PUNCT
sc-5860	80	15	layer	layer	NOUN
sc-5860	80	16	neural	neural	ADJ
sc-5860	80	17	network	network	NOUN
sc-5860	80	18	is	be	AUX
sc-5860	80	19	used	use	VERB
sc-5860	80	20	in	in	ADP
sc-5860	80	21	the	the	DET
sc-5860	80	22	study	study	NOUN
sc-5860	80	23	,	,	PUNCT
sc-5860	80	24	where	where	SCONJ
sc-5860	80	25	the	the	DET
sc-5860	80	26	first	first	ADJ
sc-5860	80	27	two	two	NUM
sc-5860	80	28	layers	layer	NOUN
sc-5860	80	29	collect	collect	VERB
sc-5860	80	30	features	feature	NOUN
sc-5860	80	31	from	from	ADP
sc-5860	80	32	different	different	ADJ
sc-5860	80	33	modalities	modality	NOUN
sc-5860	80	34	and	and	CCONJ
sc-5860	80	35	the	the	DET
sc-5860	80	36	third	third	ADJ
sc-5860	80	37	layer	layer	NOUN
sc-5860	80	38	is	be	AUX
sc-5860	80	39	used	use	VERB
sc-5860	80	40	for	for	ADP
sc-5860	80	41	classification	classification	NOUN
sc-5860	80	42	of	of	ADP
sc-5860	80	43	sentiments	sentiment	NOUN
sc-5860	80	44	.	.	PUNCT
sc-5860	81	1	3	3	X
sc-5860	81	2	.	.	X
sc-5860	81	3	material	material	NOUN
sc-5860	81	4	and	and	CCONJ
sc-5860	81	5	methods	method	NOUN
sc-5860	81	6	the	the	DET
sc-5860	81	7	human	human	ADJ
sc-5860	81	8	brain	brain	NOUN
sc-5860	81	9	consists	consist	VERB
sc-5860	81	10	of	of	ADP
sc-5860	81	11	neural	neural	ADJ
sc-5860	81	12	networks	network	NOUN
sc-5860	81	13	that	that	PRON
sc-5860	81	14	can	can	AUX
sc-5860	81	15	process	process	VERB
sc-5860	81	16	multiple	multiple	ADJ
sc-5860	81	17	modalities	modality	NOUN
sc-5860	81	18	simultaneously	simultaneously	ADV
sc-5860	81	19	.	.	PUNCT
sc-5860	82	1	for	for	ADP
sc-5860	82	2	example	example	NOUN
sc-5860	82	3	,	,	PUNCT
sc-5860	82	4	when	when	SCONJ
sc-5860	82	5	a	a	DET
sc-5860	82	6	conversation	conversation	NOUN
sc-5860	82	7	is	be	AUX
sc-5860	82	8	held	hold	VERB
sc-5860	82	9	,	,	PUNCT
sc-5860	82	10	the	the	DET
sc-5860	82	11	brain	brain	NOUN
sc-5860	82	12	’s	’s	PART
sc-5860	82	13	neural	neural	ADJ
sc-5860	82	14	networks	network	NOUN
sc-5860	82	15	process	process	NOUN
sc-5860	82	16	multimodal	multimodal	NOUN
sc-5860	82	17	input	input	NOUN
sc-5860	82	18	(	(	PUNCT
sc-5860	82	19	sound	sound	ADJ
sc-5860	82	20	,	,	PUNCT
sc-5860	82	21	image	image	NOUN
sc-5860	82	22	,	,	PUNCT
sc-5860	82	23	text	text	NOUN
sc-5860	82	24	,	,	PUNCT
sc-5860	82	25	smell	smell	NOUN
sc-5860	82	26	)	)	PUNCT
sc-5860	82	27	.	.	PUNCT
sc-5860	83	1	after	after	ADP
sc-5860	83	2	a	a	DET
sc-5860	83	3	deep	deep	ADJ
sc-5860	83	4	subconscious	subconscious	ADJ
sc-5860	83	5	modality	modality	NOUN
sc-5860	83	6	fusion	fusion	NOUN
sc-5860	83	7	,	,	PUNCT
sc-5860	83	8	we	we	PRON
sc-5860	83	9	are	be	AUX
sc-5860	83	10	able	able	ADJ
sc-5860	83	11	to	to	PART
sc-5860	83	12	reason	reason	VERB
sc-5860	83	13	about	about	ADP
sc-5860	83	14	what	what	PRON
sc-5860	83	15	our	our	PRON
sc-5860	83	16	conversational	conversational	ADJ
sc-5860	83	17	interlocutor	interlocutor	NOUN
sc-5860	83	18	is	be	AUX
sc-5860	83	19	saying	say	VERB
sc-5860	83	20	,	,	PUNCT
sc-5860	83	21	his	his	PRON
sc-5860	83	22	emotional	emotional	ADJ
sc-5860	83	23	state	state	NOUN
sc-5860	83	24	,	,	PUNCT
sc-5860	83	25	and	and	CCONJ
sc-5860	83	26	our	our	PRON
sc-5860	83	27	environment	environment	NOUN
sc-5860	83	28	.	.	PUNCT
sc-5860	84	1	this	this	DET
sc-5860	84	2	approach	approach	NOUN
sc-5860	84	3	allows	allow	VERB
sc-5860	84	4	for	for	ADP
sc-5860	84	5	a	a	DET
sc-5860	84	6	more	more	ADV
sc-5860	84	7	holistic	holistic	ADJ
sc-5860	84	8	view	view	NOUN
sc-5860	84	9	and	and	CCONJ
sc-5860	84	10	deeper	deep	ADJ
sc-5860	84	11	understanding	understanding	NOUN
sc-5860	84	12	of	of	ADP
sc-5860	84	13	the	the	DET
sc-5860	84	14	situation	situation	NOUN
sc-5860	84	15	.	.	PUNCT
sc-5860	85	1	within	within	ADP
sc-5860	85	2	the	the	DET
sc-5860	85	3	scope	scope	NOUN
sc-5860	85	4	of	of	ADP
sc-5860	85	5	artificial	artificial	ADJ
sc-5860	85	6	intelligence	intelligence	NOUN
sc-5860	85	7	’s	’s	PART
sc-5860	85	8	quest	quest	NOUN
sc-5860	85	9	to	to	PART
sc-5860	85	10	imitate	imitate	VERB
sc-5860	85	11	human	human	ADJ
sc-5860	85	12	intelligence	intelligence	NOUN
sc-5860	85	13	,	,	PUNCT
sc-5860	85	14	it	it	PRON
sc-5860	85	15	is	be	AUX
sc-5860	85	16	an	an	DET
sc-5860	85	17	inevitable	inevitable	ADJ
sc-5860	85	18	fact	fact	NOUN
sc-5860	85	19	that	that	SCONJ
sc-5860	85	20	artificial	artificial	ADJ
sc-5860	85	21	intelligence	intelligence	NOUN
sc-5860	85	22	will	will	AUX
sc-5860	85	23	learn	learn	VERB
sc-5860	85	24	to	to	PART
sc-5860	85	25	interpret	interpret	VERB
sc-5860	85	26	,	,	PUNCT
sc-5860	85	27	reason	reason	NOUN
sc-5860	85	28	and	and	CCONJ
sc-5860	85	29	combine	combine	VERB
sc-5860	85	30	multimodal	multimodal	ADJ
sc-5860	85	31	information	information	NOUN
sc-5860	85	32	to	to	PART
sc-5860	85	33	match	match	VERB
sc-5860	85	34	human	human	ADJ
sc-5860	85	35	intelligence	intelligence	NOUN
sc-5860	85	36	.	.	PUNCT
sc-5860	86	1	in	in	ADP
sc-5860	86	2	this	this	DET
sc-5860	86	3	study	study	NOUN
sc-5860	86	4	,	,	PUNCT
sc-5860	86	5	studies	study	NOUN
sc-5860	86	6	are	be	AUX
sc-5860	86	7	carried	carry	VERB
sc-5860	86	8	out	out	ADP
sc-5860	86	9	on	on	ADP
sc-5860	86	10	a	a	DET
sc-5860	86	11	deep	deep	ADJ
sc-5860	86	12	learning	learning	NOUN
sc-5860	86	13	model	model	NOUN
sc-5860	86	14	that	that	PRON
sc-5860	86	15	perceives	perceive	VERB
sc-5860	86	16	the	the	DET
sc-5860	86	17	world	world	NOUN
sc-5860	86	18	more	more	ADV
sc-5860	86	19	holistically	holistically	ADV
sc-5860	86	20	with	with	ADP
sc-5860	86	21	multi	multi	ADJ
sc-5860	86	22	-	-	ADJ
sc-5860	86	23	modal	modal	ADJ
sc-5860	86	24	analyzes	analyze	NOUN
sc-5860	86	25	within	within	ADP
sc-5860	86	26	the	the	DET
sc-5860	86	27	scope	scope	NOUN
sc-5860	86	28	of	of	ADP
sc-5860	86	29	text	text	NOUN
sc-5860	86	30	+	+	CCONJ
sc-5860	86	31	figure	figure	NOUN
sc-5860	86	32	1	1	NUM
sc-5860	86	33	.	.	PUNCT
sc-5860	86	34	architecture	architecture	NOUN
sc-5860	86	35	for	for	ADP
sc-5860	86	36	multimodal	multimodal	ADJ
sc-5860	86	37	sentiment	sentiment	NOUN
sc-5860	86	38	learning	learn	VERB
sc-5860	86	39	image	image	NOUN
sc-5860	86	40	.	.	PUNCT
sc-5860	87	1	the	the	DET
sc-5860	87	2	main	main	ADJ
sc-5860	87	3	idea	idea	NOUN
sc-5860	87	4	under	under	ADP
sc-5860	87	5	multimodal	multimodal	NOUN
sc-5860	87	6	classification	classification	NOUN
sc-5860	87	7	is	be	AUX
sc-5860	87	8	to	to	PART
sc-5860	87	9	use	use	VERB
sc-5860	87	10	different	different	ADJ
sc-5860	87	11	types	type	NOUN
sc-5860	87	12	of	of	ADP
sc-5860	87	13	high	high	ADJ
sc-5860	87	14	-	-	PUNCT
sc-5860	87	15	level	level	NOUN
sc-5860	87	16	visual	visual	ADJ
sc-5860	87	17	features	feature	NOUN
sc-5860	87	18	and	and	CCONJ
sc-5860	87	19	combine	combine	VERB
sc-5860	87	20	them	they	PRON
sc-5860	87	21	with	with	ADP
sc-5860	87	22	a	a	DET
sc-5860	87	23	textual	textual	ADJ
sc-5860	87	24	model	model	NOUN
sc-5860	87	25	.	.	PUNCT
sc-5860	88	1	the	the	DET
sc-5860	88	2	used	use	VERB
sc-5860	88	3	architecture	architecture	NOUN
sc-5860	88	4	is	be	AUX
sc-5860	88	5	shown	show	VERB
sc-5860	88	6	in	in	ADP
sc-5860	88	7	figure	figure	NOUN
sc-5860	88	8	1	1	NUM
sc-5860	88	9	and	and	CCONJ
sc-5860	88	10	it	it	PRON
sc-5860	88	11	is	be	AUX
sc-5860	88	12	basis	basis	NOUN
sc-5860	88	13	on	on	ADP
sc-5860	88	14	semlnn	semlnn	NOUN
sc-5860	88	15	[	[	X
sc-5860	88	16	14	14	NUM
sc-5860	88	17	]	]	PUNCT
sc-5860	88	18	.	.	PUNCT
sc-5860	89	1	it	it	PRON
sc-5860	89	2	combines	combine	VERB
sc-5860	89	3	several	several	ADJ
sc-5860	89	4	visual	visual	ADJ
sc-5860	89	5	features	feature	NOUN
sc-5860	89	6	with	with	ADP
sc-5860	89	7	contextual	contextual	ADJ
sc-5860	89	8	text	text	NOUN
sc-5860	89	9	features	feature	NOUN
sc-5860	89	10	to	to	PART
sc-5860	89	11	predict	predict	VERB
sc-5860	89	12	the	the	DET
sc-5860	89	13	overall	overall	ADJ
sc-5860	89	14	sentiment	sentiment	NOUN
sc-5860	89	15	accurately	accurately	ADV
sc-5860	89	16	.	.	PUNCT
sc-5860	90	1	the	the	DET
sc-5860	90	2	detailed	detailed	ADJ
sc-5860	90	3	information	information	NOUN
sc-5860	90	4	of	of	ADP
sc-5860	90	5	the	the	DET
sc-5860	90	6	architecture	architecture	NOUN
sc-5860	90	7	are	be	AUX
sc-5860	90	8	explained	explain	VERB
sc-5860	90	9	in	in	ADP
sc-5860	90	10	the	the	DET
sc-5860	90	11	subsections	subsection	NOUN
sc-5860	90	12	.	.	PUNCT
sc-5860	91	1	3.1	3.1	NUM
sc-5860	91	2	.	.	PUNCT
sc-5860	92	1	visual	visual	ADJ
sc-5860	92	2	features	feature	VERB
sc-5860	92	3	visual	visual	ADJ
sc-5860	92	4	features	feature	NOUN
sc-5860	92	5	within	within	ADP
sc-5860	92	6	the	the	DET
sc-5860	92	7	scope	scope	NOUN
sc-5860	92	8	of	of	ADP
sc-5860	92	9	the	the	DET
sc-5860	92	10	study	study	NOUN
sc-5860	92	11	;	;	PUNCT
sc-5860	92	12	it	it	PRON
sc-5860	92	13	is	be	AUX
sc-5860	92	14	examined	examine	VERB
sc-5860	92	15	in	in	ADP
sc-5860	92	16	four	four	NUM
sc-5860	92	17	categories	category	NOUN
sc-5860	92	18	:	:	PUNCT
sc-5860	92	19	object	object	VERB
sc-5860	92	20	features	feature	NOUN
sc-5860	92	21	,	,	PUNCT
sc-5860	92	22	place	place	NOUN
sc-5860	92	23	and	and	CCONJ
sc-5860	92	24	scene	scene	NOUN
sc-5860	92	25	features	feature	NOUN
sc-5860	92	26	,	,	PUNCT
sc-5860	92	27	facial	facial	ADJ
sc-5860	92	28	expressions	expression	NOUN
sc-5860	92	29	,	,	PUNCT
sc-5860	92	30	and	and	CCONJ
sc-5860	92	31	effective	effective	ADJ
sc-5860	92	32	image	image	NOUN
sc-5860	92	33	content	content	NOUN
sc-5860	92	34	.	.	PUNCT
sc-5860	93	1	object	object	NOUN
sc-5860	93	2	features	feature	NOUN
sc-5860	93	3	.	.	PUNCT
sc-5860	94	1	object	object	NOUN
sc-5860	94	2	features	feature	NOUN
sc-5860	94	3	refer	refer	VERB
sc-5860	94	4	to	to	ADP
sc-5860	94	5	the	the	DET
sc-5860	94	6	attributes	attribute	NOUN
sc-5860	94	7	or	or	CCONJ
sc-5860	94	8	characteristics	characteristic	NOUN
sc-5860	94	9	that	that	PRON
sc-5860	94	10	can	can	AUX
sc-5860	94	11	be	be	AUX
sc-5860	94	12	used	use	VERB
sc-5860	94	13	to	to	PART
sc-5860	94	14	describe	describe	VERB
sc-5860	94	15	and	and	CCONJ
sc-5860	94	16	identify	identify	VERB
sc-5860	94	17	an	an	DET
sc-5860	94	18	object	object	NOUN
sc-5860	94	19	.	.	PUNCT
sc-5860	95	1	these	these	DET
sc-5860	95	2	features	feature	NOUN
sc-5860	95	3	can	can	AUX
sc-5860	95	4	be	be	AUX
sc-5860	95	5	derived	derive	VERB
sc-5860	95	6	from	from	ADP
sc-5860	95	7	various	various	ADJ
sc-5860	95	8	sensory	sensory	ADJ
sc-5860	95	9	inputs	input	NOUN
sc-5860	95	10	,	,	PUNCT
sc-5860	95	11	including	include	VERB
sc-5860	95	12	visual	visual	ADJ
sc-5860	95	13	,	,	PUNCT
sc-5860	95	14	tactile	tactile	ADJ
sc-5860	95	15	,	,	PUNCT
sc-5860	95	16	auditory	auditory	ADJ
sc-5860	95	17	,	,	PUNCT
sc-5860	95	18	and	and	CCONJ
sc-5860	95	19	more	more	ADJ
sc-5860	95	20	.	.	PUNCT
sc-5860	96	1	in	in	ADP
sc-5860	96	2	the	the	DET
sc-5860	96	3	context	context	NOUN
sc-5860	96	4	of	of	ADP
sc-5860	96	5	visual	visual	ADJ
sc-5860	96	6	perception	perception	NOUN
sc-5860	96	7	and	and	CCONJ
sc-5860	96	8	computer	computer	NOUN
sc-5860	96	9	vision	vision	NOUN
sc-5860	96	10	,	,	PUNCT
sc-5860	96	11	object	object	NOUN
sc-5860	96	12	features	feature	NOUN
sc-5860	96	13	primarily	primarily	ADV
sc-5860	96	14	involve	involve	VERB
sc-5860	96	15	characteristics	characteristic	NOUN
sc-5860	96	16	that	that	PRON
sc-5860	96	17	can	can	AUX
sc-5860	96	18	be	be	AUX
sc-5860	96	19	observed	observe	VERB
sc-5860	96	20	and	and	CCONJ
sc-5860	96	21	analyzed	analyze	VERB
sc-5860	96	22	from	from	ADP
sc-5860	96	23	images	image	NOUN
sc-5860	96	24	or	or	CCONJ
sc-5860	96	25	visual	visual	ADJ
sc-5860	96	26	data	datum	NOUN
sc-5860	96	27	.	.	PUNCT
sc-5860	97	1	different	different	ADJ
sc-5860	97	2	objects	object	NOUN
sc-5860	97	3	in	in	ADP
sc-5860	97	4	an	an	DET
sc-5860	97	5	image	image	NOUN
sc-5860	97	6	can	can	AUX
sc-5860	97	7	evoke	evoke	VERB
sc-5860	97	8	a	a	DET
sc-5860	97	9	certain	certain	ADJ
sc-5860	97	10	sentiment	sentiment	NOUN
sc-5860	97	11	in	in	ADP
sc-5860	97	12	a	a	DET
sc-5860	97	13	person	person	NOUN
sc-5860	97	14	.	.	PUNCT
sc-5860	98	1	for	for	ADP
sc-5860	98	2	example	example	NOUN
sc-5860	98	3	,	,	PUNCT
sc-5860	98	4	while	while	SCONJ
sc-5860	98	5	the	the	DET
sc-5860	98	6	destructive	destructive	ADJ
sc-5860	98	7	effect	effect	NOUN
sc-5860	98	8	of	of	ADP
sc-5860	98	9	the	the	DET
sc-5860	98	10	earthquake	earthquake	NOUN
sc-5860	98	11	may	may	AUX
sc-5860	98	12	evoke	evoke	VERB
sc-5860	98	13	a	a	DET
sc-5860	98	14	negative	negative	ADJ
sc-5860	98	15	emotion	emotion	NOUN
sc-5860	98	16	,	,	PUNCT
sc-5860	98	17	objects	object	VERB
sc-5860	98	18	that	that	PRON
sc-5860	98	19	involve	involve	VERB
sc-5860	98	20	humanitarian	humanitarian	ADJ
sc-5860	98	21	aid	aid	NOUN
sc-5860	98	22	organizations	organization	NOUN
sc-5860	98	23	providing	provide	VERB
sc-5860	98	24	aid	aid	NOUN
sc-5860	98	25	to	to	ADP
sc-5860	98	26	earthquake	earthquake	NOUN
sc-5860	98	27	-	-	PUNCT
sc-5860	98	28	affected	affect	VERB
sc-5860	98	29	citizens	citizen	NOUN
sc-5860	98	30	(	(	PUNCT
sc-5860	98	31	tents	tent	NOUN
sc-5860	98	32	,	,	PUNCT
sc-5860	98	33	aid	aid	NOUN
sc-5860	98	34	parcels	parcel	NOUN
sc-5860	98	35	,	,	PUNCT
sc-5860	98	36	volunteer	volunteer	NOUN
sc-5860	98	37	helpful	helpful	ADJ
sc-5860	98	38	communities	community	NOUN
sc-5860	98	39	)	)	PUNCT
sc-5860	98	40	may	may	AUX
sc-5860	98	41	evoke	evoke	VERB
sc-5860	98	42	a	a	DET
sc-5860	98	43	positive	positive	ADJ
sc-5860	98	44	emotion	emotion	NOUN
sc-5860	98	45	.	.	PUNCT
sc-5860	99	1	a	a	DET
sc-5860	99	2	pre	pre	ADJ
sc-5860	99	3	-	-	ADJ
sc-5860	99	4	trained	train	VERB
sc-5860	99	5	resnet	resnet	NOUN
sc-5860	99	6	model	model	NOUN
sc-5860	99	7	on	on	ADP
sc-5860	99	8	imagenet	imagenet	NOUN
sc-5860	99	9	is	be	AUX
sc-5860	99	10	used	use	VERB
sc-5860	99	11	to	to	PART
sc-5860	99	12	encode	encode	VERB
sc-5860	99	13	objects	object	NOUN
sc-5860	99	14	and	and	CCONJ
sc-5860	99	15	overall	overall	ADJ
sc-5860	99	16	image	image	NOUN
sc-5860	99	17	content	content	NOUN
sc-5860	99	18	.	.	PUNCT
sc-5860	100	1	resnet-50	resnet-50	PROPN
sc-5860	100	2	and	and	CCONJ
sc-5860	100	3	its	its	PRON
sc-5860	100	4	final	final	ADJ
sc-5860	100	5	convolution	convolution	NOUN
sc-5860	100	6	layer	layer	NOUN
sc-5860	100	7	are	be	AUX
sc-5860	100	8	used	use	VERB
sc-5860	100	9	instead	instead	ADV
sc-5860	100	10	of	of	ADP
sc-5860	100	11	object	object	NOUN
sc-5860	100	12	categories	category	NOUN
sc-5860	100	13	(	(	PUNCT
sc-5860	100	14	last	last	ADJ
sc-5860	100	15	layer	layer	NOUN
sc-5860	100	16	)	)	PUNCT
sc-5860	100	17	to	to	PART
sc-5860	100	18	extract	extract	VERB
sc-5860	100	19	features	feature	NOUN
sc-5860	100	20	.	.	PUNCT
sc-5860	101	1	place	place	NOUN
sc-5860	101	2	and	and	CCONJ
sc-5860	101	3	scene	scene	NOUN
sc-5860	101	4	features	feature	NOUN
sc-5860	101	5	.	.	PUNCT
sc-5860	102	1	place	place	NOUN
sc-5860	102	2	and	and	CCONJ
sc-5860	102	3	scene	scene	NOUN
sc-5860	102	4	features	feature	NOUN
sc-5860	102	5	in	in	ADP
sc-5860	102	6	visual	visual	ADJ
sc-5860	102	7	perception	perception	NOUN
sc-5860	102	8	refer	refer	VERB
sc-5860	102	9	to	to	ADP
sc-5860	102	10	the	the	DET
sc-5860	102	11	attributes	attribute	NOUN
sc-5860	102	12	that	that	PRON
sc-5860	102	13	characterize	characterize	VERB
sc-5860	102	14	entire	entire	ADJ
sc-5860	102	15	environments	environment	NOUN
sc-5860	102	16	or	or	CCONJ
sc-5860	102	17	contexts	context	NOUN
sc-5860	102	18	,	,	PUNCT
sc-5860	102	19	rather	rather	ADV
sc-5860	102	20	than	than	ADP
sc-5860	102	21	individual	individual	ADJ
sc-5860	102	22	objects	object	NOUN
sc-5860	102	23	.	.	PUNCT
sc-5860	103	1	these	these	PRON
sc-5860	103	2	features	feature	VERB
sc-5860	103	3	help	help	NOUN
sc-5860	103	4	in	in	ADP
sc-5860	103	5	understanding	understand	VERB
sc-5860	103	6	the	the	DET
sc-5860	103	7	broader	broad	ADJ
sc-5860	103	8	context	context	NOUN
sc-5860	103	9	of	of	ADP
sc-5860	103	10	a	a	DET
sc-5860	103	11	visual	visual	ADJ
sc-5860	103	12	scene	scene	NOUN
sc-5860	103	13	,	,	PUNCT
sc-5860	103	14	such	such	ADJ
sc-5860	103	15	as	as	ADP
sc-5860	103	16	the	the	DET
sc-5860	103	17	type	type	NOUN
sc-5860	103	18	of	of	ADP
sc-5860	103	19	location	location	NOUN
sc-5860	103	20	,	,	PUNCT
sc-5860	103	21	the	the	DET
sc-5860	103	22	overall	overall	ADJ
sc-5860	103	23	layout	layout	NOUN
sc-5860	103	24	,	,	PUNCT
sc-5860	103	25	and	and	CCONJ
sc-5860	103	26	the	the	DET
sc-5860	103	27	relationships	relationship	NOUN
sc-5860	103	28	between	between	ADP
sc-5860	103	29	various	various	ADJ
sc-5860	103	30	elements	element	NOUN
sc-5860	103	31	within	within	ADP
sc-5860	103	32	the	the	DET
sc-5860	103	33	scene	scene	NOUN
sc-5860	103	34	.	.	PUNCT
sc-5860	104	1	a	a	DET
sc-5860	104	2	scene	scene	NOUN
sc-5860	104	3	or	or	CCONJ
sc-5860	104	4	3	3	NUM
sc-5860	104	5	eai	eai	NOUN
sc-5860	104	6	endorsed	endorse	VERB
sc-5860	104	7	transactions	transaction	NOUN
sc-5860	104	8	on	on	ADP
sc-5860	104	9	smart	smart	ADJ
sc-5860	104	10	cities	city	NOUN
sc-5860	104	11	|	|	ADV
sc-5860	104	12	volume	volume	VERB
sc-5860	104	13	7	7	NUM
sc-5860	104	14	|	|	NOUN
sc-5860	104	15	issue	issue	VERB
sc-5860	104	16	4	4	NUM
sc-5860	104	17	|	|	CCONJ
sc-5860	104	18	2023	2023	NUM
sc-5860	104	19	|	|	ADV
sc-5860	104	20	s.	s.	PROPN
sc-5860	104	21	dursun	dursun	PROPN
sc-5860	104	22	and	and	CCONJ
sc-5860	104	23	s.	s.	PROPN
sc-5860	104	24	eken	eken	VERB
sc-5860	104	25	place	place	NOUN
sc-5860	104	26	can	can	AUX
sc-5860	104	27	also	also	ADV
sc-5860	104	28	evoke	evoke	VERB
sc-5860	104	29	different	different	ADJ
sc-5860	104	30	sentiments	sentiment	NOUN
sc-5860	104	31	in	in	ADP
sc-5860	104	32	people	people	NOUN
sc-5860	104	33	.	.	PUNCT
sc-5860	105	1	as	as	SCONJ
sc-5860	105	2	can	can	AUX
sc-5860	105	3	be	be	AUX
sc-5860	105	4	seen	see	VERB
sc-5860	105	5	from	from	ADP
sc-5860	105	6	the	the	DET
sc-5860	105	7	dataset	dataset	NOUN
sc-5860	105	8	prepared	prepare	VERB
sc-5860	105	9	within	within	ADP
