id	sid	tid	token	lemma	pos
cana-5485	1	1	communications	communication	NOUN
cana-5485	1	2	on	on	ADP
cana-5485	1	3	applied	apply	VERB
cana-5485	1	4	nonlinear	nonlinear	ADJ
cana-5485	1	5	analysis	analysis	NOUN
cana-5485	1	6	issn	issn	NOUN
cana-5485	1	7	:	:	PUNCT
cana-5485	1	8	1074	1074	NUM
cana-5485	1	9	-	-	PUNCT
cana-5485	1	10	133x	133x	NUM
cana-5485	1	11	vol	vol	VERB
cana-5485	1	12	32	32	NUM
cana-5485	1	13	no	no	NOUN
cana-5485	1	14	.	.	PUNCT
cana-5485	2	1	10s	10	NOUN
cana-5485	2	2	(	(	PUNCT
cana-5485	2	3	2025	2025	NUM
cana-5485	2	4	)	)	PUNCT
cana-5485	2	5	2375	2375	NUM
cana-5485	2	6	https://internationalpubls.com	https://internationalpubls.com	X
cana-5485	2	7	hybrid	hybrid	ADJ
cana-5485	2	8	deep	deep	ADJ
cana-5485	2	9	learning	learning	NOUN
cana-5485	2	10	for	for	ADP
cana-5485	2	11	advanced	advanced	ADJ
cana-5485	2	12	fake	fake	ADJ
cana-5485	2	13	news	news	NOUN
cana-5485	2	14	detection	detection	NOUN
cana-5485	2	15	using	use	VERB
cana-5485	2	16	explainable	explainable	ADJ
cana-5485	2	17	ai	ai	NOUN
cana-5485	2	18	and	and	CCONJ
cana-5485	2	19	fast	fast	ADJ
cana-5485	2	20	text	text	NOUN
cana-5485	2	21	1malatthi	1malatthi	NUM
cana-5485	2	22	sivasundaram	sivasundaram	PROPN
cana-5485	2	23	,	,	PUNCT
cana-5485	2	24	2v	2v	X
cana-5485	2	25	.	.	PUNCT
cana-5485	3	1	prakasham	prakasham	PROPN
cana-5485	3	2	,	,	PUNCT
cana-5485	3	3	3r	3r	NUM
cana-5485	3	4	.	.	PUNCT
cana-5485	3	5	rathika	rathika	PROPN
cana-5485	3	6	,	,	PUNCT
cana-5485	3	7	4p	4p	NOUN
cana-5485	3	8	.	.	PUNCT
cana-5485	4	1	subhashini	subhashini	PROPN
cana-5485	4	2	,	,	PUNCT
cana-5485	4	3	5p	5p	NUM
cana-5485	4	4	.	.	PUNCT
cana-5485	5	1	karthi	karthi	PROPN
cana-5485	5	2	,	,	PUNCT
cana-5485	5	3	6a	6a	NOUN
cana-5485	5	4	.	.	PUNCT
cana-5485	6	1	priyadharshini	priyadharshini	PROPN
cana-5485	6	2	1	1	NUM
cana-5485	6	3	department	department	NOUN
cana-5485	6	4	of	of	ADP
cana-5485	6	5	computer	computer	NOUN
cana-5485	6	6	science	science	NOUN
cana-5485	6	7	and	and	CCONJ
cana-5485	6	8	engineering	engineering	NOUN
cana-5485	6	9	,	,	PUNCT
cana-5485	7	1	k	k	PROPN
cana-5485	7	2	s	s	PROPN
cana-5485	7	3	r	r	NOUN
cana-5485	7	4	college	college	NOUN
cana-5485	7	5	of	of	ADP
cana-5485	7	6	engineering	engineering	NOUN
cana-5485	7	7	,	,	PUNCT
cana-5485	7	8	tiruchengode	tiruchengode	NOUN
cana-5485	7	9	637215	637215	NUM
cana-5485	7	10	2	2	NUM
cana-5485	7	11	department	department	NOUN
cana-5485	7	12	of	of	ADP
cana-5485	7	13	computer	computer	NOUN
cana-5485	7	14	science	science	NOUN
cana-5485	7	15	and	and	CCONJ
cana-5485	7	16	engineering	engineering	NOUN
cana-5485	7	17	,	,	PUNCT
cana-5485	8	1	k	k	PROPN
cana-5485	8	2	s	s	PROPN
cana-5485	8	3	r	r	NOUN
cana-5485	8	4	college	college	NOUN
cana-5485	8	5	of	of	ADP
cana-5485	8	6	engineering	engineering	NOUN
cana-5485	8	7	,	,	PUNCT
cana-5485	8	8	tiruchengode	tiruchengode	NOUN
cana-5485	8	9	637215	637215	NUM
cana-5485	8	10	3	3	NUM
cana-5485	8	11	department	department	NOUN
cana-5485	8	12	of	of	ADP
cana-5485	8	13	computer	computer	NOUN
cana-5485	8	14	science	science	NOUN
cana-5485	8	15	and	and	CCONJ
cana-5485	8	16	engineering	engineering	NOUN
cana-5485	8	17	,	,	PUNCT
cana-5485	8	18	ksr	ksr	PROPN
cana-5485	8	19	college	college	PROPN
cana-5485	8	20	of	of	ADP
cana-5485	8	21	engineering	engineering	NOUN
cana-5485	8	22	tiruchengode	tiruchengode	NOUN
cana-5485	8	23	–	–	PUNCT
cana-5485	8	24	637215	637215	NUM
cana-5485	8	25	4	4	NUM
cana-5485	8	26	department	department	NOUN
cana-5485	8	27	of	of	ADP
cana-5485	8	28	computer	computer	NOUN
cana-5485	8	29	science	science	NOUN
cana-5485	8	30	and	and	CCONJ
cana-5485	8	31	engineering	engineering	NOUN
cana-5485	8	32	,	,	PUNCT
cana-5485	9	1	k	k	PROPN
cana-5485	9	2	s	s	PROPN
cana-5485	9	3	r	r	PROPN
cana-5485	9	4	institute	institute	NOUN
cana-5485	9	5	for	for	ADP
cana-5485	9	6	engineering	engineering	NOUN
cana-5485	9	7	and	and	CCONJ
cana-5485	9	8	technology	technology	NOUN
cana-5485	9	9	,	,	PUNCT
cana-5485	9	10	tiruchengode	tiruchengode	NOUN
cana-5485	9	11	–	–	PUNCT
cana-5485	9	12	637215	637215	NUM
cana-5485	9	13	5	5	NUM
cana-5485	9	14	department	department	NOUN
cana-5485	9	15	of	of	ADP
cana-5485	9	16	computer	computer	NOUN
cana-5485	9	17	science	science	NOUN
cana-5485	9	18	and	and	CCONJ
cana-5485	9	19	design	design	NOUN
cana-5485	9	20	,	,	PUNCT
cana-5485	10	1	k	k	PROPN
cana-5485	10	2	s	s	PROPN
cana-5485	10	3	r	r	NOUN
cana-5485	10	4	college	college	NOUN
cana-5485	10	5	of	of	ADP
cana-5485	10	6	engineering	engineering	NOUN
cana-5485	10	7	,	,	PUNCT
cana-5485	10	8	tiruchengode	tiruchengode	NOUN
cana-5485	10	9	–	–	PUNCT
cana-5485	10	10	637215	637215	NUM
cana-5485	10	11	6	6	NUM
cana-5485	10	12	department	department	NOUN
cana-5485	10	13	of	of	ADP
cana-5485	10	14	computer	computer	NOUN
cana-5485	10	15	science	science	NOUN
cana-5485	10	16	and	and	CCONJ
cana-5485	10	17	engineering	engineering	NOUN
cana-5485	10	18	,	,	PUNCT
cana-5485	10	19	ksr	ksr	PROPN
cana-5485	10	20	college	college	PROPN
cana-5485	10	21	of	of	ADP
cana-5485	10	22	engineering	engineering	NOUN
cana-5485	10	23	tiruchengode	tiruchengode	NOUN
cana-5485	10	24	–	–	PUNCT
cana-5485	10	25	637215	637215	NUM
cana-5485	10	26	article	article	NOUN
cana-5485	10	27	history	history	NOUN
cana-5485	10	28	:	:	PUNCT
cana-5485	10	29	received	receive	VERB
cana-5485	10	30	:	:	PUNCT
cana-5485	10	31	12	12	NUM
cana-5485	10	32	-	-	SYM
cana-5485	10	33	01	01	NUM
cana-5485	10	34	-	-	PUNCT
cana-5485	10	35	2025	2025	NUM
cana-5485	10	36	revised	revise	VERB
cana-5485	10	37	:	:	PUNCT
cana-5485	10	38	15	15	NUM
cana-5485	10	39	-	-	NUM
cana-5485	10	40	02	02	NUM
cana-5485	10	41	-	-	PUNCT
cana-5485	10	42	2025	2025	NUM
cana-5485	10	43	accepted	accept	VERB
cana-5485	10	44	:	:	PUNCT
cana-5485	10	45	01	01	NUM
cana-5485	10	46	-	-	SYM
cana-5485	10	47	03	03	NUM
cana-5485	10	48	-	-	PUNCT
cana-5485	10	49	2025	2025	NUM
cana-5485	10	50	abstract	abstract	NOUN
cana-5485	10	51	:	:	PUNCT
cana-5485	10	52	in	in	ADP
cana-5485	10	53	order	order	NOUN
cana-5485	10	54	to	to	PART
cana-5485	10	55	enhance	enhance	VERB
cana-5485	10	56	transparency	transparency	NOUN
cana-5485	10	57	and	and	CCONJ
cana-5485	10	58	interpretability	interpretability	NOUN
cana-5485	10	59	,	,	PUNCT
cana-5485	10	60	the	the	DET
cana-5485	10	61	main	main	ADJ
cana-5485	10	62	goal	goal	NOUN
cana-5485	10	63	of	of	ADP
cana-5485	10	64	this	this	DET
cana-5485	10	65	project	project	NOUN
cana-5485	10	66	is	be	AUX
cana-5485	10	67	to	to	PART
cana-5485	10	68	create	create	VERB
cana-5485	10	69	a	a	DET
cana-5485	10	70	hybrid	hybrid	ADJ
cana-5485	10	71	deep	deep	ADJ
cana-5485	10	72	learning	learning	NOUN
cana-5485	10	73	model	model	NOUN
cana-5485	10	74	for	for	ADP
cana-5485	10	75	fake	fake	ADJ
cana-5485	10	76	news	news	NOUN
cana-5485	10	77	detection	detection	NOUN
cana-5485	10	78	by	by	ADP
cana-5485	10	79	fusing	fuse	VERB
cana-5485	10	80	explainable	explainable	ADJ
cana-5485	10	81	ai	ai	NOUN
cana-5485	10	82	(	(	PUNCT
cana-5485	10	83	xai	xai	PROPN
cana-5485	10	84	)	)	PUNCT
cana-5485	10	85	techniques	technique	NOUN
cana-5485	10	86	like	like	ADP
cana-5485	10	87	shapley	shapley	NOUN
cana-5485	10	88	additive	additive	ADJ
cana-5485	10	89	explanations	explanation	NOUN
cana-5485	10	90	(	(	PUNCT
cana-5485	10	91	shap	shap	NOUN
cana-5485	10	92	)	)	PUNCT
cana-5485	10	93	with	with	ADP
cana-5485	10	94	xlnet	xlnet	PROPN
cana-5485	10	95	,	,	PUNCT
cana-5485	10	96	fasttext	fasttext	PROPN
cana-5485	10	97	,	,	PUNCT
cana-5485	10	98	and	and	CCONJ
cana-5485	10	99	cnn	cnn	PROPN
cana-5485	10	100	algorithm	algorithm	NOUN
cana-5485	10	101	.	.	PUNCT
cana-5485	11	1	introduction	introduction	NOUN
cana-5485	11	2	:	:	PUNCT
cana-5485	11	3	fake	fake	ADJ
cana-5485	11	4	news	news	NOUN
cana-5485	11	5	rapid	rapid	ADJ
cana-5485	11	6	spread	spread	NOUN
cana-5485	11	7	in	in	ADP
cana-5485	11	8	the	the	DET
cana-5485	11	9	digital	digital	ADJ
cana-5485	11	10	age	age	NOUN
cana-5485	11	11	has	have	AUX
cana-5485	11	12	turned	turn	VERB
cana-5485	11	13	into	into	ADP
cana-5485	11	14	a	a	DET
cana-5485	11	15	significant	significant	ADJ
cana-5485	11	16	issue	issue	NOUN
cana-5485	11	17	that	that	PRON
cana-5485	11	18	influences	influence	VERB
cana-5485	11	19	social	social	ADJ
cana-5485	11	20	stability	stability	NOUN
cana-5485	11	21	,	,	PUNCT
cana-5485	11	22	public	public	ADJ
cana-5485	11	23	opinion	opinion	NOUN
cana-5485	11	24	,	,	PUNCT
cana-5485	11	25	and	and	CCONJ
cana-5485	11	26	political	political	ADJ
cana-5485	11	27	outcomes	outcome	NOUN
cana-5485	11	28	.	.	PUNCT
cana-5485	12	1	false	false	ADJ
cana-5485	12	2	information	information	NOUN
cana-5485	12	3	has	have	AUX
cana-5485	12	4	spread	spread	VERB
cana-5485	12	5	by	by	ADP
cana-5485	12	6	virtue	virtue	NOUN
cana-5485	12	7	of	of	ADP
cana-5485	12	8	social	social	ADJ
cana-5485	12	9	media	medium	NOUN
cana-5485	12	10	platforms	platform	NOUN
cana-5485	12	11	'	'	PART
cana-5485	12	12	inability	inability	NOUN
cana-5485	12	13	to	to	PART
cana-5485	12	14	distinguish	distinguish	VERB
cana-5485	12	15	between	between	ADP
cana-5485	12	16	authentic	authentic	ADJ
cana-5485	12	17	and	and	CCONJ
cana-5485	12	18	fraudulent	fraudulent	ADJ
cana-5485	12	19	content	content	NOUN
cana-5485	12	20	.	.	PUNCT
cana-5485	13	1	despite	despite	SCONJ
cana-5485	13	2	their	their	PRON
cana-5485	13	3	effectiveness	effectiveness	NOUN
cana-5485	13	4	,	,	PUNCT
cana-5485	13	5	traditional	traditional	ADJ
cana-5485	13	6	fact	fact	NOUN
cana-5485	13	7	-	-	PUNCT
cana-5485	13	8	checking	check	VERB
cana-5485	13	9	methods	method	NOUN
cana-5485	13	10	are	be	AUX
cana-5485	13	11	timeconsuming	timeconsuming	ADJ
cana-5485	13	12	and	and	CCONJ
cana-5485	13	13	unable	unable	ADJ
cana-5485	13	14	to	to	PART
cana-5485	13	15	handle	handle	VERB
cana-5485	13	16	the	the	DET
cana-5485	13	17	volume	volume	NOUN
cana-5485	13	18	of	of	ADP
cana-5485	13	19	data	datum	NOUN
cana-5485	13	20	generated	generate	VERB
cana-5485	13	21	daily	daily	ADV
cana-5485	13	22	.	.	PUNCT
cana-5485	14	1	as	as	ADP
cana-5485	14	2	a	a	DET
cana-5485	14	3	result	result	NOUN
cana-5485	14	4	,	,	PUNCT
cana-5485	14	5	automated	automate	VERB
cana-5485	14	6	systems	system	NOUN
cana-5485	14	7	for	for	ADP
cana-5485	14	8	detecting	detect	VERB
cana-5485	14	9	false	false	ADJ
cana-5485	14	10	news	news	NOUN
cana-5485	14	11	that	that	PRON
cana-5485	14	12	utilize	utilize	VERB
cana-5485	14	13	advanced	advanced	ADJ
cana-5485	14	14	artificial	artificial	ADJ
cana-5485	14	15	intelligence	intelligence	NOUN
cana-5485	14	16	demonstrated	demonstrate	VERB
cana-5485	14	17	impressive	impressive	ADJ
cana-5485	14	18	performance	performance	NOUN
cana-5485	14	19	in	in	ADP
cana-5485	14	20	text	text	NOUN
cana-5485	14	21	classification	classification	NOUN
cana-5485	14	22	tasks	task	NOUN
cana-5485	14	23	,	,	PUNCT
cana-5485	14	24	such	such	ADJ
cana-5485	14	25	as	as	ADP
cana-5485	14	26	identifying	identify	VERB
cana-5485	14	27	false	false	ADJ
cana-5485	14	28	news	news	NOUN
cana-5485	14	29	.	.	PUNCT
cana-5485	15	1	it	it	PRON
cana-5485	15	2	is	be	AUX
cana-5485	15	3	challenging	challenge	VERB
cana-5485	15	4	to	to	PART
cana-5485	15	5	comprehend	comprehend	VERB
cana-5485	15	6	how	how	SCONJ
cana-5485	15	7	these	these	DET
cana-5485	15	8	models	model	NOUN
cana-5485	15	9	make	make	VERB
cana-5485	15	10	decisions	decision	NOUN
cana-5485	15	11	,	,	PUNCT
cana-5485	15	12	though	though	ADV
cana-5485	15	13	,	,	PUNCT
cana-5485	15	14	because	because	SCONJ
cana-5485	15	15	they	they	PRON
cana-5485	15	16	function	function	VERB
cana-5485	15	17	as	as	ADP
cana-5485	15	18	black	black	ADJ
cana-5485	15	19	-	-	PUNCT
cana-5485	15	20	box	box	NOUN
cana-5485	15	21	systems	system	NOUN
cana-5485	15	22	.	.	PUNCT
cana-5485	16	1	in	in	ADP
cana-5485	16	2	order	order	NOUN
cana-5485	16	3	to	to	PART
cana-5485	16	4	improve	improve	VERB
cana-5485	16	5	interpretability	interpretability	NOUN
cana-5485	16	6	,	,	PUNCT
cana-5485	16	7	explainable	explainable	ADJ
cana-5485	16	8	ai	ai	NOUN
cana-5485	16	9	(	(	PUNCT
cana-5485	16	10	xai	xai	PROPN
cana-5485	16	11	)	)	PUNCT
cana-5485	16	12	techniques	technique	NOUN
cana-5485	16	13	have	have	AUX
cana-5485	16	14	been	be	AUX
cana-5485	16	15	developed	develop	VERB
cana-5485	16	16	.	.	PUNCT
cana-5485	17	1	the	the	DET
cana-5485	17	2	shapley	shapley	ADJ
cana-5485	17	3	additive	additive	ADJ
cana-5485	17	4	explanations	explanation	NOUN
cana-5485	17	5	(	(	PUNCT
cana-5485	17	6	shap	shap	NOUN
cana-5485	17	7	)	)	PUNCT
cana-5485	17	8	method	method	NOUN
cana-5485	17	9	is	be	AUX
cana-5485	17	10	one	one	NUM
cana-5485	17	11	that	that	PRON
cana-5485	17	12	offers	offer	VERB
cana-5485	17	13	details	detail	NOUN
cana-5485	17	14	on	on	ADP
cana-5485	17	15	model	model	NOUN
cana-5485	17	16	predictions	prediction	NOUN
cana-5485	17	17	.	.	PUNCT
cana-5485	18	1	objectives	objective	NOUN
cana-5485	18	2	:	:	PUNCT
cana-5485	18	3	the	the	DET
cana-5485	18	4	objective	objective	NOUN
cana-5485	18	5	of	of	ADP
cana-5485	18	6	this	this	DET
cana-5485	18	7	project	project	NOUN
cana-5485	18	8	is	be	AUX
cana-5485	18	9	to	to	PART
cana-5485	18	10	develop	develop	VERB
cana-5485	18	11	a	a	DET
cana-5485	18	12	sophisticated	sophisticated	ADJ
cana-5485	18	13	fake	fake	ADJ
cana-5485	18	14	news	news	NOUN
cana-5485	18	15	detection	detection	NOUN
cana-5485	18	16	system	system	NOUN
cana-5485	18	17	that	that	PRON
cana-5485	18	18	combines	combine	VERB
cana-5485	18	19	advanced	advanced	ADJ
cana-5485	18	20	natural	natural	ADJ
cana-5485	18	21	language	language	NOUN
cana-5485	18	22	processing	processing	NOUN
cana-5485	18	23	and	and	CCONJ
cana-5485	18	24	machine	machine	NOUN
cana-5485	18	25	learning	learn	VERB
cana-5485	18	26	techniques	technique	NOUN
cana-5485	18	27	.	.	PUNCT
cana-5485	19	1	by	by	ADP
cana-5485	19	2	integrating	integrate	VERB
cana-5485	19	3	xlnet	xlnet	PROPN
cana-5485	19	4	for	for	ADP
cana-5485	19	5	superior	superior	ADJ
cana-5485	19	6	language	language	NOUN
cana-5485	19	7	understanding	understanding	NOUN
cana-5485	19	8	,	,	PUNCT
cana-5485	19	9	fasttext	fasttext	NOUN
cana-5485	19	10	for	for	ADP
cana-5485	19	11	efficient	efficient	ADJ
cana-5485	19	12	word	word	NOUN
cana-5485	19	13	representation	representation	NOUN
cana-5485	19	14	,	,	PUNCT
cana-5485	19	15	and	and	CCONJ
cana-5485	19	16	convolutional	convolutional	ADJ
cana-5485	19	17	neural	neural	ADJ
cana-5485	19	18	networks	network	NOUN
cana-5485	19	19	(	(	PUNCT
cana-5485	19	20	cnns	cnns	PROPN
cana-5485	19	21	)	)	PUNCT
cana-5485	19	22	for	for	ADP
cana-5485	19	23	robust	robust	ADJ
cana-5485	19	24	feature	feature	NOUN
cana-5485	19	25	extraction	extraction	NOUN
cana-5485	19	26	,	,	PUNCT
cana-5485	19	27	the	the	DET
cana-5485	19	28	system	system	NOUN
cana-5485	19	29	aims	aim	VERB
cana-5485	19	30	to	to	PART
cana-5485	19	31	enhance	enhance	VERB
cana-5485	19	32	detection	detection	NOUN
cana-5485	19	33	accuracy	accuracy	NOUN
cana-5485	19	34	.	.	PUNCT
cana-5485	20	1	additionally	additionally	ADV
cana-5485	20	2	,	,	PUNCT
cana-5485	20	3	incorporating	incorporate	VERB
cana-5485	20	4	explainable	explainable	ADJ
cana-5485	20	5	ai	ai	ADJ
cana-5485	20	6	techniques	technique	NOUN
cana-5485	20	7	,	,	PUNCT
cana-5485	20	8	particularly	particularly	ADV
cana-5485	20	9	shap	shap	NOUN
cana-5485	20	10	,	,	PUNCT
cana-5485	20	11	will	will	AUX
cana-5485	20	12	provide	provide	VERB
cana-5485	20	13	clear	clear	ADJ
cana-5485	20	14	and	and	CCONJ
cana-5485	20	15	interpretable	interpretable	ADJ
cana-5485	20	16	explanations	explanation	NOUN
cana-5485	20	17	of	of	ADP
cana-5485	20	18	the	the	DET
cana-5485	20	19	model	model	NOUN
cana-5485	20	20	's	's	PART
cana-5485	20	21	predictions	prediction	NOUN
cana-5485	20	22	.	.	PUNCT
cana-5485	21	1	this	this	DET
cana-5485	21	2	dual	dual	ADJ
cana-5485	21	3	focus	focus	NOUN
cana-5485	21	4	on	on	ADP
cana-5485	21	5	performance	performance	NOUN
cana-5485	21	6	and	and	CCONJ
cana-5485	21	7	transparency	transparency	NOUN
cana-5485	21	8	seeks	seek	VERB
cana-5485	21	9	to	to	PART
cana-5485	21	10	create	create	VERB
cana-5485	21	11	a	a	DET
cana-5485	21	12	reliable	reliable	ADJ
cana-5485	21	13	tool	tool	NOUN
cana-5485	21	14	for	for	ADP
cana-5485	21	15	identifying	identify	VERB
cana-5485	21	16	misinformation	misinformation	NOUN
cana-5485	21	17	,	,	PUNCT
cana-5485	21	18	ultimately	ultimately	ADV
cana-5485	21	19	fostering	foster	VERB
cana-5485	21	20	greater	great	ADJ
cana-5485	21	21	public	public	ADJ
cana-5485	21	22	trust	trust	NOUN
cana-5485	21	23	in	in	ADP
cana-5485	21	24	digital	digital	ADJ
cana-5485	21	25	information	information	NOUN
cana-5485	21	26	sources	source	NOUN
cana-5485	21	27	.	.	PUNCT
cana-5485	22	1	methods	method	NOUN
cana-5485	22	2	:	:	PUNCT
cana-5485	22	3	convolutional	convolutional	ADJ
cana-5485	22	4	neural	neural	ADJ
cana-5485	22	5	networks	network	NOUN
cana-5485	22	6	(	(	PUNCT
cana-5485	22	7	cnn	cnn	PROPN
cana-5485	22	8	)	)	PUNCT
cana-5485	22	9	,	,	PUNCT
cana-5485	22	10	xl	xl	PROPN
cana-5485	22	11	net	net	PROPN
cana-5485	22	12	,	,	PUNCT
cana-5485	22	13	and	and	CCONJ
cana-5485	22	14	shap	shap	NOUN
cana-5485	22	15	with	with	ADP
cana-5485	22	16	fast	fast	ADJ
cana-5485	22	17	text	text	NOUN
cana-5485	22	18	are	be	AUX
cana-5485	22	19	examples	example	NOUN
cana-5485	22	20	of	of	ADP
cana-5485	22	21	explainable	explainable	ADJ
cana-5485	22	22	ai	ai	NOUN
cana-5485	22	23	(	(	PUNCT
cana-5485	22	24	xai	xai	PROPN
cana-5485	22	25	)	)	PUNCT
cana-5485	22	26	techniques	technique	NOUN
cana-5485	22	27	that	that	PRON
cana-5485	22	28	were	be	AUX
cana-5485	22	29	used	use	VERB
cana-5485	22	30	in	in	ADP
cana-5485	22	31	the	the	DET
cana-5485	22	32	study	study	NOUN
cana-5485	22	33	's	's	PART
cana-5485	22	34	hybrid	hybrid	ADJ
cana-5485	22	35	deep	deep	ADJ
cana-5485	22	36	learning	learning	NOUN
cana-5485	22	37	methodology	methodology	NOUN
cana-5485	22	38	.	.	PUNCT
cana-5485	23	1	group	group	NOUN
cana-5485	23	2	1	1	NUM
cana-5485	23	3	:	:	PUNCT
cana-5485	23	4	robert	robert	PROPN
cana-5485	23	5	and	and	CCONJ
cana-5485	23	6	bert	bert	PROPN
cana-5485	23	7	although	although	SCONJ
cana-5485	23	8	methods	method	NOUN
cana-5485	23	9	are	be	AUX
cana-5485	23	10	effective	effective	ADJ
cana-5485	23	11	,	,	PUNCT
cana-5485	23	12	they	they	PRON
cana-5485	23	13	are	be	AUX
cana-5485	23	14	not	not	PART
cana-5485	23	15	transparent	transparent	ADJ
cana-5485	23	16	enough	enough	ADV
cana-5485	23	17	for	for	SCONJ
cana-5485	23	18	users	user	NOUN
cana-5485	23	19	to	to	PART
cana-5485	23	20	comprehend	comprehend	VERB
cana-5485	23	21	and	and	CCONJ
cana-5485	23	22	have	have	VERB
cana-5485	23	23	faith	faith	NOUN
cana-5485	23	24	in	in	ADP
cana-5485	23	25	their	their	PRON
cana-5485	23	26	predictions	prediction	NOUN
cana-5485	23	27	.	.	PUNCT
cana-5485	24	1	group	group	NOUN
cana-5485	24	2	2	2	NUM
cana-5485	24	3	:	:	PUNCT
cana-5485	24	4	explainable	explainable	ADJ
cana-5485	24	5	ai	ai	NOUN
cana-5485	24	6	and	and	CCONJ
cana-5485	24	7	fat	fat	ADJ
cana-5485	24	8	text	text	NOUN
cana-5485	24	9	were	be	AUX
cana-5485	24	10	used	use	VERB
cana-5485	24	11	in	in	ADP
cana-5485	24	12	combination	combination	NOUN
cana-5485	24	13	with	with	ADP
cana-5485	24	14	the	the	DET
cana-5485	24	15	hybrid	hybrid	ADJ
cana-5485	24	16	model	model	NOUN
cana-5485	24	17	.	.	PUNCT
cana-5485	25	1	results	result	NOUN
cana-5485	25	2	:	:	PUNCT
cana-5485	25	3	the	the	DET
cana-5485	25	4	hybrid	hybrid	ADJ
cana-5485	25	5	model	model	NOUN
cana-5485	25	6	's	's	PART
cana-5485	25	7	accuracy	accuracy	NOUN
cana-5485	25	8	of	of	ADP
cana-5485	25	9	92.3	92.3	NUM
cana-5485	25	10	%	%	NOUN
cana-5485	25	11	represents	represent	VERB
cana-5485	25	12	a	a	DET
cana-5485	25	13	5.6	5.6	NUM
cana-5485	25	14	%	%	NOUN
cana-5485	25	15	improvement	improvement	NOUN
cana-5485	25	16	over	over	ADP
cana-5485	25	17	the	the	DET
cana-5485	25	18	baseline	baseline	ADJ
cana-5485	25	19	accuracy	accuracy	NOUN
cana-5485	25	20	of	of	ADP
cana-5485	25	21	87.4	87.4	NUM
cana-5485	25	22	%	%	NOUN
cana-5485	25	23	.	.	PUNCT
cana-5485	26	1	this	this	PRON
cana-5485	26	2	shows	show	VERB
cana-5485	26	3	that	that	SCONJ
cana-5485	26	4	the	the	DET
cana-5485	26	5	hybrid	hybrid	ADJ
cana-5485	26	6	approach	approach	NOUN
cana-5485	26	7	is	be	AUX
cana-5485	26	8	more	more	ADV
cana-5485	26	9	effective	effective	ADJ
cana-5485	26	10	at	at	ADP
cana-5485	26	11	communications	communication	NOUN
cana-5485	26	12	on	on	ADP
cana-5485	26	13	applied	apply	VERB
cana-5485	26	14	nonlinear	nonlinear	ADJ
cana-5485	26	15	analysis	analysis	NOUN
cana-5485	26	16	issn	issn	NOUN
cana-5485	26	17	:	:	PUNCT
cana-5485	26	18	1074	1074	NUM
cana-5485	26	19	-	-	PUNCT
cana-5485	26	20	133x	133x	NUM
cana-5485	26	21	vol	vol	VERB
cana-5485	26	22	32	32	NUM
cana-5485	26	23	no	no	NOUN
cana-5485	26	24	.	.	PUNCT
cana-5485	27	1	10s	10	NOUN
cana-5485	27	2	(	(	PUNCT
cana-5485	27	3	2025	2025	NUM
cana-5485	27	4	)	)	PUNCT
cana-5485	27	5	2376	2376	NUM
cana-5485	27	6	https://internationalpubls.com	https://internationalpubls.com	X
cana-5485	27	7	correctly	correctly	ADV
cana-5485	27	8	distinguishing	distinguish	VERB
cana-5485	27	9	between	between	ADP
cana-5485	27	10	real	real	ADJ
cana-5485	27	11	and	and	CCONJ
cana-5485	27	12	fake	fake	ADJ
cana-5485	27	13	news	news	NOUN
cana-5485	27	14	articles	article	NOUN
cana-5485	27	15	.	.	PUNCT
cana-5485	28	1	additionally	additionally	ADV
cana-5485	28	2	,	,	PUNCT
cana-5485	28	3	the	the	DET
cana-5485	28	4	hybrid	hybrid	NOUN
cana-5485	28	5	model	model	NOUN
cana-5485	28	6	is	be	AUX
cana-5485	28	7	more	more	ADV
cana-5485	28	8	effective	effective	ADJ
cana-5485	28	9	at	at	ADP
cana-5485	28	10	reducing	reduce	VERB
cana-5485	28	11	false	false	ADJ
cana-5485	28	12	positives	positive	NOUN
cana-5485	28	13	,	,	PUNCT
cana-5485	28	14	as	as	SCONJ
cana-5485	28	15	evidenced	evidence	VERB
cana-5485	28	16	by	by	ADP
cana-5485	28	17	its	its	PRON
cana-5485	28	18	90.5	90.5	NUM
cana-5485	28	19	%	%	NOUN
cana-5485	28	20	accuracy	accuracy	NOUN
cana-5485	28	21	,	,	PUNCT
cana-5485	28	22	which	which	PRON
cana-5485	28	23	is	be	AUX
cana-5485	28	24	6.2	6.2	NUM
cana-5485	28	25	%	%	NOUN
cana-5485	28	26	higher	high	ADJ
cana-5485	28	27	than	than	ADP
cana-5485	28	28	the	the	DET
cana-5485	28	29	baseline	baseline	PROPN
cana-5485	28	30	model	model	NOUN
cana-5485	28	31	's	's	PART
cana-5485	28	32	85.2	85.2	NUM
cana-5485	28	33	%	%	NOUN
cana-5485	28	34	accuracy	accuracy	NOUN
cana-5485	28	35	.	.	PUNCT
cana-5485	29	1	similarly	similarly	ADV
cana-5485	29	2	,	,	PUNCT
cana-5485	29	3	from	from	ADP
cana-5485	29	4	86.1	86.1	NUM
cana-5485	29	5	%	%	NOUN
cana-5485	29	6	in	in	ADP
cana-5485	29	7	the	the	DET
cana-5485	29	8	baseline	baseline	NOUN
cana-5485	29	9	model	model	NOUN
cana-5485	29	10	to	to	ADP
cana-5485	29	11	91.8	91.8	NUM
cana-5485	29	12	%	%	NOUN
cana-5485	29	13	in	in	ADP
cana-5485	29	14	the	the	DET
cana-5485	29	15	hybrid	hybrid	NOUN
cana-5485	29	16	model	model	NOUN
cana-5485	29	17	,	,	PUNCT
cana-5485	29	18	the	the	DET
cana-5485	29	19	hybrid	hybrid	ADJ
cana-5485	29	20	model	model	NOUN
cana-5485	29	21	's	's	PART
cana-5485	29	22	recall	recall	NOUN
cana-5485	29	23	increases	increase	NOUN
cana-5485	29	24	by	by	ADP
cana-5485	29	25	6.6	6.6	NUM
cana-5485	29	26	%	%	NOUN
cana-5485	29	27	,	,	PUNCT
cana-5485	29	28	indicating	indicate	VERB
cana-5485	29	29	that	that	SCONJ
cana-5485	29	30	it	it	PRON
cana-5485	29	31	is	be	AUX
cana-5485	29	32	better	well	ADJ
cana-5485	29	33	at	at	ADP
cana-5485	29	34	spotting	spot	VERB
cana-5485	29	35	fake	fake	ADJ
cana-5485	29	36	news	news	NOUN
cana-5485	29	37	.	.	PUNCT
cana-5485	30	1	finally	finally	ADV
cana-5485	30	2	,	,	PUNCT
cana-5485	30	3	the	the	DET
cana-5485	30	4	f1	f1	NOUN
cana-5485	30	5	-	-	PUNCT
cana-5485	30	6	score	score	NOUN
cana-5485	30	7	,	,	PUNCT
cana-5485	30	8	which	which	PRON
cana-5485	30	9	strikes	strike	VERB
cana-5485	30	10	a	a	DET
cana-5485	30	11	balance	balance	NOUN
cana-5485	30	12	between	between	ADP
cana-5485	30	13	recall	recall	NOUN
cana-5485	30	14	and	and	CCONJ
cana-5485	30	15	precision	precision	NOUN
cana-5485	30	16	,	,	PUNCT
cana-5485	30	17	increased	increase	VERB
cana-5485	30	18	from	from	ADP