sc-5860	105	10	the	the	DET
sc-5860	105	11	scope	scope	NOUN
sc-5860	105	12	of	of	ADP
sc-5860	105	13	the	the	DET
sc-5860	105	14	study	study	NOUN
sc-5860	105	15	,	,	PUNCT
sc-5860	105	16	while	while	SCONJ
sc-5860	105	17	people	people	NOUN
sc-5860	105	18	stranded	strand	VERB
sc-5860	105	19	in	in	ADP
sc-5860	105	20	a	a	DET
sc-5860	105	21	flood	flood	NOUN
sc-5860	105	22	disaster	disaster	NOUN
sc-5860	105	23	can	can	AUX
sc-5860	105	24	evoke	evoke	VERB
sc-5860	105	25	a	a	DET
sc-5860	105	26	negative	negative	ADJ
sc-5860	105	27	emotion	emotion	NOUN
sc-5860	105	28	,	,	PUNCT
sc-5860	105	29	the	the	DET
sc-5860	105	30	aid	aid	NOUN
sc-5860	105	31	teams	team	NOUN
sc-5860	105	32	that	that	PRON
sc-5860	105	33	come	come	VERB
sc-5860	105	34	to	to	PART
sc-5860	105	35	save	save	VERB
sc-5860	105	36	these	these	DET
sc-5860	105	37	people	people	NOUN
sc-5860	105	38	with	with	ADP
sc-5860	105	39	boats	boat	NOUN
sc-5860	105	40	can	can	AUX
sc-5860	105	41	evoke	evoke	VERB
sc-5860	105	42	a	a	DET
sc-5860	105	43	positive	positive	ADJ
sc-5860	105	44	emotion	emotion	NOUN
sc-5860	105	45	.	.	PUNCT
sc-5860	106	1	to	to	PART
sc-5860	106	2	encode	encode	VERB
sc-5860	106	3	the	the	DET
sc-5860	106	4	scene	scene	NOUN
sc-5860	106	5	information	information	NOUN
sc-5860	106	6	of	of	ADP
sc-5860	106	7	an	an	DET
sc-5860	106	8	image	image	NOUN
sc-5860	106	9	,	,	PUNCT
sc-5860	106	10	resnet-101	resnet-101	NOUN
sc-5860	106	11	architecture	architecture	NOUN
sc-5860	106	12	pre	pre	VERB
sc-5860	106	13	-	-	VERB
sc-5860	106	14	trained	train	VERB
sc-5860	106	15	on	on	ADP
sc-5860	106	16	places365	places365	PROPN
sc-5860	106	17	[	[	X
sc-5860	106	18	29	29	NUM
sc-5860	106	19	]	]	PUNCT
sc-5860	106	20	is	be	AUX
sc-5860	106	21	used	use	VERB
sc-5860	106	22	.	.	PUNCT
sc-5860	107	1	facial	facial	ADJ
sc-5860	107	2	expressions	expression	NOUN
sc-5860	107	3	.	.	PUNCT
sc-5860	108	1	facial	facial	ADJ
sc-5860	108	2	expressions	expression	NOUN
sc-5860	108	3	are	be	AUX
sc-5860	108	4	a	a	DET
sc-5860	108	5	subset	subset	NOUN
sc-5860	108	6	of	of	ADP
sc-5860	108	7	visual	visual	ADJ
sc-5860	108	8	features	feature	NOUN
sc-5860	108	9	that	that	PRON
sc-5860	108	10	convey	convey	VERB
sc-5860	108	11	emotions	emotion	NOUN
sc-5860	108	12	,	,	PUNCT
sc-5860	108	13	intentions	intention	NOUN
sc-5860	108	14	,	,	PUNCT
sc-5860	108	15	and	and	CCONJ
sc-5860	108	16	reactions	reaction	NOUN
sc-5860	108	17	.	.	PUNCT
sc-5860	109	1	they	they	PRON
sc-5860	109	2	are	be	AUX
sc-5860	109	3	crucial	crucial	ADJ
sc-5860	109	4	for	for	ADP
sc-5860	109	5	social	social	ADJ
sc-5860	109	6	interactions	interaction	NOUN
sc-5860	109	7	and	and	CCONJ
sc-5860	109	8	communication	communication	NOUN
sc-5860	109	9	.	.	PUNCT
sc-5860	110	1	the	the	DET
sc-5860	110	2	presence	presence	NOUN
sc-5860	110	3	of	of	ADP
sc-5860	110	4	faces	face	NOUN
sc-5860	110	5	and	and	CCONJ
sc-5860	110	6	facial	facial	ADJ
sc-5860	110	7	expressions	expression	NOUN
sc-5860	110	8	(	(	PUNCT
sc-5860	110	9	such	such	ADJ
sc-5860	110	10	as	as	ADP
sc-5860	110	11	smiling	smile	VERB
sc-5860	110	12	and	and	CCONJ
sc-5860	110	13	sad	sad	ADJ
sc-5860	110	14	)	)	PUNCT
sc-5860	110	15	in	in	ADP
sc-5860	110	16	an	an	DET
sc-5860	110	17	image	image	NOUN
sc-5860	110	18	can	can	AUX
sc-5860	110	19	also	also	ADV
sc-5860	110	20	affect	affect	VERB
sc-5860	110	21	an	an	DET
sc-5860	110	22	observer	observer	NOUN
sc-5860	110	23	’s	’s	PART
sc-5860	110	24	emotions	emotion	NOUN
sc-5860	110	25	.	.	PUNCT
sc-5860	111	1	in	in	ADP
sc-5860	111	2	the	the	DET
sc-5860	111	3	images	image	NOUN
sc-5860	111	4	in	in	ADP
sc-5860	111	5	the	the	DET
sc-5860	111	6	prepared	prepare	VERB
sc-5860	111	7	data	datum	NOUN
sc-5860	111	8	set	set	VERB
sc-5860	111	9	,	,	PUNCT
sc-5860	111	10	it	it	PRON
sc-5860	111	11	is	be	AUX
sc-5860	111	12	observed	observe	VERB
sc-5860	111	13	that	that	SCONJ
sc-5860	111	14	facial	facial	ADJ
sc-5860	111	15	expressions	expression	NOUN
sc-5860	111	16	have	have	VERB
sc-5860	111	17	a	a	DET
sc-5860	111	18	direct	direct	ADJ
sc-5860	111	19	effect	effect	NOUN
sc-5860	111	20	on	on	ADP
sc-5860	111	21	the	the	DET
sc-5860	111	22	intensity	intensity	NOUN
sc-5860	111	23	of	of	ADP
sc-5860	111	24	emotion	emotion	NOUN
sc-5860	111	25	in	in	ADP
sc-5860	111	26	data	datum	NOUN
sc-5860	111	27	containing	contain	VERB
sc-5860	111	28	facial	facial	ADJ
sc-5860	111	29	images	image	NOUN
sc-5860	111	30	.	.	PUNCT
sc-5860	112	1	pre	pre	VERB
sc-5860	112	2	-	-	ADJ
sc-5860	112	3	trained	train	VERB
sc-5860	112	4	large	large	ADJ
sc-5860	112	5	ensemble	ensemble	ADJ
sc-5860	112	6	-	-	PUNCT
sc-5860	112	7	based	base	VERB
sc-5860	112	8	convolutional	convolutional	ADJ
sc-5860	112	9	neural	neural	ADJ
sc-5860	112	10	networks	network	NOUN
sc-5860	112	11	[	[	X
sc-5860	112	12	30	30	NUM
sc-5860	112	13	]	]	PUNCT
sc-5860	112	14	are	be	AUX
sc-5860	112	15	used	use	VERB
sc-5860	112	16	to	to	PART
sc-5860	112	17	encode	encode	VERB
sc-5860	112	18	facial	facial	ADJ
sc-5860	112	19	expression	expression	NOUN
sc-5860	112	20	information	information	NOUN
sc-5860	112	21	.	.	PUNCT
sc-5860	113	1	within	within	ADP
sc-5860	113	2	the	the	DET
sc-5860	113	3	scope	scope	NOUN
sc-5860	113	4	of	of	ADP
sc-5860	113	5	these	these	DET
sc-5860	113	6	networks	network	NOUN
sc-5860	113	7	,	,	PUNCT
sc-5860	113	8	ensembles	ensemble	NOUN
sc-5860	113	9	with	with	ADP
sc-5860	113	10	shared	share	VERB
sc-5860	113	11	representations	representation	NOUN
sc-5860	113	12	based	base	VERB
sc-5860	113	13	on	on	ADP
sc-5860	113	14	convolutional	convolutional	ADJ
sc-5860	113	15	networks	network	NOUN
sc-5860	113	16	[	[	X
sc-5860	113	17	31	31	NUM
sc-5860	113	18	]	]	PUNCT
sc-5860	113	19	have	have	AUX
sc-5860	113	20	been	be	AUX
sc-5860	113	21	studied	study	VERB
sc-5860	113	22	to	to	ADP
sc-5860	113	23	quantitatively	quantitatively	ADV
sc-5860	113	24	and	and	CCONJ
sc-5860	113	25	qualitatively	qualitatively	ADV
sc-5860	113	26	demonstrate	demonstrate	VERB
sc-5860	113	27	their	their	PRON
sc-5860	113	28	data	datum	NOUN
sc-5860	113	29	processing	processing	NOUN
sc-5860	113	30	efficiency	efficiency	NOUN
sc-5860	113	31	and	and	CCONJ
sc-5860	113	32	scalability	scalability	NOUN
sc-5860	113	33	to	to	ADP
sc-5860	113	34	large	large	ADJ
sc-5860	113	35	-	-	PUNCT
sc-5860	113	36	scale	scale	NOUN
sc-5860	113	37	facial	facial	ADJ
sc-5860	113	38	expression	expression	NOUN
sc-5860	113	39	datasets	dataset	NOUN
sc-5860	113	40	.	.	PUNCT
sc-5860	114	1	in	in	ADP
sc-5860	114	2	this	this	DET
sc-5860	114	3	study	study	NOUN
sc-5860	114	4	,	,	PUNCT
sc-5860	114	5	facial	facial	ADJ
sc-5860	114	6	expression	expression	NOUN
sc-5860	114	7	analysis	analysis	NOUN
sc-5860	114	8	achieves	achieve	VERB
sc-5860	114	9	human	human	ADJ
sc-5860	114	10	-	-	PUNCT
sc-5860	114	11	level	level	NOUN
sc-5860	114	12	performance	performance	NOUN
sc-5860	114	13	,	,	PUNCT
sc-5860	114	14	outperforming	outperform	VERB
sc-5860	114	15	state	state	NOUN
sc-5860	114	16	-	-	PUNCT
sc-5860	114	17	of	of	ADP
sc-5860	114	18	-	-	PUNCT
sc-5860	114	19	theart	theart	NOUN
sc-5860	114	20	methods	method	NOUN
sc-5860	114	21	in	in	ADP
sc-5860	114	22	facial	facial	ADJ
sc-5860	114	23	expression	expression	NOUN
sc-5860	114	24	recognition	recognition	NOUN
sc-5860	114	25	using	use	VERB
sc-5860	114	26	emotion	emotion	NOUN
sc-5860	114	27	and	and	CCONJ
sc-5860	114	28	affect	affect	VERB
sc-5860	114	29	concepts	concept	NOUN
sc-5860	114	30	.	.	PUNCT
sc-5860	115	1	effective	effective	ADJ
sc-5860	115	2	image	image	NOUN
sc-5860	115	3	content	content	NOUN
sc-5860	115	4	.	.	PUNCT
sc-5860	116	1	effective	effective	ADJ
sc-5860	116	2	image	image	NOUN
sc-5860	116	3	content	content	NOUN
sc-5860	116	4	as	as	SCONJ
sc-5860	116	5	visual	visual	ADJ
sc-5860	116	6	features	feature	NOUN
sc-5860	116	7	refers	refer	VERB
sc-5860	116	8	to	to	ADP
sc-5860	116	9	the	the	DET
sc-5860	116	10	attributes	attribute	NOUN
sc-5860	116	11	and	and	CCONJ
sc-5860	116	12	elements	element	NOUN
sc-5860	116	13	within	within	ADP
sc-5860	116	14	an	an	DET
sc-5860	116	15	image	image	NOUN
sc-5860	116	16	that	that	PRON
sc-5860	116	17	make	make	VERB
sc-5860	116	18	it	it	PRON
sc-5860	116	19	engaging	engage	VERB
sc-5860	116	20	,	,	PUNCT
sc-5860	116	21	informative	informative	ADJ
sc-5860	116	22	,	,	PUNCT
sc-5860	116	23	and	and	CCONJ
sc-5860	116	24	aesthetically	aesthetically	ADV
sc-5860	116	25	pleasing	pleasing	ADJ
sc-5860	116	26	.	.	PUNCT
sc-5860	117	1	these	these	DET
sc-5860	117	2	features	feature	NOUN
sc-5860	117	3	contribute	contribute	VERB
sc-5860	117	4	to	to	ADP
sc-5860	117	5	the	the	DET
sc-5860	117	6	overall	overall	ADJ
sc-5860	117	7	impact	impact	NOUN
sc-5860	117	8	and	and	CCONJ
sc-5860	117	9	communicative	communicative	ADJ
sc-5860	117	10	power	power	NOUN
sc-5860	117	11	of	of	ADP
sc-5860	117	12	an	an	DET
sc-5860	117	13	image	image	NOUN
sc-5860	117	14	.	.	PUNCT
sc-5860	118	1	overall	overall	ADJ
sc-5860	118	2	effective	effective	ADJ
sc-5860	118	3	image	image	NOUN
sc-5860	118	4	content	content	NOUN
sc-5860	118	5	may	may	AUX
sc-5860	118	6	also	also	ADV
sc-5860	118	7	be	be	AUX
sc-5860	118	8	important	important	ADJ
sc-5860	118	9	for	for	ADP
sc-5860	118	10	multimodal	multimodal	NOUN
sc-5860	118	11	emotion	emotion	NOUN
sc-5860	118	12	detection	detection	NOUN
sc-5860	118	13	.	.	PUNCT
sc-5860	119	1	this	this	DET
sc-5860	119	2	field	field	NOUN
sc-5860	119	3	has	have	AUX
sc-5860	119	4	made	make	VERB
sc-5860	119	5	rapid	rapid	ADJ
sc-5860	119	6	progress	progress	NOUN
sc-5860	119	7	in	in	ADP
sc-5860	119	8	recent	recent	ADJ
sc-5860	119	9	years	year	NOUN
sc-5860	119	10	,	,	PUNCT
sc-5860	119	11	with	with	ADP
sc-5860	119	12	datasets	dataset	NOUN
sc-5860	119	13	taken	take	VERB
sc-5860	119	14	from	from	ADP
sc-5860	119	15	popular	popular	ADJ
sc-5860	119	16	social	social	ADJ
sc-5860	119	17	media	medium	NOUN
sc-5860	119	18	image	image	NOUN
sc-5860	119	19	sharing	sharing	NOUN
sc-5860	119	20	platforms	platform	NOUN
sc-5860	119	21	such	such	ADJ
sc-5860	119	22	as	as	ADP
sc-5860	119	23	flickr	flickr	PROPN
sc-5860	119	24	and	and	CCONJ
sc-5860	119	25	instagram	instagram	PROPN
sc-5860	119	26	.	.	PUNCT
sc-5860	120	1	to	to	PART
sc-5860	120	2	encode	encode	VERB
sc-5860	120	3	the	the	DET
sc-5860	120	4	overall	overall	ADJ
sc-5860	120	5	emotion	emotion	NOUN
sc-5860	120	6	,	,	PUNCT
sc-5860	120	7	a	a	DET
sc-5860	120	8	resnet-50	resnet-50	PROPN
sc-5860	120	9	imagenet	imagenet	ADJ
sc-5860	120	10	model	model	NOUN
sc-5860	120	11	is	be	AUX
sc-5860	120	12	first	first	ADJ
sc-5860	120	13	fine	fine	ADV
sc-5860	120	14	-	-	PUNCT
sc-5860	120	15	tuned	tune	VERB
sc-5860	120	16	on	on	ADP
sc-5860	120	17	the	the	DET
sc-5860	120	18	publicly	publicly	ADV
sc-5860	120	19	available	available	ADJ
sc-5860	120	20	fi	fi	NOUN
sc-5860	120	21	(	(	PUNCT
sc-5860	120	22	flickr	flickr	PROPN
sc-5860	120	23	&	&	CCONJ
sc-5860	120	24	instagram	instagram	PROPN
sc-5860	120	25	)	)	PUNCT
sc-5860	120	26	dataset	dataset	NOUN
sc-5860	120	27	,	,	PUNCT
sc-5860	120	28	and	and	CCONJ
sc-5860	120	29	last	last	ADJ
sc-5860	120	30	layer	layer	NOUN
sc-5860	120	31	convolution	convolution	NOUN
sc-5860	120	32	features	feature	NOUN
sc-5860	120	33	are	be	AUX
sc-5860	120	34	extracted	extract	VERB
sc-5860	120	35	for	for	ADP
sc-5860	120	36	object	object	NOUN
sc-5860	120	37	and	and	CCONJ
sc-5860	120	38	scene	scene	NOUN
sc-5860	120	39	embeddings	embedding	NOUN
sc-5860	120	40	.	.	PUNCT
sc-5860	121	1	the	the	DET
sc-5860	121	2	dataset	dataset	NOUN
sc-5860	121	3	used	use	VERB
sc-5860	121	4	consists	consist	NOUN
sc-5860	121	5	of	of	ADP
sc-5860	121	6	approximately	approximately	ADV
sc-5860	121	7	23,000	23,000	NUM
sc-5860	121	8	training	training	NOUN
sc-5860	121	9	images	image	NOUN
sc-5860	121	10	and	and	CCONJ
sc-5860	121	11	eight	eight	NUM
sc-5860	121	12	emotion	emotion	NOUN
sc-5860	121	13	classes	class	NOUN
sc-5860	121	14	:	:	PUNCT
sc-5860	121	15	amusement	amusement	NOUN
sc-5860	121	16	,	,	PUNCT
sc-5860	121	17	anger	anger	NOUN
sc-5860	121	18	,	,	PUNCT
sc-5860	121	19	awe	awe	PROPN
sc-5860	121	20	,	,	PUNCT
sc-5860	121	21	satisfaction	satisfaction	NOUN
sc-5860	121	22	,	,	PUNCT
sc-5860	121	23	disgust	disgust	NOUN
sc-5860	121	24	,	,	PUNCT
sc-5860	121	25	excitement	excitement	NOUN
sc-5860	121	26	,	,	PUNCT
sc-5860	121	27	fear	fear	NOUN
sc-5860	121	28	,	,	PUNCT
sc-5860	121	29	and	and	CCONJ
sc-5860	121	30	sadness	sadness	NOUN
sc-5860	121	31	.	.	PUNCT
sc-5860	122	1	3.2	3.2	NUM
sc-5860	122	2	.	.	PUNCT
sc-5860	123	1	textual	textual	ADJ
sc-5860	123	2	features	feature	NOUN
sc-5860	123	3	textual	textual	ADJ
sc-5860	123	4	features	feature	NOUN
sc-5860	123	5	refer	refer	VERB
sc-5860	123	6	to	to	ADP
sc-5860	123	7	the	the	DET
sc-5860	123	8	characteristics	characteristic	NOUN
sc-5860	123	9	and	and	CCONJ
sc-5860	123	10	attributes	attribute	NOUN
sc-5860	123	11	of	of	ADP
sc-5860	123	12	text	text	NOUN
sc-5860	123	13	that	that	PRON
sc-5860	123	14	can	can	AUX
sc-5860	123	15	be	be	AUX
sc-5860	123	16	analyzed	analyze	VERB
sc-5860	123	17	and	and	CCONJ
sc-5860	123	18	used	use	VERB
sc-5860	123	19	for	for	ADP
sc-5860	123	20	various	various	ADJ
sc-5860	123	21	purposes	purpose	NOUN
sc-5860	123	22	such	such	ADJ
sc-5860	123	23	as	as	ADP
sc-5860	123	24	natural	natural	ADJ
sc-5860	123	25	language	language	NOUN
sc-5860	123	26	processing	processing	NOUN
sc-5860	123	27	(	(	PUNCT
sc-5860	123	28	nlp	nlp	NOUN
sc-5860	123	29	)	)	PUNCT
sc-5860	123	30	,	,	PUNCT
sc-5860	123	31	text	text	NOUN
sc-5860	123	32	mining	mining	NOUN
sc-5860	123	33	,	,	PUNCT
sc-5860	123	34	information	information	NOUN
sc-5860	123	35	retrieval	retrieval	NOUN
sc-5860	123	36	,	,	PUNCT
sc-5860	123	37	and	and	CCONJ
sc-5860	123	38	more	more	ADJ
sc-5860	123	39	.	.	PUNCT
sc-5860	124	1	these	these	PRON
sc-5860	124	2	features	feature	VERB
sc-5860	124	3	help	help	NOUN
sc-5860	124	4	in	in	ADP
sc-5860	124	5	understanding	understanding	NOUN
sc-5860	124	6	,	,	PUNCT
sc-5860	124	7	processing	processing	NOUN
sc-5860	124	8	,	,	PUNCT
sc-5860	124	9	and	and	CCONJ
sc-5860	124	10	deriving	derive	VERB
sc-5860	124	11	insights	insight	NOUN
sc-5860	124	12	from	from	ADP
sc-5860	124	13	text	text	NOUN
sc-5860	124	14	data	datum	NOUN
sc-5860	124	15	.	.	PUNCT
sc-5860	125	1	since	since	SCONJ
sc-5860	125	2	the	the	DET
sc-5860	125	3	context	context	NOUN
sc-5860	125	4	and	and	CCONJ
sc-5860	125	5	meaning	meaning	NOUN
sc-5860	125	6	of	of	ADP
sc-5860	125	7	words	word	NOUN
sc-5860	125	8	are	be	AUX
sc-5860	125	9	equally	equally	ADV
sc-5860	125	10	important	important	ADJ
sc-5860	125	11	for	for	ADP
sc-5860	125	12	the	the	DET
sc-5860	125	13	emotional	emotional	ADJ
sc-5860	125	14	impact	impact	NOUN
sc-5860	125	15	of	of	ADP
sc-5860	125	16	the	the	DET
sc-5860	125	17	entire	entire	ADJ
sc-5860	125	18	sentence	sentence	NOUN
sc-5860	125	19	,	,	PUNCT
sc-5860	125	20	roberta	roberta	NOUN
sc-5860	125	21	-	-	PUNCT
sc-5860	125	22	base	base	NOUN
sc-5860	125	23	[	[	X
sc-5860	125	24	32	32	NUM
sc-5860	125	25	]	]	PUNCT
sc-5860	125	26	is	be	AUX
sc-5860	125	27	used	use	VERB
sc-5860	125	28	to	to	PART
sc-5860	125	29	extract	extract	VERB
sc-5860	125	30	contextual	contextual	ADJ
sc-5860	125	31	word	word	NOUN
sc-5860	125	32	embeddings	embedding	NOUN
sc-5860	125	33	.	.	PUNCT
sc-5860	126	1	data	datum	NOUN
sc-5860	126	2	preprocessing	preprocessing	NOUN
sc-5860	126	3	.	.	PUNCT
sc-5860	127	1	emojis	emojis	PROPN
sc-5860	127	2	,	,	PUNCT
sc-5860	127	3	unnecessary	unnecessary	ADJ
sc-5860	127	4	non	non	ADJ
sc-5860	127	5	-	-	ADJ
sc-5860	127	6	ascii	ascii	ADJ
sc-5860	127	7	characters	character	NOUN
sc-5860	127	8	,	,	PUNCT
sc-5860	127	9	numbers	number	NOUN
sc-5860	127	10	,	,	PUNCT
sc-5860	127	11	url	url	NOUN
sc-5860	127	12	,	,	PUNCT
sc-5860	127	13	hashtag	hashtag	NOUN
sc-5860	127	14	sign	sign	NOUN
sc-5860	127	15	and	and	CCONJ
sc-5860	127	16	other	other	ADJ
sc-5860	127	17	characters	character	NOUN
sc-5860	127	18	in	in	ADP
sc-5860	127	19	the	the	DET
sc-5860	127	20	tweet	tweet	NOUN
sc-5860	127	21	text	text	NOUN
sc-5860	127	22	column	column	NOUN
sc-5860	127	23	of	of	ADP
sc-5860	127	24	the	the	DET
sc-5860	127	25	labeled	label	VERB
sc-5860	127	26	dataset	dataset	NOUN
sc-5860	127	27	have	have	AUX
sc-5860	127	28	been	be	AUX
sc-5860	127	29	removed	remove	VERB
sc-5860	127	30	from	from	ADP
sc-5860	127	31	the	the	DET
sc-5860	127	32	text	text	NOUN
sc-5860	127	33	content	content	NOUN
sc-5860	127	34	due	due	ADP
sc-5860	127	35	to	to	ADP
sc-5860	127	36	the	the	DET
sc-5860	127	37	inference	inference	NOUN
sc-5860	127	38	that	that	PRON
sc-5860	127	39	they	they	PRON
sc-5860	127	40	do	do	AUX
sc-5860	127	41	not	not	PART
sc-5860	127	42	help	help	VERB
sc-5860	127	43	the	the	DET
sc-5860	127	44	model	model	NOUN
sc-5860	127	45	perform	perform	VERB
sc-5860	127	46	better	well	ADV
sc-5860	127	47	.	.	PUNCT
sc-5860	128	1	text	text	NOUN
sc-5860	128	2	data	datum	NOUN
sc-5860	128	3	labeling	labeling	NOUN
sc-5860	128	4	.	.	PUNCT
sc-5860	129	1	textblob	textblob	NOUN
sc-5860	129	2	library	library	NOUN
sc-5860	129	3	[	[	X
sc-5860	129	4	33	33	NUM
sc-5860	129	5	]	]	PUNCT
sc-5860	129	6	is	be	AUX
sc-5860	129	7	used	use	VERB
sc-5860	129	8	to	to	PART
sc-5860	129	9	label	label	VERB
sc-5860	129	10	the	the	DET
sc-5860	129	11	textual	textual	ADJ
sc-5860	129	12	data	datum	NOUN
sc-5860	129	13	in	in	ADP
sc-5860	129	14	the	the	DET
sc-5860	129	15	dataset	dataset	NOUN
sc-5860	129	16	.	.	PUNCT
sc-5860	130	1	textblob	textblob	NOUN
sc-5860	130	2	is	be	AUX
sc-5860	130	3	a	a	DET
sc-5860	130	4	python	python	NOUN
sc-5860	130	5	library	library	NOUN
sc-5860	130	6	for	for	ADP
sc-5860	130	7	textual	textual	ADJ
sc-5860	130	8	information	information	NOUN
sc-5860	130	9	that	that	PRON
sc-5860	130	10	provides	provide	VERB