cana-5485	30	19	85.6	85.6	NUM
cana-5485	30	20	%	%	NOUN
cana-5485	30	21	to	to	ADP
cana-5485	30	22	91.1	91.1	NUM
cana-5485	30	23	%	%	NOUN
cana-5485	30	24	,	,	PUNCT
cana-5485	30	25	a	a	DET
cana-5485	30	26	6.4	6.4	NUM
cana-5485	30	27	%	%	NOUN
cana-5485	30	28	improvement	improvement	NOUN
cana-5485	30	29	.	.	PUNCT
cana-5485	31	1	conclusions	conclusion	NOUN
cana-5485	31	2	:	:	PUNCT
cana-5485	31	3	by	by	ADP
cana-5485	31	4	combining	combine	VERB
cana-5485	31	5	xl	xl	PROPN
cana-5485	31	6	net	net	NOUN
cana-5485	31	7	,	,	PUNCT
cana-5485	31	8	fast	fast	ADJ
cana-5485	31	9	text	text	NOUN
cana-5485	31	10	,	,	PUNCT
cana-5485	31	11	cnn	cnn	PROPN
cana-5485	31	12	,	,	PUNCT
cana-5485	31	13	and	and	CCONJ
cana-5485	31	14	explainable	explainable	ADJ
cana-5485	31	15	ai	ai	ADJ
cana-5485	31	16	techniques	technique	NOUN
cana-5485	31	17	,	,	PUNCT
cana-5485	31	18	the	the	DET
cana-5485	31	19	proposed	propose	VERB
cana-5485	31	20	hybrid	hybrid	NOUN
cana-5485	31	21	deep	deep	ADJ
cana-5485	31	22	learning	learning	NOUN
cana-5485	31	23	model	model	NOUN
cana-5485	31	24	significantly	significantly	ADV
cana-5485	31	25	increases	increase	VERB
cana-5485	31	26	the	the	DET
cana-5485	31	27	accuracy	accuracy	NOUN
cana-5485	31	28	of	of	ADP
cana-5485	31	29	fake	fake	ADJ
cana-5485	31	30	news	news	NOUN
cana-5485	31	31	detection	detection	NOUN
cana-5485	31	32	while	while	SCONJ
cana-5485	31	33	maintaining	maintain	VERB
cana-5485	31	34	interpretability	interpretability	NOUN
cana-5485	31	35	.	.	PUNCT
cana-5485	32	1	this	this	DET
cana-5485	32	2	tactic	tactic	NOUN
cana-5485	32	3	provides	provide	VERB
cana-5485	32	4	a	a	DET
cana-5485	32	5	robust	robust	ADJ
cana-5485	32	6	and	and	CCONJ
cana-5485	32	7	transparent	transparent	ADJ
cana-5485	32	8	framework	framework	NOUN
cana-5485	32	9	for	for	ADP
cana-5485	32	10	effectively	effectively	ADV
cana-5485	32	11	combating	combat	VERB
cana-5485	32	12	misinformation	misinformation	NOUN
cana-5485	32	13	.	.	PUNCT
cana-5485	33	1	keywords	keyword	NOUN
cana-5485	33	2	:	:	PUNCT
cana-5485	33	3	fake	fake	ADJ
cana-5485	33	4	news	news	NOUN
cana-5485	33	5	detection	detection	NOUN
cana-5485	33	6	,	,	PUNCT
cana-5485	33	7	xl	xl	PROPN
cana-5485	33	8	net	net	PROPN
cana-5485	33	9	,	,	PUNCT
cana-5485	33	10	explainable	explainable	ADJ
cana-5485	33	11	ai	ai	NOUN
cana-5485	33	12	,	,	PUNCT
cana-5485	33	13	shap	shap	NOUN
cana-5485	33	14	,	,	PUNCT
cana-5485	33	15	fast	fast	ADJ
cana-5485	33	16	text	text	NOUN
cana-5485	33	17	,	,	PUNCT
cana-5485	33	18	hybrid	hybrid	ADJ
cana-5485	33	19	deep	deep	ADJ
cana-5485	33	20	learning	learning	NOUN
cana-5485	33	21	,	,	PUNCT
cana-5485	33	22	natural	natural	ADJ
cana-5485	33	23	language	language	NOUN
cana-5485	33	24	processing	processing	NOUN
cana-5485	33	25	,	,	PUNCT
cana-5485	33	26	robert	robert	PROPN
cana-5485	33	27	,	,	PUNCT
cana-5485	33	28	bert	bert	PROPN
cana-5485	33	29	,	,	PUNCT
cana-5485	33	30	convolutional	convolutional	ADJ
cana-5485	33	31	neural	neural	ADJ
cana-5485	33	32	network	network	NOUN
cana-5485	33	33	,	,	PUNCT
cana-5485	33	34	word	word	NOUN
cana-5485	33	35	representation	representation	NOUN
cana-5485	33	36	,	,	PUNCT
cana-5485	33	37	misinformation	misinformation	NOUN
cana-5485	33	38	.	.	PUNCT
cana-5485	34	1	1	1	X
cana-5485	34	2	.	.	X
cana-5485	34	3	introduction	introduction	NOUN
cana-5485	34	4	fake	fake	ADJ
cana-5485	34	5	news	news	NOUN
cana-5485	34	6	rapid	rapid	ADJ
cana-5485	34	7	spread	spread	NOUN
cana-5485	34	8	in	in	ADP
cana-5485	34	9	the	the	DET
cana-5485	34	10	digital	digital	ADJ
cana-5485	34	11	age	age	NOUN
cana-5485	34	12	has	have	AUX
cana-5485	34	13	turned	turn	VERB
cana-5485	34	14	into	into	ADP
cana-5485	34	15	a	a	DET
cana-5485	34	16	significant	significant	ADJ
cana-5485	34	17	issue	issue	NOUN
cana-5485	34	18	that	that	PRON
cana-5485	34	19	influences	influence	VERB
cana-5485	34	20	social	social	ADJ
cana-5485	34	21	stability	stability	NOUN
cana-5485	34	22	,	,	PUNCT
cana-5485	34	23	public	public	ADJ
cana-5485	34	24	opinion	opinion	NOUN
cana-5485	34	25	,	,	PUNCT
cana-5485	34	26	and	and	CCONJ
cana-5485	34	27	political	political	ADJ
cana-5485	34	28	outcomes	outcome	NOUN
cana-5485	34	29	.	.	PUNCT
cana-5485	35	1	false	false	ADJ
cana-5485	35	2	information	information	NOUN
cana-5485	35	3	has	have	AUX
cana-5485	35	4	spread	spread	VERB
cana-5485	35	5	by	by	ADP
cana-5485	35	6	virtue	virtue	NOUN
cana-5485	35	7	of	of	ADP
cana-5485	35	8	social	social	ADJ
cana-5485	35	9	media	medium	NOUN
cana-5485	35	10	platforms	platform	NOUN
cana-5485	35	11	'	'	PART
cana-5485	35	12	inability	inability	NOUN
cana-5485	35	13	to	to	PART
cana-5485	35	14	distinguish	distinguish	VERB
cana-5485	35	15	between	between	ADP
cana-5485	35	16	authentic	authentic	ADJ
cana-5485	35	17	and	and	CCONJ
cana-5485	35	18	fraudulent	fraudulent	ADJ
cana-5485	35	19	content	content	NOUN
cana-5485	35	20	.	.	PUNCT
cana-5485	36	1	despite	despite	SCONJ
cana-5485	36	2	their	their	PRON
cana-5485	36	3	effectiveness	effectiveness	NOUN
cana-5485	36	4	,	,	PUNCT
cana-5485	36	5	traditional	traditional	ADJ
cana-5485	36	6	fact	fact	NOUN
cana-5485	36	7	-	-	PUNCT
cana-5485	36	8	checking	check	VERB
cana-5485	36	9	methods	method	NOUN
cana-5485	36	10	are	be	AUX
cana-5485	36	11	time	time	NOUN
cana-5485	36	12	-	-	PUNCT
cana-5485	36	13	consuming	consume	VERB
cana-5485	36	14	and	and	CCONJ
cana-5485	36	15	unable	unable	ADJ
cana-5485	36	16	to	to	PART
cana-5485	36	17	handle	handle	VERB
cana-5485	36	18	the	the	DET
cana-5485	36	19	volume	volume	NOUN
cana-5485	36	20	of	of	ADP
cana-5485	36	21	data	datum	NOUN
cana-5485	36	22	generated	generate	VERB
cana-5485	36	23	daily	daily	ADV
cana-5485	36	24	.	.	PUNCT
cana-5485	37	1	as	as	ADP
cana-5485	37	2	a	a	DET
cana-5485	37	3	result	result	NOUN
cana-5485	37	4	,	,	PUNCT
cana-5485	37	5	automated	automate	VERB
cana-5485	37	6	systems	system	NOUN
cana-5485	37	7	for	for	ADP
cana-5485	37	8	detecting	detect	VERB
cana-5485	37	9	false	false	ADJ
cana-5485	37	10	news	news	NOUN
cana-5485	37	11	that	that	PRON
cana-5485	37	12	utilize	utilize	VERB
cana-5485	37	13	advanced	advanced	ADJ
cana-5485	37	14	artificial	artificial	ADJ
cana-5485	37	15	intelligence	intelligence	NOUN
cana-5485	37	16	demonstrated	demonstrate	VERB
cana-5485	37	17	impressive	impressive	ADJ
cana-5485	37	18	performance	performance	NOUN
cana-5485	37	19	in	in	ADP
cana-5485	37	20	text	text	NOUN
cana-5485	37	21	classification	classification	NOUN
cana-5485	37	22	tasks	task	NOUN
cana-5485	37	23	,	,	PUNCT
cana-5485	37	24	such	such	ADJ
cana-5485	37	25	as	as	ADP
cana-5485	37	26	identifying	identify	VERB
cana-5485	37	27	false	false	ADJ
cana-5485	37	28	news	news	NOUN
cana-5485	37	29	.	.	PUNCT
cana-5485	38	1	it	it	PRON
cana-5485	38	2	is	be	AUX
cana-5485	38	3	challenging	challenge	VERB
cana-5485	38	4	to	to	PART
cana-5485	38	5	comprehend	comprehend	VERB
cana-5485	38	6	how	how	SCONJ
cana-5485	38	7	these	these	DET
cana-5485	38	8	models	model	NOUN
cana-5485	38	9	make	make	VERB
cana-5485	38	10	decisions	decision	NOUN
cana-5485	38	11	,	,	PUNCT
cana-5485	38	12	though	though	ADV
cana-5485	38	13	,	,	PUNCT
cana-5485	38	14	because	because	SCONJ
cana-5485	38	15	they	they	PRON
cana-5485	38	16	function	function	VERB
cana-5485	38	17	as	as	ADP
cana-5485	38	18	black	black	ADJ
cana-5485	38	19	-	-	PUNCT
cana-5485	38	20	box	box	NOUN
cana-5485	38	21	systems	system	NOUN
cana-5485	38	22	.	.	PUNCT
cana-5485	39	1	in	in	ADP
cana-5485	39	2	order	order	NOUN
cana-5485	39	3	to	to	PART
cana-5485	39	4	improve	improve	VERB
cana-5485	39	5	interpretability	interpretability	NOUN
cana-5485	39	6	,	,	PUNCT
cana-5485	39	7	explainable	explainable	ADJ
cana-5485	39	8	ai	ai	NOUN
cana-5485	39	9	(	(	PUNCT
cana-5485	39	10	xai	xai	PROPN
cana-5485	39	11	)	)	PUNCT
cana-5485	39	12	techniques	technique	NOUN
cana-5485	39	13	have	have	AUX
cana-5485	39	14	been	be	AUX
cana-5485	39	15	developed	develop	VERB
cana-5485	39	16	.	.	PUNCT
cana-5485	40	1	the	the	DET
cana-5485	40	2	shapley	shapley	ADJ
cana-5485	40	3	additive	additive	ADJ
cana-5485	40	4	explanations	explanation	NOUN
cana-5485	40	5	(	(	PUNCT
cana-5485	40	6	shap	shap	NOUN
cana-5485	40	7	)	)	PUNCT
cana-5485	40	8	method	method	NOUN
cana-5485	40	9	is	be	AUX
cana-5485	40	10	one	one	NUM
cana-5485	40	11	that	that	PRON
cana-5485	40	12	offers	offer	VERB
cana-5485	40	13	details	detail	NOUN
cana-5485	40	14	on	on	ADP
cana-5485	40	15	model	model	NOUN
cana-5485	40	16	predictions	prediction	NOUN
cana-5485	40	17	.	.	PUNCT
cana-5485	41	1	2	2	X
cana-5485	41	2	.	.	X
cana-5485	41	3	objectives	objective	VERB
cana-5485	41	4	the	the	DET
cana-5485	41	5	objective	objective	NOUN
cana-5485	41	6	of	of	ADP
cana-5485	41	7	this	this	DET
cana-5485	41	8	project	project	NOUN
cana-5485	41	9	is	be	AUX
cana-5485	41	10	to	to	PART
cana-5485	41	11	develop	develop	VERB
cana-5485	41	12	a	a	DET
cana-5485	41	13	sophisticated	sophisticated	ADJ
cana-5485	41	14	fake	fake	ADJ
cana-5485	41	15	news	news	NOUN
cana-5485	41	16	detection	detection	NOUN
cana-5485	41	17	system	system	NOUN
cana-5485	41	18	that	that	PRON
cana-5485	41	19	combines	combine	VERB
cana-5485	41	20	advanced	advanced	ADJ
cana-5485	41	21	natural	natural	ADJ
cana-5485	41	22	language	language	NOUN
cana-5485	41	23	processing	processing	NOUN
cana-5485	41	24	and	and	CCONJ
cana-5485	41	25	machine	machine	NOUN
cana-5485	41	26	learning	learn	VERB
cana-5485	41	27	techniques	technique	NOUN
cana-5485	41	28	.	.	PUNCT
cana-5485	42	1	by	by	ADP
cana-5485	42	2	integrating	integrate	VERB
cana-5485	42	3	xlnet	xlnet	PROPN
cana-5485	42	4	for	for	ADP
cana-5485	42	5	superior	superior	ADJ
cana-5485	42	6	language	language	NOUN
cana-5485	42	7	understanding	understanding	NOUN
cana-5485	42	8	,	,	PUNCT
cana-5485	42	9	fasttext	fasttext	NOUN
cana-5485	42	10	for	for	ADP
cana-5485	42	11	efficient	efficient	ADJ
cana-5485	42	12	word	word	NOUN
cana-5485	42	13	representation	representation	NOUN
cana-5485	42	14	,	,	PUNCT
cana-5485	42	15	and	and	CCONJ
cana-5485	42	16	convolutional	convolutional	ADJ
cana-5485	42	17	neural	neural	ADJ
cana-5485	42	18	networks	network	NOUN
cana-5485	42	19	(	(	PUNCT
cana-5485	42	20	cnns	cnns	PROPN
cana-5485	42	21	)	)	PUNCT
cana-5485	42	22	for	for	ADP
cana-5485	42	23	robust	robust	ADJ
cana-5485	42	24	feature	feature	NOUN
cana-5485	42	25	extraction	extraction	NOUN
cana-5485	42	26	,	,	PUNCT
cana-5485	42	27	the	the	DET
cana-5485	42	28	system	system	NOUN
cana-5485	42	29	aims	aim	VERB
cana-5485	42	30	to	to	PART
cana-5485	42	31	enhance	enhance	VERB
cana-5485	42	32	detection	detection	NOUN
cana-5485	42	33	accuracy	accuracy	NOUN
cana-5485	42	34	.	.	PUNCT
cana-5485	43	1	additionally	additionally	ADV
cana-5485	43	2	,	,	PUNCT
cana-5485	43	3	incorporating	incorporate	VERB
cana-5485	43	4	explainable	explainable	ADJ
cana-5485	43	5	ai	ai	ADJ
cana-5485	43	6	techniques	technique	NOUN
cana-5485	43	7	,	,	PUNCT
cana-5485	43	8	particularly	particularly	ADV
cana-5485	43	9	shap	shap	NOUN
cana-5485	43	10	,	,	PUNCT
cana-5485	43	11	will	will	AUX
cana-5485	43	12	provide	provide	VERB
cana-5485	43	13	clear	clear	ADJ
cana-5485	43	14	and	and	CCONJ
cana-5485	43	15	interpretable	interpretable	ADJ
cana-5485	43	16	explanations	explanation	NOUN
cana-5485	43	17	of	of	ADP
cana-5485	43	18	the	the	DET
cana-5485	43	19	model	model	NOUN
cana-5485	43	20	's	's	PART
cana-5485	43	21	predictions	prediction	NOUN
cana-5485	43	22	.	.	PUNCT
cana-5485	44	1	this	this	DET
cana-5485	44	2	dual	dual	ADJ
cana-5485	44	3	focus	focus	NOUN
cana-5485	44	4	on	on	ADP
cana-5485	44	5	performance	performance	NOUN
cana-5485	44	6	and	and	CCONJ
cana-5485	44	7	transparency	transparency	NOUN
cana-5485	44	8	seeks	seek	VERB
cana-5485	44	9	to	to	PART
cana-5485	44	10	create	create	VERB
cana-5485	44	11	a	a	DET
cana-5485	44	12	reliable	reliable	ADJ
cana-5485	44	13	tool	tool	NOUN
cana-5485	44	14	for	for	ADP
cana-5485	44	15	identifying	identify	VERB
cana-5485	44	16	misinformation	misinformation	NOUN
cana-5485	44	17	,	,	PUNCT
cana-5485	44	18	ultimately	ultimately	ADV