sc-5860	130	11	a	a	DET
sc-5860	130	12	simple	simple	ADJ
sc-5860	130	13	api	api	NOUN
sc-5860	130	14	to	to	PART
sc-5860	130	15	access	access	VERB
sc-5860	130	16	nlp	nlp	NOUN
sc-5860	130	17	activities	activity	NOUN
sc-5860	130	18	.	.	PUNCT
sc-5860	131	1	using	use	VERB
sc-5860	131	2	the	the	DET
sc-5860	131	3	tokenization	tokenization	NOUN
sc-5860	131	4	,	,	PUNCT
sc-5860	131	5	lemmatization	lemmatization	NOUN
sc-5860	131	6	,	,	PUNCT
sc-5860	131	7	speech	speech	NOUN
sc-5860	131	8	tagging	tagging	NOUN
sc-5860	131	9	and	and	CCONJ
sc-5860	131	10	noun	noun	NOUN
sc-5860	131	11	expression	expression	NOUN
sc-5860	131	12	extraction	extraction	NOUN
sc-5860	131	13	features	feature	NOUN
sc-5860	131	14	of	of	ADP
sc-5860	131	15	the	the	DET
sc-5860	131	16	textblob	textblob	NOUN
sc-5860	131	17	library	library	NOUN
sc-5860	131	18	,	,	PUNCT
sc-5860	131	19	each	each	DET
sc-5860	131	20	textual	textual	ADJ
sc-5860	131	21	data	datum	NOUN
sc-5860	131	22	line	line	NOUN
sc-5860	131	23	was	be	AUX
sc-5860	131	24	labeled	label	VERB
sc-5860	131	25	as	as	ADP
sc-5860	131	26	0	0	NUM
sc-5860	131	27	(	(	PUNCT
sc-5860	131	28	neutral	neutral	ADJ
sc-5860	131	29	)	)	PUNCT
sc-5860	131	30	,	,	PUNCT
sc-5860	131	31	1	1	NUM
sc-5860	131	32	(	(	PUNCT
sc-5860	131	33	positive	positive	ADJ
sc-5860	131	34	)	)	PUNCT
sc-5860	131	35	,	,	PUNCT
sc-5860	131	36	2	2	NUM
sc-5860	131	37	(	(	PUNCT
sc-5860	131	38	negative	negative	ADJ
sc-5860	131	39	)	)	PUNCT
sc-5860	131	40	.	.	PUNCT
sc-5860	132	1	3.3	3.3	NUM
sc-5860	132	2	.	.	PUNCT
sc-5860	133	1	multimodal	multimodal	NOUN
sc-5860	133	2	features	feature	VERB
sc-5860	133	3	multimodal	multimodal	NOUN
sc-5860	133	4	features	feature	NOUN
sc-5860	133	5	refer	refer	VERB
sc-5860	133	6	to	to	ADP
sc-5860	133	7	the	the	DET
sc-5860	133	8	combination	combination	NOUN
sc-5860	133	9	of	of	ADP
sc-5860	133	10	features	feature	NOUN
sc-5860	133	11	from	from	ADP
sc-5860	133	12	multiple	multiple	ADJ
sc-5860	133	13	types	type	NOUN
sc-5860	133	14	of	of	ADP
sc-5860	133	15	data	datum	NOUN
sc-5860	133	16	,	,	PUNCT
sc-5860	133	17	such	such	ADJ
sc-5860	133	18	as	as	ADP
sc-5860	133	19	text	text	NOUN
sc-5860	133	20	,	,	PUNCT
sc-5860	133	21	images	image	NOUN
sc-5860	133	22	,	,	PUNCT
sc-5860	133	23	audio	audio	NOUN
sc-5860	133	24	,	,	PUNCT
sc-5860	133	25	and	and	CCONJ
sc-5860	133	26	video	video	NOUN
sc-5860	133	27	,	,	PUNCT
sc-5860	133	28	to	to	PART
sc-5860	133	29	create	create	VERB
sc-5860	133	30	a	a	DET
sc-5860	133	31	more	more	ADV
sc-5860	133	32	comprehensive	comprehensive	ADJ
sc-5860	133	33	understanding	understanding	NOUN
sc-5860	133	34	of	of	ADP
sc-5860	133	35	content	content	NOUN
sc-5860	133	36	.	.	PUNCT
sc-5860	134	1	these	these	DET
sc-5860	134	2	features	feature	NOUN
sc-5860	134	3	are	be	AUX
sc-5860	134	4	essential	essential	ADJ
sc-5860	134	5	in	in	ADP
sc-5860	134	6	tasks	task	NOUN
sc-5860	134	7	that	that	PRON
sc-5860	134	8	require	require	VERB
sc-5860	134	9	the	the	DET
sc-5860	134	10	integration	integration	NOUN
sc-5860	134	11	of	of	ADP
sc-5860	134	12	information	information	NOUN
sc-5860	134	13	from	from	ADP
sc-5860	134	14	different	different	ADJ
sc-5860	134	15	modalities	modality	NOUN
sc-5860	134	16	to	to	PART
sc-5860	134	17	improve	improve	VERB
sc-5860	134	18	the	the	DET
sc-5860	134	19	accuracy	accuracy	NOUN
sc-5860	134	20	and	and	CCONJ
sc-5860	134	21	richness	richness	NOUN
sc-5860	134	22	of	of	ADP
sc-5860	134	23	the	the	DET
sc-5860	134	24	analysis	analysis	NOUN
sc-5860	134	25	.	.	PUNCT
sc-5860	135	1	for	for	ADP
sc-5860	135	2	the	the	DET
sc-5860	135	3	extraction	extraction	NOUN
sc-5860	135	4	of	of	ADP
sc-5860	135	5	multimodal	multimodal	NOUN
sc-5860	135	6	features	feature	NOUN
sc-5860	135	7	,	,	PUNCT
sc-5860	135	8	the	the	DET
sc-5860	135	9	multimodal	multimodal	ADJ
sc-5860	135	10	clip	clip	NOUN
sc-5860	135	11	[	[	X
sc-5860	135	12	12	12	NUM
sc-5860	135	13	]	]	PUNCT
sc-5860	135	14	model	model	NOUN
sc-5860	135	15	,	,	PUNCT
sc-5860	135	16	trained	train	VERB
sc-5860	135	17	on	on	ADP
sc-5860	135	18	400	400	NUM
sc-5860	135	19	million	million	NUM
sc-5860	135	20	image	image	NOUN
sc-5860	135	21	-	-	PUNCT
sc-5860	135	22	text	text	NOUN
sc-5860	135	23	pairs	pair	NOUN
sc-5860	135	24	collected	collect	VERB
sc-5860	135	25	from	from	ADP
sc-5860	135	26	the	the	DET
sc-5860	135	27	internet	internet	NOUN
sc-5860	135	28	,	,	PUNCT
sc-5860	135	29	is	be	AUX
sc-5860	135	30	used	use	VERB
sc-5860	135	31	.	.	PUNCT
sc-5860	136	1	the	the	DET
sc-5860	136	2	model	model	NOUN
sc-5860	136	3	is	be	AUX
sc-5860	136	4	trained	train	VERB
sc-5860	136	5	to	to	PART
sc-5860	136	6	predict	predict	VERB
sc-5860	136	7	which	which	DET
sc-5860	136	8	text	text	NOUN
sc-5860	136	9	goes	go	VERB
sc-5860	136	10	with	with	ADP
sc-5860	136	11	which	which	DET
sc-5860	136	12	image	image	NOUN
sc-5860	136	13	,	,	PUNCT
sc-5860	136	14	and	and	CCONJ
sc-5860	136	15	in	in	ADP
sc-5860	136	16	doing	do	VERB
sc-5860	136	17	so	so	ADV
sc-5860	136	18	,	,	PUNCT
sc-5860	136	19	it	it	PRON
sc-5860	136	20	learns	learn	VERB
sc-5860	136	21	meaningful	meaningful	ADJ
sc-5860	136	22	image	image	NOUN
sc-5860	136	23	representation	representation	NOUN
sc-5860	136	24	without	without	ADP
sc-5860	136	25	the	the	DET
sc-5860	136	26	need	need	NOUN
sc-5860	136	27	for	for	ADP
sc-5860	136	28	millions	million	NOUN
sc-5860	136	29	of	of	ADP
sc-5860	136	30	labeled	label	VERB
sc-5860	136	31	training	training	NOUN
sc-5860	136	32	examples	example	NOUN
sc-5860	136	33	.	.	PUNCT
sc-5860	137	1	compared	compare	VERB
sc-5860	137	2	to	to	ADP
sc-5860	137	3	multimodal	multimodal	NOUN
sc-5860	137	4	transformers	transformer	NOUN
sc-5860	137	5	,	,	PUNCT
sc-5860	137	6	the	the	DET
sc-5860	137	7	model	model	NOUN
sc-5860	137	8	uses	use	VERB
sc-5860	137	9	binary	binary	NOUN
sc-5860	137	10	learning	learning	NOUN
sc-5860	137	11	over	over	ADP
sc-5860	137	12	n	n	CCONJ
sc-5860	137	13	-	-	PUNCT
sc-5860	137	14	pair	pair	NOUN
sc-5860	137	15	images	image	NOUN
sc-5860	137	16	and	and	CCONJ
sc-5860	137	17	text	text	NOUN
sc-5860	137	18	,	,	PUNCT
sc-5860	137	19	and	and	CCONJ
sc-5860	137	20	no	no	DET
sc-5860	137	21	cross	cross	ADJ
sc-5860	137	22	-	-	ADJ
sc-5860	137	23	attention	attention	ADJ
sc-5860	137	24	mechanisms	mechanism	NOUN
sc-5860	137	25	are	be	AUX
sc-5860	137	26	used	use	VERB
sc-5860	137	27	to	to	PART
sc-5860	137	28	learn	learn	VERB
sc-5860	137	29	multimodal	multimodal	NOUN
sc-5860	137	30	features	feature	NOUN
sc-5860	137	31	.	.	PUNCT
sc-5860	138	1	this	this	PRON
sc-5860	138	2	makes	make	VERB
sc-5860	138	3	the	the	DET
sc-5860	138	4	model	model	NOUN
sc-5860	138	5	easier	easy	ADJ
sc-5860	138	6	to	to	PART
sc-5860	138	7	use	use	VERB
sc-5860	138	8	,	,	PUNCT
sc-5860	138	9	as	as	SCONJ
sc-5860	138	10	image	image	NOUN
sc-5860	138	11	and	and	CCONJ
sc-5860	138	12	text	text	NOUN
sc-5860	138	13	embeddings	embedding	NOUN
sc-5860	138	14	can	can	AUX
sc-5860	138	15	be	be	AUX
sc-5860	138	16	calculated	calculate	VERB
sc-5860	138	17	independently	independently	ADV
sc-5860	138	18	of	of	ADP
sc-5860	138	19	the	the	DET
sc-5860	138	20	respective	respective	ADJ
sc-5860	138	21	image	image	NOUN
sc-5860	138	22	and	and	CCONJ
sc-5860	138	23	text	text	NOUN
sc-5860	138	24	encoders	encoder	NOUN
sc-5860	138	25	.	.	PUNCT
sc-5860	139	1	due	due	ADP
sc-5860	139	2	to	to	ADP
sc-5860	139	3	the	the	DET
sc-5860	139	4	diversity	diversity	NOUN
sc-5860	139	5	and	and	CCONJ
sc-5860	139	6	large	large	ADJ
sc-5860	139	7	amount	amount	NOUN
sc-5860	139	8	of	of	ADP
sc-5860	139	9	data	datum	NOUN
sc-5860	139	10	,	,	PUNCT
sc-5860	139	11	the	the	DET
sc-5860	139	12	model	model	NOUN
sc-5860	139	13	demonstrates	demonstrate	VERB
sc-5860	139	14	competitive	competitive	ADJ
sc-5860	139	15	zero	zero	NUM
sc-5860	139	16	-	-	PUNCT
sc-5860	139	17	shot	shot	NOUN
sc-5860	139	18	recognition	recognition	NOUN
sc-5860	139	19	performance	performance	NOUN
sc-5860	139	20	on	on	ADP
sc-5860	139	21	30	30	NUM
sc-5860	139	22	different	different	ADJ
sc-5860	139	23	computer	computer	NOUN
sc-5860	139	24	vision	vision	NOUN
sc-5860	139	25	datasets	dataset	NOUN
sc-5860	139	26	compared	compare	VERB
sc-5860	139	27	to	to	ADP
sc-5860	139	28	its	its	PRON
sc-5860	139	29	supervised	supervised	ADJ
sc-5860	139	30	baselines	baseline	NOUN
sc-5860	139	31	.	.	PUNCT
sc-5860	140	1	this	this	PRON
sc-5860	140	2	shows	show	VERB
sc-5860	140	3	that	that	SCONJ
sc-5860	140	4	the	the	DET
sc-5860	140	5	amount	amount	NOUN
sc-5860	140	6	and	and	CCONJ
sc-5860	140	7	quality	quality	NOUN
sc-5860	140	8	of	of	ADP
sc-5860	140	9	visual	visual	ADJ
sc-5860	140	10	information	information	NOUN
sc-5860	140	11	encoded	encode	VERB
sc-5860	140	12	in	in	ADP
sc-5860	140	13	the	the	DET
sc-5860	140	14	model	model	NOUN
sc-5860	140	15	’s	’s	PART
sc-5860	140	16	visual	visual	ADJ
sc-5860	140	17	features	feature	NOUN
sc-5860	140	18	are	be	AUX
sc-5860	140	19	much	much	ADV
sc-5860	140	20	better	well	ADJ
sc-5860	140	21	than	than	ADP
sc-5860	140	22	imagenet	imagenet	NOUN
sc-5860	140	23	and	and	CCONJ
sc-5860	140	24	places365	places365	PROPN
sc-5860	140	25	supervised	supervise	VERB
sc-5860	140	26	pre	pre	ADJ
sc-5860	140	27	-	-	ADJ
sc-5860	140	28	trained	train	VERB
sc-5860	140	29	models	model	NOUN
sc-5860	140	30	.	.	PUNCT
sc-5860	141	1	both	both	DET
sc-5860	141	2	image	image	NOUN
sc-5860	141	3	and	and	CCONJ
sc-5860	141	4	text	text	NOUN
sc-5860	141	5	features	feature	NOUN
sc-5860	141	6	are	be	AUX
sc-5860	141	7	extracted	extract	VERB
sc-5860	141	8	using	use	VERB
sc-5860	141	9	an	an	DET
sc-5860	141	10	open	open	ADJ
sc-5860	141	11	-	-	PUNCT
sc-5860	141	12	source	source	NOUN
sc-5860	141	13	clip	clip	NOUN
sc-5860	141	14	model	model	NOUN
sc-5860	141	15	,	,	PUNCT
sc-5860	141	16	where	where	SCONJ
sc-5860	141	17	both	both	DET
sc-5860	141	18	image	image	NOUN
sc-5860	141	19	and	and	CCONJ
sc-5860	141	20	text	text	NOUN
sc-5860	141	21	embeddings	embedding	NOUN
sc-5860	141	22	are	be	AUX
sc-5860	141	23	512	512	NUM
sc-5860	141	24	-	-	PUNCT
sc-5860	141	25	dimensional	dimensional	ADJ
sc-5860	141	26	vectors	vector	NOUN
sc-5860	141	27	.	.	PUNCT
sc-5860	142	1	4	4	X
sc-5860	142	2	.	.	X
sc-5860	142	3	experimental	experimental	ADJ
sc-5860	142	4	results	result	NOUN
sc-5860	142	5	4.1	4.1	NUM
sc-5860	142	6	.	.	PUNCT
sc-5860	143	1	test	test	NOUN
sc-5860	143	2	environment	environment	NOUN
sc-5860	143	3	the	the	DET
sc-5860	143	4	training	training	NOUN
sc-5860	143	5	and	and	CCONJ
sc-5860	143	6	testing	testing	NOUN
sc-5860	143	7	phases	phase	NOUN
sc-5860	143	8	of	of	ADP
sc-5860	143	9	the	the	DET
sc-5860	143	10	studies	study	NOUN
sc-5860	143	11	are	be	AUX
sc-5860	143	12	carried	carry	VERB
sc-5860	143	13	out	out	ADP
sc-5860	143	14	on	on	ADP
sc-5860	143	15	google	google	PROPN
sc-5860	143	16	colab	colab	PROPN
sc-5860	143	17	pro	pro	X
sc-5860	143	18	on	on	ADP
sc-5860	143	19	virtual	virtual	ADJ
sc-5860	143	20	servers	server	NOUN
sc-5860	143	21	with	with	ADP
sc-5860	143	22	the	the	DET
sc-5860	143	23	following	follow	VERB
sc-5860	143	24	features	feature	NOUN
sc-5860	143	25	:	:	PUNCT
sc-5860	143	26	pc	pc	NOUN
sc-5860	143	27	1	1	NUM
sc-5860	143	28	(	(	PUNCT
sc-5860	143	29	gpu	gpu	X
sc-5860	143	30	virtual	virtual	ADJ
sc-5860	143	31	machine	machine	NOUN
sc-5860	143	32	):	):	PUNCT
sc-5860	143	33	4	4	NUM
sc-5860	143	34	eai	eai	NOUN
sc-5860	143	35	endorsed	endorse	VERB
sc-5860	143	36	transactions	transaction	NOUN
sc-5860	143	37	on	on	ADP
sc-5860	143	38	smart	smart	ADJ
sc-5860	143	39	cities	city	NOUN
sc-5860	143	40	|	|	ADV
sc-5860	143	41	volume	volume	VERB
sc-5860	143	42	7	7	NUM
sc-5860	143	43	|	|	NOUN
sc-5860	143	44	issue	issue	VERB
sc-5860	143	45	4	4	NUM
sc-5860	143	46	|	|	CCONJ
sc-5860	143	47	2023	2023	NUM
sc-5860	143	48	|	|	NOUN
sc-5860	143	49	multimodal	multimodal	ADJ
sc-5860	143	50	sentiment	sentiment	NOUN
sc-5860	143	51	analysis	analysis	NOUN
sc-5860	143	52	in	in	ADP
sc-5860	143	53	natural	natural	ADJ
sc-5860	143	54	disaster	disaster	NOUN
sc-5860	143	55	data	datum	NOUN
sc-5860	143	56	on	on	ADP
sc-5860	143	57	social	social	ADJ
sc-5860	143	58	media	medium	NOUN
sc-5860	143	59	operating	operating	NOUN
sc-5860	143	60	system	system	NOUN
sc-5860	143	61	:	:	PUNCT
sc-5860	143	62	ubuntu	ubuntu	ADJ
sc-5860	143	63	18.04.6	18.04.6	NUM
sc-5860	143	64	lts	lts	NOUN
sc-5860	143	65	,	,	PUNCT
sc-5860	143	66	cpu	cpu	PROPN
sc-5860	143	67	:	:	PUNCT
sc-5860	143	68	intel(r	intel(r	PROPN
sc-5860	143	69	)	)	PUNCT
sc-5860	143	70	xeon(r	xeon(r	PROPN
sc-5860	143	71	)	)	PUNCT
sc-5860	143	72	cpu	cpu	NOUN
sc-5860	143	73	@	@	ADP
sc-5860	143	74	2.20ghz	2.20ghz	NUM
sc-5860	143	75	,	,	PUNCT
sc-5860	143	76	gpu	gpu	PROPN
sc-5860	143	77	:	:	PUNCT
sc-5860	143	78	tesla	tesla	PROPN
sc-5860	143	79	v100	v100	PROPN
sc-5860	143	80	,	,	PUNCT
sc-5860	143	81	ram	ram	NOUN
sc-5860	143	82	:	:	PUNCT
sc-5860	143	83	16	16	NUM
sc-5860	143	84	gb	gb	NOUN
sc-5860	143	85	,	,	PUNCT
sc-5860	143	86	disk	disk	NOUN
sc-5860	143	87	:	:	PUNCT
sc-5860	143	88	108	108	NUM
sc-5860	143	89	gb	gb	NOUN
sc-5860	143	90	;	;	PUNCT
sc-5860	143	91	pc	pc	NOUN
sc-5860	143	92	2	2	NUM
sc-5860	143	93	(	(	PUNCT
sc-5860	143	94	tpu	tpu	PROPN
sc-5860	143	95	virtual	virtual	ADJ
sc-5860	143	96	machine	machine	NOUN
sc-5860	143	97	):	):	PUNCT
sc-5860	143	98	operating	operate	VERB
sc-5860	143	99	system	system	NOUN
sc-5860	143	100	:	:	PUNCT
sc-5860	143	101	ubuntu	ubuntu	ADJ
sc-5860	143	102	18.04.6	18.04.6	NUM
sc-5860	143	103	lts	lts	NOUN
sc-5860	143	104	,	,	PUNCT
sc-5860	143	105	cpu	cpu	PROPN
sc-5860	143	106	:	:	PUNCT
sc-5860	144	1	intel(r	intel(r	PROPN
sc-5860	144	2	)	)	PUNCT
sc-5860	144	3	xeon(r	xeon(r	PROPN
sc-5860	144	4	)	)	PUNCT
sc-5860	144	5	cpu	cpu	NOUN
sc-5860	144	6	@	@	ADP
sc-5860	144	7	2.20ghz	2.20ghz	NUM
sc-5860	144	8	,	,	PUNCT
sc-5860	144	9	ram	ram	NOUN
sc-5860	144	10	:	:	PUNCT
sc-5860	144	11	16	16	NUM
sc-5860	144	12	gb	gb	NOUN
sc-5860	144	13	,	,	PUNCT
sc-5860	144	14	disk	disk	NOUN
sc-5860	144	15	:	:	PUNCT
sc-5860	144	16	108	108	NUM
sc-5860	144	17	gb	gb	NOUN
sc-5860	144	18	.	.	PUNCT
sc-5860	145	1	4.2	4.2	NUM
sc-5860	145	2	.	.	PUNCT
sc-5860	146	1	dataset	dataset	NOUN
sc-5860	146	2	and	and	CCONJ
sc-5860	146	3	training	training	NOUN
sc-5860	146	4	parameters	parameter	NOUN
sc-5860	146	5	in	in	ADP
sc-5860	146	6	the	the	DET
sc-5860	146	7	study	study	NOUN
sc-5860	146	8	,	,	PUNCT
sc-5860	146	9	we	we	PRON
sc-5860	146	10	create	create	VERB
sc-5860	146	11	a	a	DET
sc-5860	146	12	new	new	ADJ
sc-5860	146	13	dataset	dataset	NOUN
sc-5860	146	14	containing	contain	VERB
sc-5860	146	15	1,000	1,000	NUM
sc-5860	146	16	images	image	NOUN
sc-5860	146	17	and	and	CCONJ
sc-5860	146	18	1,000	1,000	NUM
sc-5860	146	19	texts	text	NOUN
sc-5860	146	20	by	by	ADP
sc-5860	146	21	filtering	filter	VERB
sc-5860	146	22	the	the	DET
sc-5860	146	23	natural	natural	ADJ
sc-5860	146	24	disaster	disaster	NOUN
sc-5860	146	25	hashtags	hashtag	NOUN
sc-5860	146	26	and	and	CCONJ
sc-5860	146	27	meaningful	meaningful	ADJ
sc-5860	146	28	tweets	tweet	NOUN
sc-5860	146	29	containing	contain	VERB
sc-5860	146	30	images	image	NOUN
sc-5860	146	31	+	+	CCONJ
sc-5860	146	32	text	text	NOUN
sc-5860	146	33	determined	determine	VERB
sc-5860	146	34	on	on	ADP
sc-5860	146	35	twitter	twitter	NOUN
sc-5860	146	36	.	.	PUNCT
sc-5860	147	1	the	the	DET
sc-5860	147	2	determined	determined	ADJ
sc-5860	147	3	natural	natural	ADJ
sc-5860	147	4	disaster	disaster	NOUN
sc-5860	147	5	hashtags	hashtag	NOUN
sc-5860	147	6	are	be	AUX
sc-5860	147	7	as	as	SCONJ
sc-5860	147	8	follows	follow	VERB
sc-5860	147	9	:	:	PUNCT
sc-5860	147	10	earthquakes	earthquake	NOUN
sc-5860	147	11	,	,	PUNCT
sc-5860	147	12	landslides	landslide	NOUN
sc-5860	147	13	,	,	PUNCT
sc-5860	147	14	droughts	drought	NOUN
sc-5860	147	15	,	,	PUNCT
sc-5860	147	16	famines	famine	NOUN
sc-5860	147	17	,	,	PUNCT
sc-5860	147	18	poverty	poverty	NOUN
sc-5860	147	19	,	,	PUNCT
sc-5860	147	20	hurricanes	hurricane	NOUN
sc-5860	147	21	,	,	PUNCT
sc-5860	147	22	extreme	extreme	ADJ
sc-5860	147	23	precipitation	precipitation	NOUN
sc-5860	147	24	and	and	CCONJ
sc-5860	147	25	flood	flood	NOUN
sc-5860	147	26	,	,	PUNCT
sc-5860	147	27	wildfires	wildfire	VERB
sc-5860	147	28	.	.	PUNCT
sc-5860	148	1	table	table	NOUN
sc-5860	148	2	1	1	NUM
sc-5860	148	3	shows	show	VERB
sc-5860	148	4	the	the	DET
sc-5860	148	5	distribution	distribution	NOUN
sc-5860	148	6	in	in	ADP
sc-5860	148	7	the	the	DET
sc-5860	148	8	dataset	dataset	NOUN
sc-5860	148	9	.	.	PUNCT
sc-5860	149	1	figure	figure	NOUN
sc-5860	149	2	2	2	NUM
sc-5860	149	3	contains	contain	VERB
sc-5860	149	4	examples	example	NOUN
sc-5860	149	5	of	of	ADP
sc-5860	149	6	natural	natural	ADJ
sc-5860	149	7	disasters	disaster	NOUN
sc-5860	149	8	in	in	ADP
sc-5860	149	9	the	the	DET
sc-5860	149	10	dataset	dataset	NOUN
sc-5860	149	11	.	.	PUNCT
sc-5860	150	1	table	table	NOUN
sc-5860	150	2	1	1	NUM
sc-5860	150	3	.	.	PUNCT
sc-5860	151	1	distribution	distribution	NOUN
sc-5860	151	2	of	of	ADP
sc-5860	151	3	natural	natural	ADJ
sc-5860	151	4	disasters	disaster	NOUN
sc-5860	151	5	in	in	ADP
sc-5860	151	6	the	the	DET
sc-5860	151	7	dataset	dataset	VERB
sc-5860	151	8	natural	natural	ADJ
sc-5860	151	9	disaster	disaster	NOUN
sc-5860	151	10	type	type	NOUN
sc-5860	151	11	number	number	NOUN
sc-5860	151	12	of	of	ADP
sc-5860	151	13	sample	sample	NOUN
sc-5860	151	14	earthquake	earthquake	NOUN
sc-5860	151	15	245	245	NUM
sc-5860	151	16	landslide	landslide	NOUN
sc-5860	151	17	95	95	NUM
sc-5860	151	18	droughts	drought	NOUN
sc-5860	151	19	/	/	SYM
sc-5860	151	20	famines	famine	NOUN
sc-5860	151	21	118	118	NUM
sc-5860	151	22	poverty	poverty	NOUN
sc-5860	151	23	92	92	NUM
sc-5860	151	24	hurricane	hurricane	NOUN
sc-5860	151	25	136	136	NUM
sc-5860	151	26	flood	flood	NOUN
sc-5860	151	27	180	180	NUM
sc-5860	151	28	wildfire	wildfire	NOUN
sc-5860	151	29	134	134	NUM
sc-5860	151	30	data	datum	NOUN
sc-5860	151	31	is	be	AUX
sc-5860	151	32	collected	collect	VERB
sc-5860	151	33	manually	manually	ADV
sc-5860	151	34	,	,	PUNCT
sc-5860	151	35	via	via	ADP
sc-5860	151	36	python	python	NOUN
sc-5860	151	37	tweepy	tweepy	PROPN
sc-5860	151	38	service	service	NOUN
sc-5860	151	39	and	and	CCONJ
sc-5860	151	40	twitter	twitter	NOUN
sc-5860	151	41	apis	apis	ADJ
sc-5860	151	42	.	.	PUNCT
sc-5860	152	1	positive	positive	ADJ
sc-5860	152	2	,	,	PUNCT
sc-5860	152	3	negative	negative	ADJ
sc-5860	152	4	and	and	CCONJ
sc-5860	152	5	neutral	neutral	ADJ
sc-5860	152	6	categories	category	NOUN
sc-5860	152	7	are	be	AUX
sc-5860	152	8	used	use	VERB
sc-5860	152	9	in	in	ADP
sc-5860	152	10	the	the	DET
sc-5860	152	11	labeling	labeling	NOUN
sc-5860	152	12	of	of	ADP
sc-5860	152	13	the	the	DET
sc-5860	152	14	data	datum	NOUN
sc-5860	152	15	.	.	PUNCT
sc-5860	153	1	the	the	DET
sc-5860	153	2	class	class	NOUN
sc-5860	153	3	labels	label	NOUN
sc-5860	153	4	assigned	assign	VERB
sc-5860	153	5	to	to	ADP
sc-5860	153	6	each	each	DET
sc-5860	153	7	pair	pair	NOUN
sc-5860	153	8	(	(	PUNCT
sc-5860	153	9	image+text	image+text	PROPN
sc-5860	153	10	)	)	PUNCT
sc-5860	153	11	are	be	AUX
sc-5860	153	12	collected	collect	VERB
sc-5860	153	13	under	under	ADP
sc-5860	153	14	three	three	NUM
sc-5860	153	15	cases	case	NOUN
sc-5860	153	16	:	:	PUNCT
sc-5860	153	17	1	1	X
sc-5860	153	18	)	)	PUNCT
sc-5860	153	19	if	if	SCONJ
sc-5860	153	20	both	both	PRON
sc-5860	153	21	have	have	VERB
sc-5860	153	22	the	the	DET
sc-5860	153	23	same	same	ADJ
sc-5860	153	24	label	label	NOUN
sc-5860	153	25	,	,	PUNCT
sc-5860	153	26	they	they	PRON
sc-5860	153	27	are	be	AUX
sc-5860	153	28	valid	valid	ADJ
sc-5860	153	29	and	and	CCONJ
sc-5860	153	30	the	the	DET
sc-5860	153	31	same	same	ADJ
sc-5860	153	32	.	.	PUNCT
sc-5860	154	1	2	2	X
sc-5860	154	2	)	)	PUNCT
sc-5860	154	3	if	if	SCONJ
sc-5860	154	4	one	one	NUM
sc-5860	154	5	label	label	NOUN
sc-5860	154	6	is	be	AUX
sc-5860	154	7	positive	positive	ADJ
sc-5860	154	8	or	or	CCONJ
sc-5860	154	9	negative	negative	ADJ
sc-5860	154	10	and	and	CCONJ
sc-5860	154	11	the	the	DET
sc-5860	154	12	other	other	ADJ
sc-5860	154	13	is	be	AUX
sc-5860	154	14	neutral	neutral	ADJ
sc-5860	154	15	,	,	PUNCT
sc-5860	154	16	it	it	PRON
sc-5860	154	17	is	be	AUX
sc-5860	154	18	a	a	DET
sc-5860	154	19	polar	polar	ADJ
sc-5860	154	20	(	(	PUNCT
sc-5860	154	21	positive	positive	ADJ
sc-5860	154	22	/	/	SYM
sc-5860	154	23	negative	negative	ADJ
sc-5860	154	24	)	)	PUNCT
sc-5860	154	25	label	label	NOUN
sc-5860	154	26	.	.	PUNCT
sc-5860	155	1	3	3	X
sc-5860	155	2	)	)	PUNCT
sc-5860	155	3	if	if	SCONJ
sc-5860	155	4	the	the	DET
sc-5860	155	5	image	image	NOUN
sc-5860	155	6	and	and	CCONJ
sc-5860	155	7	text	text	NOUN
sc-5860	155	8	have	have	VERB
sc-5860	155	9	opposite	opposite	ADJ
sc-5860	155	10	polarity	polarity	NOUN
sc-5860	155	11	tags	tag	NOUN
sc-5860	155	12	,	,	PUNCT
sc-5860	155	13	the	the	DET
sc-5860	155	14	tweet	tweet	NOUN
sc-5860	155	15	is	be	AUX
sc-5860	155	16	a	a	DET
sc-5860	155	17	conflict	conflict	NOUN
sc-5860	155	18	.	.	PUNCT
sc-5860	156	1	the	the	DET
sc-5860	156	2	distribution	distribution	NOUN
sc-5860	156	3	based	base	VERB
sc-5860	156	4	on	on	ADP
sc-5860	156	5	sentiment	sentiment	NOUN
sc-5860	156	6	in	in	ADP
sc-5860	156	7	the	the	DET
sc-5860	156	8	dataset	dataset	NOUN
sc-5860	156	9	,	,	PUNCT
sc-5860	156	10	images	image	NOUN
sc-5860	156	11	and	and	CCONJ
sc-5860	156	12	text	text	NOUN
sc-5860	156	13	,	,	PUNCT
sc-5860	156	14	is	be	AUX
sc-5860	156	15	as	as	SCONJ
sc-5860	156	16	shown	show	VERB
sc-5860	156	17	in	in	ADP
sc-5860	156	18	table	table	NOUN
sc-5860	156	19	2	2	NUM
sc-5860	156	20	.	.	PUNCT
sc-5860	157	1	in	in	ADP
sc-5860	157	2	the	the	DET
sc-5860	157	3	training	training	NOUN
sc-5860	157	4	phase	phase	NOUN
sc-5860	157	5	,	,	PUNCT
sc-5860	157	6	adam	adam	PROPN
sc-5860	157	7	(	(	PUNCT
sc-5860	157	8	adaptive	adaptive	ADJ
sc-5860	157	9	moment	moment	NOUN
sc-5860	157	10	estimation	estimation	NOUN
sc-5860	157	11	)	)	PUNCT
sc-5860	157	12	is	be	AUX
sc-5860	157	13	used	use	VERB
sc-5860	157	14	to	to	PART
sc-5860	157	15	update	update	VERB
sc-5860	157	16	the	the	DET
sc-5860	157	17	cross	cross	NOUN
sc-5860	157	18	entropy	entropy	PROPN
sc-5860	157	19	and	and	CCONJ
sc-5860	157	20	neural	neural	ADJ
sc-5860	157	21	network	network	NOUN
sc-5860	157	22	parameters	parameter	NOUN
sc-5860	157	23	as	as	ADP
sc-5860	157	24	the	the	DET
sc-5860	157	25	objective	objective	ADJ
sc-5860	157	26	function	function	NOUN
sc-5860	157	27	.	.	PUNCT
sc-5860	158	1	the	the	DET
sc-5860	158	2	learning	learning	NOUN
sc-5860	158	3	rate	rate	NOUN
sc-5860	158	4	is	be	AUX
sc-5860	158	5	set	set	VERB
sc-5860	158	6	to	to	ADP
sc-5860	158	7	2	2	NUM
sc-5860	158	8	×	×	NOUN
sc-5860	158	9	10	10	NUM
sc-5860	158	10	-	-	SYM
sc-5860	158	11	5	5	NUM
sc-5860	158	12	and	and	CCONJ
sc-5860	158	13	all	all	DET
sc-5860	158	14	models	model	NOUN
sc-5860	158	15	are	be	AUX
sc-5860	158	16	trained	train	VERB
sc-5860	158	17	for	for	ADP
sc-5860	158	18	100	100	NUM
sc-5860	158	19	epochs	epoch	NOUN
sc-5860	158	20	.	.	PUNCT
sc-5860	159	1	if	if	SCONJ
sc-5860	159	2	the	the	DET
sc-5860	159	3	validation	validation	NOUN
sc-5860	159	4	loss	loss	NOUN
sc-5860	159	5	does	do	AUX
sc-5860	159	6	not	not	PART
sc-5860	159	7	decrease	decrease	VERB
sc-5860	159	8	for	for	ADP
sc-5860	159	9	five	five	NUM
sc-5860	159	10	epochs	epoch	NOUN
sc-5860	159	11	,	,	PUNCT
sc-5860	159	12	the	the	DET
sc-5860	159	13	learning	learning	NOUN
sc-5860	159	14	rate	rate	NOUN
sc-5860	159	15	is	be	AUX
sc-5860	159	16	optimized	optimize	VERB
sc-5860	159	17	to	to	PART
sc-5860	159	18	decrease	decrease	VERB
sc-5860	159	19	by	by	ADP
sc-5860	159	20	10	10	NUM
sc-5860	159	21	times	time	NOUN
sc-5860	159	22	.	.	PUNCT
sc-5860	160	1	to	to	PART
sc-5860	160	2	prevent	prevent	VERB
sc-5860	160	3	overfitting	overfitting	NOUN
sc-5860	160	4	,	,	PUNCT
sc-5860	160	5	a	a	DET
sc-5860	160	6	dropout	dropout	NOUN
sc-5860	160	7	of	of	ADP
sc-5860	160	8	0.5	0.5	NUM
sc-5860	160	9	is	be	AUX
sc-5860	160	10	applied	apply	VERB
sc-5860	160	11	after	after	ADP
sc-5860	160	12	all	all	DET
sc-5860	160	13	intermediate	intermediate	ADJ
sc-5860	160	14	linear	linear	ADJ
sc-5860	160	15	table	table	NOUN
sc-5860	160	16	2	2	NUM
sc-5860	160	17	.	.	PUNCT
sc-5860	161	1	distribution	distribution	NOUN
sc-5860	161	2	of	of	ADP
sc-5860	161	3	sentiment	sentiment	NOUN
sc-5860	161	4	in	in	ADP
sc-5860	161	5	the	the	DET
sc-5860	161	6	dataset	dataset	VERB
sc-5860	161	7	positive	positive	ADJ
sc-5860	161	8	negative	negative	ADJ
sc-5860	161	9	neutral	neutral	ADJ
sc-5860	161	10	image	image	NOUN
sc-5860	161	11	181	181	NUM
sc-5860	161	12	617	617	NUM
sc-5860	161	13	202	202	NUM
sc-5860	161	14	text	text	NOUN
sc-5860	161	15	289	289	NUM
sc-5860	161	16	478	478	NUM
sc-5860	161	17	233	233	NUM
sc-5860	161	18	layers	layer	NOUN
sc-5860	161	19	.	.	PUNCT
sc-5860	162	1	during	during	ADP
sc-5860	162	2	training	training	NOUN
sc-5860	162	3	phase	phase	NOUN
sc-5860	162	4	,	,	PUNCT
sc-5860	162	5	the	the	DET
sc-5860	162	6	pytorch	pytorch	NOUN
sc-5860	162	7	library	library	NOUN
sc-5860	162	8	is	be	AUX
sc-5860	162	9	used	use	VERB
sc-5860	162	10	and	and	CCONJ
sc-5860	162	11	object	object	VERB
sc-5860	162	12	attributes	attribute	NOUN
sc-5860	162	13	are	be	AUX
sc-5860	162	14	extracted	extract	VERB
sc-5860	162	15	from	from	ADP
sc-5860	162	16	the	the	DET
sc-5860	162	17	publicly	publicly	ADV
sc-5860	162	18	available	available	ADJ
sc-5860	162	19	imagenet	imagenet	NOUN
sc-5860	162	20	pre	pre	ADJ
sc-5860	162	21	-	-	ADJ
sc-5860	162	22	trained	train	VERB
sc-5860	162	23	resnet-50	resnet-50	PROPN
sc-5860	162	24	model	model	NOUN
sc-5860	162	25	.	.	PUNCT
sc-5860	163	1	the	the	DET
sc-5860	163	2	resnet-101	resnet-101	PROPN
sc-5860	163	3	model	model	NOUN
sc-5860	163	4	,	,	PUNCT
sc-5860	163	5	pre	pre	VERB
sc-5860	163	6	-	-	VERB
sc-5860	163	7	trained	train	VERB
sc-5860	163	8	from	from	ADP
sc-5860	163	9	the	the	DET
sc-5860	163	10	places365	places365	PROPN
sc-5860	163	11	dataset	dataset	PROPN
sc-5860	163	12	,	,	PUNCT
sc-5860	163	13	is	be	AUX
sc-5860	163	14	used	use	VERB
sc-5860	163	15	to	to	PART
sc-5860	163	16	extract	extract	VERB
sc-5860	163	17	scene	scene	NOUN
sc-5860	163	18	features	feature	NOUN
sc-5860	163	19	.	.	PUNCT
sc-5860	164	1	4.3	4.3	NUM
sc-5860	164	2	.	.	PUNCT
sc-5860	164	3	performance	performance	NOUN
sc-5860	164	4	metrics	metric	NOUN
sc-5860	164	5	a	a	DET
sc-5860	164	6	10	10	NUM
sc-5860	164	7	-	-	ADJ
sc-5860	164	8	fold	fold	ADJ
sc-5860	164	9	cross	cross	ADJ
sc-5860	164	10	-	-	ADJ
sc-5860	164	11	validation	validation	ADJ
sc-5860	164	12	metric	metric	NOUN
sc-5860	164	13	is	be	AUX
sc-5860	164	14	used	use	VERB
sc-5860	164	15	in	in	ADP
sc-5860	164	16	the	the	DET
sc-5860	164	17	training	training	NOUN
sc-5860	164	18	,	,	PUNCT
sc-5860	164	19	validation	validation	NOUN
sc-5860	164	20	and	and	CCONJ
sc-5860	164	21	test	test	NOUN
sc-5860	164	22	set	set	NOUN
sc-5860	164	23	,	,	PUNCT
sc-5860	164	24	where	where	SCONJ
sc-5860	164	25	each	each	DET
sc-5860	164	26	bin	bin	NOUN
sc-5860	164	27	resulted	result	VERB
sc-5860	164	28	in	in	ADP
sc-5860	164	29	an	an	DET
sc-5860	164	30	8:1:1	8:1:1	NUM
sc-5860	164	31	ratio	ratio	NOUN
sc-5860	164	32	of	of	ADP
sc-5860	164	33	data	datum	NOUN
sc-5860	164	34	and	and	CCONJ
sc-5860	164	35	the	the	DET
sc-5860	164	36	same	same	ADJ
sc-5860	164	37	label	label	NOUN
sc-5860	164	38	distribution	distribution	NOUN
sc-5860	164	39	.	.	PUNCT
sc-5860	165	1	during	during	ADP
sc-5860	165	2	the	the	DET
sc-5860	165	3	training	training	NOUN
sc-5860	165	4	phase	phase	NOUN
sc-5860	165	5	,	,	PUNCT
sc-5860	165	6	training	training	NOUN
sc-5860	165	7	loss	loss	NOUN
sc-5860	165	8	,	,	PUNCT
sc-5860	165	9	validation	validation	NOUN
sc-5860	165	10	loss	loss	NOUN
sc-5860	165	11	,	,	PUNCT
sc-5860	165	12	training	training	NOUN
sc-5860	165	13	accuracy	accuracy	NOUN
sc-5860	165	14	,	,	PUNCT
sc-5860	165	15	validation	validation	NOUN
sc-5860	165	16	accuracy	accuracy	NOUN
sc-5860	165	17	and	and	CCONJ
sc-5860	165	18	f1	f1	NOUN
sc-5860	165	19	-	-	PUNCT
sc-5860	165	20	score	score	NOUN
sc-5860	165	21	values	value	NOUN
sc-5860	165	22	are	be	AUX
sc-5860	165	23	reported	report	VERB
sc-5860	165	24	in	in	ADP
sc-5860	165	25	each	each	DET
sc-5860	165	26	epoch	epoch	NOUN
sc-5860	165	27	.	.	PUNCT
sc-5860	166	1	4.4	4.4	NUM
sc-5860	166	2	.	.	PUNCT
sc-5860	167	1	learning	learn	VERB
sc-5860	167	2	models	model	NOUN
sc-5860	167	3	to	to	PART
sc-5860	167	4	encode	encode	VERB
sc-5860	167	5	objects	object	NOUN
sc-5860	167	6	and	and	CCONJ
sc-5860	167	7	overall	overall	ADJ
sc-5860	167	8	image	image	NOUN
sc-5860	167	9	content	content	NOUN
sc-5860	167	10	,	,	PUNCT
sc-5860	167	11	a	a	DET
sc-5860	167	12	pre	pre	ADJ
sc-5860	167	13	-	-	ADJ
sc-5860	167	14	trained	train	VERB
sc-5860	167	15	resnet-50	resnet-50	NOUN
sc-5860	167	16	model	model	NOUN
sc-5860	167	17	is	be	AUX
sc-5860	167	18	used	use	VERB
sc-5860	167	19	to	to	PART
sc-5860	167	20	extract	extract	VERB
sc-5860	167	21	features	feature	NOUN
sc-5860	167	22	on	on	ADP
sc-5860	167	23	imagenet	imagenet	NOUN
sc-5860	167	24	.	.	PUNCT
sc-5860	168	1	to	to	PART
sc-5860	168	2	encode	encode	VERB
sc-5860	168	3	scene	scene	NOUN
sc-5860	168	4	information	information	NOUN
sc-5860	168	5	,	,	PUNCT
sc-5860	168	6	features	feature	NOUN
sc-5860	168	7	are	be	AUX
sc-5860	168	8	extracted	extract	VERB
sc-5860	168	9	from	from	ADP
sc-5860	168	10	a	a	DET
sc-5860	168	11	resnet-101	resnet-101	NOUN
sc-5860	168	12	model	model	NOUN
sc-5860	168	13	pretrained	pretraine	VERB
sc-5860	168	14	on	on	ADP
sc-5860	168	15	places365	places365	PROPN
sc-5860	168	16	.	.	PUNCT
sc-5860	169	1	pre	pre	VERB
sc-5860	169	2	-	-	ADJ
sc-5860	169	3	trained	train	VERB
sc-5860	169	4	wide	wide	ADJ
sc-5860	169	5	ensemblebased	ensemblebase	VERB
sc-5860	169	6	convolutional	convolutional	ADJ
sc-5860	169	7	neural	neural	ADJ
sc-5860	169	8	networks	network	NOUN
sc-5860	169	9	are	be	AUX
sc-5860	169	10	used	use	VERB
sc-5860	169	11	to	to	PART
sc-5860	169	12	encode	encode	VERB
sc-5860	169	13	facial	facial	ADJ
sc-5860	169	14	expression	expression	NOUN
sc-5860	169	15	information	information	NOUN
sc-5860	169	16	.	.	PUNCT
sc-5860	170	1	ensembles	ensemble	NOUN
sc-5860	170	2	with	with	ADP
sc-5860	170	3	shared	share	VERB
sc-5860	170	4	representations	representation	NOUN
sc-5860	170	5	(	(	PUNCT
sc-5860	170	6	esrs	esr	NOUN
sc-5860	170	7	)	)	PUNCT
sc-5860	170	8	based	base	VERB
sc-5860	170	9	on	on	ADP
sc-5860	170	10	convolutional	convolutional	ADJ
sc-5860	170	11	networks	network	NOUN
sc-5860	170	12	have	have	AUX
sc-5860	170	13	been	be	AUX
sc-5860	170	14	studied	study	VERB
sc-5860	170	15	to	to	ADP
sc-5860	170	16	quantitatively	quantitatively	ADV
sc-5860	170	17	and	and	CCONJ
sc-5860	170	18	qualitatively	qualitatively	ADV
sc-5860	170	19	demonstrate	demonstrate	VERB
sc-5860	170	20	their	their	PRON
sc-5860	170	21	scalability	scalability	NOUN
sc-5860	170	22	to	to	ADP
sc-5860	170	23	facial	facial	ADJ
sc-5860	170	24	expression	expression	NOUN
sc-5860	170	25	datasets	dataset	NOUN
sc-5860	170	26	.	.	PUNCT
sc-5860	171	1	roberta	roberta	NOUN
sc-5860	171	2	-	-	PUNCT
sc-5860	171	3	base	base	NOUN
sc-5860	171	4	is	be	AUX
sc-5860	171	5	used	use	VERB
sc-5860	171	6	to	to	PART
sc-5860	171	7	extract	extract	VERB
sc-5860	171	8	contextual	contextual	ADJ
sc-5860	171	9	word	word	NOUN
sc-5860	171	10	embeddings	embedding	NOUN
sc-5860	171	11	.	.	PUNCT
sc-5860	172	1	the	the	DET
sc-5860	172	2	multimodal	multimodal	NOUN
sc-5860	172	3	clip	clip	NOUN
sc-5860	172	4	model	model	NOUN
sc-5860	172	5	is	be	AUX
sc-5860	172	6	used	use	VERB
sc-5860	172	7	to	to	PART
sc-5860	172	8	extract	extract	VERB
sc-5860	172	9	multimodal	multimodal	NOUN
sc-5860	172	10	features	feature	NOUN
sc-5860	172	11	.	.	PUNCT
sc-5860	173	1	both	both	DET
sc-5860	173	2	image	image	NOUN
sc-5860	173	3	and	and	CCONJ
sc-5860	173	4	text	text	NOUN
sc-5860	173	5	features	feature	NOUN
sc-5860	173	6	are	be	AUX
sc-5860	173	7	extracted	extract	VERB
sc-5860	173	8	using	use	VERB
sc-5860	173	9	an	an	DET
sc-5860	173	10	opensource	opensource	NOUN
sc-5860	173	11	clip	clip	NOUN
sc-5860	173	12	model	model	NOUN
sc-5860	173	13	,	,	PUNCT
sc-5860	173	14	where	where	SCONJ
sc-5860	173	15	both	both	DET
sc-5860	173	16	image	image	NOUN
sc-5860	173	17	and	and	CCONJ
sc-5860	173	18	text	text	NOUN
sc-5860	173	19	embeddings	embedding	NOUN
sc-5860	173	20	are	be	AUX
sc-5860	173	21	512	512	NUM
sc-5860	173	22	-	-	PUNCT
sc-5860	173	23	dimensional	dimensional	ADJ
sc-5860	173	24	vectors	vector	NOUN
sc-5860	173	25	.	.	PUNCT
sc-5860	174	1	4.5	4.5	NUM
sc-5860	174	2	.	.	PUNCT
sc-5860	174	3	performance	performance	NOUN
sc-5860	174	4	results	result	VERB
sc-5860	174	5	the	the	DET
sc-5860	174	6	dataset	dataset	NOUN
sc-5860	174	7	is	be	AUX
sc-5860	174	8	divided	divide	VERB
sc-5860	174	9	into	into	ADP
sc-5860	174	10	80	80	NUM
sc-5860	174	11	%	%	NOUN
sc-5860	174	12	training	training	NOUN
sc-5860	174	13	,	,	PUNCT
sc-5860	174	14	10	10	NUM
sc-5860	174	15	%	%	NOUN
sc-5860	174	16	validation	validation	NOUN
sc-5860	174	17	/	/	SYM
sc-5860	174	18	validation	validation	NOUN
sc-5860	174	19	,	,	PUNCT
sc-5860	174	20	and	and	CCONJ
sc-5860	174	21	10	10	NUM
sc-5860	174	22	%	%	NOUN
sc-5860	174	23	test	test	NOUN
sc-5860	174	24	data	datum	NOUN
sc-5860	174	25	in	in	ADP
sc-5860	174	26	order	order	NOUN
sc-5860	174	27	to	to	PART
sc-5860	174	28	perform	perform	VERB
sc-5860	174	29	performance	performance	NOUN
sc-5860	174	30	tests	test	NOUN
sc-5860	174	31	.	.	PUNCT
sc-5860	175	1	the	the	DET
sc-5860	175	2	10	10	NUM
sc-5860	175	3	-	-	ADJ
sc-5860	175	4	fold	fold	ADJ
sc-5860	175	5	cross	cross	NOUN
sc-5860	175	6	validation	validation	NOUN
sc-5860	175	7	method	method	NOUN
sc-5860	175	8	is	be	AUX
sc-5860	175	9	used	use	VERB
sc-5860	175	10	to	to	PART
sc-5860	175	11	verify	verify	VERB
sc-5860	175	12	the	the	DET
sc-5860	175	13	performance	performance	NOUN
sc-5860	175	14	values	value	NOUN
sc-5860	175	15	.	.	PUNCT
sc-5860	176	1	table	table	NOUN
sc-5860	176	2	3	3	NUM
sc-5860	176	3	presents	present	VERB
sc-5860	176	4	the	the	DET
sc-5860	176	5	evaluation	evaluation	NOUN
sc-5860	176	6	results	result	NOUN
sc-5860	176	7	of	of	ADP
sc-5860	176	8	unimodal	unimodal	ADJ
sc-5860	176	9	textual	textual	ADJ
sc-5860	176	10	and	and	CCONJ
sc-5860	176	11	visual	visual	ADJ
sc-5860	176	12	features	feature	NOUN
sc-5860	176	13	for	for	ADP
sc-5860	176	14	the	the	DET
sc-5860	176	15	dataset	dataset	NOUN
sc-5860	176	16	.	.	PUNCT
sc-5860	177	1	(	(	PUNCT
sc-5860	177	2	accuracy	accuracy	NOUN
sc-5860	177	3	and	and	CCONJ
sc-5860	177	4	f1	f1	NOUN
sc-5860	177	5	-	-	PUNCT
sc-5860	177	6	scores	score	NOUN
sc-5860	177	7	are	be	AUX
sc-5860	177	8	averaged	average	VERB
sc-5860	177	9	over	over	ADP
sc-5860	177	10	10	10	NUM
sc-5860	177	11	-	-	ADJ
sc-5860	177	12	fold	fold	ADJ
sc-5860	177	13	cross	cross	NOUN
sc-5860	177	14	-	-	NOUN
sc-5860	177	15	validation	validation	ADJ
sc-5860	177	16	.	.	PUNCT
sc-5860	177	17	)	)	PUNCT
sc-5860	177	18	table	table	NOUN
sc-5860	177	19	3	3	NUM
sc-5860	177	20	.	.	PUNCT
sc-5860	177	21	unimodal	unimodal	ADJ
sc-5860	177	22	visual	visual	ADJ
sc-5860	177	23	and	and	CCONJ
sc-5860	177	24	textual	textual	ADJ
sc-5860	177	25	feature	feature	NOUN
sc-5860	177	26	results	result	NOUN
sc-5860	177	27	for	for	ADP
sc-5860	177	28	sentiments	sentiment	NOUN
sc-5860	177	29	on	on	ADP
sc-5860	177	30	our	our	PRON
sc-5860	177	31	own	own	ADJ
sc-5860	177	32	dataset	dataset	NOUN
sc-5860	177	33	features	feature	NOUN