cana-5485	44	19	fostering	foster	VERB
cana-5485	44	20	greater	great	ADJ
cana-5485	44	21	public	public	ADJ
cana-5485	44	22	trust	trust	NOUN
cana-5485	44	23	in	in	ADP
cana-5485	44	24	digital	digital	ADJ
cana-5485	44	25	information	information	NOUN
cana-5485	44	26	sources	source	NOUN
cana-5485	44	27	.	.	PUNCT
cana-5485	45	1	3	3	X
cana-5485	45	2	.	.	X
cana-5485	45	3	methods	method	NOUN
cana-5485	45	4	high	high	ADJ
cana-5485	45	5	-	-	PUNCT
cana-5485	45	6	performance	performance	NOUN
cana-5485	45	7	gpus	gpu	NOUN
cana-5485	45	8	efficiently	efficiently	ADV
cana-5485	45	9	handled	handle	VERB
cana-5485	45	10	deep	deep	ADJ
cana-5485	45	11	learning	learning	NOUN
cana-5485	45	12	computations	computation	NOUN
cana-5485	45	13	during	during	ADP
cana-5485	45	14	the	the	DET
cana-5485	45	15	tests	test	NOUN
cana-5485	45	16	,	,	PUNCT
cana-5485	45	17	which	which	PRON
cana-5485	45	18	were	be	AUX
cana-5485	45	19	conducted	conduct	VERB
cana-5485	45	20	in	in	ADP
cana-5485	45	21	the	the	DET
cana-5485	45	22	antenna	antenna	NOUN
cana-5485	45	23	lab	lab	NOUN
cana-5485	45	24	at	at	ADP
cana-5485	45	25	the	the	DET
cana-5485	45	26	ksr	ksr	PROPN
cana-5485	45	27	college	college	PROPN
cana-5485	45	28	of	of	ADP
cana-5485	45	29	engineering	engineering	NOUN
cana-5485	45	30	.	.	PUNCT
cana-5485	46	1	the	the	DET
cana-5485	46	2	model	model	NOUN
cana-5485	46	3	was	be	AUX
cana-5485	46	4	evaluated	evaluate	VERB
cana-5485	46	5	using	use	VERB
cana-5485	46	6	metrics	metric	NOUN
cana-5485	46	7	such	such	ADJ
cana-5485	46	8	as	as	ADP
cana-5485	46	9	accuracy	accuracy	NOUN
cana-5485	46	10	,	,	PUNCT
cana-5485	46	11	precision	precision	NOUN
cana-5485	46	12	,	,	PUNCT
cana-5485	46	13	recall	recall	NOUN
cana-5485	46	14	,	,	PUNCT
cana-5485	46	15	and	and	CCONJ
cana-5485	46	16	f1	f1	NOUN
cana-5485	46	17	-	-	PUNCT
cana-5485	46	18	score	score	NOUN
cana-5485	46	19	,	,	PUNCT
cana-5485	46	20	and	and	CCONJ
cana-5485	46	21	baseline	baseline	NOUN
cana-5485	46	22	models	model	NOUN
cana-5485	46	23	such	such	ADJ
cana-5485	46	24	as	as	ADP
cana-5485	46	25	bert	bert	PROPN
cana-5485	46	26	and	and	CCONJ
cana-5485	46	27	roberta	roberta	PROPN
cana-5485	46	28	were	be	AUX
cana-5485	46	29	contrasted	contrast	VERB
cana-5485	46	30	.	.	PUNCT
cana-5485	47	1	the	the	DET
cana-5485	47	2	hybrid	hybrid	ADJ
cana-5485	47	3	approach	approach	NOUN
cana-5485	47	4	demonstrated	demonstrate	VERB
cana-5485	47	5	outstanding	outstanding	ADJ
cana-5485	47	6	performance	performance	NOUN
cana-5485	47	7	in	in	ADP
cana-5485	47	8	categorizing	categorize	VERB
cana-5485	47	9	fake	fake	ADJ
cana-5485	47	10	news	news	NOUN
cana-5485	47	11	while	while	SCONJ
cana-5485	47	12	maintaining	maintain	VERB
cana-5485	47	13	explainability	explainability	NOUN
cana-5485	47	14	,	,	PUNCT
cana-5485	47	15	making	make	VERB
cana-5485	47	16	it	it	PRON
cana-5485	47	17	a	a	DET
cana-5485	47	18	reliable	reliable	ADJ
cana-5485	47	19	technique	technique	NOUN
cana-5485	47	20	for	for	ADP
cana-5485	47	21	identifying	identify	VERB
cana-5485	47	22	misleading	mislead	VERB
cana-5485	47	23	information	information	NOUN
cana-5485	47	24	.	.	PUNCT
cana-5485	48	1	the	the	DET
cana-5485	48	2	accuracy	accuracy	NOUN
cana-5485	48	3	rate	rate	NOUN
cana-5485	48	4	of	of	ADP
cana-5485	48	5	the	the	DET
cana-5485	48	6	model	model	NOUN
cana-5485	48	7	was	be	AUX
cana-5485	48	8	80	80	NUM
cana-5485	48	9	%	%	NOUN
cana-5485	48	10	with	with	ADP
cana-5485	48	11	a	a	DET
cana-5485	48	12	95	95	NUM
cana-5485	48	13	%	%	NOUN
cana-5485	48	14	confidence	confidence	NOUN
cana-5485	48	15	interval	interval	NOUN
cana-5485	48	16	and	and	CCONJ
cana-5485	48	17	a	a	DET
cana-5485	48	18	0.05	0.05	NUM
cana-5485	48	19	%	%	NOUN
cana-5485	48	20	confidence	confidence	NOUN
cana-5485	48	21	threshold	threshold	NOUN
cana-5485	48	22	.	.	PUNCT
cana-5485	49	1	communications	communication	NOUN
cana-5485	49	2	on	on	ADP
cana-5485	49	3	applied	apply	VERB
cana-5485	49	4	nonlinear	nonlinear	ADJ
cana-5485	49	5	analysis	analysis	NOUN
cana-5485	49	6	issn	issn	NOUN
cana-5485	49	7	:	:	PUNCT
cana-5485	49	8	1074	1074	NUM
cana-5485	49	9	-	-	PUNCT
cana-5485	49	10	133x	133x	NUM
cana-5485	49	11	vol	vol	VERB
cana-5485	49	12	32	32	NUM
cana-5485	49	13	no	no	NOUN
cana-5485	49	14	.	.	PUNCT
cana-5485	50	1	10s	10	NOUN
cana-5485	50	2	(	(	PUNCT
cana-5485	50	3	2025	2025	NUM
cana-5485	50	4	)	)	PUNCT
cana-5485	50	5	2377	2377	NUM
cana-5485	50	6	https://internationalpubls.com	https://internationalpubls.com	X
cana-5485	51	1	the	the	DET
cana-5485	51	2	experimental	experimental	ADJ
cana-5485	51	3	setup	setup	NOUN
cana-5485	51	4	consists	consist	VERB
cana-5485	51	5	of	of	ADP
cana-5485	51	6	two	two	NUM
cana-5485	51	7	groups	group	NOUN
cana-5485	51	8	.	.	PUNCT
cana-5485	52	1	by	by	ADP
cana-5485	52	2	employing	employ	VERB
cana-5485	52	3	pre	pre	ADJ
cana-5485	52	4	-	-	VERB
cana-5485	52	5	trained	train	VERB
cana-5485	52	6	xlnet	xlnet	NOUN
cana-5485	52	7	with	with	ADP
cana-5485	52	8	fine	fine	ADV
cana-5485	52	9	-	-	PUNCT
cana-5485	52	10	tuning	tuning	NOUN
cana-5485	52	11	on	on	ADP
cana-5485	52	12	the	the	DET
cana-5485	52	13	collected	collect	VERB
cana-5485	52	14	dataset	dataset	NOUN
cana-5485	52	15	,	,	PUNCT
cana-5485	52	16	group	group	NOUN
cana-5485	52	17	1	1	NUM
cana-5485	52	18	gains	gain	NOUN
cana-5485	52	19	access	access	NOUN
cana-5485	52	20	to	to	ADP
cana-5485	52	21	xlnet	xlnet	PROPN
cana-5485	52	22	's	's	PART
cana-5485	52	23	permutation	permutation	NOUN
cana-5485	52	24	-	-	PUNCT
cana-5485	52	25	based	base	VERB
cana-5485	52	26	training	training	NOUN
cana-5485	52	27	methodology	methodology	NOUN
cana-5485	52	28	,	,	PUNCT
cana-5485	52	29	which	which	PRON
cana-5485	52	30	captures	capture	VERB
cana-5485	52	31	complex	complex	ADJ
cana-5485	52	32	contextual	contextual	ADJ
cana-5485	52	33	links	link	NOUN
cana-5485	52	34	in	in	ADP
cana-5485	52	35	news	news	NOUN
cana-5485	52	36	articles	article	NOUN
cana-5485	52	37	.	.	PUNCT
cana-5485	53	1	group	group	NOUN
cana-5485	53	2	2	2	NUM
cana-5485	53	3	uses	use	VERB
cana-5485	53	4	the	the	DET
cana-5485	53	5	hybrid	hybrid	NOUN
cana-5485	53	6	model	model	NOUN
cana-5485	53	7	,	,	PUNCT
cana-5485	53	8	where	where	SCONJ
cana-5485	53	9	fasttext	fasttext	NOUN
cana-5485	53	10	generates	generate	VERB
cana-5485	53	11	word	word	NOUN
cana-5485	53	12	embeddings	embedding	NOUN
cana-5485	53	13	,	,	PUNCT
cana-5485	53	14	cnn	cnn	PROPN
cana-5485	53	15	recovers	recover	VERB
cana-5485	53	16	local	local	ADJ
cana-5485	53	17	text	text	NOUN
cana-5485	53	18	patterns	pattern	NOUN
cana-5485	53	19	,	,	PUNCT
cana-5485	53	20	and	and	CCONJ
cana-5485	53	21	xlnet	xlnet	PROPN
cana-5485	53	22	enhances	enhance	VERB
cana-5485	53	23	classification	classification	NOUN
cana-5485	53	24	performance	performance	NOUN
cana-5485	53	25	by	by	ADP
cana-5485	53	26	understanding	understand	VERB
cana-5485	53	27	bidirectional	bidirectional	ADJ
cana-5485	53	28	context	context	NOUN
cana-5485	53	29	.	.	PUNCT
cana-5485	54	1	group	group	NOUN
cana-5485	54	2	1	1	NUM
cana-5485	54	3	conventional	conventional	ADJ
cana-5485	54	4	machine	machine	NOUN
cana-5485	54	5	learning	learning	NOUN
cana-5485	54	6	models	model	NOUN
cana-5485	54	7	and	and	CCONJ
cana-5485	54	8	rule	rule	NOUN
cana-5485	54	9	-	-	PUNCT
cana-5485	54	10	based	base	VERB
cana-5485	54	11	approaches	approach	NOUN
cana-5485	54	12	are	be	AUX
cana-5485	54	13	the	the	DET
cana-5485	54	14	main	main	ADJ
cana-5485	54	15	components	component	NOUN
cana-5485	54	16	of	of	ADP
cana-5485	54	17	group	group	NOUN
cana-5485	54	18	i	i	PRON
cana-5485	54	19	methods	method	VERB
cana-5485	54	20	.	.	PUNCT
cana-5485	55	1	to	to	PART
cana-5485	55	2	determine	determine	VERB
cana-5485	55	3	whether	whether	SCONJ
cana-5485	55	4	news	news	NOUN
cana-5485	55	5	is	be	AUX
cana-5485	55	6	authentic	authentic	ADJ
cana-5485	55	7	or	or	CCONJ
cana-5485	55	8	fraudulent	fraudulent	ADJ
cana-5485	55	9	,	,	PUNCT
cana-5485	55	10	these	these	DET
cana-5485	55	11	methods	method	NOUN
cana-5485	55	12	use	use	VERB
cana-5485	55	13	manually	manually	ADV
cana-5485	55	14	constructed	construct	VERB
cana-5485	55	15	features	feature	NOUN
cana-5485	55	16	like	like	ADP
cana-5485	55	17	sentiment	sentiment	NOUN
cana-5485	55	18	analysis	analysis	NOUN
cana-5485	55	19	,	,	PUNCT
cana-5485	55	20	linguistic	linguistic	ADJ
cana-5485	55	21	patterns	pattern	NOUN
cana-5485	55	22	,	,	PUNCT
cana-5485	55	23	and	and	CCONJ
cana-5485	55	24	statistical	statistical	ADJ
cana-5485	55	25	indicators	indicator	NOUN
cana-5485	55	26	.	.	PUNCT
cana-5485	56	1	this	this	DET
cana-5485	56	2	category	category	NOUN
cana-5485	56	3	frequently	frequently	ADV
cana-5485	56	4	uses	use	VERB
cana-5485	56	5	traditional	traditional	ADJ
cana-5485	56	6	machine	machine	NOUN
cana-5485	56	7	learning	learning	NOUN
cana-5485	56	8	models	model	NOUN
cana-5485	56	9	such	such	ADJ
cana-5485	56	10	as	as	ADP
cana-5485	56	11	random	random	ADJ
cana-5485	56	12	forest	forest	NOUN
cana-5485	56	13	,	,	PUNCT
cana-5485	56	14	naïve	naïve	ADJ
cana-5485	56	15	bayes	bayes	NOUN
cana-5485	56	16	,	,	PUNCT
cana-5485	56	17	decision	decision	NOUN
cana-5485	56	18	trees	tree	NOUN
cana-5485	56	19	,	,	PUNCT
cana-5485	56	20	and	and	CCONJ
cana-5485	56	21	support	support	VERB
cana-5485	56	22	vector	vector	NOUN
cana-5485	56	23	machines	machine	NOUN
cana-5485	56	24	(	(	PUNCT
cana-5485	56	25	svm	svm	PROPN
cana-5485	56	26	)	)	PUNCT
cana-5485	56	27	.	.	PUNCT
cana-5485	57	1	these	these	DET
cana-5485	57	2	models	model	NOUN
cana-5485	57	3	can	can	AUX
cana-5485	57	4	be	be	AUX
cana-5485	57	5	useful	useful	ADJ
cana-5485	57	6	,	,	PUNCT
cana-5485	57	7	but	but	CCONJ
cana-5485	57	8	they	they	PRON
cana-5485	57	9	are	be	AUX
cana-5485	57	10	frequently	frequently	ADV
cana-5485	57	11	not	not	PART
cana-5485	57	12	flexible	flexible	ADJ
cana-5485	57	13	enough	enough	ADV
cana-5485	57	14	to	to	PART
cana-5485	57	15	recognize	recognize	VERB
cana-5485	57	16	intricate	intricate	ADJ
cana-5485	57	17	and	and	CCONJ
cana-5485	57	18	changing	change	VERB
cana-5485	57	19	disinformation	disinformation	NOUN
cana-5485	57	20	strategies	strategy	NOUN
cana-5485	57	21	.	.	PUNCT
cana-5485	58	1	additionally	additionally	ADV
cana-5485	58	2	,	,	PUNCT
cana-5485	58	3	rule	rule	NOUN
cana-5485	58	4	-	-	PUNCT
cana-5485	58	5	based	base	VERB
cana-5485	58	6	methods	method	NOUN
cana-5485	58	7	may	may	AUX
cana-5485	58	8	not	not	PART
cana-5485	58	9	be	be	AUX
cana-5485	58	10	able	able	ADJ
cana-5485	58	11	to	to	PART
cana-5485	58	12	detect	detect	VERB
cana-5485	58	13	subtle	subtle	ADJ
cana-5485	58	14	and	and	CCONJ
cana-5485	58	15	context	context	NOUN
cana-5485	58	16	-	-	PUNCT
cana-5485	58	17	dependent	dependent	ADJ
cana-5485	58	18	fake	fake	ADJ
cana-5485	58	19	news	news	NOUN
cana-5485	58	20	because	because	SCONJ
cana-5485	58	21	they	they	PRON
cana-5485	58	22	rely	rely	VERB
cana-5485	58	23	on	on	ADP
cana-5485	58	24	predefined	predefine	VERB
cana-5485	58	25	heuristics	heuristic	NOUN
cana-5485	58	26	and	and	CCONJ
cana-5485	58	27	keyword	keyword	NOUN
cana-5485	58	28	matching	matching	NOUN
cana-5485	58	29	.	.	PUNCT
cana-5485	59	1	group	group	NOUN
cana-5485	59	2	2	2	NUM
cana-5485	59	3	demonstrates	demonstrate	VERB
cana-5485	59	4	more	more	ADV
cana-5485	59	5	sophisticated	sophisticated	ADJ
cana-5485	59	6	methods	method	NOUN
cana-5485	59	7	that	that	PRON
cana-5485	59	8	improve	improve	VERB
cana-5485	59	9	fake	fake	ADJ
cana-5485	59	10	news	news	NOUN
cana-5485	59	11	detection	detection	NOUN
cana-5485	59	12	by	by	ADP
cana-5485	59	13	utilizing	utilize	VERB
cana-5485	59	14	deep	deep	ADJ
cana-5485	59	15	learning	learning	NOUN
cana-5485	59	16	and	and	CCONJ
cana-5485	59	17	hybrid	hybrid	ADJ
cana-5485	59	18	ai	ai	PROPN
cana-5485	59	19	models	model	NOUN
cana-5485	59	20	.	.	PUNCT
cana-5485	60	1	this	this	PRON
cana-5485	60	2	comprises	comprise	VERB
cana-5485	60	3	transformer	transformer	NOUN
cana-5485	60	4	-	-	PUNCT
cana-5485	60	5	based	base	VERB
cana-5485	60	6	architectures	architecture	NOUN
cana-5485	60	7	like	like	ADP
cana-5485	60	8	bert	bert	NOUN