sc-5860	177	34	accuracy	accuracy	NOUN
sc-5860	177	35	f1	f1	ADJ
sc-5860	177	36	-	-	PUNCT
sc-5860	177	37	score	score	NOUN
sc-5860	177	38	object	object	NOUN
sc-5860	177	39	features	feature	VERB
sc-5860	177	40	0.632	0.632	NUM
sc-5860	177	41	0.601	0.601	NUM
sc-5860	177	42	facial	facial	ADJ
sc-5860	177	43	expressions	expression	NOUN
sc-5860	177	44	0.615	0.615	NUM
sc-5860	177	45	0.502	0.502	NUM
sc-5860	177	46	place	place	NOUN
sc-5860	177	47	and	and	CCONJ
sc-5860	177	48	scene	scene	NOUN
sc-5860	177	49	features	feature	VERB
sc-5860	177	50	0.630	0.630	NUM
sc-5860	177	51	0.611	0.611	NUM
sc-5860	177	52	effective	effective	ADJ
sc-5860	177	53	image	image	NOUN
sc-5860	177	54	content	content	NOUN
sc-5860	177	55	0.635	0.635	NUM
sc-5860	177	56	0.617	0.617	NUM
sc-5860	177	57	clip	clip	NOUN
sc-5860	177	58	(	(	PUNCT
sc-5860	177	59	image	image	NOUN
sc-5860	177	60	)	)	PUNCT
sc-5860	177	61	0.715	0.715	NOUN
sc-5860	177	62	0.695	0.695	NUM
sc-5860	177	63	clip	clip	NOUN
sc-5860	177	64	(	(	PUNCT
sc-5860	177	65	text	text	NOUN
sc-5860	177	66	)	)	PUNCT
sc-5860	178	1	0.708	0.708	NUM
sc-5860	178	2	0.684	0.684	NUM
sc-5860	178	3	roberta	roberta	NOUN
sc-5860	178	4	-	-	PUNCT
sc-5860	178	5	base	base	NOUN
sc-5860	178	6	(	(	PUNCT
sc-5860	178	7	text	text	NOUN
sc-5860	178	8	)	)	PUNCT
sc-5860	178	9	0.662	0.662	NUM
sc-5860	178	10	0.636	0.636	NUM
sc-5860	178	11	5	5	NUM
sc-5860	178	12	eai	eai	NOUN
sc-5860	178	13	endorsed	endorse	VERB
sc-5860	178	14	transactions	transaction	NOUN
sc-5860	178	15	on	on	ADP
sc-5860	178	16	smart	smart	ADJ
sc-5860	178	17	cities	city	NOUN
sc-5860	178	18	|	|	ADV
sc-5860	178	19	volume	volume	VERB
sc-5860	178	20	7	7	NUM
sc-5860	178	21	|	|	NOUN
sc-5860	178	22	issue	issue	VERB
sc-5860	178	23	4	4	NUM
sc-5860	178	24	|	|	CCONJ
sc-5860	178	25	2023	2023	NUM
sc-5860	178	26	|	|	ADV
sc-5860	178	27	s.	s.	PROPN
sc-5860	178	28	dursun	dursun	PROPN
sc-5860	178	29	and	and	CCONJ
sc-5860	178	30	s.	s.	PROPN
sc-5860	178	31	eken	eken	VERB
sc-5860	178	32	figure	figure	NOUN
sc-5860	178	33	2	2	NUM
sc-5860	178	34	.	.	PUNCT
sc-5860	178	35	sample	sample	NOUN
sc-5860	178	36	images	image	NOUN
sc-5860	178	37	and	and	CCONJ
sc-5860	178	38	texts	text	NOUN
sc-5860	178	39	from	from	ADP
sc-5860	178	40	natural	natural	ADJ
sc-5860	178	41	disasters	disaster	NOUN
sc-5860	178	42	in	in	ADP
sc-5860	178	43	the	the	DET
sc-5860	178	44	dataset	dataset	NOUN
sc-5860	178	45	in	in	ADP
sc-5860	178	46	order	order	NOUN
sc-5860	178	47	to	to	PART
sc-5860	178	48	compare	compare	VERB
sc-5860	178	49	the	the	DET
sc-5860	178	50	performance	performance	NOUN
sc-5860	178	51	results	result	NOUN
sc-5860	178	52	,	,	PUNCT
sc-5860	178	53	different	different	ADJ
sc-5860	178	54	models	model	NOUN
sc-5860	178	55	are	be	AUX
sc-5860	178	56	used	use	VERB
sc-5860	178	57	to	to	PART
sc-5860	178	58	extract	extract	VERB
sc-5860	178	59	image	image	NOUN
sc-5860	178	60	and	and	CCONJ
sc-5860	178	61	text	text	NOUN
sc-5860	178	62	features	feature	NOUN
sc-5860	178	63	and	and	CCONJ
sc-5860	178	64	the	the	DET
sc-5860	178	65	results	result	NOUN
sc-5860	178	66	are	be	AUX
sc-5860	178	67	summarized	summarize	VERB
sc-5860	178	68	in	in	ADP
sc-5860	178	69	table	table	NOUN
sc-5860	178	70	4	4	NUM
sc-5860	178	71	.	.	PUNCT
sc-5860	179	1	figure	figure	VERB
sc-5860	179	2	3	3	NUM
sc-5860	179	3	shows	show	VERB
sc-5860	179	4	examples	example	NOUN
sc-5860	179	5	of	of	ADP
sc-5860	179	6	unimodal	unimodal	ADJ
sc-5860	179	7	and	and	CCONJ
sc-5860	179	8	multimodal	multimodal	NOUN
sc-5860	179	9	estimation	estimation	NOUN
sc-5860	179	10	results	result	NOUN
sc-5860	179	11	.	.	PUNCT
sc-5860	180	1	the	the	DET
sc-5860	180	2	highlighted	highlight	VERB
sc-5860	180	3	text	text	NOUN
sc-5860	180	4	parts	part	NOUN
sc-5860	180	5	in	in	ADP
sc-5860	180	6	yellow	yellow	ADJ
sc-5860	180	7	are	be	AUX
sc-5860	180	8	the	the	DET
sc-5860	180	9	word(s	word(s	PROPN
sc-5860	180	10	)	)	PUNCT
sc-5860	180	11	that	that	PRON
sc-5860	180	12	indicate	indicate	VERB
sc-5860	180	13	the	the	DET
sc-5860	180	14	sentiment	sentiment	NOUN
sc-5860	180	15	in	in	ADP
sc-5860	180	16	the	the	DET
sc-5860	180	17	text	text	NOUN
sc-5860	180	18	.	.	PUNCT
sc-5860	181	1	as	as	SCONJ
sc-5860	181	2	seen	see	VERB
sc-5860	181	3	in	in	ADP
sc-5860	181	4	table	table	NOUN
sc-5860	181	5	4	4	NUM
sc-5860	181	6	,	,	PUNCT
sc-5860	181	7	the	the	DET
sc-5860	181	8	highest	high	ADJ
sc-5860	181	9	performance	performance	NOUN
sc-5860	181	10	values	value	NOUN
sc-5860	181	11	are	be	AUX
sc-5860	181	12	achieved	achieve	VERB
sc-5860	181	13	by	by	ADP
sc-5860	181	14	using	use	VERB
sc-5860	181	15	the	the	DET
sc-5860	181	16	clip	clip	NOUN
sc-5860	181	17	model	model	NOUN
sc-5860	181	18	in	in	ADP
sc-5860	181	19	the	the	DET
sc-5860	181	20	image	image	NOUN
sc-5860	181	21	and	and	CCONJ
sc-5860	181	22	the	the	DET
sc-5860	181	23	roberta	roberta	PROPN
sc-5860	181	24	model	model	NOUN
sc-5860	181	25	in	in	ADP
sc-5860	181	26	the	the	DET
sc-5860	181	27	text	text	NOUN
sc-5860	181	28	(	(	PUNCT
sc-5860	181	29	accuracy	accuracy	NOUN
sc-5860	181	30	77.5	77.5	NUM
sc-5860	181	31	%	%	NOUN
sc-5860	181	32	;	;	PUNCT
sc-5860	181	33	f1score	f1score	NOUN
sc-5860	181	34	71	71	NUM
sc-5860	181	35	%	%	NOUN
sc-5860	181	36	)	)	PUNCT
sc-5860	181	37	.	.	PUNCT
sc-5860	182	1	the	the	DET
sc-5860	182	2	issues	issue	NOUN
sc-5860	182	3	that	that	PRON
sc-5860	182	4	caused	cause	VERB
sc-5860	182	5	this	this	DET
sc-5860	182	6	limited	limited	ADJ
sc-5860	182	7	success	success	NOUN
sc-5860	182	8	performance	performance	NOUN
sc-5860	182	9	can	can	AUX
sc-5860	182	10	be	be	AUX
sc-5860	182	11	attributed	attribute	VERB
sc-5860	182	12	as	as	SCONJ
sc-5860	182	13	follows	follow	VERB
sc-5860	182	14	:	:	PUNCT
sc-5860	182	15	•	•	ADV
sc-5860	182	16	since	since	SCONJ
sc-5860	182	17	the	the	DET
sc-5860	182	18	subject	subject	NOUN
sc-5860	182	19	of	of	ADP
sc-5860	182	20	the	the	DET
sc-5860	182	21	study	study	NOUN
sc-5860	182	22	is	be	AUX
sc-5860	182	23	disaster	disaster	NOUN
sc-5860	182	24	data	datum	NOUN
sc-5860	182	25	,	,	PUNCT
sc-5860	182	26	the	the	DET
sc-5860	182	27	number	number	NOUN
sc-5860	182	28	of	of	ADP
sc-5860	182	29	negative	negative	ADJ
sc-5860	182	30	examples	example	NOUN
sc-5860	182	31	is	be	AUX
sc-5860	182	32	quite	quite	ADV
sc-5860	182	33	high	high	ADJ
sc-5860	182	34	.	.	PUNCT
sc-5860	183	1	•	•	NUM
sc-5860	183	2	insufficient	insufficient	ADJ
sc-5860	183	3	label	label	NOUN
sc-5860	183	4	pooling	pool	VERB
sc-5860	183	5	strategy	strategy	NOUN
sc-5860	183	6	used	use	VERB
sc-5860	183	7	to	to	ADP
sc-5860	183	8	pool	pool	NOUN
sc-5860	183	9	labels	label	NOUN
sc-5860	183	10	,	,	PUNCT
sc-5860	183	11	favoring	favor	VERB
sc-5860	183	12	positive	positive	ADJ
sc-5860	183	13	or	or	CCONJ
sc-5860	183	14	negative	negative	ADJ
sc-5860	183	15	over	over	ADP
sc-5860	183	16	neutral	neutral	ADJ
sc-5860	183	17	labels	label	NOUN
sc-5860	183	18	,	,	PUNCT
sc-5860	183	19	resulting	result	VERB
sc-5860	183	20	in	in	ADP
sc-5860	183	21	a	a	DET
sc-5860	183	22	higher	high	ADJ
sc-5860	183	23	number	number	NOUN
sc-5860	183	24	of	of	ADP
sc-5860	183	25	controversial	controversial	ADJ
sc-5860	183	26	labels	label	NOUN
sc-5860	183	27	.	.	PUNCT
sc-5860	184	1	•	•	NUM
sc-5860	184	2	the	the	DET
sc-5860	184	3	model	model	NOUN
sc-5860	184	4	can	can	AUX
sc-5860	184	5	not	not	PART
sc-5860	184	6	capture	capture	VERB
sc-5860	184	7	interactions	interaction	NOUN
sc-5860	184	8	and	and	CCONJ
sc-5860	184	9	can	can	AUX
sc-5860	184	10	not	not	PART
sc-5860	184	11	distinguish	distinguish	VERB
sc-5860	184	12	between	between	ADP
sc-5860	184	13	neutral	neutral	ADJ
sc-5860	184	14	and	and	CCONJ
sc-5860	184	15	polar	polar	ADJ
sc-5860	184	16	samples	sample	NOUN
sc-5860	184	17	.	.	PUNCT
sc-5860	185	1	•	•	NUM
sc-5860	185	2	the	the	DET
sc-5860	185	3	data	datum	NOUN
sc-5860	185	4	set	set	VERB
sc-5860	185	5	is	be	AUX
sc-5860	185	6	limited	limit	VERB
sc-5860	185	7	to	to	ADP
sc-5860	185	8	1,000	1,000	NUM
sc-5860	185	9	samples	sample	NOUN
sc-5860	185	10	.	.	PUNCT
sc-5860	186	1	5	5	X
sc-5860	186	2	.	.	X
sc-5860	186	3	conclusions	conclusion	NOUN
sc-5860	186	4	and	and	CCONJ
sc-5860	186	5	future	future	ADJ
sc-5860	186	6	works	work	NOUN
sc-5860	186	7	in	in	ADP
sc-5860	186	8	this	this	DET
sc-5860	186	9	study	study	NOUN
sc-5860	186	10	,	,	PUNCT
sc-5860	186	11	an	an	DET
sc-5860	186	12	experimental	experimental	ADJ
sc-5860	186	13	evaluation	evaluation	NOUN
sc-5860	186	14	of	of	ADP
sc-5860	186	15	visual	visual	ADJ
sc-5860	186	16	,	,	PUNCT
sc-5860	186	17	textual	textual	ADJ
sc-5860	186	18	and	and	CCONJ
sc-5860	186	19	multimodal	multimodal	NOUN
sc-5860	186	20	features	feature	NOUN
sc-5860	186	21	is	be	AUX
sc-5860	186	22	presented	present	VERB
sc-5860	186	23	for	for	ADP
sc-5860	186	24	msa	msa	PROPN
sc-5860	186	25	of	of	ADP
sc-5860	186	26	natural	natural	ADJ
sc-5860	186	27	disasters	disaster	NOUN
sc-5860	186	28	-	-	PUNCT
sc-5860	186	29	themed	theme	VERB
sc-5860	186	30	tweets	tweet	NOUN
sc-5860	186	31	collected	collect	VERB
sc-5860	186	32	on	on	ADP
sc-5860	186	33	twitter	twitter	NOUN
sc-5860	186	34	.	.	PUNCT
sc-5860	187	1	experience	experience	NOUN
sc-5860	187	2	has	have	AUX
sc-5860	187	3	revealed	reveal	VERB
sc-5860	187	4	that	that	SCONJ
sc-5860	187	5	clip	clip	NOUN
sc-5860	187	6	embeddings	embedding	NOUN
sc-5860	187	7	can	can	AUX
sc-5860	187	8	6	6	NUM
sc-5860	187	9	eai	eai	NOUN
sc-5860	187	10	endorsed	endorse	VERB
sc-5860	187	11	transactions	transaction	NOUN
sc-5860	187	12	on	on	ADP
sc-5860	187	13	smart	smart	ADJ
sc-5860	187	14	cities	city	NOUN
sc-5860	187	15	|	|	ADV
sc-5860	187	16	volume	volume	VERB
sc-5860	187	17	7	7	NUM
sc-5860	187	18	|	|	NOUN
sc-5860	187	19	issue	issue	VERB
sc-5860	187	20	4	4	NUM
sc-5860	187	21	|	|	CCONJ
sc-5860	187	22	2023	2023	NUM
sc-5860	187	23	|	|	NOUN
sc-5860	187	24	multimodal	multimodal	ADJ
sc-5860	187	25	sentiment	sentiment	NOUN
sc-5860	187	26	analysis	analysis	NOUN
sc-5860	187	27	in	in	ADP
sc-5860	187	28	natural	natural	ADJ
sc-5860	187	29	disaster	disaster	NOUN
sc-5860	187	30	data	datum	NOUN
sc-5860	187	31	on	on	ADP
sc-5860	187	32	social	social	ADJ
sc-5860	187	33	media	medium	NOUN
sc-5860	187	34	table	table	NOUN
sc-5860	187	35	4	4	NUM
sc-5860	187	36	.	.	PUNCT
sc-5860	187	37	multimodal	multimodal	ADJ
sc-5860	187	38	visual	visual	ADJ
sc-5860	187	39	and	and	CCONJ
sc-5860	187	40	textual	textual	ADJ
sc-5860	187	41	feature	feature	NOUN
sc-5860	187	42	results	result	NOUN
sc-5860	187	43	for	for	ADP
sc-5860	187	44	sentiments	sentiment	NOUN
sc-5860	187	45	on	on	ADP
sc-5860	187	46	our	our	PRON
sc-5860	187	47	own	own	ADJ
sc-5860	187	48	dataset	dataset	NOUN
sc-5860	187	49	accuracy	accuracy	NOUN
sc-5860	187	50	f1	f1	NOUN
sc-5860	187	51	-	-	PUNCT
sc-5860	187	52	score	score	NOUN
sc-5860	187	53	accuracy	accuracy	NOUN
sc-5860	187	54	f1	f1	NOUN
sc-5860	187	55	-	-	PUNCT
sc-5860	187	56	score	score	NOUN
sc-5860	187	57	roberta	roberta	PROPN
sc-5860	187	58	clip	clip	PROPN
sc-5860	187	59	(	(	PUNCT
sc-5860	187	60	text	text	NOUN
sc-5860	187	61	)	)	PUNCT
sc-5860	187	62	effective	effective	ADJ
sc-5860	187	63	image	image	NOUN
sc-5860	187	64	content	content	NOUN
sc-5860	187	65	(	(	PUNCT
sc-5860	187	66	imagenet	imagenet	NOUN
sc-5860	187	67	)	)	PUNCT
sc-5860	187	68	0.700	0.700	NUM
sc-5860	187	69	0.685	0.685	NUM
sc-5860	187	70	0.729	0.729	NUM
sc-5860	187	71	0.700	0.700	NUM
sc-5860	187	72	clip	clip	NOUN
sc-5860	187	73	(	(	PUNCT
sc-5860	187	74	image	image	NOUN
sc-5860	187	75	)	)	PUNCT
sc-5860	187	76	0.775	0.775	NUM
sc-5860	187	77	0.710	0.710	NUM
sc-5860	187	78	0.701	0.701	NUM
sc-5860	187	79	0.750	0.750	NUM
sc-5860	187	80	figure	figure	NOUN
sc-5860	187	81	3	3	NUM
sc-5860	187	82	.	.	PUNCT
sc-5860	187	83	examples	example	NOUN
sc-5860	187	84	for	for	ADP
sc-5860	187	85	unimodal	unimodal	ADJ
sc-5860	187	86	and	and	CCONJ
sc-5860	187	87	multimodal	multimodal	NOUN
sc-5860	187	88	prediction	prediction	NOUN
sc-5860	187	89	results	result	NOUN
sc-5860	187	90	serve	serve	VERB
sc-5860	187	91	as	as	ADP
sc-5860	187	92	a	a	DET
sc-5860	187	93	strong	strong	ADJ
sc-5860	187	94	basis	basis	NOUN
sc-5860	187	95	for	for	ADP
sc-5860	187	96	the	the	DET
sc-5860	187	97	task	task	NOUN
sc-5860	187	98	of	of	ADP
sc-5860	187	99	multimodal	multimodal	ADJ
sc-5860	187	100	sentiment	sentiment	NOUN
sc-5860	187	101	prediction	prediction	NOUN
sc-5860	187	102	in	in	ADP
sc-5860	187	103	tweets	tweet	NOUN
sc-5860	187	104	.	.	PUNCT
sc-5860	188	1	it	it	PRON
sc-5860	188	2	is	be	AUX
sc-5860	188	3	envisaged	envisage	VERB
sc-5860	188	4	that	that	SCONJ
sc-5860	188	5	the	the	DET
sc-5860	188	6	natural	natural	ADJ
sc-5860	188	7	disasters	disaster	NOUN
sc-5860	188	8	themed	theme	VERB
sc-5860	188	9	dataset	dataset	VERB
sc-5860	188	10	prepared	prepare	VERB
sc-5860	188	11	by	by	ADP
sc-5860	188	12	collecting	collect	VERB
sc-5860	188	13	from	from	ADP
sc-5860	188	14	social	social	ADJ
sc-5860	188	15	media	medium	NOUN
sc-5860	188	16	within	within	ADP
sc-5860	188	17	the	the	DET
sc-5860	188	18	scope	scope	NOUN
sc-5860	188	19	of	of	ADP
sc-5860	188	20	multimodality	multimodality	NOUN
sc-5860	188	21	(	(	PUNCT
sc-5860	188	22	image	image	NOUN
sc-5860	188	23	and	and	CCONJ
sc-5860	188	24	text	text	NOUN
sc-5860	188	25	)	)	PUNCT
sc-5860	188	26	may	may	AUX
sc-5860	188	27	be	be	AUX
sc-5860	188	28	useful	useful	ADJ
sc-5860	188	29	for	for	ADP
sc-5860	188	30	future	future	ADJ
sc-5860	188	31	research	research	NOUN
sc-5860	188	32	in	in	ADP
sc-5860	188	33	this	this	DET
sc-5860	188	34	field	field	NOUN
sc-5860	188	35	.	.	PUNCT
sc-5860	189	1	while	while	SCONJ
sc-5860	189	2	msa	msa	PROPN
sc-5860	189	3	can	can	AUX
sc-5860	189	4	provide	provide	VERB
sc-5860	189	5	richer	rich	ADJ
sc-5860	189	6	insights	insight	NOUN
sc-5860	189	7	compared	compare	VERB
sc-5860	189	8	to	to	ADP
sc-5860	189	9	unimodal	unimodal	ADJ
sc-5860	189	10	analysis	analysis	NOUN
sc-5860	189	11	,	,	PUNCT
sc-5860	189	12	it	it	PRON
sc-5860	189	13	also	also	ADV
sc-5860	189	14	faces	face	VERB
sc-5860	189	15	several	several	ADJ
sc-5860	189	16	limitations	limitation	NOUN
sc-5860	189	17	.	.	PUNCT
sc-5860	190	1	social	social	ADJ
sc-5860	190	2	media	medium	NOUN
sc-5860	190	3	data	datum	NOUN
sc-5860	190	4	is	be	AUX
sc-5860	190	5	noisy	noisy	ADJ
sc-5860	190	6	,	,	PUNCT
sc-5860	190	7	with	with	ADP
sc-5860	190	8	many	many	ADJ
sc-5860	190	9	irrelevant	irrelevant	ADJ
sc-5860	190	10	or	or	CCONJ
sc-5860	190	11	low	low	ADJ
sc-5860	190	12	-	-	PUNCT
sc-5860	190	13	quality	quality	NOUN
sc-5860	190	14	posts	post	NOUN
sc-5860	190	15	.	.	PUNCT
sc-5860	191	1	there	there	PRON
sc-5860	191	2	is	be	VERB
sc-5860	191	3	often	often	ADV
sc-5860	191	4	a	a	DET
sc-5860	191	5	lack	lack	NOUN
sc-5860	191	6	of	of	ADP
sc-5860	191	7	labeled	label	VERB
sc-5860	191	8	datasets	dataset	NOUN
sc-5860	191	9	that	that	PRON
sc-5860	191	10	contain	contain	VERB
sc-5860	191	11	multimodal	multimodal	ADJ
sc-5860	191	12	information	information	NOUN
sc-5860	191	13	specific	specific	ADJ
sc-5860	191	14	to	to	ADP
sc-5860	191	15	natural	natural	ADJ
sc-5860	191	16	disasters	disaster	NOUN
sc-5860	191	17	.	.	PUNCT
sc-5860	192	1	sentiments	sentiment	NOUN
sc-5860	192	2	during	during	ADP
sc-5860	192	3	a	a	DET
sc-5860	192	4	natural	natural	ADJ
sc-5860	192	5	disaster	disaster	NOUN
sc-5860	192	6	can	can	AUX
sc-5860	192	7	change	change	VERB
sc-5860	192	8	rapidly	rapidly	ADV
sc-5860	192	9	,	,	PUNCT
sc-5860	192	10	requiring	require	VERB
sc-5860	192	11	models	model	NOUN
sc-5860	192	12	to	to	PART
sc-5860	192	13	account	account	VERB
sc-5860	192	14	for	for	ADP
sc-5860	192	15	temporal	temporal	ADJ
sc-5860	192	16	dynamics	dynamic	NOUN
sc-5860	192	17	.	.	PUNCT
sc-5860	193	1	using	use	VERB
sc-5860	193	2	social	social	ADJ
sc-5860	193	3	media	medium	NOUN
sc-5860	193	4	data	datum	NOUN
sc-5860	193	5	raises	raise	VERB
sc-5860	193	6	concerns	concern	NOUN
sc-5860	193	7	about	about	ADP
sc-5860	193	8	user	user	NOUN
sc-5860	193	9	privacy	privacy	NOUN
sc-5860	193	10	and	and	CCONJ
sc-5860	193	11	data	datum	NOUN
sc-5860	193	12	security	security	NOUN
sc-5860	193	13	.	.	PUNCT
sc-5860	194	1	future	future	ADJ
sc-5860	194	2	work	work	NOUN
sc-5860	194	3	can	can	AUX
sc-5860	194	4	address	address	VERB
sc-5860	194	5	these	these	DET
sc-5860	194	6	challenges	challenge	NOUN
sc-5860	194	7	to	to	PART
sc-5860	194	8	improve	improve	VERB
sc-5860	194	9	the	the	DET
sc-5860	194	10	effectiveness	effectiveness	NOUN
sc-5860	194	11	and	and	CCONJ
sc-5860	194	12	reliability	reliability	NOUN
sc-5860	194	13	of	of	ADP
sc-5860	194	14	msa	msa	PROPN
sc-5860	194	15	in	in	ADP
sc-5860	194	16	this	this	DET
sc-5860	194	17	context	context	NOUN
sc-5860	194	18	.	.	PUNCT
sc-5860	195	1	in	in	ADP
sc-5860	195	2	the	the	DET
sc-5860	195	3	future	future	NOUN
sc-5860	195	4	,	,	PUNCT
sc-5860	195	5	it	it	PRON
sc-5860	195	6	is	be	AUX
sc-5860	195	7	planned	plan	VERB
sc-5860	195	8	to	to	PART
sc-5860	195	9	collect	collect	VERB
sc-5860	195	10	more	more	ADJ
sc-5860	195	11	images	image	NOUN
sc-5860	195	12	and	and	CCONJ
sc-5860	195	13	text	text	NOUN
sc-5860	195	14	from	from	ADP
sc-5860	195	15	different	different	ADJ
sc-5860	195	16	social	social	ADJ
sc-5860	195	17	media	medium	NOUN
sc-5860	195	18	platforms	platform	NOUN
sc-5860	195	19	and	and	CCONJ
sc-5860	195	20	carry	carry	VERB
sc-5860	195	21	out	out	ADP
sc-5860	195	22	7	7	NUM
sc-5860	195	23	eai	eai	NOUN
sc-5860	195	24	endorsed	endorse	VERB
sc-5860	195	25	transactions	transaction	NOUN
sc-5860	195	26	on	on	ADP
sc-5860	195	27	smart	smart	ADJ
sc-5860	195	28	cities	city	NOUN
sc-5860	195	29	|	|	ADV
sc-5860	195	30	volume	volume	VERB
sc-5860	195	31	7	7	NUM
sc-5860	195	32	|	|	NOUN
sc-5860	195	33	issue	issue	VERB
sc-5860	195	34	4	4	NUM
sc-5860	195	35	|	|	CCONJ