cana-5485	60	9	and	and	CCONJ
cana-5485	60	10	gpt	gpt	NOUN
cana-5485	60	11	,	,	PUNCT
cana-5485	60	12	as	as	ADV
cana-5485	60	13	well	well	ADV
cana-5485	60	14	as	as	ADP
cana-5485	60	15	convolutional	convolutional	ADJ
cana-5485	60	16	neural	neural	ADJ
cana-5485	60	17	networks	network	NOUN
cana-5485	60	18	(	(	PUNCT
cana-5485	60	19	cnns	cnns	PROPN
cana-5485	60	20	)	)	PUNCT
cana-5485	60	21	and	and	CCONJ
cana-5485	60	22	recurrent	recurrent	ADJ
cana-5485	60	23	neural	neural	ADJ
cana-5485	60	24	networks	network	NOUN
cana-5485	60	25	(	(	PUNCT
cana-5485	60	26	rnns	rnns	PROPN
cana-5485	60	27	)	)	PUNCT
cana-5485	60	28	.	.	PUNCT
cana-5485	61	1	these	these	DET
cana-5485	61	2	models	model	NOUN
cana-5485	61	3	greatly	greatly	ADV
cana-5485	61	4	increase	increase	VERB
cana-5485	61	5	detection	detection	NOUN
cana-5485	61	6	accuracy	accuracy	NOUN
cana-5485	61	7	by	by	ADP
cana-5485	61	8	capturing	capture	VERB
cana-5485	61	9	complex	complex	ADJ
cana-5485	61	10	textual	textual	ADJ
cana-5485	61	11	features	feature	NOUN
cana-5485	61	12	and	and	CCONJ
cana-5485	61	13	contextual	contextual	ADJ
cana-5485	61	14	relationships	relationship	NOUN
cana-5485	61	15	.	.	PUNCT
cana-5485	62	1	furthermore	furthermore	ADV
cana-5485	62	2	,	,	PUNCT
cana-5485	62	3	hybrid	hybrid	ADJ
cana-5485	62	4	models	model	NOUN
cana-5485	62	5	integrate	integrate	VERB
cana-5485	62	6	several	several	ADJ
cana-5485	62	7	methods	method	NOUN
cana-5485	62	8	,	,	PUNCT
cana-5485	62	9	like	like	ADP
cana-5485	62	10	combining	combine	VERB
cana-5485	62	11	machine	machine	NOUN
cana-5485	62	12	learning	learn	VERB
cana-5485	62	13	classifiers	classifier	NOUN
cana-5485	62	14	with	with	ADP
cana-5485	62	15	deep	deep	ADJ
cana-5485	62	16	learning	learning	NOUN
cana-5485	62	17	frameworks	framework	NOUN
cana-5485	62	18	for	for	ADP
cana-5485	62	19	increased	increase	VERB
cana-5485	62	20	robustness	robustness	NOUN
cana-5485	62	21	or	or	CCONJ
cana-5485	62	22	cnns	cnn	NOUN
cana-5485	62	23	for	for	ADP
cana-5485	62	24	feature	feature	NOUN
cana-5485	62	25	extraction	extraction	NOUN
cana-5485	62	26	with	with	ADP
cana-5485	62	27	rnns	rnn	NOUN
cana-5485	62	28	for	for	ADP
cana-5485	62	29	sequential	sequential	ADJ
cana-5485	62	30	text	text	NOUN
cana-5485	62	31	analysis	analysis	NOUN
cana-5485	62	32	.	.	PUNCT
cana-5485	63	1	explainable	explainable	ADJ
cana-5485	63	2	ai	ai	NOUN
cana-5485	63	3	(	(	PUNCT
cana-5485	63	4	xai	xai	PROPN
cana-5485	63	5	)	)	PUNCT
cana-5485	63	6	is	be	AUX
cana-5485	63	7	used	use	VERB
cana-5485	63	8	to	to	PART
cana-5485	63	9	further	far	ADV
cana-5485	63	10	improve	improve	VERB
cana-5485	63	11	these	these	DET
cana-5485	63	12	methods	method	NOUN
cana-5485	63	13	by	by	ADP
cana-5485	63	14	offering	offer	VERB
cana-5485	63	15	interpretability	interpretability	NOUN
cana-5485	63	16	and	and	CCONJ
cana-5485	63	17	transparency	transparency	NOUN
cana-5485	63	18	in	in	ADP
cana-5485	63	19	decision	decision	NOUN
cana-5485	63	20	-	-	PUNCT
cana-5485	63	21	making	making	NOUN
cana-5485	63	22	,	,	PUNCT
cana-5485	63	23	which	which	PRON
cana-5485	63	24	aids	aid	VERB
cana-5485	63	25	in	in	ADP
cana-5485	63	26	understanding	understand	VERB
cana-5485	63	27	the	the	DET
cana-5485	63	28	reasons	reason	NOUN
cana-5485	63	29	behind	behind	ADP
cana-5485	63	30	a	a	DET
cana-5485	63	31	given	give	VERB
cana-5485	63	32	news	news	NOUN
cana-5485	63	33	item	item	NOUN
cana-5485	63	34	's	's	PART
cana-5485	63	35	classification	classification	NOUN
cana-5485	63	36	as	as	ADP
cana-5485	63	37	real	real	ADJ
cana-5485	63	38	or	or	CCONJ
cana-5485	63	39	fake	fake	ADJ
cana-5485	63	40	.	.	PUNCT
cana-5485	64	1	fig	fig	NOUN
cana-5485	64	2	.	.	PUNCT
cana-5485	65	1	1	1	X
cana-5485	65	2	.	.	X
cana-5485	65	3	hybrid	hybrid	ADJ
cana-5485	65	4	deep	deep	ADJ
cana-5485	65	5	learning	learning	NOUN
cana-5485	65	6	for	for	ADP
cana-5485	65	7	advanced	advanced	ADJ
cana-5485	65	8	fake	fake	ADJ
cana-5485	65	9	news	news	NOUN
cana-5485	65	10	detection	detection	NOUN
cana-5485	65	11	using	use	VERB
cana-5485	65	12	explainable	explainable	ADJ
cana-5485	65	13	ai	ai	NOUN
cana-5485	65	14	and	and	CCONJ
cana-5485	65	15	fasttext	fasttext	PROPN
cana-5485	65	16	communications	communication	NOUN
cana-5485	65	17	on	on	ADP
cana-5485	65	18	applied	apply	VERB
cana-5485	65	19	nonlinear	nonlinear	ADJ
cana-5485	65	20	analysis	analysis	NOUN
cana-5485	65	21	issn	issn	NOUN
cana-5485	65	22	:	:	PUNCT
cana-5485	65	23	1074	1074	NUM
cana-5485	65	24	-	-	PUNCT
cana-5485	65	25	133x	133x	NUM
cana-5485	65	26	vol	vol	VERB
cana-5485	65	27	32	32	NUM
cana-5485	65	28	no	no	NOUN
cana-5485	65	29	.	.	PUNCT
cana-5485	66	1	10s	10	NOUN
cana-5485	66	2	(	(	PUNCT
cana-5485	66	3	2025	2025	NUM
cana-5485	66	4	)	)	PUNCT
cana-5485	66	5	2378	2378	NUM
cana-5485	66	6	https://internationalpubls.com	https://internationalpubls.com	X
cana-5485	66	7	4	4	NUM
cana-5485	66	8	.	.	PUNCT
cana-5485	66	9	results	result	NOUN
cana-5485	66	10	when	when	SCONJ
cana-5485	66	11	compared	compare	VERB
cana-5485	66	12	to	to	ADP
cana-5485	66	13	other	other	ADJ
cana-5485	66	14	models	model	NOUN
cana-5485	66	15	like	like	ADP
cana-5485	66	16	bert	bert	PROPN
cana-5485	66	17	and	and	CCONJ
cana-5485	66	18	roberta	roberta	PROPN
cana-5485	66	19	,	,	PUNCT
cana-5485	66	20	the	the	DET
cana-5485	66	21	outcomes	outcome	NOUN
cana-5485	66	22	of	of	ADP
cana-5485	66	23	our	our	PRON
cana-5485	66	24	suggested	suggest	VERB
cana-5485	66	25	hybrid	hybrid	ADJ
cana-5485	66	26	deep	deep	ADJ
cana-5485	66	27	learning	learning	NOUN
cana-5485	66	28	model	model	NOUN
cana-5485	66	29	for	for	ADP
cana-5485	66	30	fake	fake	ADJ
cana-5485	66	31	news	news	NOUN
cana-5485	66	32	identification	identification	NOUN
cana-5485	66	33	show	show	VERB
cana-5485	66	34	notable	notable	ADJ
cana-5485	66	35	improvements	improvement	NOUN
cana-5485	66	36	.	.	PUNCT
cana-5485	67	1	the	the	DET
cana-5485	67	2	model	model	NOUN
cana-5485	67	3	's	's	PART
cana-5485	67	4	performance	performance	NOUN
cana-5485	67	5	is	be	AUX
cana-5485	67	6	evaluated	evaluate	VERB
cana-5485	67	7	using	use	VERB
cana-5485	67	8	metrics	metric	NOUN
cana-5485	67	9	such	such	ADJ
cana-5485	67	10	as	as	ADP
cana-5485	67	11	f1	f1	NOUN
cana-5485	67	12	-	-	PUNCT
cana-5485	67	13	score	score	NOUN
cana-5485	67	14	,	,	PUNCT
cana-5485	67	15	recall	recall	NOUN
cana-5485	67	16	,	,	PUNCT
cana-5485	67	17	accuracy	accuracy	NOUN
cana-5485	67	18	,	,	PUNCT
cana-5485	67	19	and	and	CCONJ
cana-5485	67	20	precision	precision	NOUN
cana-5485	67	21	.	.	PUNCT
cana-5485	68	1	the	the	DET
cana-5485	68	2	bert	bert	PROPN
cana-5485	68	3	model	model	PROPN
cana-5485	68	4	's	's	PART
cana-5485	68	5	f1	f1	NOUN
cana-5485	68	6	-	-	PUNCT
cana-5485	68	7	score	score	NOUN
cana-5485	68	8	,	,	PUNCT
cana-5485	68	9	recall	recall	NOUN
cana-5485	68	10	,	,	PUNCT
cana-5485	68	11	accuracy	accuracy	NOUN
cana-5485	68	12	,	,	PUNCT
cana-5485	68	13	and	and	CCONJ
cana-5485	68	14	precision	precision	NOUN
cana-5485	68	15	were	be	AUX
cana-5485	68	16	86.5	86.5	NUM
cana-5485	68	17	%	%	NOUN
cana-5485	68	18	,	,	PUNCT
cana-5485	68	19	85.9	85.9	NUM
cana-5485	68	20	%	%	NOUN
cana-5485	68	21	,	,	PUNCT
cana-5485	68	22	and	and	CCONJ
cana-5485	68	23	88.4	88.4	NUM
cana-5485	68	24	%	%	NOUN
cana-5485	68	25	,	,	PUNCT
cana-5485	68	26	in	in	ADP
cana-5485	68	27	that	that	DET
cana-5485	68	28	order	order	NOUN
cana-5485	68	29	.	.	PUNCT
cana-5485	69	1	with	with	ADP
cana-5485	69	2	an	an	DET
cana-5485	69	3	f1	f1	NOUN
cana-5485	69	4	-	-	PUNCT
cana-5485	69	5	score	score	NOUN
cana-5485	69	6	of	of	ADP
cana-5485	69	7	88.6	88.6	NUM
cana-5485	69	8	%	%	NOUN
cana-5485	69	9	,	,	PUNCT
cana-5485	69	10	recall	recall	NOUN
cana-5485	69	11	of	of	ADP
cana-5485	69	12	88.0	88.0	NUM
cana-5485	69	13	%	%	NOUN
cana-5485	69	14	,	,	PUNCT
cana-5485	69	15	accuracy	accuracy	NOUN
cana-5485	69	16	of	of	ADP
cana-5485	69	17	90.2	90.2	NUM
cana-5485	69	18	%	%	NOUN
cana-5485	69	19	,	,	PUNCT
cana-5485	69	20	and	and	CCONJ
cana-5485	69	21	precision	precision	NOUN
cana-5485	69	22	of	of	ADP
cana-5485	69	23	89.3	89.3	NUM
cana-5485	69	24	%	%	NOUN
cana-5485	69	25	,	,	PUNCT
cana-5485	69	26	roberta	roberta	PROPN
cana-5485	69	27	,	,	PUNCT
cana-5485	69	28	an	an	DET
cana-5485	69	29	enhanced	enhanced	ADJ
cana-5485	69	30	version	version	NOUN
cana-5485	69	31	of	of	ADP
cana-5485	69	32	bert	bert	PROPN
cana-5485	69	33	,	,	PUNCT
cana-5485	69	34	outperformed	outperform	VERB
cana-5485	69	35	.	.	PUNCT
cana-5485	70	1	these	these	DET
cana-5485	70	2	models	model	NOUN
cana-5485	70	3	are	be	AUX
cana-5485	70	4	effective	effective	ADJ
cana-5485	70	5	,	,	PUNCT
cana-5485	70	6	but	but	CCONJ
cana-5485	70	7	because	because	SCONJ
cana-5485	70	8	they	they	PRON
cana-5485	70	9	can	can	AUX
cana-5485	70	10	not	not	PART
cana-5485	70	11	be	be	AUX
cana-5485	70	12	interpreted	interpret	VERB
cana-5485	70	13	,	,	PUNCT
cana-5485	70	14	it	it	PRON
cana-5485	70	15	is	be	AUX
cana-5485	70	16	difficult	difficult	ADJ
cana-5485	70	17	to	to	PART
cana-5485	70	18	substantiate	substantiate	VERB
cana-5485	70	19	their	their	PRON
cana-5485	70	20	predictions.90.9	predictions.90.9	ADJ
cana-5485	70	21	%	%	NOUN
cana-5485	70	22	recall	recall	NOUN
cana-5485	70	23	,	,	PUNCT
cana-5485	70	24	92.8	92.8	NUM
cana-5485	70	25	%	%	NOUN
cana-5485	70	26	accuracy	accuracy	NOUN
cana-5485	70	27	,	,	PUNCT
cana-5485	70	28	91.5	91.5	NUM
cana-5485	70	29	%	%	NOUN
cana-5485	70	30	precision	precision	NOUN
cana-5485	70	31	,	,	PUNCT
cana-5485	70	32	and	and	CCONJ
cana-5485	70	33	91.2	91.2	NUM
cana-5485	70	34	%	%	NOUN
cana-5485	70	35	f1	f1	NOUN
cana-5485	70	36	-	-	PUNCT
cana-5485	70	37	score	score	NOUN
cana-5485	70	38	were	be	AUX
cana-5485	70	39	attained	attain	VERB
cana-5485	70	40	by	by	ADP
cana-5485	70	41	the	the	DET
cana-5485	70	42	hybrid	hybrid	NOUN
cana-5485	70	43	model	model	NOUN
cana-5485	70	44	.	.	PUNCT
cana-5485	71	1	cnn	cnn	PROPN
cana-5485	71	2	was	be	AUX
cana-5485	71	3	able	able	ADJ
cana-5485	71	4	to	to	PART
cana-5485	71	5	recognize	recognize	VERB
cana-5485	71	6	patterns	pattern	NOUN
cana-5485	71	7	that	that	PRON
cana-5485	71	8	suggested	suggest	VERB
cana-5485	71	9	fake	fake	ADJ
cana-5485	71	10	news	news	NOUN
cana-5485	71	11	,	,	PUNCT
cana-5485	71	12	even	even	ADV
cana-5485	71	13	though	though	SCONJ
cana-5485	71	14	the	the	DET
cana-5485	71	15	fasttext	fasttext	PROPN
cana-5485	71	16	component	component	NOUN
cana-5485	71	17	helped	help	VERB
cana-5485	71	18	to	to	PART
cana-5485	71	19	improve	improve	VERB
cana-5485	71	20	word	word	NOUN
cana-5485	71	21	representation.table	representation.table	ADJ
cana-5485	71	22	1	1	NUM
cana-5485	71	23	shows	show	VERB
cana-5485	71	24	the	the	DET
cana-5485	71	25	classification	classification	NOUN
cana-5485	71	26	performance	performance	NOUN
cana-5485	71	27	of	of	ADP
cana-5485	71	28	the	the	DET
cana-5485	71	29	suggested	suggest	VERB
cana-5485	71	30	hybrid	hybrid	ADJ
cana-5485	71	31	deep	deep	ADJ
cana-5485	71	32	learning	learning	NOUN
cana-5485	71	33	model	model	NOUN
cana-5485	71	34	,	,	PUNCT
cana-5485	71	35	which	which	PRON
cana-5485	71	36	blends	blend	VERB
cana-5485	71	37	explainable	explainable	ADJ
cana-5485	71	38	ai	ai	NOUN
cana-5485	71	39	(	(	PUNCT
cana-5485	71	40	xai	xai	PROPN
cana-5485	71	41	)	)	PUNCT
cana-5485	71	42	methods	method	NOUN
cana-5485	71	43	like	like	ADP
cana-5485	71	44	shap	shap	PROPN
cana-5485	71	45	with	with	ADP
cana-5485	71	46	cnn	cnn	PROPN
cana-5485	71	47	and	and	CCONJ
cana-5485	71	48	fasttext	fasttext	PROPN
cana-5485	71	49	.	.	PUNCT
cana-5485	72	1	table	table	NOUN
cana-5485	72	2	2	2	NUM
cana-5485	72	3	summarizes	summarize	NOUN
cana-5485	72	4	a	a	DET
cana-5485	72	5	comparison	comparison	NOUN
cana-5485	72	6	study	study	NOUN
cana-5485	72	7	that	that	PRON
cana-5485	72	8	used	use	VERB
cana-5485	72	9	t	t	NOUN
cana-5485	72	10	-	-	PUNCT
cana-5485	72	11	tests	test	NOUN
cana-5485	72	12	to	to	PART
cana-5485	72	13	assess	assess	VERB
cana-5485	72	14	the	the	DET
cana-5485	72	15	performance	performance	NOUN
cana-5485	72	16	differences	difference	NOUN
cana-5485	72	17	between	between	ADP
cana-5485	72	18	xlnet	xlnet	PROPN
cana-5485	72	19	and	and	CCONJ
cana-5485	72	20	the	the	DET
cana-5485	72	21	hybrid	hybrid	ADJ
cana-5485	72	22	cnn	cnn	PROPN
cana-5485	72	23	-	-	PUNCT
cana-5485	72	24	fasttext	fasttext	PROPN
cana-5485	72	25	model.a	model.a	VERB
cana-5485	72	26	statistically	statistically	ADV
cana-5485	72	27	significant	significant	ADJ
cana-5485	72	28	difference	difference	NOUN
cana-5485	72	29	(	(	PUNCT
cana-5485	72	30	p	p	X
cana-5485	72	31	<	<	X
cana-5485	72	32	0.05	0.05	NUM
cana-5485	72	33	)	)	PUNCT
cana-5485	72	34	between	between	ADP
cana-5485	72	35	the	the	DET
cana-5485	72	36	suggested	suggest	VERB
cana-5485	72	37	hybrid	hybrid	NOUN
cana-5485	72	38	model	model	NOUN
cana-5485	72	39	and	and	CCONJ
cana-5485	72	40	current	current	ADJ
cana-5485	72	41	approaches	approach	NOUN
cana-5485	72	42	is	be	AUX
cana-5485	72	43	also	also	ADV
cana-5485	72	44	shown	show	VERB
cana-5485	72	45	in	in	ADP
cana-5485	72	46	table	table	NOUN
cana-5485	72	47	3	3	NUM
cana-5485	72	48	,	,	PUNCT
cana-5485	72	49	which	which	PRON
cana-5485	72	50	displays	display	VERB
cana-5485	72	51	the	the	DET
cana-5485	72	52	classification	classification	NOUN
cana-5485	72	53	performance	performance	NOUN
cana-5485	72	54	's	's	PART
cana-5485	72	55	mean	mean	ADJ
cana-5485	72	56	,	,	PUNCT
cana-5485	72	57	standard	standard	ADJ
cana-5485	72	58	deviation	deviation	NOUN
cana-5485	72	59	,	,	PUNCT
cana-5485	72	60	and	and	CCONJ
cana-5485	72	61	significance	significance	NOUN
cana-5485	72	62	level	level	NOUN
cana-5485	72	63	.	.	PUNCT
cana-5485	73	1	convolutional	convolutional	ADJ
cana-5485	73	2	layers	layer	NOUN
cana-5485	73	3	are	be	AUX
cana-5485	73	4	used	use	VERB
cana-5485	73	5	for	for	ADP
cana-5485	73	6	feature	feature	NOUN
cana-5485	73	7	extraction	extraction	NOUN
cana-5485	73	8	,	,	PUNCT
cana-5485	73	9	and	and	CCONJ
cana-5485	73	10	fasttext	fasttext	PROPN
cana-5485	73	11	embeddings	embedding	NOUN
cana-5485	73	12	are	be	AUX
cana-5485	73	13	used	use	VERB
cana-5485	73	14	for	for	ADP
cana-5485	73	15	efficient	efficient	ADJ
cana-5485	73	16	word	word	NOUN
cana-5485	73	17	representation	representation	NOUN
cana-5485	73	18	in	in	ADP
cana-5485	73	19	the	the	DET
cana-5485	73	20	architecture	architecture	NOUN
cana-5485	73	21	of	of	ADP
cana-5485	73	22	the	the	DET
cana-5485	73	23	proposed	propose	VERB
cana-5485	73	24	a	a	DET
cana-5485	73	25	combination	combination	NOUN
cana-5485	73	26	deep	deep	ADJ
cana-5485	73	27	learning	learning	NOUN
cana-5485	73	28	structure	structure	NOUN
cana-5485	73	29	,	,	PUNCT
cana-5485	73	30	which	which	PRON
cana-5485	73	31	is	be	AUX
cana-5485	73	32	displayed	display	VERB
cana-5485	73	33	in	in	ADP
cana-5485	73	34	fig	fig	NOUN
cana-5485	73	35	.	.	PUNCT
cana-5485	74	1	2	2	X
cana-5485	74	2	.	.	X
cana-5485	74	3	the	the	DET
cana-5485	74	4	precision	precision	NOUN
cana-5485	74	5	and	and	CCONJ
cana-5485	74	6	recall	recall	NOUN
cana-5485	74	7	metrics	metric	NOUN
cana-5485	74	8	of	of	ADP
cana-5485	74	9	the	the	DET
cana-5485	74	10	model	model	NOUN
cana-5485	74	11	's	's	PART
cana-5485	74	12	classification	classification	NOUN
cana-5485	74	13	report	report	NOUN
cana-5485	74	14	are	be	AUX
cana-5485	74	15	significantly	significantly	ADV
cana-5485	74	16	improved	improve	VERB
cana-5485	74	17	when	when	SCONJ
cana-5485	74	18	compared	compare	VERB
cana-5485	74	19	to	to	ADP
cana-5485	74	20	more	more	ADV
cana-5485	74	21	traditional	traditional	ADJ
cana-5485	74	22	transformer	transformer	NOUN
cana-5485	74	23	-	-	PUNCT
cana-5485	74	24	based	base	VERB
cana-5485	74	25	techniques	technique	NOUN
cana-5485	74	26	(	(	PUNCT
cana-5485	74	27	fig	fig	NOUN
cana-5485	74	28	.	.	PUNCT
cana-5485	75	1	3	3	NUM
cana-5485	75	2	)	)	PUNCT
cana-5485	75	3	.	.	PUNCT
cana-5485	76	1	the	the	DET
cana-5485	76	2	evaluation	evaluation	NOUN
cana-5485	76	3	's	's	PART
cana-5485	76	4	conclusions	conclusion	NOUN
cana-5485	76	5	,	,	PUNCT
cana-5485	76	6	as	as	SCONJ
cana-5485	76	7	illustrated	illustrate	VERB
cana-5485	76	8	in	in	ADP
cana-5485	76	9	fig	fig	NOUN
cana-5485	76	10	.	.	PUNCT
cana-5485	77	1	4	4	NUM
cana-5485	77	2	,	,	PUNCT
cana-5485	77	3	provide	provide	VERB
cana-5485	77	4	a	a	DET
cana-5485	77	5	clear	clear	ADJ
cana-5485	77	6	understanding	understanding	NOUN
cana-5485	77	7	of	of	ADP
cana-5485	77	8	the	the	DET
cana-5485	77	9	procedure	procedure	NOUN
cana-5485	77	10	for	for	ADP
cana-5485	77	11	making	make	VERB
cana-5485	77	12	choices	choice	NOUN
cana-5485	77	13	and	and	CCONJ
cana-5485	77	14	show	show	VERB
cana-5485	77	15	how	how	SCONJ
cana-5485	77	16	the	the	DET
cana-5485	77	17	application	application	NOUN
cana-5485	77	18	of	of	ADP
cana-5485	77	19	explainable	explainable	ADJ
cana-5485	77	20	ai	ai	NOUN
cana-5485	77	21	techniques	technique	NOUN
cana-5485	77	22	can	can	AUX
cana-5485	77	23	improve	improve	VERB
cana-5485	77	24	accuracy	accuracy	NOUN
cana-5485	77	25	.	.	PUNCT
cana-5485	78	1	compared	compare	VERB
cana-5485	78	2	to	to	ADP
cana-5485	78	3	models	model	NOUN
cana-5485	78	4	like	like	ADP
cana-5485	78	5	bert	bert	PROPN
cana-5485	78	6	and	and	CCONJ
cana-5485	78	7	roberta	roberta	PROPN
cana-5485	78	8	,	,	PUNCT
cana-5485	78	9	the	the	DET
cana-5485	78	10	hybrid	hybrid	NOUN
cana-5485	78	11	model	model	NOUN
cana-5485	78	12	achieves	achieve	VERB
cana-5485	78	13	a	a	DET
cana-5485	78	14	better	well	ADJ
cana-5485	78	15	balance	balance	NOUN
cana-5485	78	16	between	between	ADP
cana-5485	78	17	explainability	explainability	NOUN
cana-5485	78	18	and	and	CCONJ
cana-5485	78	19	performance	performance	NOUN
cana-5485	78	20	.	.	PUNCT
cana-5485	79	1	fig	fig	NOUN
cana-5485	79	2	.	.	PUNCT
cana-5485	80	1	2	2	X
cana-5485	80	2	.	.	X
cana-5485	80	3	the	the	DET
cana-5485	80	4	relatively	relatively	ADV
cana-5485	80	5	close	close	ADJ
cana-5485	80	6	median	median	NOUN
cana-5485	80	7	(	(	PUNCT
cana-5485	80	8	middle	middle	ADJ
cana-5485	80	9	line	line	NOUN
cana-5485	80	10	in	in	ADP
cana-5485	80	11	each	each	DET
cana-5485	80	12	box	box	NOUN
cana-5485	80	13	)	)	PUNCT
cana-5485	80	14	for	for	ADP
cana-5485	80	15	each	each	DET
cana-5485	80	16	type	type	NOUN
cana-5485	80	17	of	of	ADP
cana-5485	80	18	news	news	NOUN
cana-5485	80	19	indicates	indicate	VERB
cana-5485	80	20	that	that	SCONJ
cana-5485	80	21	the	the	DET
cana-5485	80	22	normal	normal	ADJ
cana-5485	80	23	lengths	length	NOUN
cana-5485	80	24	of	of	ADP
cana-5485	80	25	fake	fake	ADJ
cana-5485	80	26	and	and	CCONJ
cana-5485	80	27	true	true	ADJ
cana-5485	80	28	news	news	NOUN
cana-5485	80	29	stories	story	NOUN
cana-5485	80	30	are	be	AUX
cana-5485	80	31	similar	similar	ADJ
cana-5485	80	32	.	.	PUNCT
cana-5485	81	1	the	the	DET
cana-5485	81	2	box	box	NOUN
cana-5485	81	3	displays	display	VERB
cana-5485	81	4	the	the	DET
cana-5485	81	5	iqr	iqr	PROPN
cana-5485	81	6	,	,	PUNCT
cana-5485	81	7	or	or	CCONJ
cana-5485	81	8	middle	middle	ADJ
cana-5485	81	9	50	50	NUM
cana-5485	81	10	%	%	NOUN
cana-5485	81	11	of	of	ADP
cana-5485	81	12	the	the	DET
cana-5485	81	13	data	datum	NOUN
cana-5485	81	14	.	.	PUNCT
cana-5485	82	1	the	the	DET
cana-5485	82	2	somewhat	somewhat	ADV
cana-5485	82	3	larger	large	ADJ
cana-5485	82	4	iqr	iqr	NOUN
cana-5485	82	5	for	for	ADP
cana-5485	82	6	true	true	ADJ
cana-5485	82	7	news	news	NOUN
cana-5485	82	8	articles	article	NOUN
cana-5485	82	9	compared	compare	VERB
cana-5485	82	10	to	to	ADP
cana-5485	82	11	fake	fake	ADJ
cana-5485	82	12	articles	article	NOUN
cana-5485	82	13	suggests	suggest	VERB
cana-5485	82	14	that	that	SCONJ
cana-5485	82	15	true	true	ADJ
cana-5485	82	16	news	news	NOUN
cana-5485	82	17	articles	article	NOUN
cana-5485	82	18	typically	typically	ADV
cana-5485	82	19	have	have	VERB
cana-5485	82	20	more	more	ADJ
cana-5485	82	21	variance	variance	NOUN
cana-5485	82	22	in	in	ADP
cana-5485	82	23	duration	duration	NOUN
cana-5485	82	24	.	.	PUNCT
cana-5485	83	1	communications	communication	NOUN
cana-5485	83	2	on	on	ADP
cana-5485	83	3	applied	apply	VERB
cana-5485	83	4	nonlinear	nonlinear	ADJ
cana-5485	83	5	analysis	analysis	NOUN
cana-5485	83	6	issn	issn	NOUN
cana-5485	83	7	:	:	PUNCT
cana-5485	83	8	1074	1074	NUM
cana-5485	83	9	-	-	PUNCT
cana-5485	83	10	133x	133x	NUM
cana-5485	83	11	vol	vol	VERB
cana-5485	83	12	32	32	NUM
cana-5485	83	13	no	no	NOUN
cana-5485	83	14	.	.	PUNCT
cana-5485	84	1	10s	10	NOUN
cana-5485	84	2	(	(	PUNCT
cana-5485	84	3	2025	2025	NUM
cana-5485	84	4	)	)	PUNCT
cana-5485	84	5	2379	2379	NUM
cana-5485	84	6	https://internationalpubls.com	https://internationalpubls.com	X
cana-5485	84	7	fig	fig	NOUN
cana-5485	84	8	.	.	PUNCT
cana-5485	85	1	3	3	X
cana-5485	85	2	.	.	PUNCT
cana-5485	86	1	this	this	DET
cana-5485	86	2	bar	bar	NOUN
cana-5485	86	3	chart	chart	NOUN
cana-5485	86	4	displays	display	VERB
cana-5485	86	5	the	the	DET
cana-5485	86	6	most	most	ADV
cana-5485	86	7	prevalent	prevalent	ADJ
cana-5485	86	8	words	word	NOUN
cana-5485	86	9	in	in	ADP
cana-5485	86	10	the	the	DET
cana-5485	86	11	dataset	dataset	NOUN
cana-5485	86	12	along	along	ADP
cana-5485	86	13	with	with	ADP
cana-5485	86	14	the	the	DET
cana-5485	86	15	frequencies	frequency	NOUN
cana-5485	86	16	that	that	PRON
cana-5485	86	17	correspond	correspond	VERB
cana-5485	86	18	to	to	ADP
cana-5485	86	19	them	they	PRON
cana-5485	86	20	.	.	PUNCT
cana-5485	87	1	donald	donald	PROPN
cana-5485	87	2	trump	trump	PROPN
cana-5485	87	3	is	be	AUX
cana-5485	87	4	associated	associate	VERB
cana-5485	87	5	with	with	ADP
cana-5485	87	6	a	a	DET
cana-5485	87	7	significant	significant	ADJ
cana-5485	87	8	portion	portion	NOUN
cana-5485	87	9	of	of	ADP
cana-5485	87	10	the	the	DET
cana-5485	87	11	dataset	dataset	NOUN
cana-5485	87	12	,	,	PUNCT
cana-5485	87	13	as	as	SCONJ
cana-5485	87	14	evidenced	evidence	VERB
cana-5485	87	15	by	by	ADP
cana-5485	87	16	the	the	DET
cana-5485	87	17	fact	fact	NOUN
cana-5485	87	18	that	that	SCONJ
cana-5485	87	19	the	the	DET
cana-5485	87	20	word	word	NOUN
cana-5485	87	21	"	"	PUNCT
cana-5485	87	22	trump	trump	NOUN
cana-5485	87	23	"	"	PUNCT
cana-5485	87	24	appears	appear	VERB
cana-5485	87	25	far	far	ADV
cana-5485	87	26	more	more	ADV
cana-5485	87	27	frequently	frequently	ADV
cana-5485	87	28	than	than	ADP
cana-5485	87	29	any	any	DET
cana-5485	87	30	other	other	ADJ
cana-5485	87	31	word	word	NOUN
cana-5485	87	32	.	.	PUNCT
cana-5485	88	1	this	this	PRON
cana-5485	88	2	might	might	AUX
cana-5485	88	3	indicate	indicate	VERB
cana-5485	88	4	that	that	SCONJ
cana-5485	88	5	the	the	DET
cana-5485	88	6	dataset	dataset	NOUN
cana-5485	88	7	contains	contain	VERB
cana-5485	88	8	a	a	DET
cana-5485	88	9	large	large	ADJ
cana-5485	88	10	number	number	NOUN
cana-5485	88	11	of	of	ADP
cana-5485	88	12	news	news	NOUN
cana-5485	88	13	items	item	NOUN
cana-5485	88	14	,	,	PUNCT
cana-5485	88	15	discussions	discussion	NOUN
cana-5485	88	16	,	,	PUNCT
cana-5485	88	17	or	or	CCONJ
cana-5485	88	18	debates	debate	NOUN
cana-5485	88	19	about	about	ADP
cana-5485	88	20	trump	trump	NOUN
cana-5485	88	21	,	,	PUNCT
cana-5485	88	22	possibly	possibly	ADV
cana-5485	88	23	related	relate	VERB
cana-5485	88	24	to	to	ADP
cana-5485	88	25	politics	politic	NOUN
cana-5485	88	26	or	or	CCONJ
cana-5485	88	27	elections	election	NOUN
cana-5485	88	28	.	.	PUNCT
cana-5485	89	1	words	word	NOUN
cana-5485	89	2	like	like	ADP
cana-5485	89	3	"	"	PUNCT
cana-5485	89	4	said	say	VERB
cana-5485	89	5	,	,	PUNCT
cana-5485	89	6	"	"	PUNCT
cana-5485	89	7	"	"	PUNCT
cana-5485	89	8	president	president	NOUN
cana-5485	89	9	,	,	PUNCT
cana-5485	89	10	"	"	PUNCT
cana-5485	89	11	"	"	PUNCT
cana-5485	89	12	people	people	NOUN
cana-5485	89	13	,	,	PUNCT
cana-5485	89	14	"	"	PUNCT
cana-5485	89	15	and	and	CCONJ
cana-5485	89	16	"	"	PUNCT
cana-5485	89	17	just	just	ADV
cana-5485	89	18	"	"	PUNCT
cana-5485	89	19	are	be	AUX
cana-5485	89	20	also	also	ADV
cana-5485	89	21	commonly	commonly	ADV
cana-5485	89	22	used	use	VERB
cana-5485	89	23	.	.	PUNCT
cana-5485	90	1	these	these	PRON
cana-5485	90	2	are	be	AUX
cana-5485	90	3	general	general	ADJ
cana-5485	90	4	terms	term	NOUN
cana-5485	90	5	that	that	PRON
cana-5485	90	6	are	be	AUX