sc-5860	195	36	2023	2023	NUM
sc-5860	195	37	|	|	ADV
sc-5860	195	38	s.	s.	PROPN
sc-5860	195	39	dursun	dursun	PROPN
sc-5860	195	40	and	and	CCONJ
sc-5860	195	41	s.	s.	PROPN
sc-5860	195	42	eken	eken	VERB
sc-5860	195	43	activities	activity	NOUN
sc-5860	195	44	to	to	PART
sc-5860	195	45	reduce	reduce	VERB
sc-5860	195	46	bias	bias	NOUN
sc-5860	195	47	.	.	PUNCT
sc-5860	196	1	also	also	ADV
sc-5860	196	2	,	,	PUNCT
sc-5860	196	3	we	we	PRON
sc-5860	196	4	will	will	AUX
sc-5860	196	5	develop	develop	VERB
sc-5860	196	6	models	model	NOUN
sc-5860	196	7	that	that	PRON
sc-5860	196	8	combine	combine	VERB
sc-5860	196	9	early	early	ADV
sc-5860	196	10	,	,	PUNCT
sc-5860	196	11	late	late	ADV
sc-5860	196	12	,	,	PUNCT
sc-5860	196	13	and	and	CCONJ
sc-5860	196	14	hybrid	hybrid	ADJ
sc-5860	196	15	fusion	fusion	NOUN
sc-5860	196	16	techniques	technique	NOUN
sc-5860	196	17	to	to	PART
sc-5860	196	18	better	well	ADV
sc-5860	196	19	integrate	integrate	VERB
sc-5860	196	20	multimodal	multimodal	NOUN
sc-5860	196	21	features	feature	NOUN
sc-5860	196	22	.	.	PUNCT
sc-5860	197	1	moreover	moreover	ADV
sc-5860	197	2	,	,	PUNCT
sc-5860	197	3	it	it	PRON
sc-5860	197	4	is	be	AUX
sc-5860	197	5	important	important	ADJ
sc-5860	197	6	to	to	PART
sc-5860	197	7	capture	capture	VERB
sc-5860	197	8	temporal	temporal	ADJ
sc-5860	197	9	dynamics	dynamic	NOUN
sc-5860	197	10	and	and	CCONJ
sc-5860	197	11	changes	change	NOUN
sc-5860	197	12	in	in	ADP
sc-5860	197	13	sentiment	sentiment	NOUN
sc-5860	197	14	over	over	ADP
sc-5860	197	15	time	time	NOUN
sc-5860	197	16	.	.	PUNCT
sc-5860	198	1	data	datum	NOUN
sc-5860	198	2	and	and	CCONJ
sc-5860	198	3	code	code	NOUN
sc-5860	198	4	availability	availability	NOUN
sc-5860	198	5	code	code	NOUN
sc-5860	198	6	underlying	underlie	VERB
sc-5860	198	7	this	this	DET
sc-5860	198	8	paper	paper	NOUN
sc-5860	198	9	are	be	AUX
sc-5860	198	10	available	available	ADJ
sc-5860	198	11	at	at	ADP
sc-5860	198	12	github	github	PROPN
sc-5860	198	13	repo	repo	PROPN
sc-5860	198	14	.	.	PUNCT
sc-5860	199	1	references	reference	NOUN
sc-5860	199	2	[	[	X
sc-5860	199	3	1	1	NUM
sc-5860	199	4	]	]	PUNCT
sc-5860	199	5	chandrasekaran	chandrasekaran	VERB
sc-5860	199	6	,	,	PUNCT
sc-5860	199	7	g.	g.	PROPN
sc-5860	199	8	,	,	PUNCT
sc-5860	199	9	nguyen	nguyen	PROPN
sc-5860	199	10	,	,	PUNCT
sc-5860	199	11	t.n	t.n	PROPN
sc-5860	199	12	.	.	PROPN
sc-5860	199	13	and	and	CCONJ
sc-5860	199	14	hemanth	hemanth	PROPN
sc-5860	199	15	d	d	PROPN
sc-5860	199	16	,	,	PUNCT
sc-5860	199	17	j.	j.	PROPN
sc-5860	199	18	(	(	PUNCT
sc-5860	199	19	2021	2021	NUM
sc-5860	199	20	)	)	PUNCT
sc-5860	199	21	multimodal	multimodal	ADJ
sc-5860	199	22	sentimental	sentimental	ADJ
sc-5860	199	23	analysis	analysis	NOUN
sc-5860	199	24	for	for	ADP
sc-5860	199	25	social	social	ADJ
sc-5860	199	26	media	medium	NOUN
sc-5860	199	27	applications	application	NOUN
sc-5860	199	28	:	:	PUNCT
sc-5860	199	29	a	a	DET
sc-5860	199	30	comprehensive	comprehensive	ADJ
sc-5860	199	31	review	review	NOUN
sc-5860	199	32	.	.	PUNCT
sc-5860	200	1	wiley	wiley	PROPN
sc-5860	200	2	interdisciplinary	interdisciplinary	ADJ
sc-5860	200	3	reviews	review	NOUN
sc-5860	200	4	:	:	PUNCT
sc-5860	200	5	data	datum	NOUN
sc-5860	200	6	mining	mining	NOUN
sc-5860	200	7	and	and	CCONJ
sc-5860	200	8	knowledge	knowledge	NOUN
sc-5860	200	9	discovery	discovery	PROPN
sc-5860	200	10	11(5	11(5	NUM
sc-5860	200	11	):	):	PUNCT
sc-5860	200	12	e1415	e1415	ADJ
sc-5860	200	13	.	.	PUNCT
sc-5860	201	1	[	[	X
sc-5860	201	2	2	2	NUM
sc-5860	201	3	]	]	X
sc-5860	201	4	yavuz	yavuz	PROPN
sc-5860	201	5	,	,	PUNCT
sc-5860	201	6	a.	a.	PROPN
sc-5860	201	7	and	and	CCONJ
sc-5860	201	8	eken	eken	PROPN
sc-5860	201	9	,	,	PUNCT
sc-5860	201	10	s.	s.	PROPN
sc-5860	201	11	(	(	PUNCT
sc-5860	201	12	2023	2023	NUM
sc-5860	201	13	)	)	PUNCT
sc-5860	201	14	gold	gold	NOUN
sc-5860	201	15	returns	return	NOUN
sc-5860	201	16	prediction	prediction	NOUN
sc-5860	201	17	:	:	PUNCT
sc-5860	201	18	assessment	assessment	NOUN
sc-5860	201	19	based	base	VERB
sc-5860	201	20	on	on	ADP
sc-5860	201	21	major	major	ADJ
sc-5860	201	22	events	event	NOUN
sc-5860	201	23	.	.	PUNCT
sc-5860	202	1	eai	eai	PROPN
sc-5860	202	2	endorsed	endorse	VERB
sc-5860	202	3	transactions	transaction	NOUN
sc-5860	202	4	on	on	ADP
sc-5860	202	5	scalable	scalable	ADJ
sc-5860	202	6	information	information	NOUN
sc-5860	202	7	systems	system	NOUN
sc-5860	202	8	10(5	10(5	NOUN
sc-5860	202	9	)	)	PUNCT
sc-5860	202	10	.	.	PUNCT
sc-5860	203	1	[	[	X
sc-5860	203	2	3	3	NUM
sc-5860	203	3	]	]	X
sc-5860	203	4	balta	balta	PROPN
sc-5860	203	5	kaç	kaç	PROPN
sc-5860	203	6	,	,	PUNCT
sc-5860	203	7	s.	s.	PROPN
sc-5860	203	8	and	and	CCONJ
sc-5860	203	9	eken	eken	PROPN
sc-5860	203	10	,	,	PUNCT
sc-5860	203	11	s.	s.	PROPN
sc-5860	203	12	(	(	PUNCT
sc-5860	203	13	2023	2023	NUM
sc-5860	203	14	)	)	PUNCT
sc-5860	203	15	customer	customer	NOUN
sc-5860	203	16	complaintsbased	complaintsbase	VERB
sc-5860	203	17	water	water	NOUN
sc-5860	203	18	quality	quality	NOUN
sc-5860	203	19	analysis	analysis	NOUN
sc-5860	203	20	.	.	PUNCT
sc-5860	204	1	water	water	NOUN
sc-5860	204	2	15(18	15(18	NUM
sc-5860	204	3	):	):	PUNCT
sc-5860	204	4	3171	3171	NUM
sc-5860	204	5	.	.	PUNCT
sc-5860	205	1	[	[	X
sc-5860	205	2	4	4	X
sc-5860	205	3	]	]	PUNCT
sc-5860	205	4	yurtsever	yurtsever	NOUN
sc-5860	205	5	,	,	PUNCT
sc-5860	205	6	m.m.e	m.m.e	PROPN
sc-5860	205	7	.	.	PROPN
sc-5860	205	8	,	,	PUNCT
sc-5860	205	9	shiraz	shiraz	PROPN
sc-5860	205	10	,	,	PUNCT
sc-5860	205	11	m.	m.	NOUN
sc-5860	205	12	,	,	PUNCT
sc-5860	205	13	ekinci	ekinci	PROPN
sc-5860	205	14	,	,	PUNCT
sc-5860	205	15	e.	e.	PROPN
sc-5860	205	16	and	and	CCONJ
sc-5860	205	17	eken	eken	PROPN
sc-5860	205	18	,	,	PUNCT
sc-5860	205	19	s.	s.	PROPN
sc-5860	205	20	(	(	PUNCT
sc-5860	205	21	2023	2023	NUM
sc-5860	205	22	)	)	PUNCT
sc-5860	205	23	comparing	compare	VERB
sc-5860	205	24	covid-19	covid-19	PROPN
sc-5860	205	25	vaccine	vaccine	NOUN
sc-5860	205	26	passports	passport	VERB
sc-5860	205	27	attitudes	attitude	NOUN
sc-5860	205	28	across	across	ADP
sc-5860	205	29	countries	country	NOUN
sc-5860	205	30	by	by	ADP
sc-5860	205	31	analysing	analyse	VERB
sc-5860	205	32	reddit	reddit	NOUN
sc-5860	205	33	comments	comment	NOUN
sc-5860	205	34	.	.	PUNCT
sc-5860	206	1	journal	journal	NOUN
sc-5860	206	2	of	of	ADP
sc-5860	206	3	information	information	NOUN
sc-5860	206	4	science	science	NOUN
sc-5860	206	5	:	:	PUNCT
sc-5860	206	6	01655515221148356	01655515221148356	NUM
sc-5860	206	7	.	.	PUNCT
sc-5860	207	1	[	[	X
sc-5860	207	2	5	5	NUM
sc-5860	207	3	]	]	PUNCT
sc-5860	207	4	yurtsever	yurtsever	NOUN
sc-5860	207	5	,	,	PUNCT
sc-5860	207	6	m.m.e	m.m.e	PROPN
sc-5860	207	7	.	.	PROPN
sc-5860	207	8	,	,	PUNCT
sc-5860	207	9	ekinci	ekinci	PROPN
sc-5860	207	10	,	,	PUNCT
sc-5860	207	11	e.	e.	PROPN
sc-5860	207	12	and	and	CCONJ
sc-5860	207	13	eken	eken	PROPN
sc-5860	207	14	,	,	PUNCT
sc-5860	207	15	s.	s.	PROPN
sc-5860	207	16	(	(	PUNCT
sc-5860	207	17	2023	2023	NUM
sc-5860	207	18	)	)	PUNCT
sc-5860	207	19	covid-19	covid-19	PROPN
sc-5860	207	20	and	and	CCONJ
sc-5860	207	21	behavioral	behavioral	ADJ
sc-5860	207	22	analytics	analytic	NOUN
sc-5860	207	23	:	:	PUNCT
sc-5860	207	24	deep	deep	ADJ
sc-5860	207	25	learning	learning	NOUN
sc-5860	207	26	-	-	PUNCT
sc-5860	207	27	based	base	VERB
sc-5860	207	28	work	work	NOUN
sc-5860	207	29	-	-	PUNCT
sc-5860	207	30	from	from	ADP
sc-5860	207	31	-	-	PUNCT
sc-5860	207	32	home	home	NOUN
sc-5860	207	33	sensing	sensing	NOUN
sc-5860	207	34	from	from	ADP
sc-5860	207	35	reddit	reddit	NOUN
sc-5860	207	36	comments	comment	NOUN
sc-5860	207	37	.	.	PUNCT
sc-5860	208	1	in	in	ADP
sc-5860	208	2	international	international	ADJ
sc-5860	208	3	conference	conference	NOUN
sc-5860	208	4	on	on	ADP
sc-5860	208	5	computing	computing	NOUN
sc-5860	208	6	,	,	PUNCT
sc-5860	208	7	intelligence	intelligence	NOUN
sc-5860	208	8	and	and	CCONJ
sc-5860	208	9	data	datum	NOUN
sc-5860	208	10	analytics	analytic	NOUN
sc-5860	208	11	(	(	PUNCT
sc-5860	208	12	springer	springer	NOUN
sc-5860	208	13	):	):	PUNCT
sc-5860	208	14	143–155	143–155	NUM
sc-5860	208	15	.	.	PUNCT
sc-5860	209	1	[	[	X
sc-5860	209	2	6	6	NUM
sc-5860	209	3	]	]	X
sc-5860	209	4	köroğlu	köroğlu	PROPN
sc-5860	209	5	,	,	PUNCT
sc-5860	209	6	f.e	f.e	PROPN
sc-5860	209	7	.	.	PROPN
sc-5860	209	8	,	,	PUNCT
sc-5860	209	9	çakmak	çakmak	PROPN
sc-5860	209	10	,	,	PUNCT
sc-5860	209	11	s.	s.	PROPN
sc-5860	209	12	,	,	PUNCT
sc-5860	209	13	yurtsever	yurtsever	NOUN
sc-5860	209	14	,	,	PUNCT
sc-5860	209	15	m.m.e	m.m.e	PROPN
sc-5860	209	16	.	.	PUNCT
sc-5860	209	17	and	and	CCONJ
sc-5860	209	18	eken	eken	VERB
sc-5860	209	19	,	,	PUNCT
sc-5860	209	20	s.	s.	PROPN
sc-5860	209	21	(	(	PUNCT
sc-5860	209	22	2024	2024	NUM
sc-5860	209	23	)	)	PUNCT
sc-5860	209	24	smart	smart	ADJ
sc-5860	209	25	waste	waste	NOUN
sc-5860	209	26	management	management	NOUN
sc-5860	209	27	:	:	PUNCT
sc-5860	209	28	a	a	DET
sc-5860	209	29	case	case	NOUN
sc-5860	209	30	study	study	NOUN
sc-5860	209	31	on	on	ADP
sc-5860	209	32	garbage	garbage	NOUN
sc-5860	209	33	container	container	NOUN
sc-5860	209	34	detection	detection	NOUN
sc-5860	209	35	.	.	PUNCT
sc-5860	210	1	in	in	ADP
sc-5860	210	2	6th	6th	ADJ
sc-5860	210	3	mediterranean	mediterranean	PROPN
sc-5860	210	4	conference	conference	NOUN
sc-5860	210	5	on	on	ADP
sc-5860	210	6	pattern	pattern	NOUN
sc-5860	210	7	recognition	recognition	NOUN
sc-5860	210	8	and	and	CCONJ
sc-5860	210	9	artificial	artificial	ADJ
sc-5860	210	10	intelligence	intelligence	NOUN
sc-5860	210	11	,	,	PUNCT
sc-5860	210	12	medprai	medprai	PROPN
sc-5860	210	13	2024	2024	NUM
sc-5860	210	14	,	,	PUNCT
sc-5860	210	15	i̇stanbul	i̇stanbul	PROPN
sc-5860	210	16	,	,	PUNCT
sc-5860	210	17	türkiye	türkiye	PROPN
sc-5860	210	18	,	,	PUNCT
sc-5860	210	19	october	october	PROPN
sc-5860	210	20	18	18	NUM
sc-5860	210	21	-	-	SYM
sc-5860	210	22	19	19	NUM
sc-5860	210	23	,	,	PUNCT
sc-5860	210	24	2024	2024	NUM
sc-5860	210	25	,	,	PUNCT
sc-5860	210	26	proceedings	proceeding	NOUN
sc-5860	210	27	(	(	PUNCT
sc-5860	210	28	springer	springer	NOUN
sc-5860	210	29	)	)	PUNCT
sc-5860	210	30	.	.	PUNCT
sc-5860	211	1	[	[	X
sc-5860	211	2	7	7	X
sc-5860	211	3	]	]	X
sc-5860	211	4	medhat	medhat	PROPN
sc-5860	211	5	,	,	PUNCT
sc-5860	211	6	w.	w.	PROPN
sc-5860	211	7	,	,	PUNCT
sc-5860	211	8	hassan	hassan	PROPN
sc-5860	211	9	,	,	PUNCT
sc-5860	211	10	a.	a.	NOUN
sc-5860	211	11	and	and	CCONJ
sc-5860	211	12	korashy	korashy	PROPN
sc-5860	211	13	,	,	PUNCT
sc-5860	211	14	h.	h.	PROPN
sc-5860	211	15	(	(	PUNCT
sc-5860	211	16	2014	2014	NUM
sc-5860	211	17	)	)	PUNCT
sc-5860	211	18	sentiment	sentiment	NOUN
sc-5860	211	19	analysis	analysis	NOUN
sc-5860	211	20	algorithms	algorithm	NOUN
sc-5860	211	21	and	and	CCONJ
sc-5860	211	22	applications	application	NOUN
sc-5860	211	23	:	:	PUNCT
sc-5860	211	24	a	a	DET
sc-5860	211	25	survey	survey	NOUN
sc-5860	211	26	.	.	PUNCT
sc-5860	212	1	ain	ain	PROPN
sc-5860	212	2	shams	sham	VERB
sc-5860	212	3	engineering	engineering	PROPN
sc-5860	212	4	journal	journal	PROPN
sc-5860	212	5	5(4	5(4	NOUN
sc-5860	212	6	):	):	PUNCT
sc-5860	212	7	1093–1113	1093–1113	NUM
sc-5860	213	1	.	.	PUNCT
sc-5860	214	1	[	[	X
sc-5860	214	2	8	8	NUM
sc-5860	214	3	]	]	X
sc-5860	214	4	alam	alam	PROPN
sc-5860	214	5	,	,	PUNCT
sc-5860	214	6	f.	f.	PROPN
sc-5860	214	7	,	,	PUNCT
sc-5860	214	8	ofli	ofli	PROPN
sc-5860	214	9	,	,	PUNCT
sc-5860	214	10	f.	f.	PROPN
sc-5860	214	11	and	and	CCONJ
sc-5860	214	12	imran	imran	PROPN
sc-5860	214	13	,	,	PUNCT
sc-5860	214	14	m.	m.	NOUN
sc-5860	214	15	(	(	PUNCT
sc-5860	214	16	2018	2018	NUM
sc-5860	214	17	)	)	PUNCT
sc-5860	214	18	crisismmd	crisismmd	NOUN
sc-5860	214	19	:	:	PUNCT
sc-5860	214	20	multimodal	multimodal	ADJ
sc-5860	214	21	twitter	twitter	NOUN
sc-5860	214	22	datasets	dataset	NOUN
sc-5860	214	23	from	from	ADP
sc-5860	214	24	natural	natural	ADJ
sc-5860	214	25	disasters	disaster	NOUN
sc-5860	214	26	.	.	PUNCT
sc-5860	215	1	in	in	ADP
sc-5860	215	2	proceedings	proceeding	NOUN
sc-5860	215	3	of	of	ADP
sc-5860	215	4	the	the	DET
sc-5860	215	5	international	international	ADJ
sc-5860	215	6	aaai	aaai	PROPN
sc-5860	215	7	conference	conference	PROPN
sc-5860	215	8	on	on	ADP
sc-5860	215	9	web	web	NOUN
sc-5860	215	10	and	and	CCONJ
sc-5860	215	11	social	social	ADJ
sc-5860	215	12	media	medium	NOUN
sc-5860	215	13	,	,	PUNCT
sc-5860	215	14	12	12	NUM
sc-5860	215	15	.	.	PUNCT
sc-5860	216	1	[	[	X
sc-5860	216	2	9	9	NUM
sc-5860	216	3	]	]	SYM
sc-5860	216	4	kaç	kaç	X
sc-5860	216	5	,	,	PUNCT
sc-5860	216	6	s.b	s.b	PROPN
sc-5860	216	7	.	.	PROPN
sc-5860	216	8	,	,	PUNCT
sc-5860	216	9	eken	eken	VERB
sc-5860	216	10	,	,	PUNCT
sc-5860	216	11	s.	s.	PROPN
sc-5860	216	12	,	,	PUNCT
sc-5860	216	13	balta	balta	PROPN
sc-5860	216	14	,	,	PUNCT
sc-5860	216	15	d.d	d.d	PROPN
sc-5860	216	16	.	.	PROPN
sc-5860	216	17	,	,	PUNCT
sc-5860	216	18	balta	balta	PROPN
sc-5860	216	19	,	,	PUNCT
sc-5860	216	20	m.	m.	NOUN
sc-5860	216	21	,	,	PUNCT
sc-5860	216	22	i̇skefiyeli	i̇skefiyeli	NOUN
sc-5860	216	23	,	,	PUNCT
sc-5860	216	24	m.	m.	NOUN
sc-5860	216	25	and	and	CCONJ
sc-5860	216	26	özçelik	özçelik	NOUN
sc-5860	216	27	,	,	PUNCT
sc-5860	216	28	i̇.	i̇.	PROPN
sc-5860	216	29	(	(	PUNCT
sc-5860	216	30	2024	2024	NUM
sc-5860	216	31	)	)	PUNCT
sc-5860	216	32	image	image	NOUN
sc-5860	216	33	-	-	PUNCT
sc-5860	216	34	based	base	VERB
sc-5860	216	35	security	security	NOUN
sc-5860	216	36	techniques	technique	NOUN
sc-5860	216	37	for	for	ADP
sc-5860	216	38	water	water	NOUN
sc-5860	216	39	critical	critical	ADJ
sc-5860	216	40	infrastructure	infrastructure	NOUN
sc-5860	216	41	surveillance	surveillance	NOUN
sc-5860	216	42	.	.	PUNCT
sc-5860	216	43	applied	apply	VERB
sc-5860	216	44	soft	soft	ADJ
sc-5860	216	45	computing	computing	NOUN
sc-5860	216	46	161	161	NUM
sc-5860	216	47	:	:	SYM
sc-5860	216	48	111730	111730	NUM
sc-5860	216	49	.	.	PUNCT
sc-5860	217	1	[	[	X
sc-5860	217	2	10	10	NUM
sc-5860	217	3	]	]	X
sc-5860	217	4	jayagopal	jayagopal	ADJ
sc-5860	217	5	,	,	PUNCT
sc-5860	217	6	a.	a.	NOUN
sc-5860	217	7	,	,	PUNCT
sc-5860	217	8	aiswarya	aiswarya	ADJ
sc-5860	217	9	,	,	PUNCT
sc-5860	217	10	a.m.	a.m.	ADV
sc-5860	217	11	,	,	PUNCT
sc-5860	217	12	garg	garg	NOUN
sc-5860	217	13	,	,	PUNCT
sc-5860	217	14	a.	a.	NOUN
sc-5860	217	15	and	and	CCONJ
sc-5860	217	16	nandakumar	nandakumar	PROPN
sc-5860	217	17	,	,	PUNCT
sc-5860	217	18	s.k	s.k	PROPN
sc-5860	217	19	.	.	PROPN
sc-5860	217	20	(	(	PUNCT
sc-5860	217	21	2022	2022	NUM
sc-5860	217	22	)	)	PUNCT
sc-5860	217	23	multimodal	multimodal	NOUN
sc-5860	217	24	representation	representation	NOUN
sc-5860	217	25	learning	learn	VERB
sc-5860	217	26	with	with	ADP
sc-5860	217	27	text	text	NOUN
sc-5860	217	28	and	and	CCONJ
sc-5860	217	29	images	image	NOUN
sc-5860	217	30	.	.	PUNCT
sc-5860	218	1	arxiv	arxiv	PROPN
sc-5860	218	2	preprint	preprint	VERB
sc-5860	218	3	arxiv:2205.00142	arxiv:2205.00142	PRON
sc-5860	218	4	.	.	PUNCT
sc-5860	219	1	[	[	X
sc-5860	219	2	11	11	NUM
sc-5860	219	3	]	]	X
sc-5860	219	4	zhang	zhang	PROPN
sc-5860	219	5	,	,	PUNCT
sc-5860	219	6	y.	y.	PROPN
sc-5860	219	7	,	,	PUNCT
sc-5860	219	8	jiang	jiang	PROPN
sc-5860	219	9	,	,	PUNCT
sc-5860	219	10	h.	h.	PROPN
sc-5860	219	11	,	,	PUNCT
sc-5860	219	12	miura	miura	PROPN
sc-5860	219	13	,	,	PUNCT
sc-5860	219	14	y.	y.	PROPN
sc-5860	219	15	,	,	PUNCT
sc-5860	219	16	manning	manning	NOUN
sc-5860	219	17	,	,	PUNCT
sc-5860	219	18	c.d	c.d	PROPN
sc-5860	219	19	.	.	PROPN
sc-5860	219	20	and	and	CCONJ
sc-5860	219	21	langlotz	langlotz	PROPN
sc-5860	219	22	,	,	PUNCT
sc-5860	219	23	c.p	c.p	PROPN
sc-5860	219	24	.	.	PROPN
sc-5860	219	25	(	(	PUNCT
sc-5860	219	26	2022	2022	NUM
sc-5860	219	27	)	)	PUNCT
sc-5860	219	28	contrastive	contrastive	ADJ
sc-5860	219	29	learning	learning	NOUN
sc-5860	219	30	of	of	ADP
sc-5860	219	31	medical	medical	ADJ
sc-5860	219	32	visual	visual	ADJ
sc-5860	219	33	representations	representation	NOUN
sc-5860	219	34	from	from	ADP
sc-5860	219	35	paired	pair	VERB
sc-5860	219	36	images	image	NOUN
sc-5860	219	37	and	and	CCONJ
sc-5860	219	38	text	text	NOUN
sc-5860	219	39	.	.	PUNCT
sc-5860	220	1	in	in	ADP
sc-5860	220	2	machine	machine	NOUN
sc-5860	220	3	learning	learn	VERB
sc-5860	220	4	for	for	ADP
sc-5860	220	5	healthcare	healthcare	PROPN
sc-5860	220	6	conference	conference	NOUN
sc-5860	220	7	(	(	PUNCT
sc-5860	220	8	pmlr	pmlr	NOUN
sc-5860	220	9	):	):	PUNCT
sc-5860	220	10	2	2	NUM
sc-5860	220	11	–	–	PUNCT
sc-5860	220	12	25	25	NUM
sc-5860	220	13	.	.	PUNCT
sc-5860	221	1	[	[	X
sc-5860	221	2	12	12	NUM
sc-5860	221	3	]	]	X
sc-5860	221	4	radford	radford	PROPN
sc-5860	221	5	,	,	PUNCT
sc-5860	221	6	a.	a.	PROPN
sc-5860	221	7	,	,	PUNCT
sc-5860	221	8	kim	kim	PROPN
sc-5860	221	9	,	,	PUNCT
sc-5860	221	10	j.w	j.w	PROPN
sc-5860	221	11	.	.	PROPN
sc-5860	221	12	,	,	PUNCT
sc-5860	221	13	hallacy	hallacy	PROPN
sc-5860	221	14	,	,	PUNCT
sc-5860	221	15	c.	c.	PROPN
sc-5860	221	16	,	,	PUNCT
sc-5860	221	17	ramesh	ramesh	PROPN
sc-5860	221	18	,	,	PUNCT
sc-5860	221	19	a.	a.	PROPN
sc-5860	221	20	,	,	PUNCT
sc-5860	221	21	goh	goh	PROPN
sc-5860	221	22	,	,	PUNCT
sc-5860	221	23	g.	g.	PROPN
sc-5860	221	24	,	,	PUNCT
sc-5860	221	25	agarwal	agarwal	PROPN
sc-5860	221	26	,	,	PUNCT
sc-5860	221	27	s.	s.	PROPN
sc-5860	221	28	,	,	PUNCT
sc-5860	221	29	sastry	sastry	PROPN
sc-5860	221	30	,	,	PUNCT
sc-5860	221	31	g.	g.	PROPN
sc-5860	221	32	et	et	PROPN
sc-5860	221	33	al	al	PROPN
sc-5860	221	34	.	.	PROPN
sc-5860	222	1	(	(	PUNCT
sc-5860	222	2	2021	2021	NUM
sc-5860	222	3	)	)	PUNCT
sc-5860	222	4	learning	learn	VERB
sc-5860	222	5	transferable	transferable	ADJ
sc-5860	222	6	visual	visual	ADJ
sc-5860	222	7	models	model	NOUN
sc-5860	222	8	from	from	ADP
sc-5860	222	9	natural	natural	ADJ