cana-5485	90	7	commonly	commonly	ADV
cana-5485	90	8	used	use	VERB
cana-5485	90	9	in	in	ADP
cana-5485	90	10	news	news	NOUN
cana-5485	90	11	reporting	reporting	NOUN
cana-5485	90	12	,	,	PUNCT
cana-5485	90	13	even	even	ADV
cana-5485	90	14	though	though	SCONJ
cana-5485	90	15	"	"	PUNCT
cana-5485	90	16	said	say	VERB
cana-5485	90	17	"	"	PUNCT
cana-5485	90	18	is	be	AUX
cana-5485	90	19	often	often	ADV
cana-5485	90	20	used	use	VERB
cana-5485	90	21	to	to	PART
cana-5485	90	22	quote	quote	VERB
cana-5485	90	23	statements	statement	NOUN
cana-5485	90	24	.	.	PUNCT
cana-5485	91	1	fig	fig	NOUN
cana-5485	91	2	.	.	PUNCT
cana-5485	92	1	4	4	X
cana-5485	92	2	.	.	X
cana-5485	92	3	the	the	DET
cana-5485	92	4	chart	chart	NOUN
cana-5485	92	5	below	below	ADV
cana-5485	92	6	shows	show	VERB
cana-5485	92	7	the	the	DET
cana-5485	92	8	training	training	NOUN
cana-5485	92	9	and	and	CCONJ
cana-5485	92	10	validation	validation	NOUN
cana-5485	92	11	loss	loss	NOUN
cana-5485	92	12	for	for	ADP
cana-5485	92	13	a	a	DET
cana-5485	92	14	machine	machine	NOUN
cana-5485	92	15	learning	learn	VERB
cana-5485	92	16	model	model	NOUN
cana-5485	92	17	over	over	ADP
cana-5485	92	18	four	four	NUM
cana-5485	92	19	periods	period	NOUN
cana-5485	92	20	.	.	PUNCT
cana-5485	93	1	this	this	DET
cana-5485	93	2	graph	graph	NOUN
cana-5485	93	3	shows	show	VERB
cana-5485	93	4	the	the	DET
cana-5485	93	5	loss	loss	NOUN
cana-5485	93	6	during	during	ADP
cana-5485	93	7	training	training	NOUN
cana-5485	93	8	and	and	CCONJ
cana-5485	93	9	validation	validation	NOUN
cana-5485	93	10	for	for	ADP
cana-5485	93	11	a	a	DET
cana-5485	93	12	machine	machine	NOUN
cana-5485	93	13	learning	learn	VERB
cana-5485	93	14	model	model	NOUN
cana-5485	93	15	over	over	ADP
cana-5485	93	16	four	four	NUM
cana-5485	93	17	epochs	epoch	NOUN
cana-5485	93	18	.	.	PUNCT
cana-5485	94	1	the	the	DET
cana-5485	94	2	green	green	ADJ
cana-5485	94	3	line	line	NOUN
cana-5485	94	4	,	,	PUNCT
cana-5485	94	5	which	which	PRON
cana-5485	94	6	remains	remain	VERB
cana-5485	94	7	constant	constant	ADJ
cana-5485	94	8	at	at	ADP
cana-5485	94	9	a	a	DET
cana-5485	94	10	relatively	relatively	ADV
cana-5485	94	11	high	high	ADJ
cana-5485	94	12	value	value	NOUN
cana-5485	94	13	,	,	PUNCT
cana-5485	94	14	indicates	indicate	VERB
cana-5485	94	15	the	the	DET
cana-5485	94	16	validation	validation	NOUN
cana-5485	94	17	loss	loss	NOUN
cana-5485	94	18	.	.	PUNCT
cana-5485	95	1	this	this	PRON
cana-5485	95	2	indicates	indicate	VERB
cana-5485	95	3	that	that	SCONJ
cana-5485	95	4	although	although	SCONJ
cana-5485	95	5	the	the	DET
cana-5485	95	6	model	model	NOUN
cana-5485	95	7	performs	perform	VERB
cana-5485	95	8	more	more	ADV
cana-5485	95	9	effectively	effectively	ADV
cana-5485	95	10	on	on	ADP
cana-5485	95	11	the	the	DET
cana-5485	95	12	training	training	NOUN
cana-5485	95	13	set	set	NOUN
cana-5485	95	14	,	,	PUNCT
cana-5485	95	15	it	it	PRON
cana-5485	95	16	does	do	AUX
cana-5485	95	17	not	not	PART
cana-5485	95	18	improve	improve	VERB
cana-5485	95	19	on	on	ADP
cana-5485	95	20	the	the	DET
cana-5485	95	21	validation	validation	NOUN
cana-5485	95	22	set	set	NOUN
cana-5485	95	23	.	.	PUNCT
cana-5485	96	1	communications	communication	NOUN
cana-5485	96	2	on	on	ADP
cana-5485	96	3	applied	apply	VERB
cana-5485	96	4	nonlinear	nonlinear	ADJ
cana-5485	96	5	analysis	analysis	NOUN
cana-5485	96	6	issn	issn	NOUN
cana-5485	96	7	:	:	PUNCT
cana-5485	96	8	1074	1074	NUM
cana-5485	96	9	-	-	PUNCT
cana-5485	96	10	133x	133x	NUM
cana-5485	96	11	vol	vol	VERB
cana-5485	96	12	32	32	NUM
cana-5485	96	13	no	no	NOUN
cana-5485	96	14	.	.	PUNCT
cana-5485	97	1	10s	10	NOUN
cana-5485	97	2	(	(	PUNCT
cana-5485	97	3	2025	2025	NUM
cana-5485	97	4	)	)	PUNCT
cana-5485	97	5	2380	2380	NUM
cana-5485	97	6	https://internationalpubls.com	https://internationalpubls.com	X
cana-5485	97	7	5	5	X
cana-5485	97	8	.	.	PUNCT
cana-5485	97	9	discussion	discussion	NOUN
cana-5485	97	10	with	with	ADP
cana-5485	97	11	its	its	PRON
cana-5485	97	12	high	high	ADJ
cana-5485	97	13	interpretability	interpretability	NOUN
cana-5485	97	14	and	and	CCONJ
cana-5485	97	15	accuracy	accuracy	NOUN
cana-5485	97	16	,	,	PUNCT
cana-5485	97	17	the	the	DET
cana-5485	97	18	explainable	explainable	ADJ
cana-5485	97	19	ai	ai	VERB
cana-5485	97	20	hybrid	hybrid	ADJ
cana-5485	97	21	deep	deep	ADJ
cana-5485	97	22	learning	learning	NOUN
cana-5485	97	23	model	model	NOUN
cana-5485	97	24	—	—	PUNCT
cana-5485	97	25	which	which	PRON
cana-5485	97	26	combines	combine	VERB
cana-5485	97	27	cnn	cnn	PROPN
cana-5485	97	28	,	,	PUNCT
cana-5485	97	29	xlnet	xlnet	PROPN
cana-5485	97	30	,	,	PUNCT
cana-5485	97	31	and	and	CCONJ
cana-5485	97	32	fasttext	fasttext	NOUN
cana-5485	97	33	with	with	ADP
cana-5485	97	34	explainable	explainable	ADJ
cana-5485	97	35	ai	ai	NOUN
cana-5485	97	36	—	—	PUNCT
cana-5485	97	37	meets	meet	VERB
cana-5485	97	38	the	the	DET
cana-5485	97	39	crucial	crucial	ADJ
cana-5485	97	40	need	need	NOUN
cana-5485	97	41	for	for	ADP
cana-5485	97	42	transparency	transparency	NOUN
cana-5485	97	43	in	in	ADP
cana-5485	97	44	automated	automate	VERB
cana-5485	97	45	disinformation	disinformation	NOUN
cana-5485	97	46	detection	detection	NOUN
cana-5485	97	47	and	and	CCONJ
cana-5485	97	48	greatly	greatly	ADV
cana-5485	97	49	enhances	enhance	VERB
cana-5485	97	50	the	the	DET
cana-5485	97	51	detection	detection	NOUN
cana-5485	97	52	of	of	ADP
cana-5485	97	53	fake	fake	ADJ
cana-5485	97	54	news.the	news.the	DET
cana-5485	97	55	hybrid	hybrid	ADJ
cana-5485	97	56	method	method	NOUN
cana-5485	97	57	performs	perform	VERB
cana-5485	97	58	better	well	ADJ
cana-5485	97	59	than	than	ADP
cana-5485	97	60	current	current	ADJ
cana-5485	97	61	transformer	transformer	NOUN
cana-5485	97	62	-	-	PUNCT
cana-5485	97	63	based	base	VERB
cana-5485	97	64	models	model	NOUN
cana-5485	97	65	in	in	ADP
cana-5485	97	66	terms	term	NOUN
cana-5485	97	67	of	of	ADP
cana-5485	97	68	accuracy	accuracy	NOUN
cana-5485	97	69	,	,	PUNCT
cana-5485	97	70	precision	precision	NOUN
cana-5485	97	71	,	,	PUNCT
cana-5485	97	72	recall	recall	NOUN
cana-5485	97	73	,	,	PUNCT
cana-5485	97	74	and	and	CCONJ
cana-5485	97	75	f1	f1	NOUN
cana-5485	97	76	-	-	PUNCT
cana-5485	97	77	score	score	NOUN
cana-5485	97	78	,	,	PUNCT
cana-5485	97	79	according	accord	VERB
cana-5485	97	80	to	to	ADP
cana-5485	97	81	experimental	experimental	ADJ
cana-5485	97	82	results	result	NOUN
cana-5485	97	83	on	on	ADP
cana-5485	97	84	benchmark	benchmark	ADJ
cana-5485	97	85	datasets	dataset	NOUN
cana-5485	97	86	.	.	PUNCT
cana-5485	98	1	the	the	DET
cana-5485	98	2	hybrid	hybrid	NOUN
cana-5485	98	3	model	model	NOUN
cana-5485	98	4	not	not	PART
cana-5485	98	5	only	only	ADV
cana-5485	98	6	enhances	enhance	VERB
cana-5485	98	7	classification	classification	NOUN
cana-5485	98	8	performance	performance	NOUN
cana-5485	98	9	but	but	CCONJ
cana-5485	98	10	also	also	ADV
cana-5485	98	11	produces	produce	VERB
cana-5485	98	12	comprehensible	comprehensible	ADJ
cana-5485	98	13	predictions	prediction	NOUN
cana-5485	98	14	,	,	PUNCT
cana-5485	98	15	which	which	PRON
cana-5485	98	16	is	be	AUX
cana-5485	98	17	crucial	crucial	ADJ
cana-5485	98	18	for	for	ADP
cana-5485	98	19	misinformation	misinformation	NOUN
cana-5485	98	20	detection	detection	NOUN
cana-5485	98	21	systems	system	NOUN
cana-5485	98	22	in	in	ADP
cana-5485	98	23	the	the	DET
cana-5485	98	24	real	real	ADJ
cana-5485	98	25	world	world	NOUN
cana-5485	98	26	.	.	PUNCT
cana-5485	99	1	while	while	SCONJ
cana-5485	99	2	cnn	cnn	PROPN
cana-5485	99	3	effectively	effectively	ADV
cana-5485	99	4	recognizes	recognize	VERB
cana-5485	99	5	local	local	ADJ
cana-5485	99	6	patterns	pattern	NOUN
cana-5485	99	7	from	from	ADP
cana-5485	99	8	textual	textual	ADJ
cana-5485	99	9	data	datum	NOUN
cana-5485	99	10	,	,	PUNCT
cana-5485	99	11	including	include	VERB
cana-5485	99	12	phrases	phrase	NOUN
cana-5485	99	13	and	and	CCONJ
cana-5485	99	14	structures	structure	NOUN
cana-5485	99	15	suggestive	suggestive	ADJ
cana-5485	99	16	of	of	ADP
cana-5485	99	17	disinformation	disinformation	NOUN
cana-5485	99	18	,	,	PUNCT
cana-5485	99	19	xlnet	xlnet	PROPN
cana-5485	99	20	uses	use	VERB
cana-5485	99	21	a	a	DET
cana-5485	99	22	permutation	permutation	NOUN
cana-5485	99	23	-	-	PUNCT
cana-5485	99	24	based	base	VERB
cana-5485	99	25	training	training	NOUN
cana-5485	99	26	strategy	strategy	NOUN
cana-5485	99	27	to	to	PART
cana-5485	99	28	improve	improve	VERB
cana-5485	99	29	contextual	contextual	ADJ
cana-5485	99	30	comprehension	comprehension	NOUN
cana-5485	99	31	.	.	PUNCT
cana-5485	100	1	when	when	SCONJ
cana-5485	100	2	combined	combine	VERB
cana-5485	100	3	,	,	PUNCT
cana-5485	100	4	they	they	PRON
cana-5485	100	5	yield	yield	VERB
cana-5485	100	6	a	a	DET
cana-5485	100	7	more	more	ADV
cana-5485	100	8	accurate	accurate	ADJ
cana-5485	100	9	and	and	CCONJ
cana-5485	100	10	resilient	resilient	ADJ
cana-5485	100	11	categorization	categorization	NOUN
cana-5485	100	12	model	model	NOUN
cana-5485	100	13	than	than	ADP
cana-5485	100	14	traditional	traditional	ADJ
cana-5485	100	15	techniques	technique	NOUN
cana-5485	100	16	.	.	PUNCT
cana-5485	101	1	explainable	explainable	ADJ
cana-5485	101	2	ai	ai	PROPN
cana-5485	101	3	(	(	PUNCT
cana-5485	101	4	xai	xai	PROPN
cana-5485	101	5	)	)	PUNCT
cana-5485	101	6	in	in	ADP
cana-5485	101	7	combination	combination	NOUN
cana-5485	101	8	with	with	ADP
cana-5485	101	9	cnn	cnn	PROPN
cana-5485	101	10	,	,	PUNCT
cana-5485	101	11	xlnet	xlnet	PROPN
cana-5485	101	12	,	,	PUNCT
cana-5485	101	13	and	and	CCONJ
cana-5485	101	14	fasttext	fasttext	PROPN
cana-5485	101	15	has	have	AUX
cana-5485	101	16	greatly	greatly	ADV
cana-5485	101	17	improved	improve	VERB
cana-5485	101	18	the	the	DET
cana-5485	101	19	interpretability	interpretability	NOUN
cana-5485	101	20	and	and	CCONJ
cana-5485	101	21	accuracy	accuracy	NOUN
cana-5485	101	22	of	of	ADP
cana-5485	101	23	fake	fake	ADJ
cana-5485	101	24	news	news	NOUN
cana-5485	101	25	identification	identification	NOUN
cana-5485	101	26	.	.	PUNCT
cana-5485	102	1	unlike	unlike	ADP
cana-5485	102	2	traditional	traditional	ADJ
cana-5485	102	3	deep	deep	ADJ
cana-5485	102	4	learning	learning	NOUN
cana-5485	102	5	models	model	NOUN
cana-5485	102	6	like	like	ADP
cana-5485	102	7	bert	bert	PROPN
cana-5485	102	8	and	and	CCONJ
cana-5485	102	9	roberta	roberta	PROPN
cana-5485	102	10	,	,	PUNCT
cana-5485	102	11	which	which	PRON
cana-5485	102	12	have	have	AUX
cana-5485	102	13	demonstrated	demonstrate	VERB
cana-5485	102	14	exceptional	exceptional	ADJ
cana-5485	102	15	accuracy	accuracy	NOUN
cana-5485	102	16	but	but	CCONJ
cana-5485	102	17	lack	lack	VERB
cana-5485	102	18	transparency	transparency	NOUN
cana-5485	102	19	in	in	ADP
cana-5485	102	20	decision	decision	NOUN
cana-5485	102	21	-	-	PUNCT
cana-5485	102	22	making	making	NOUN
cana-5485	102	23	,	,	PUNCT
cana-5485	102	24	our	our	PRON
cana-5485	102	25	proposed	propose	VERB
cana-5485	102	26	hybrid	hybrid	NOUN
cana-5485	102	27	approach	approach	NOUN
cana-5485	102	28	overcomes	overcome	VERB
cana-5485	102	29	this	this	DET
cana-5485	102	30	restriction	restriction	NOUN
cana-5485	102	31	by	by	ADP
cana-5485	102	32	incorporating	incorporate	VERB
cana-5485	102	33	shap	shap	NOUN
cana-5485	102	34	values	value	NOUN
cana-5485	102	35	for	for	ADP
cana-5485	102	36	explainability	explainability	NOUN
cana-5485	102	37	.	.	PUNCT
cana-5485	103	1	previous	previous	ADJ
cana-5485	103	2	studies	study	NOUN
cana-5485	103	3	have	have	AUX
cana-5485	103	4	shown	show	VERB
cana-5485	103	5	that	that	SCONJ
cana-5485	103	6	because	because	SCONJ
cana-5485	103	7	transformerbased	transformerbase	VERB
cana-5485	103	8	models	model	NOUN
cana-5485	103	9	,	,	PUNCT
cana-5485	103	10	like	like	ADP
cana-5485	103	11	xlnet	xlnet	PROPN
cana-5485	103	12	,	,	PUNCT
cana-5485	103	13	can	can	AUX
cana-5485	103	14	capture	capture	VERB
cana-5485	103	15	contextual	contextual	ADJ
cana-5485	103	16	relationships	relationship	NOUN
cana-5485	103	17	,	,	PUNCT
cana-5485	103	18	they	they	PRON
cana-5485	103	19	outperform	outperform	VERB