sc-5860	222	10	language	language	NOUN
sc-5860	222	11	supervision	supervision	NOUN
sc-5860	222	12	.	.	PUNCT
sc-5860	223	1	in	in	ADP
sc-5860	223	2	international	international	ADJ
sc-5860	223	3	conference	conference	NOUN
sc-5860	223	4	on	on	ADP
sc-5860	223	5	machine	machine	NOUN
sc-5860	223	6	learning	learning	NOUN
sc-5860	223	7	(	(	PUNCT
sc-5860	223	8	pmlr	pmlr	NOUN
sc-5860	223	9	):	):	PUNCT
sc-5860	223	10	8748–8763	8748–8763	NUM
sc-5860	223	11	.	.	PUNCT
sc-5860	224	1	[	[	X
sc-5860	224	2	13	13	NUM
sc-5860	224	3	]	]	X
sc-5860	224	4	lin	lin	PROPN
sc-5860	224	5	,	,	PUNCT
sc-5860	224	6	z.	z.	PROPN
sc-5860	224	7	,	,	PUNCT
sc-5860	224	8	bas	bas	PROPN
sc-5860	224	9	,	,	PUNCT
sc-5860	224	10	e.	e.	PROPN
sc-5860	224	11	,	,	PUNCT
sc-5860	224	12	singh	singh	PROPN
sc-5860	224	13	,	,	PUNCT
sc-5860	224	14	k.y	k.y	PROPN
sc-5860	224	15	.	.	PROPN
sc-5860	224	16	,	,	PUNCT
sc-5860	224	17	swaminathan	swaminathan	ADV
sc-5860	224	18	,	,	PUNCT
sc-5860	224	19	g.	g.	PROPN
sc-5860	224	20	and	and	CCONJ
sc-5860	224	21	bhotika	bhotika	PROPN
sc-5860	224	22	,	,	PUNCT
sc-5860	224	23	r.	r.	PROPN
sc-5860	224	24	(	(	PUNCT
sc-5860	224	25	2023	2023	NUM
sc-5860	224	26	)	)	PUNCT
sc-5860	224	27	relaxing	relax	VERB
sc-5860	224	28	contrastiveness	contrastiveness	NOUN
sc-5860	224	29	in	in	ADP
sc-5860	224	30	multimodal	multimodal	NOUN
sc-5860	224	31	representation	representation	NOUN
sc-5860	224	32	learning	learning	NOUN
sc-5860	224	33	.	.	PUNCT
sc-5860	225	1	in	in	ADP
sc-5860	225	2	proceedings	proceeding	NOUN
sc-5860	225	3	of	of	ADP
sc-5860	225	4	the	the	DET
sc-5860	225	5	ieee	ieee	NOUN
sc-5860	225	6	/	/	SYM
sc-5860	225	7	cvf	cvf	NOUN
sc-5860	225	8	winter	winter	NOUN
sc-5860	225	9	conference	conference	NOUN
sc-5860	225	10	on	on	ADP
sc-5860	225	11	applications	application	NOUN
sc-5860	225	12	of	of	ADP
sc-5860	225	13	computer	computer	NOUN
sc-5860	225	14	vision	vision	NOUN
sc-5860	225	15	:	:	PUNCT
sc-5860	225	16	2227–2236	2227–2236	NUM
sc-5860	225	17	.	.	PUNCT
sc-5860	226	1	[	[	X
sc-5860	226	2	14	14	NUM
sc-5860	226	3	]	]	X
sc-5860	226	4	cheema	cheema	PROPN
sc-5860	226	5	,	,	PUNCT
sc-5860	226	6	g.s	g.s	PROPN
sc-5860	226	7	.	.	PROPN
sc-5860	226	8	,	,	PUNCT
sc-5860	226	9	hakimov	hakimov	PROPN
sc-5860	226	10	,	,	PUNCT
sc-5860	226	11	s.	s.	PROPN
sc-5860	226	12	,	,	PUNCT
sc-5860	226	13	müller	müller	PROPN
sc-5860	226	14	-	-	PUNCT
sc-5860	226	15	budack	budack	NOUN
sc-5860	226	16	,	,	PUNCT
sc-5860	226	17	e.	e.	PROPN
sc-5860	226	18	and	and	CCONJ
sc-5860	226	19	ewerth	ewerth	PROPN
sc-5860	226	20	,	,	PUNCT
sc-5860	226	21	r.	r.	PROPN
sc-5860	226	22	(	(	PUNCT
sc-5860	226	23	2021	2021	NUM
sc-5860	226	24	)	)	PUNCT
sc-5860	226	25	a	a	DET
sc-5860	226	26	fair	fair	ADJ
sc-5860	226	27	and	and	CCONJ
sc-5860	226	28	comprehensive	comprehensive	ADJ
sc-5860	226	29	comparison	comparison	NOUN
sc-5860	226	30	of	of	ADP
sc-5860	226	31	multimodal	multimodal	NOUN
sc-5860	226	32	tweet	tweet	NOUN
sc-5860	226	33	sentiment	sentiment	NOUN
sc-5860	226	34	analysis	analysis	NOUN
sc-5860	226	35	methods	method	NOUN
sc-5860	226	36	.	.	PUNCT
sc-5860	227	1	in	in	ADP
sc-5860	227	2	proceedings	proceeding	NOUN
sc-5860	227	3	of	of	ADP
sc-5860	227	4	the	the	DET
sc-5860	227	5	2021	2021	NUM
sc-5860	227	6	workshop	workshop	NOUN
sc-5860	227	7	on	on	ADP
sc-5860	227	8	multi	multi	ADJ
sc-5860	227	9	-	-	ADJ
sc-5860	227	10	modal	modal	ADJ
sc-5860	227	11	pretraining	pretraining	NOUN
sc-5860	227	12	for	for	ADP
sc-5860	227	13	multimedia	multimedia	NOUN
sc-5860	227	14	understanding	understanding	NOUN
sc-5860	227	15	:	:	PUNCT
sc-5860	227	16	37–45	37–45	NUM
sc-5860	227	17	.	.	PUNCT
sc-5860	228	1	[	[	X
sc-5860	228	2	15	15	NUM
sc-5860	228	3	]	]	X
sc-5860	228	4	thelwall	thelwall	NOUN
sc-5860	228	5	,	,	PUNCT
sc-5860	228	6	m.	m.	NOUN
sc-5860	228	7	,	,	PUNCT
sc-5860	228	8	buckley	buckley	NOUN
sc-5860	228	9	,	,	PUNCT
sc-5860	228	10	k.	k.	PROPN
sc-5860	228	11	,	,	PUNCT
sc-5860	228	12	paltoglou	paltoglou	PROPN
sc-5860	228	13	,	,	PUNCT
sc-5860	228	14	g.	g.	PROPN
sc-5860	228	15	,	,	PUNCT
sc-5860	228	16	cai	cai	PROPN
sc-5860	228	17	,	,	PUNCT
sc-5860	228	18	d.	d.	PROPN
sc-5860	228	19	and	and	CCONJ
sc-5860	228	20	kappas	kappas	PROPN
sc-5860	228	21	,	,	PUNCT
sc-5860	228	22	a.	a.	NOUN
sc-5860	228	23	(	(	PUNCT
sc-5860	228	24	2010	2010	NUM
sc-5860	228	25	)	)	PUNCT
sc-5860	228	26	sentiment	sentiment	NOUN
sc-5860	228	27	strength	strength	NOUN
sc-5860	228	28	detection	detection	NOUN
sc-5860	228	29	in	in	ADP
sc-5860	228	30	short	short	ADJ
sc-5860	228	31	informal	informal	ADJ
sc-5860	228	32	text	text	NOUN
sc-5860	228	33	.	.	PUNCT
sc-5860	229	1	journal	journal	NOUN
sc-5860	229	2	of	of	ADP
sc-5860	229	3	the	the	DET
sc-5860	229	4	american	american	ADJ
sc-5860	229	5	society	society	NOUN
sc-5860	229	6	for	for	ADP
sc-5860	229	7	information	information	NOUN
sc-5860	229	8	science	science	NOUN
sc-5860	229	9	and	and	CCONJ
sc-5860	229	10	technology	technology	NOUN
sc-5860	229	11	61(12	61(12	NOUN
sc-5860	229	12	):	):	PUNCT
sc-5860	229	13	2544–2558	2544–2558	NUM
sc-5860	229	14	.	.	PUNCT
sc-5860	230	1	[	[	X
sc-5860	230	2	16	16	NUM
sc-5860	230	3	]	]	X
sc-5860	230	4	saif	saif	PROPN
sc-5860	230	5	,	,	PUNCT
sc-5860	230	6	h.	h.	PROPN
sc-5860	230	7	,	,	PUNCT
sc-5860	230	8	bashevoy	bashevoy	PROPN
sc-5860	230	9	,	,	PUNCT
sc-5860	230	10	m.	m.	NOUN
sc-5860	230	11	,	,	PUNCT
sc-5860	230	12	taylor	taylor	PROPN
sc-5860	230	13	,	,	PUNCT
sc-5860	230	14	s.	s.	PROPN
sc-5860	230	15	,	,	PUNCT
sc-5860	230	16	fernandez	fernandez	PROPN
sc-5860	230	17	,	,	PUNCT
sc-5860	230	18	m.	m.	NOUN
sc-5860	230	19	and	and	CCONJ
sc-5860	230	20	alani	alani	PROPN
sc-5860	230	21	,	,	PUNCT
sc-5860	230	22	h.	h.	PROPN
sc-5860	230	23	(	(	PUNCT
sc-5860	230	24	2016	2016	NUM
sc-5860	230	25	)	)	PUNCT
sc-5860	230	26	senticircles	senticircle	NOUN
sc-5860	230	27	:	:	PUNCT
sc-5860	230	28	a	a	DET
sc-5860	230	29	platform	platform	NOUN
sc-5860	230	30	for	for	ADP
sc-5860	230	31	contextual	contextual	ADJ
sc-5860	230	32	and	and	CCONJ
sc-5860	230	33	conceptual	conceptual	ADJ
sc-5860	230	34	sentiment	sentiment	NOUN
sc-5860	230	35	analysis	analysis	NOUN
sc-5860	230	36	.	.	PUNCT
sc-5860	231	1	in	in	ADP
sc-5860	231	2	the	the	DET
sc-5860	231	3	semantic	semantic	ADJ
sc-5860	231	4	web	web	NOUN
sc-5860	231	5	:	:	PUNCT
sc-5860	231	6	eswc	eswc	PROPN
sc-5860	231	7	2016	2016	NUM
sc-5860	231	8	satellite	satellite	NOUN
sc-5860	231	9	events	event	NOUN
sc-5860	231	10	,	,	PUNCT
sc-5860	231	11	heraklion	heraklion	NOUN
sc-5860	231	12	,	,	PUNCT
sc-5860	231	13	crete	crete	PROPN
sc-5860	231	14	,	,	PUNCT
sc-5860	231	15	greece	greece	PROPN
sc-5860	231	16	,	,	PUNCT
sc-5860	231	17	may	may	AUX
sc-5860	231	18	29	29	NUM
sc-5860	231	19	–	–	PUNCT
sc-5860	231	20	june	june	PROPN
sc-5860	231	21	2	2	NUM
sc-5860	231	22	,	,	PUNCT
sc-5860	231	23	2016	2016	NUM
sc-5860	231	24	,	,	PUNCT
sc-5860	231	25	revised	revise	VERB
sc-5860	231	26	selected	select	VERB
sc-5860	231	27	papers	paper	NOUN
sc-5860	231	28	13	13	NUM
sc-5860	231	29	(	(	PUNCT
sc-5860	231	30	springer	springer	NOUN
sc-5860	231	31	):	):	PUNCT
sc-5860	231	32	140–145	140–145	NUM
sc-5860	231	33	.	.	PUNCT
sc-5860	232	1	[	[	X
sc-5860	232	2	17	17	NUM
sc-5860	232	3	]	]	X
sc-5860	232	4	you	you	PRON
sc-5860	232	5	,	,	PUNCT
sc-5860	232	6	q.	q.	PROPN
sc-5860	232	7	,	,	PUNCT
sc-5860	232	8	luo	luo	PROPN
sc-5860	232	9	,	,	PUNCT
sc-5860	232	10	j.	j.	PROPN
sc-5860	232	11	,	,	PUNCT
sc-5860	232	12	jin	jin	PROPN
sc-5860	232	13	,	,	PUNCT
sc-5860	232	14	h.	h.	PROPN
sc-5860	232	15	and	and	CCONJ
sc-5860	232	16	yang	yang	PROPN
sc-5860	232	17	,	,	PUNCT
sc-5860	232	18	j.	j.	PROPN
sc-5860	232	19	(	(	PUNCT
sc-5860	232	20	2015	2015	NUM
sc-5860	232	21	)	)	PUNCT
sc-5860	232	22	robust	robust	ADJ
sc-5860	232	23	image	image	NOUN
sc-5860	232	24	sentiment	sentiment	NOUN
sc-5860	232	25	analysis	analysis	NOUN
sc-5860	232	26	using	use	VERB
sc-5860	232	27	progressively	progressively	ADV
sc-5860	232	28	trained	train	VERB
sc-5860	232	29	and	and	CCONJ
sc-5860	232	30	domain	domain	NOUN
sc-5860	232	31	transferred	transfer	VERB
sc-5860	232	32	deep	deep	ADJ
sc-5860	232	33	networks	network	NOUN
sc-5860	232	34	.	.	PUNCT
sc-5860	233	1	in	in	ADP
sc-5860	233	2	proceedings	proceeding	NOUN
sc-5860	233	3	of	of	ADP
sc-5860	233	4	the	the	DET
sc-5860	233	5	aaai	aaai	PROPN
sc-5860	233	6	conference	conference	NOUN
sc-5860	233	7	on	on	ADP
sc-5860	233	8	artificial	artificial	ADJ
sc-5860	233	9	intelligence	intelligence	NOUN
sc-5860	233	10	,	,	PUNCT
sc-5860	233	11	29	29	NUM
sc-5860	233	12	.	.	PUNCT
sc-5860	234	1	[	[	X
sc-5860	234	2	18	18	NUM
sc-5860	234	3	]	]	PUNCT
sc-5860	234	4	you	you	PRON
sc-5860	234	5	,	,	PUNCT
sc-5860	234	6	q.	q.	PROPN
sc-5860	234	7	,	,	PUNCT
sc-5860	234	8	luo	luo	PROPN
sc-5860	234	9	,	,	PUNCT
sc-5860	234	10	j.	j.	PROPN
sc-5860	234	11	,	,	PUNCT
sc-5860	234	12	jin	jin	PROPN
sc-5860	234	13	,	,	PUNCT
sc-5860	234	14	h.	h.	PROPN
sc-5860	234	15	and	and	CCONJ
sc-5860	234	16	yang	yang	PROPN
sc-5860	234	17	,	,	PUNCT
sc-5860	234	18	j.	j.	PROPN
sc-5860	234	19	(	(	PUNCT
sc-5860	234	20	2016	2016	NUM
sc-5860	234	21	)	)	PUNCT
sc-5860	234	22	building	build	VERB
sc-5860	234	23	a	a	DET
sc-5860	234	24	large	large	ADJ
sc-5860	234	25	scale	scale	NOUN
sc-5860	234	26	dataset	dataset	NOUN
sc-5860	234	27	for	for	ADP
sc-5860	234	28	image	image	NOUN
sc-5860	234	29	emotion	emotion	NOUN
sc-5860	234	30	recognition	recognition	NOUN
sc-5860	234	31	:	:	PUNCT
sc-5860	234	32	the	the	DET
sc-5860	234	33	fine	fine	ADJ
sc-5860	234	34	print	print	NOUN
sc-5860	234	35	and	and	CCONJ
sc-5860	234	36	the	the	DET
sc-5860	234	37	benchmark	benchmark	NOUN
sc-5860	234	38	.	.	PUNCT
sc-5860	235	1	in	in	ADP
sc-5860	235	2	proceedings	proceeding	NOUN
sc-5860	235	3	of	of	ADP
sc-5860	235	4	the	the	DET
sc-5860	235	5	aaai	aaai	PROPN
sc-5860	235	6	conference	conference	NOUN
sc-5860	235	7	on	on	ADP
sc-5860	235	8	artificial	artificial	ADJ
sc-5860	235	9	intelligence	intelligence	NOUN
sc-5860	235	10	,	,	PUNCT
sc-5860	235	11	30	30	NUM
sc-5860	235	12	.	.	PUNCT
sc-5860	236	1	[	[	X
sc-5860	236	2	19	19	NUM
sc-5860	236	3	]	]	X
sc-5860	236	4	jiang	jiang	PROPN
sc-5860	236	5	,	,	PUNCT
sc-5860	236	6	t.	t.	PROPN
sc-5860	236	7	,	,	PUNCT
sc-5860	236	8	wang	wang	PROPN
sc-5860	236	9	,	,	PUNCT
sc-5860	236	10	j.	j.	PROPN
sc-5860	236	11	,	,	PUNCT
sc-5860	236	12	liu	liu	PROPN
sc-5860	236	13	,	,	PUNCT
sc-5860	236	14	z.	z.	PROPN
sc-5860	236	15	and	and	CCONJ
sc-5860	236	16	ling	ling	PROPN
sc-5860	236	17	,	,	PUNCT
sc-5860	236	18	y.	y.	PROPN
sc-5860	236	19	(	(	PUNCT
sc-5860	236	20	2020	2020	NUM
sc-5860	236	21	)	)	PUNCT
sc-5860	236	22	fusionextraction	fusionextraction	NOUN
sc-5860	236	23	network	network	NOUN
sc-5860	236	24	for	for	ADP
sc-5860	236	25	multimodal	multimodal	ADJ
sc-5860	236	26	sentiment	sentiment	NOUN
sc-5860	236	27	analysis	analysis	NOUN
sc-5860	236	28	.	.	PUNCT
sc-5860	237	1	in	in	ADP
sc-5860	237	2	advances	advance	NOUN
sc-5860	237	3	in	in	ADP
sc-5860	237	4	knowledge	knowledge	NOUN
sc-5860	237	5	discovery	discovery	NOUN
sc-5860	237	6	and	and	CCONJ
sc-5860	237	7	data	datum	NOUN
sc-5860	237	8	mining	mining	NOUN
sc-5860	237	9	:	:	PUNCT
sc-5860	237	10	24th	24th	ADJ
sc-5860	237	11	pacific	pacific	PROPN
sc-5860	237	12	-	-	PUNCT
sc-5860	237	13	asia	asia	PROPN
sc-5860	237	14	conference	conference	NOUN
sc-5860	237	15	,	,	PUNCT
sc-5860	237	16	pakdd	pakdd	VERB
sc-5860	237	17	2020	2020	NUM
sc-5860	237	18	,	,	PUNCT
sc-5860	237	19	singapore	singapore	PROPN
sc-5860	237	20	,	,	PUNCT
sc-5860	237	21	may	may	PROPN
sc-5860	237	22	11–14	11–14	NUM
sc-5860	237	23	,	,	PUNCT
sc-5860	237	24	2020	2020	NUM
sc-5860	237	25	,	,	PUNCT
sc-5860	237	26	proceedings	proceeding	NOUN
sc-5860	237	27	,	,	PUNCT
sc-5860	237	28	part	part	NOUN
sc-5860	237	29	ii	ii	PROPN
sc-5860	237	30	24	24	NUM
sc-5860	237	31	(	(	PUNCT
sc-5860	237	32	springer	springer	NOUN
sc-5860	237	33	):	):	PUNCT
sc-5860	237	34	785	785	NUM
sc-5860	237	35	–	–	PUNCT
sc-5860	237	36	797	797	NUM
sc-5860	237	37	.	.	PUNCT
sc-5860	238	1	[	[	X
sc-5860	238	2	20	20	NUM
sc-5860	238	3	]	]	X
sc-5860	238	4	russakovsky	russakovsky	NOUN
sc-5860	238	5	,	,	PUNCT
sc-5860	238	6	o.	o.	PROPN
sc-5860	238	7	,	,	PUNCT
sc-5860	238	8	deng	deng	PROPN
sc-5860	238	9	,	,	PUNCT
sc-5860	238	10	j.	j.	PROPN
sc-5860	238	11	,	,	PUNCT
sc-5860	238	12	su	su	PROPN
sc-5860	238	13	,	,	PUNCT
sc-5860	238	14	h.	h.	PROPN
sc-5860	238	15	,	,	PUNCT
sc-5860	238	16	krause	krause	PROPN
sc-5860	238	17	,	,	PUNCT
sc-5860	238	18	j.	j.	PROPN
sc-5860	238	19	,	,	PUNCT
sc-5860	238	20	satheesh	satheesh	PROPN
sc-5860	238	21	,	,	PUNCT
sc-5860	238	22	s.	s.	PROPN
sc-5860	238	23	,	,	PUNCT
sc-5860	238	24	ma	ma	PROPN
sc-5860	238	25	,	,	PUNCT
sc-5860	238	26	s.	s.	PROPN
sc-5860	238	27	,	,	PUNCT
sc-5860	238	28	huang	huang	PROPN
sc-5860	238	29	,	,	PUNCT
sc-5860	238	30	z.	z.	PROPN
sc-5860	238	31	et	et	PROPN
sc-5860	238	32	al	al	PROPN
sc-5860	238	33	.	.	PROPN
sc-5860	239	1	(	(	PUNCT
sc-5860	239	2	2015	2015	NUM
sc-5860	239	3	)	)	PUNCT
sc-5860	239	4	imagenet	imagenet	NOUN
sc-5860	239	5	large	large	ADJ
sc-5860	239	6	scale	scale	NOUN
sc-5860	239	7	visual	visual	ADJ
sc-5860	239	8	recognition	recognition	NOUN
sc-5860	239	9	challenge	challenge	NOUN
sc-5860	239	10	.	.	PUNCT
sc-5860	240	1	international	international	ADJ
sc-5860	240	2	journal	journal	NOUN
sc-5860	240	3	of	of	ADP
sc-5860	240	4	computer	computer	NOUN
sc-5860	240	5	vision	vision	NOUN
sc-5860	240	6	115	115	NUM
sc-5860	240	7	:	:	PUNCT
sc-5860	240	8	211–252	211–252	NUM
sc-5860	240	9	.	.	PUNCT
sc-5860	241	1	[	[	X
sc-5860	241	2	21	21	NUM
sc-5860	241	3	]	]	X
sc-5860	241	4	he	he	PRON
sc-5860	241	5	,	,	PUNCT
sc-5860	241	6	k.	k.	PROPN
sc-5860	241	7	,	,	PUNCT
sc-5860	241	8	zhang	zhang	PROPN
sc-5860	241	9	,	,	PUNCT
sc-5860	241	10	x.	x.	PROPN
sc-5860	241	11	,	,	PUNCT
sc-5860	241	12	ren	ren	PROPN
sc-5860	241	13	,	,	PUNCT
sc-5860	241	14	s.	s.	PROPN
sc-5860	241	15	and	and	CCONJ
sc-5860	241	16	sun	sun	PROPN
sc-5860	241	17	,	,	PUNCT
sc-5860	241	18	j.	j.	PROPN
sc-5860	241	19	(	(	PUNCT
sc-5860	241	20	2016	2016	NUM
sc-5860	241	21	)	)	PUNCT
sc-5860	241	22	deep	deep	ADJ
sc-5860	241	23	residual	residual	ADJ
sc-5860	241	24	learning	learning	NOUN
sc-5860	241	25	for	for	ADP
sc-5860	241	26	image	image	NOUN
sc-5860	241	27	recognition	recognition	NOUN
sc-5860	241	28	.	.	PUNCT
sc-5860	242	1	in	in	ADP
sc-5860	242	2	proceedings	proceeding	NOUN
sc-5860	242	3	of	of	ADP
sc-5860	242	4	the	the	DET
sc-5860	242	5	ieee	ieee	NOUN
sc-5860	242	6	conference	conference	NOUN
sc-5860	242	7	on	on	ADP
sc-5860	242	8	computer	computer	NOUN
sc-5860	242	9	vision	vision	NOUN
sc-5860	242	10	and	and	CCONJ
sc-5860	242	11	pattern	pattern	NOUN
sc-5860	242	12	recognition	recognition	NOUN
sc-5860	242	13	:	:	PUNCT
sc-5860	243	1	770–778	770–778	NUM
sc-5860	243	2	.	.	PUNCT
sc-5860	244	1	[	[	X
sc-5860	244	2	22	22	NUM
sc-5860	244	3	]	]	X
sc-5860	244	4	pennington	pennington	PROPN
sc-5860	244	5	,	,	PUNCT
sc-5860	244	6	j.	j.	PROPN
sc-5860	244	7	,	,	PUNCT
sc-5860	244	8	socher	socher	PROPN
sc-5860	244	9	,	,	PUNCT
sc-5860	244	10	r.	r.	PROPN
sc-5860	244	11	and	and	CCONJ
sc-5860	244	12	manning	manning	PROPN
sc-5860	244	13	,	,	PUNCT
sc-5860	244	14	c.d	c.d	PROPN
sc-5860	244	15	.	.	PROPN
sc-5860	245	1	(	(	PUNCT
sc-5860	245	2	2014	2014	NUM
sc-5860	245	3	)	)	PUNCT
sc-5860	245	4	glove	glove	NOUN
sc-5860	245	5	:	:	PUNCT
sc-5860	245	6	global	global	ADJ
sc-5860	245	7	vectors	vector	NOUN
sc-5860	245	8	for	for	ADP
sc-5860	245	9	word	word	NOUN
sc-5860	245	10	representation	representation	NOUN
sc-5860	245	11	.	.	PUNCT
sc-5860	246	1	in	in	ADP
sc-5860	246	2	proceedings	proceeding	NOUN
sc-5860	246	3	of	of	ADP
sc-5860	246	4	the	the	DET
sc-5860	246	5	2014	2014	NUM
sc-5860	246	6	conference	conference	NOUN
sc-5860	246	7	on	on	ADP
sc-5860	246	8	empirical	empirical	ADJ
sc-5860	246	9	methods	method	NOUN
sc-5860	246	10	in	in	ADP
sc-5860	246	11	natural	natural	ADJ
sc-5860	246	12	language	language	NOUN
sc-5860	246	13	processing	processing	NOUN
sc-5860	246	14	(	(	PUNCT
sc-5860	246	15	emnlp	emnlp	ADJ
sc-5860	246	16	):	):	PUNCT
sc-5860	246	17	1532–1543	1532–1543	NUM
sc-5860	246	18	.	.	PUNCT
sc-5860	247	1	[	[	X
sc-5860	247	2	23	23	NUM
sc-5860	247	3	]	]	X
sc-5860	247	4	devlin	devlin	PROPN
sc-5860	247	5	,	,	PUNCT
sc-5860	247	6	j.	j.	PROPN
sc-5860	247	7	,	,	PUNCT
sc-5860	247	8	chang	chang	PROPN
sc-5860	247	9	,	,	PUNCT
sc-5860	247	10	m.w	m.w	PROPN
sc-5860	247	11	.	.	PROPN
sc-5860	247	12	,	,	PUNCT
sc-5860	247	13	lee	lee	PROPN
sc-5860	247	14	,	,	PUNCT
sc-5860	247	15	k.	k.	PROPN
sc-5860	247	16	and	and	CCONJ
sc-5860	247	17	toutanova	toutanova	PROPN
sc-5860	247	18	,	,	PUNCT
sc-5860	247	19	k.	k.	PROPN
sc-5860	247	20	(	(	PUNCT
sc-5860	247	21	2018	2018	NUM
sc-5860	247	22	)	)	PUNCT
sc-5860	247	23	bert	bert	PROPN
sc-5860	247	24	:	:	PUNCT
sc-5860	247	25	pre	pre	ADJ
sc-5860	247	26	-	-	NOUN
sc-5860	247	27	training	training	NOUN
sc-5860	247	28	of	of	ADP
sc-5860	247	29	deep	deep	ADJ
sc-5860	247	30	bidirectional	bidirectional	ADJ
sc-5860	247	31	transformers	transformer	NOUN
sc-5860	247	32	for	for	ADP
sc-5860	247	33	language	language	NOUN
sc-5860	247	34	understanding	understanding	NOUN
sc-5860	247	35	.	.	PUNCT
sc-5860	248	1	arxiv	arxiv	PROPN
sc-5860	248	2	preprint	preprint	VERB
sc-5860	248	3	arxiv:1810.04805	arxiv:1810.04805	PROPN
sc-5860	248	4	.	.	PUNCT
sc-5860	249	1	[	[	X
sc-5860	249	2	24	24	NUM
sc-5860	249	3	]	]	PUNCT
sc-5860	249	4	bica	bica	PROPN
sc-5860	249	5	,	,	PUNCT
sc-5860	249	6	m.	m.	NOUN
sc-5860	249	7	,	,	PUNCT
sc-5860	249	8	palen	palen	PROPN
sc-5860	249	9	,	,	PUNCT
sc-5860	249	10	l.	l.	PROPN
sc-5860	249	11	and	and	CCONJ
sc-5860	249	12	bopp	bopp	PROPN
sc-5860	249	13	,	,	PUNCT
sc-5860	249	14	c.	c.	PROPN
sc-5860	249	15	(	(	PUNCT
sc-5860	249	16	2017	2017	NUM
sc-5860	249	17	)	)	PUNCT
sc-5860	249	18	visual	visual	ADJ
sc-5860	249	19	representations	representation	NOUN
sc-5860	249	20	of	of	ADP
sc-5860	249	21	disaster	disaster	NOUN
sc-5860	249	22	.	.	PUNCT
sc-5860	250	1	in	in	ADP
sc-5860	250	2	proceedings	proceeding	NOUN
sc-5860	250	3	of	of	ADP
sc-5860	250	4	the	the	DET
sc-5860	250	5	2017	2017	NUM
sc-5860	250	6	acm	acm	NOUN
sc-5860	250	7	conference	conference	NOUN
sc-5860	250	8	on	on	ADP
sc-5860	250	9	computer	computer	NOUN
sc-5860	250	10	supported	support	VERB
sc-5860	250	11	cooperative	cooperative	ADJ
sc-5860	250	12	work	work	NOUN
sc-5860	250	13	and	and	CCONJ
sc-5860	250	14	social	social	ADJ
sc-5860	250	15	computing	computing	NOUN
sc-5860	250	16	:	:	PUNCT
sc-5860	250	17	1262–1276	1262–1276	NUM
sc-5860	250	18	.	.	PUNCT
sc-5860	251	1	[	[	X
sc-5860	251	2	25	25	NUM
sc-5860	251	3	]	]	X
sc-5860	251	4	sit	sit	PROPN
sc-5860	251	5	,	,	PUNCT
sc-5860	251	6	m.a	m.a	PROPN
sc-5860	251	7	.	.	PROPN
sc-5860	251	8	,	,	PUNCT
sc-5860	251	9	koylu	koylu	PROPN
sc-5860	251	10	,	,	PUNCT
sc-5860	251	11	c.	c.	PROPN
sc-5860	251	12	and	and	CCONJ
sc-5860	251	13	demir	demir	PROPN
sc-5860	251	14	,	,	PUNCT
sc-5860	251	15	i.	i.	PROPN
sc-5860	251	16	(	(	PUNCT
sc-5860	251	17	2020	2020	NUM
sc-5860	251	18	)	)	PUNCT
sc-5860	251	19	identifying	identify	VERB
sc-5860	251	20	disaster	disaster	NOUN
sc-5860	251	21	-	-	PUNCT
sc-5860	251	22	related	relate	VERB
sc-5860	251	23	tweets	tweet	NOUN
sc-5860	251	24	and	and	CCONJ
sc-5860	251	25	their	their	PRON
sc-5860	251	26	semantic	semantic	ADJ
sc-5860	251	27	,	,	PUNCT
sc-5860	251	28	spatial	spatial	ADJ
sc-5860	251	29	and	and	CCONJ
sc-5860	251	30	temporal	temporal	ADJ
sc-5860	251	31	context	context	NOUN
sc-5860	251	32	using	use	VERB
sc-5860	251	33	deep	deep	ADJ
sc-5860	251	34	learning	learning	NOUN
sc-5860	251	35	,	,	PUNCT
sc-5860	251	36	natural	natural	ADJ
sc-5860	251	37	language	language	NOUN
sc-5860	251	38	processing	processing	NOUN
sc-5860	251	39	and	and	CCONJ
sc-5860	251	40	spatial	spatial	ADJ
sc-5860	251	41	analysis	analysis	NOUN
sc-5860	251	42	:	:	PUNCT
sc-5860	251	43	a	a	DET
sc-5860	251	44	case	case	NOUN
sc-5860	251	45	study	study	NOUN
sc-5860	251	46	of	of	ADP
sc-5860	251	47	hurricane	hurricane	PROPN
sc-5860	251	48	irma	irma	PROPN
sc-5860	251	49	.	.	PUNCT
sc-5860	252	1	in	in	ADP
sc-5860	252	2	social	social	ADJ
sc-5860	252	3	sensing	sensing	NOUN
sc-5860	252	4	and	and	CCONJ
sc-5860	252	5	big	big	ADJ
sc-5860	252	6	data	datum	NOUN
sc-5860	252	7	computing	compute	VERB
sc-5860	252	8	for	for	ADP
sc-5860	252	9	8	8	NUM
sc-5860	252	10	eai	eai	NOUN
sc-5860	252	11	endorsed	endorse	VERB
sc-5860	252	12	transactions	transaction	NOUN
sc-5860	252	13	on	on	ADP
sc-5860	252	14	smart	smart	ADJ
sc-5860	252	15	cities	city	NOUN
sc-5860	252	16	|	|	ADV
sc-5860	252	17	volume	volume	VERB
sc-5860	252	18	7	7	NUM
sc-5860	252	19	|	|	NOUN
sc-5860	252	20	issue	issue	VERB
sc-5860	252	21	4	4	NUM
sc-5860	252	22	|	|	CCONJ
sc-5860	252	23	2023	2023	NUM
sc-5860	253	1	|	|	ADV
sc-5860	253	2	https://github.com/devsdtr/multimodal	https://github.com/devsdtr/multimodal	PROPN
sc-5860	253	3	https://github.com/devsdtr/multimodal	https://github.com/devsdtr/multimodal	PROPN
sc-5860	253	4	multimodal	multimodal	PROPN
sc-5860	253	5	sentiment	sentiment	NOUN
sc-5860	253	6	analysis	analysis	NOUN
sc-5860	253	7	in	in	ADP
sc-5860	253	8	natural	natural	ADJ
sc-5860	253	9	disaster	disaster	NOUN
sc-5860	253	10	data	datum	NOUN
sc-5860	253	11	on	on	ADP
sc-5860	253	12	social	social	ADJ
sc-5860	253	13	media	medium	NOUN
sc-5860	253	14	disaster	disaster	NOUN
sc-5860	253	15	management	management	NOUN
sc-5860	253	16	(	(	PUNCT
sc-5860	253	17	routledge	routledge	PROPN
sc-5860	253	18	)	)	PUNCT
sc-5860	253	19	,	,	PUNCT
sc-5860	253	20	8–32	8–32	NOUN
sc-5860	253	21	.	.	PUNCT
sc-5860	254	1	[	[	X
sc-5860	254	2	26	26	NUM
sc-5860	254	3	]	]	X
sc-5860	254	4	hassan	hassan	PROPN
sc-5860	254	5	,	,	PUNCT
sc-5860	254	6	s.z	s.z	PROPN
sc-5860	254	7	.	.	PROPN
sc-5860	254	8	,	,	PUNCT
sc-5860	254	9	ahmad	ahmad	PROPN
sc-5860	254	10	,	,	PUNCT
sc-5860	254	11	k.	k.	PROPN
sc-5860	254	12	,	,	PUNCT
sc-5860	254	13	hicks	hicks	PROPN
sc-5860	254	14	,	,	PUNCT
sc-5860	254	15	s.	s.	PROPN
sc-5860	254	16	,	,	PUNCT
sc-5860	254	17	halvorsen	halvorsen	PROPN
sc-5860	254	18	,	,	PUNCT
sc-5860	254	19	p.	p.	PROPN
sc-5860	254	20	,	,	PUNCT
sc-5860	254	21	alfuqaha	alfuqaha	PROPN
sc-5860	254	22	,	,	PUNCT
sc-5860	254	23	a.	a.	NOUN
sc-5860	254	24	,	,	PUNCT
sc-5860	254	25	conci	conci	PROPN
sc-5860	254	26	,	,	PUNCT
sc-5860	254	27	n.	n.	NOUN
sc-5860	254	28	and	and	CCONJ
sc-5860	254	29	riegler	riegler	NOUN
sc-5860	254	30	,	,	PUNCT
sc-5860	254	31	m.	m.	NOUN
sc-5860	254	32	(	(	PUNCT
sc-5860	254	33	2022	2022	NUM
sc-5860	254	34	)	)	PUNCT
sc-5860	254	35	visual	visual	ADJ
sc-5860	254	36	sentiment	sentiment	NOUN
sc-5860	254	37	analysis	analysis	NOUN
sc-5860	254	38	from	from	ADP
sc-5860	254	39	disaster	disaster	NOUN
sc-5860	254	40	images	image	NOUN
sc-5860	254	41	in	in	ADP
sc-5860	254	42	social	social	ADJ
sc-5860	254	43	media	medium	NOUN
sc-5860	254	44	.	.	PUNCT
sc-5860	255	1	sensors	sensor	NOUN
sc-5860	255	2	22(10	22(10	NUM
sc-5860	255	3	):	):	PUNCT
sc-5860	255	4	3628	3628	NUM
sc-5860	255	5	.	.	PUNCT
sc-5860	256	1	[	[	X
sc-5860	256	2	27	27	NUM
sc-5860	256	3	]	]	X
sc-5860	256	4	hassan	hassan	PROPN
sc-5860	256	5	,	,	PUNCT
sc-5860	256	6	s.z	s.z	PROPN
sc-5860	256	7	.	.	PROPN
sc-5860	256	8	,	,	PUNCT
sc-5860	256	9	ahmad	ahmad	PROPN
sc-5860	256	10	,	,	PUNCT
sc-5860	256	11	k.	k.	PROPN
sc-5860	256	12	,	,	PUNCT
sc-5860	256	13	riegler	riegler	NOUN
sc-5860	256	14	,	,	PUNCT
sc-5860	256	15	m.a	m.a	PROPN
sc-5860	256	16	.	.	PROPN
sc-5860	256	17	,	,	PUNCT
sc-5860	256	18	hicks	hicks	PROPN
sc-5860	256	19	,	,	PUNCT
sc-5860	256	20	s.	s.	PROPN
sc-5860	256	21	,	,	PUNCT
sc-5860	256	22	conci	conci	PROPN
sc-5860	256	23	,	,	PUNCT
sc-5860	256	24	n.	n.	NOUN
sc-5860	256	25	,	,	PUNCT
sc-5860	256	26	halvorsen	halvorsen	PROPN
sc-5860	256	27	,	,	PUNCT
sc-5860	256	28	p.	p.	NOUN
sc-5860	256	29	and	and	CCONJ
sc-5860	256	30	al	al	PROPN
sc-5860	256	31	-	-	PUNCT
sc-5860	256	32	fuqaha	fuqaha	PROPN
sc-5860	256	33	,	,	PUNCT
sc-5860	256	34	a.	a.	NOUN
sc-5860	256	35	(	(	PUNCT
sc-5860	256	36	2021	2021	NUM
sc-5860	256	37	)	)	PUNCT
sc-5860	256	38	visual	visual	ADJ
sc-5860	256	39	sentiment	sentiment	NOUN
sc-5860	256	40	analysis	analysis	NOUN
sc-5860	256	41	:	:	PUNCT
sc-5860	256	42	a	a	DET
sc-5860	256	43	natural	natural	ADJ
sc-5860	256	44	disasteruse	disasteruse	NOUN
sc-5860	256	45	-	-	PUNCT
sc-5860	256	46	case	case	NOUN
sc-5860	256	47	task	task	NOUN
sc-5860	256	48	at	at	ADP
sc-5860	256	49	mediaeval	mediaeval	ADJ
sc-5860	256	50	2021	2021	NUM
sc-5860	256	51	.	.	PUNCT
sc-5860	257	1	arxiv	arxiv	PROPN
sc-5860	257	2	preprint	preprint	VERB
sc-5860	257	3	arxiv:2111.11471	arxiv:2111.11471	NOUN
sc-5860	257	4	.	.	PUNCT
sc-5860	258	1	[	[	X
sc-5860	258	2	28	28	NUM
sc-5860	258	3	]	]	X
sc-5860	258	4	niu	niu	PROPN
sc-5860	258	5	,	,	PUNCT
sc-5860	258	6	t.	t.	PROPN
sc-5860	258	7	,	,	PUNCT
sc-5860	258	8	zhu	zhu	PROPN
sc-5860	258	9	,	,	PUNCT
sc-5860	258	10	s.	s.	PROPN
sc-5860	258	11	,	,	PUNCT
sc-5860	258	12	pang	pang	NOUN
sc-5860	258	13	,	,	PUNCT
sc-5860	258	14	l.	l.	PROPN
sc-5860	258	15	and	and	CCONJ
sc-5860	258	16	el	el	PROPN
sc-5860	258	17	saddik	saddik	PROPN
sc-5860	258	18	,	,	PUNCT
sc-5860	258	19	a.	a.	NOUN
sc-5860	258	20	(	(	PUNCT
sc-5860	258	21	2016	2016	NUM
sc-5860	258	22	)	)	PUNCT
sc-5860	258	23	sentiment	sentiment	NOUN
sc-5860	258	24	analysis	analysis	NOUN
sc-5860	258	25	on	on	ADP
sc-5860	258	26	multi	multi	ADJ
sc-5860	258	27	-	-	ADJ
sc-5860	258	28	view	view	ADJ
sc-5860	258	29	social	social	ADJ
sc-5860	258	30	data	datum	NOUN
sc-5860	258	31	.	.	PUNCT
sc-5860	259	1	in	in	ADP
sc-5860	259	2	multimedia	multimedia	NOUN
sc-5860	259	3	modeling	modeling	NOUN
sc-5860	259	4	:	:	PUNCT
sc-5860	259	5	22nd	22nd	ADJ
sc-5860	259	6	international	international	ADJ
sc-5860	259	7	conference	conference	NOUN
sc-5860	259	8	,	,	PUNCT
sc-5860	259	9	mmm	mmm	PROPN
sc-5860	259	10	2016	2016	NUM
sc-5860	259	11	,	,	PUNCT
sc-5860	259	12	miami	miami	PROPN
sc-5860	259	13	,	,	PUNCT
sc-5860	259	14	fl	fl	PROPN
sc-5860	259	15	,	,	PUNCT
sc-5860	259	16	usa	usa	PROPN
sc-5860	259	17	,	,	PUNCT
sc-5860	259	18	january	january	PROPN
sc-5860	259	19	4	4	NUM
sc-5860	259	20	-	-	SYM
sc-5860	259	21	6	6	NUM
sc-5860	259	22	,	,	PUNCT
sc-5860	259	23	2016	2016	NUM
sc-5860	259	24	,	,	PUNCT
sc-5860	259	25	proceedings	proceeding	NOUN
sc-5860	259	26	,	,	PUNCT
sc-5860	259	27	part	part	NOUN
sc-5860	259	28	ii	ii	PROPN
sc-5860	259	29	22	22	NUM
sc-5860	259	30	(	(	PUNCT
sc-5860	259	31	springer	springer	NOUN
sc-5860	259	32	):	):	PUNCT
sc-5860	259	33	15–27	15–27	NUM
sc-5860	259	34	.	.	PUNCT
sc-5860	260	1	[	[	X
sc-5860	260	2	29	29	NUM
sc-5860	260	3	]	]	X
sc-5860	260	4	zhou	zhou	PROPN
sc-5860	260	5	,	,	PUNCT
sc-5860	260	6	b.	b.	PROPN
sc-5860	260	7	,	,	PUNCT
sc-5860	260	8	lapedriza	lapedriza	PROPN
sc-5860	260	9	,	,	PUNCT
sc-5860	260	10	a.	a.	NOUN
sc-5860	260	11	,	,	PUNCT
sc-5860	260	12	khosla	khosla	PROPN
sc-5860	260	13	,	,	PUNCT
sc-5860	260	14	a.	a.	PROPN
sc-5860	260	15	,	,	PUNCT
sc-5860	260	16	oliva	oliva	PROPN
sc-5860	260	17	,	,	PUNCT
sc-5860	260	18	a.	a.	NOUN
sc-5860	260	19	and	and	CCONJ
sc-5860	260	20	torralba	torralba	PROPN
sc-5860	260	21	,	,	PUNCT
sc-5860	260	22	a.	a.	NOUN
sc-5860	260	23	(	(	PUNCT
sc-5860	260	24	2017	2017	NUM
sc-5860	260	25	)	)	PUNCT
sc-5860	260	26	places	place	NOUN
sc-5860	260	27	:	:	PUNCT
sc-5860	260	28	a	a	DET
sc-5860	260	29	10	10	NUM
sc-5860	260	30	million	million	NUM
sc-5860	260	31	image	image	NOUN
sc-5860	260	32	database	database	NOUN
sc-5860	260	33	for	for	ADP
sc-5860	260	34	scene	scene	NOUN
sc-5860	260	35	recognition	recognition	NOUN
sc-5860	260	36	.	.	PUNCT
sc-5860	261	1	ieee	ieee	NOUN
sc-5860	261	2	transactions	transaction	NOUN
sc-5860	261	3	on	on	ADP
sc-5860	261	4	pattern	pattern	NOUN
sc-5860	261	5	analysis	analysis	NOUN
sc-5860	261	6	and	and	CCONJ
sc-5860	261	7	machine	machine	NOUN
sc-5860	261	8	intelligence	intelligence	NOUN
sc-5860	261	9	40(6	40(6	PROPN
sc-5860	261	10	):	):	PUNCT
sc-5860	261	11	1452–1464	1452–1464	NUM
sc-5860	261	12	.	.	PUNCT
sc-5860	262	1	[	[	X
sc-5860	262	2	30	30	NUM
sc-5860	262	3	]	]	X
sc-5860	262	4	siqueira	siqueira	NOUN
sc-5860	262	5	,	,	PUNCT
sc-5860	262	6	h.	h.	PROPN
sc-5860	262	7	,	,	PUNCT
sc-5860	262	8	magg	magg	PROPN
sc-5860	262	9	,	,	PUNCT
sc-5860	262	10	s.	s.	PROPN
sc-5860	262	11	and	and	CCONJ
sc-5860	262	12	wermter	wermter	PROPN
sc-5860	262	13	,	,	PUNCT
sc-5860	262	14	s.	s.	PROPN
sc-5860	262	15	(	(	PUNCT
sc-5860	262	16	2020	2020	NUM
sc-5860	262	17	)	)	PUNCT
sc-5860	262	18	efficient	efficient	ADJ
sc-5860	262	19	facial	facial	ADJ
sc-5860	262	20	feature	feature	NOUN
sc-5860	262	21	learning	learn	VERB
sc-5860	262	22	with	with	ADP
sc-5860	262	23	wide	wide	ADJ
sc-5860	262	24	ensemblebased	ensemblebase	VERB
sc-5860	262	25	convolutional	convolutional	ADJ
sc-5860	262	26	neural	neural	ADJ
sc-5860	262	27	networks	network	NOUN
sc-5860	262	28	.	.	PUNCT
sc-5860	263	1	in	in	ADP
sc-5860	263	2	proceedings	proceeding	NOUN
sc-5860	263	3	of	of	ADP
sc-5860	263	4	the	the	DET
sc-5860	263	5	aaai	aaai	PROPN
sc-5860	263	6	conference	conference	NOUN
sc-5860	263	7	on	on	ADP
sc-5860	263	8	artificial	artificial	ADJ
sc-5860	263	9	intelligence	intelligence	NOUN
sc-5860	263	10	,	,	PUNCT
sc-5860	263	11	34	34	NUM
sc-5860	263	12	:	:	SYM
sc-5860	263	13	5800	5800	NUM
sc-5860	263	14	–	–	PUNCT
sc-5860	263	15	5809	5809	NUM
sc-5860	263	16	.	.	PUNCT
sc-5860	264	1	[	[	X
sc-5860	264	2	31	31	NUM
sc-5860	264	3	]	]	SYM
sc-5860	264	4	yi	yi	PROPN
sc-5860	264	5	,	,	PUNCT
sc-5860	264	6	d.	d.	PROPN
sc-5860	264	7	,	,	PUNCT
sc-5860	264	8	lei	lei	PROPN
sc-5860	264	9	,	,	PUNCT
sc-5860	264	10	z.	z.	PROPN
sc-5860	264	11	and	and	CCONJ
sc-5860	264	12	li	li	PROPN
sc-5860	264	13	,	,	PUNCT
sc-5860	264	14	s.z	s.z	PROPN
sc-5860	264	15	.	.	PUNCT
sc-5860	264	16	(	(	PUNCT
sc-5860	264	17	2015	2015	NUM
sc-5860	264	18	)	)	PUNCT
sc-5860	264	19	shared	share	VERB
sc-5860	264	20	representation	representation	NOUN
sc-5860	264	21	learning	learn	VERB
sc-5860	264	22	for	for	ADP
sc-5860	264	23	heterogenous	heterogenous	ADJ
sc-5860	264	24	face	face	NOUN
sc-5860	264	25	recognition	recognition	NOUN
sc-5860	264	26	.	.	PUNCT
sc-5860	265	1	in	in	ADP
sc-5860	265	2	2015	2015	NUM
sc-5860	265	3	11th	11th	NOUN
sc-5860	265	4	ieee	ieee	NOUN
sc-5860	265	5	international	international	ADJ
sc-5860	265	6	conference	conference	NOUN
sc-5860	265	7	and	and	CCONJ
sc-5860	265	8	workshops	workshop	NOUN
sc-5860	265	9	on	on	ADP
sc-5860	265	10	automatic	automatic	ADJ
sc-5860	265	11	face	face	NOUN
sc-5860	265	12	and	and	CCONJ
sc-5860	265	13	gesture	gesture	NOUN
sc-5860	265	14	recognition	recognition	NOUN
sc-5860	265	15	(	(	PUNCT
sc-5860	265	16	fg	fg	PROPN
sc-5860	265	17	)	)	PUNCT
sc-5860	265	18	(	(	PUNCT
sc-5860	265	19	ieee	ieee	NOUN
sc-5860	265	20	)	)	PUNCT
sc-5860	265	21	,	,	PUNCT
sc-5860	265	22	1	1	NUM
sc-5860	265	23	:	:	PUNCT
sc-5860	265	24	1–7	1–7	X
sc-5860	265	25	.	.	PUNCT
sc-5860	266	1	[	[	X
sc-5860	266	2	32	32	NUM
sc-5860	266	3	]	]	X
sc-5860	266	4	liu	liu	PROPN
sc-5860	266	5	,	,	PUNCT
sc-5860	266	6	y.	y.	PROPN
sc-5860	266	7	,	,	PUNCT
sc-5860	266	8	ott	ott	PROPN
sc-5860	266	9	,	,	PUNCT
sc-5860	266	10	m.	m.	NOUN
sc-5860	266	11	,	,	PUNCT
sc-5860	266	12	goyal	goyal	PROPN
sc-5860	266	13	,	,	PUNCT
sc-5860	266	14	n.	n.	NOUN
sc-5860	266	15	,	,	PUNCT
sc-5860	266	16	du	du	PROPN
sc-5860	266	17	,	,	PUNCT
sc-5860	266	18	j.	j.	PROPN
sc-5860	266	19	,	,	PUNCT
sc-5860	266	20	joshi	joshi	PROPN
sc-5860	266	21	,	,	PUNCT
sc-5860	266	22	m.	m.	NOUN
sc-5860	266	23	,	,	PUNCT
sc-5860	266	24	chen	chen	PROPN
sc-5860	266	25	,	,	PUNCT
sc-5860	266	26	d.	d.	PROPN
sc-5860	266	27	,	,	PUNCT
sc-5860	266	28	levy	levy	NOUN
sc-5860	266	29	,	,	PUNCT
sc-5860	266	30	o.	o.	NOUN
sc-5860	266	31	et	et	PROPN
sc-5860	266	32	al	al	PROPN
sc-5860	266	33	.	.	PROPN
sc-5860	267	1	(	(	PUNCT
sc-5860	267	2	2019	2019	NUM
sc-5860	267	3	)	)	PUNCT
sc-5860	268	1	roberta	roberta	NOUN
sc-5860	268	2	:	:	PUNCT
sc-5860	268	3	a	a	DET
sc-5860	268	4	robustly	robustly	ADV
sc-5860	268	5	optimized	optimize	VERB
sc-5860	268	6	bert	bert	NOUN
sc-5860	268	7	pretraining	pretraine	VERB
sc-5860	268	8	approach	approach	NOUN
sc-5860	268	9	.	.	PUNCT
sc-5860	269	1	arxiv	arxiv	PROPN
sc-5860	269	2	preprint	preprint	NOUN
sc-5860	269	3	arxiv:1907.11692	arxiv:1907.11692	NOUN
sc-5860	269	4	.	.	PUNCT
sc-5860	270	1	[	[	X
sc-5860	270	2	33	33	NUM
sc-5860	270	3	]	]	PUNCT
sc-5860	270	4	loria	loria	PROPN
sc-5860	270	5	,	,	PUNCT
sc-5860	270	6	s.	s.	PROPN
sc-5860	270	7	et	et	PROPN
sc-5860	270	8	al	al	PROPN
sc-5860	270	9	.	.	PROPN
sc-5860	271	1	(	(	PUNCT
sc-5860	271	2	2018	2018	NUM
sc-5860	271	3	)	)	PUNCT
sc-5860	271	4	textblob	textblob	NOUN
sc-5860	271	5	documentation	documentation	NOUN
sc-5860	271	6	.	.	PUNCT
sc-5860	272	1	release	release	NOUN
sc-5860	272	2	0.15	0.15	NUM
sc-5860	272	3	2(8	2(8	NUM
sc-5860	272	4	):	):	PUNCT
sc-5860	272	5	269	269	NUM
sc-5860	272	6	.	.	NOUN
sc-5860	272	7	9	9	NUM
sc-5860	272	8	eai	eai	NOUN
sc-5860	272	9	endorsed	endorse	VERB
sc-5860	272	10	transactions	transaction	NOUN
sc-5860	272	11	on	on	ADP
sc-5860	272	12	smart	smart	ADJ
sc-5860	272	13	cities	city	NOUN
sc-5860	272	14	|	|	ADV
sc-5860	272	15	volume	volume	VERB
sc-5860	272	16	7	7	NUM
sc-5860	272	17	|	|	NOUN
sc-5860	272	18	issue	issue	VERB
sc-5860	272	19	4	4	NUM
sc-5860	272	20	|	|	CCONJ
sc-5860	272	21	2023	2023	NUM
sc-5860	273	1	|	|	NOUN
sc-5860	273	2	1	1	NUM
sc-5860	273	3	introduction	introduction	NOUN
sc-5860	273	4	2	2	NUM
sc-5860	273	5	related	relate	VERB
sc-5860	273	6	works	work	NOUN
sc-5860	273	7	3	3	NUM
sc-5860	273	8	material	material	NOUN
sc-5860	273	9	and	and	CCONJ
sc-5860	273	10	methods	method	NOUN
sc-5860	273	11	3.1	3.1	NUM
sc-5860	273	12	visual	visual	ADJ
sc-5860	273	13	features	feature	NOUN
sc-5860	273	14	object	object	VERB
sc-5860	273	15	features	feature	VERB
sc-5860	273	16	place	place	NOUN
sc-5860	273	17	and	and	CCONJ
sc-5860	273	18	scene	scene	NOUN
sc-5860	273	19	features	feature	VERB
sc-5860	273	20	facial	facial	ADJ
sc-5860	273	21	expressions	expression	NOUN
sc-5860	273	22	effective	effective	ADJ
sc-5860	273	23	image	image	NOUN
sc-5860	273	24	content	content	NOUN
sc-5860	273	25	3.2	3.2	NUM
sc-5860	273	26	textual	textual	ADJ
sc-5860	273	27	features	feature	VERB
sc-5860	273	28	data	datum	NOUN
sc-5860	273	29	preprocessing	preprocesse	VERB
sc-5860	273	30	text	text	NOUN
sc-5860	273	31	data	datum	NOUN
sc-5860	273	32	labeling	label	VERB
sc-5860	273	33	3.3	3.3	NUM
sc-5860	273	34	multimodal	multimodal	NOUN
sc-5860	273	35	features	feature	VERB
sc-5860	273	36	4	4	NUM
sc-5860	273	37	experimental	experimental	ADJ
sc-5860	273	38	results	result	NOUN
sc-5860	273	39	4.1	4.1	NUM
sc-5860	273	40	test	test	NOUN
sc-5860	273	41	environment	environment	NOUN
sc-5860	273	42	4.2	4.2	NUM
sc-5860	273	43	dataset	dataset	NOUN
sc-5860	273	44	and	and	CCONJ
sc-5860	273	45	training	training	NOUN
sc-5860	273	46	parameters	parameter	NOUN
sc-5860	273	47	4.3	4.3	NUM
sc-5860	273	48	performance	performance	NOUN
sc-5860	273	49	metrics	metric	NOUN
sc-5860	273	50	4.4	4.4	NUM
sc-5860	273	51	learning	learning	NOUN
sc-5860	273	52	models	model	NOUN
sc-5860	273	53	4.5	4.5	NUM
sc-5860	273	54	performance	performance	NOUN
sc-5860	273	55	results	result	VERB
sc-5860	273	56	5	5	NUM
sc-5860	273	57	conclusions	conclusion	NOUN
sc-5860	273	58	and	and	CCONJ
sc-5860	273	59	future	future	ADJ
sc-5860	273	60	works	work	NOUN