cana-5485	103	20	conventional	conventional	ADJ
cana-5485	103	21	machine	machine	NOUN
cana-5485	103	22	learning	learn	VERB
cana-5485	103	23	techniques	technique	NOUN
cana-5485	103	24	in	in	ADP
cana-5485	103	25	text	text	NOUN
cana-5485	103	26	classification	classification	NOUN
cana-5485	103	27	tasks	task	NOUN
cana-5485	103	28	.	.	PUNCT
cana-5485	104	1	the	the	DET
cana-5485	104	2	main	main	ADJ
cana-5485	104	3	reason	reason	NOUN
cana-5485	104	4	for	for	ADP
cana-5485	104	5	the	the	DET
cana-5485	104	6	proposed	propose	VERB
cana-5485	104	7	fake	fake	ADJ
cana-5485	104	8	news	news	NOUN
cana-5485	104	9	detection	detection	NOUN
cana-5485	104	10	model	model	NOUN
cana-5485	104	11	's	's	PART
cana-5485	104	12	limitations	limitation	NOUN
cana-5485	104	13	is	be	AUX
cana-5485	104	14	the	the	DET
cana-5485	104	15	hybrid	hybrid	ADJ
cana-5485	104	16	deep	deep	ADJ
cana-5485	104	17	learning	learning	NOUN
cana-5485	104	18	approach	approach	NOUN
cana-5485	104	19	's	's	PART
cana-5485	104	20	limited	limited	ADJ
cana-5485	104	21	ability	ability	NOUN
cana-5485	104	22	to	to	PART
cana-5485	104	23	adapt	adapt	VERB
cana-5485	104	24	to	to	ADP
cana-5485	104	25	constantly	constantly	ADV
cana-5485	104	26	changing	change	VERB
cana-5485	104	27	misinformation	misinformation	NOUN
cana-5485	104	28	patterns	pattern	NOUN
cana-5485	104	29	.	.	PUNCT
cana-5485	105	1	although	although	SCONJ
cana-5485	105	2	combining	combine	VERB
cana-5485	105	3	cnn	cnn	PROPN
cana-5485	105	4	,	,	PUNCT
cana-5485	105	5	xlnet	xlnet	PROPN
cana-5485	105	6	,	,	PUNCT
cana-5485	105	7	and	and	CCONJ
cana-5485	105	8	fasttext	fasttext	NOUN
cana-5485	105	9	enhances	enhance	VERB
cana-5485	105	10	the	the	DET
cana-5485	105	11	model	model	NOUN
cana-5485	105	12	's	's	PART
cana-5485	105	13	processing	processing	NOUN
cana-5485	105	14	of	of	ADP
cana-5485	105	15	various	various	ADJ
cana-5485	105	16	linguistic	linguistic	ADJ
cana-5485	105	17	structures	structure	NOUN
cana-5485	105	18	and	and	CCONJ
cana-5485	105	19	contextual	contextual	ADJ
cana-5485	105	20	subtleties	subtlety	NOUN
cana-5485	105	21	,	,	PUNCT
cana-5485	105	22	the	the	DET
cana-5485	105	23	dynamic	dynamic	ADJ
cana-5485	105	24	nature	nature	NOUN
cana-5485	105	25	of	of	ADP
cana-5485	105	26	fake	fake	ADJ
cana-5485	105	27	news	news	NOUN
cana-5485	105	28	necessitates	necessitate	VERB
cana-5485	105	29	frequent	frequent	ADJ
cana-5485	105	30	dataset	dataset	NOUN
cana-5485	105	31	updates	update	NOUN
cana-5485	105	32	and	and	CCONJ
cana-5485	105	33	model	model	NOUN
cana-5485	105	34	retraining	retrain	VERB
cana-5485	105	35	to	to	PART
cana-5485	105	36	preserve	preserve	VERB
cana-5485	105	37	accuracy	accuracy	NOUN
cana-5485	105	38	.	.	PUNCT
cana-5485	106	1	furthermore	furthermore	ADV
cana-5485	106	2	,	,	PUNCT
cana-5485	106	3	even	even	ADV
cana-5485	106	4	though	though	SCONJ
cana-5485	106	5	shap	shap	PROPN
cana-5485	106	6	's	's	PART
cana-5485	106	7	explainability	explainability	NOUN
cana-5485	106	8	feature	feature	NOUN
cana-5485	106	9	increases	increase	VERB
cana-5485	106	10	transparency	transparency	NOUN
cana-5485	106	11	,	,	PUNCT
cana-5485	106	12	it	it	PRON
cana-5485	106	13	also	also	ADV
cana-5485	106	14	adds	add	VERB
cana-5485	106	15	computational	computational	ADJ
cana-5485	106	16	overhead	overhead	NOUN
cana-5485	106	17	,	,	PUNCT
cana-5485	106	18	which	which	PRON
cana-5485	106	19	lengthens	lengthen	VERB
cana-5485	106	20	execution	execution	NOUN
cana-5485	106	21	times	time	NOUN
cana-5485	106	22	,	,	PUNCT
cana-5485	106	23	especially	especially	ADV
cana-5485	106	24	for	for	ADP
cana-5485	106	25	real	real	ADJ
cana-5485	106	26	-	-	PUNCT
cana-5485	106	27	time	time	NOUN
cana-5485	106	28	apps	app	NOUN
cana-5485	106	29	.	.	PUNCT
cana-5485	107	1	refrences	refrence	VERB
cana-5485	107	2	[	[	X
cana-5485	107	3	1	1	NUM
cana-5485	107	4	]	]	X
cana-5485	107	5	yin	yin	PROPN
cana-5485	107	6	,	,	PUNCT
cana-5485	107	7	j.	j.	PROPN
cana-5485	107	8	,	,	PUNCT
cana-5485	107	9	li	li	PROPN
cana-5485	107	10	,	,	PUNCT
cana-5485	107	11	q.	q.	PROPN
cana-5485	107	12	,	,	PUNCT
cana-5485	107	13	&	&	CCONJ
cana-5485	107	14	wei	wei	PROPN
cana-5485	107	15	,	,	PUNCT
cana-5485	107	16	c.	c.	PROPN
cana-5485	107	17	(	(	PUNCT
cana-5485	107	18	2021	2021	NUM
cana-5485	107	19	)	)	PUNCT
cana-5485	107	20	.	.	PUNCT
cana-5485	108	1	"	"	PUNCT
cana-5485	108	2	detecting	detect	VERB
cana-5485	108	3	fake	fake	ADJ
cana-5485	108	4	news	news	NOUN
cana-5485	108	5	on	on	ADP
cana-5485	108	6	social	social	ADJ
cana-5485	108	7	media	medium	NOUN
cana-5485	108	8	from	from	ADP
cana-5485	108	9	a	a	DET
cana-5485	108	10	data	data	NOUN
cana-5485	108	11	mining	mining	NOUN
cana-5485	108	12	perspective	perspective	NOUN
cana-5485	108	13	.	.	PUNCT
cana-5485	108	14	"	"	PUNCT
cana-5485	109	1	acm	acm	PROPN
cana-5485	109	2	computing	computing	NOUN
cana-5485	109	3	surveys	survey	NOUN
cana-5485	109	4	(	(	PUNCT
cana-5485	109	5	csur	csur	NOUN
cana-5485	109	6	)	)	PUNCT
cana-5485	109	7	,	,	PUNCT
cana-5485	109	8	54(5	54(5	NOUN
cana-5485	109	9	)	)	PUNCT
cana-5485	109	10	,	,	PUNCT
cana-5485	109	11	1	1	NUM
cana-5485	109	12	-	-	SYM
cana-5485	109	13	35	35	NUM
cana-5485	109	14	.	.	PUNCT
cana-5485	110	1	doi	doi	NOUN
cana-5485	110	2	:	:	PUNCT
cana-5485	110	3	10.1145/3462037	10.1145/3462037	NUM
cana-5485	110	4	.	.	PUNCT
cana-5485	111	1	[	[	X
cana-5485	111	2	2	2	NUM
cana-5485	111	3	]	]	X
cana-5485	111	4	zhou	zhou	PROPN
cana-5485	111	5	,	,	PUNCT
cana-5485	111	6	x.	x.	PROPN
cana-5485	111	7	,	,	PUNCT
cana-5485	111	8	&	&	CCONJ
cana-5485	111	9	zafarani	zafarani	PROPN
cana-5485	111	10	,	,	PUNCT
cana-5485	111	11	r.	r.	PROPN
cana-5485	111	12	(	(	PUNCT
cana-5485	111	13	2021	2021	NUM
cana-5485	111	14	)	)	PUNCT
cana-5485	111	15	.	.	PUNCT
cana-5485	112	1	"	"	PUNCT
cana-5485	112	2	a	a	DET
cana-5485	112	3	comprehensive	comprehensive	ADJ
cana-5485	112	4	survey	survey	NOUN
cana-5485	112	5	of	of	ADP
cana-5485	112	6	the	the	DET
cana-5485	112	7	state	state	NOUN
cana-5485	112	8	-	-	PUNCT
cana-5485	112	9	of	of	ADP
cana-5485	112	10	-	-	PUNCT
cana-5485	112	11	the	the	DET
cana-5485	112	12	-	-	PUNCT
cana-5485	112	13	art	art	NOUN
cana-5485	112	14	in	in	ADP
cana-5485	112	15	fake	fake	ADJ
cana-5485	112	16	news	news	NOUN
cana-5485	112	17	detection	detection	NOUN
cana-5485	112	18	.	.	PUNCT
cana-5485	112	19	"	"	PUNCT
cana-5485	113	1	acm	acm	PROPN
cana-5485	113	2	computing	computing	NOUN
cana-5485	113	3	surveys	survey	NOUN
cana-5485	113	4	(	(	PUNCT
cana-5485	113	5	csur	csur	NOUN
cana-5485	113	6	)	)	PUNCT
cana-5485	113	7	,	,	PUNCT
cana-5485	113	8	54(5	54(5	NOUN
cana-5485	113	9	)	)	PUNCT
cana-5485	113	10	,	,	PUNCT
cana-5485	113	11	1	1	NUM
cana-5485	113	12	-	-	SYM
cana-5485	113	13	36	36	NUM
cana-5485	113	14	.	.	PUNCT
cana-5485	114	1	doi	doi	NOUN
cana-5485	114	2	:	:	PUNCT
cana-5485	114	3	10.1145/3462034	10.1145/3462034	NUM
cana-5485	114	4	.	.	PUNCT
cana-5485	115	1	[	[	X
cana-5485	115	2	3	3	NUM
cana-5485	115	3	]	]	X
cana-5485	115	4	xie	xie	PROPN
cana-5485	115	5	,	,	PUNCT
cana-5485	115	6	p.	p.	PROPN
cana-5485	115	7	,	,	PUNCT
cana-5485	115	8	&	&	CCONJ
cana-5485	115	9	xu	xu	PROPN
cana-5485	115	10	,	,	PUNCT
cana-5485	115	11	l.	l.	PROPN
cana-5485	115	12	(	(	PUNCT
cana-5485	115	13	2020	2020	NUM
cana-5485	115	14	)	)	PUNCT
cana-5485	115	15	.	.	PUNCT
cana-5485	116	1	"	"	PUNCT
cana-5485	116	2	an	an	DET
cana-5485	116	3	integrated	integrated	ADJ
cana-5485	116	4	deep	deep	ADJ
cana-5485	116	5	learning	learning	NOUN
cana-5485	116	6	model	model	NOUN
cana-5485	116	7	for	for	ADP
cana-5485	116	8	detecting	detect	VERB
cana-5485	116	9	fake	fake	ADJ
cana-5485	116	10	news	news	NOUN
cana-5485	116	11	.	.	PUNCT
cana-5485	116	12	"	"	PUNCT
cana-5485	116	13	ieee	ieee	NOUN
cana-5485	116	14	access	access	NOUN
cana-5485	116	15	,	,	PUNCT
cana-5485	116	16	8	8	NUM
cana-5485	116	17	,	,	PUNCT
cana-5485	116	18	148436	148436	NUM
cana-5485	116	19	-	-	SYM
cana-5485	116	20	148445	148445	NUM
cana-5485	116	21	.	.	PUNCT
cana-5485	117	1	doi	doi	NOUN
cana-5485	117	2	:	:	PUNCT
cana-5485	117	3	10.1109	10.1109	NUM
cana-5485	117	4	/	/	SYM
cana-5485	117	5	access.2020.3011322	access.2020.3011322	NOUN
cana-5485	117	6	.	.	PUNCT
cana-5485	118	1	[	[	X
cana-5485	118	2	4	4	NUM
cana-5485	118	3	]	]	X
cana-5485	118	4	devlin	devlin	PROPN
cana-5485	118	5	,	,	PUNCT
cana-5485	118	6	j.	j.	PROPN
cana-5485	118	7	,	,	PUNCT
cana-5485	118	8	chang	chang	PROPN
cana-5485	118	9	,	,	PUNCT
cana-5485	118	10	m.	m.	PROPN
cana-5485	118	11	w.	w.	PROPN
cana-5485	118	12	,	,	PUNCT
cana-5485	118	13	lee	lee	PROPN
cana-5485	118	14	,	,	PUNCT
cana-5485	118	15	k.	k.	PROPN
cana-5485	118	16	,	,	PUNCT
cana-5485	118	17	&	&	CCONJ
cana-5485	118	18	toutanova	toutanova	PROPN
cana-5485	118	19	,	,	PUNCT
cana-5485	118	20	k.	k.	PROPN
cana-5485	118	21	(	(	PUNCT
cana-5485	118	22	2019	2019	NUM
cana-5485	118	23	)	)	PUNCT
cana-5485	118	24	.	.	PUNCT
cana-5485	119	1	bert	bert	PROPN
cana-5485	119	2	:	:	PUNCT
cana-5485	119	3	pre	pre	ADJ
cana-5485	119	4	-	-	NOUN
cana-5485	119	5	training	training	NOUN
cana-5485	119	6	of	of	ADP
cana-5485	119	7	deep	deep	ADJ
cana-5485	119	8	bidirectional	bidirectional	ADJ
cana-5485	119	9	transformers	transformer	NOUN
cana-5485	119	10	for	for	ADP
cana-5485	119	11	language	language	NOUN
cana-5485	119	12	understanding	understanding	NOUN
cana-5485	119	13	.	.	PUNCT
cana-5485	120	1	in	in	ADP
cana-5485	120	2	naacl	naacl	PROPN
cana-5485	120	3	-	-	PUNCT
cana-5485	120	4	hlt	hlt	PROPN
cana-5485	120	5	(	(	PUNCT
cana-5485	120	6	pp	pp	ADJ
cana-5485	120	7	.	.	PUNCT
cana-5485	121	1	4171–4186	4171–4186	NUM
cana-5485	121	2	)	)	PUNCT
cana-5485	121	3	.	.	PUNCT
cana-5485	122	1	[	[	X
cana-5485	122	2	5]li	5]li	NUM
cana-5485	122	3	,	,	PUNCT
cana-5485	122	4	h.	h.	PROPN
cana-5485	122	5	,	,	PUNCT
cana-5485	122	6	&	&	CCONJ
cana-5485	122	7	chang	chang	PROPN
cana-5485	122	8	,	,	PUNCT
cana-5485	122	9	k.	k.	PROPN
cana-5485	122	10	(	(	PUNCT
cana-5485	122	11	2022	2022	NUM
cana-5485	122	12	)	)	PUNCT
cana-5485	122	13	.	.	PUNCT
cana-5485	123	1	a	a	DET
cana-5485	123	2	comparative	comparative	ADJ
cana-5485	123	3	study	study	NOUN
cana-5485	123	4	of	of	ADP
cana-5485	123	5	bert	bert	PROPN
cana-5485	123	6	and	and	CCONJ
cana-5485	123	7	roberta	roberta	PROPN
cana-5485	123	8	for	for	ADP
cana-5485	123	9	fake	fake	ADJ
cana-5485	123	10	news	news	NOUN
cana-5485	123	11	detection	detection	NOUN
cana-5485	123	12	.	.	PUNCT
cana-5485	124	1	journal	journal	NOUN
cana-5485	124	2	of	of	ADP
cana-5485	124	3	computational	computational	ADJ
cana-5485	124	4	linguistics	linguistic	NOUN
cana-5485	124	5	and	and	CCONJ
cana-5485	124	6	chinese	chinese	ADJ
cana-5485	124	7	language	language	NOUN
cana-5485	124	8	processing	processing	NOUN
cana-5485	124	9	,	,	PUNCT
cana-5485	124	10	27(2	27(2	NUM
cana-5485	124	11	)	)	PUNCT
cana-5485	124	12	,	,	PUNCT
cana-5485	124	13	101–120	101–120	NUM
cana-5485	124	14	.	.	PUNCT
cana-5485	125	1	[	[	X
cana-5485	125	2	6]chen	6]chen	ADV
cana-5485	125	3	,	,	PUNCT
cana-5485	125	4	m.	m.	NOUN
cana-5485	125	5	,	,	PUNCT
cana-5485	125	6	&	&	CCONJ
cana-5485	125	7	li	li	PROPN
cana-5485	125	8	,	,	PUNCT
cana-5485	125	9	y.	y.	PROPN
cana-5485	125	10	(	(	PUNCT
cana-5485	125	11	2021	2021	NUM
cana-5485	125	12	)	)	PUNCT
cana-5485	125	13	.	.	PUNCT
cana-5485	126	1	interpretability	interpretability	NOUN
cana-5485	126	2	of	of	ADP
cana-5485	126	3	deep	deep	ADJ
cana-5485	126	4	learning	learning	NOUN
cana-5485	126	5	models	model	NOUN
cana-5485	126	6	:	:	PUNCT
cana-5485	126	7	a	a	DET
cana-5485	126	8	survey	survey	NOUN
cana-5485	126	9	of	of	ADP
cana-5485	126	10	recent	recent	ADJ
cana-5485	126	11	advances	advance	NOUN
cana-5485	126	12	.	.	PUNCT
cana-5485	127	1	ieee	ieee	NOUN
cana-5485	127	2	access	access	NOUN
cana-5485	127	3	,	,	PUNCT
cana-5485	127	4	9	9	NUM
cana-5485	127	5	,	,	PUNCT
cana-5485	127	6	106486–106504	106486–106504	NUM
cana-5485	127	7	.	.	PUNCT
cana-5485	128	1	communications	communication	NOUN
cana-5485	128	2	on	on	ADP
cana-5485	128	3	applied	apply	VERB
cana-5485	128	4	nonlinear	nonlinear	ADJ
cana-5485	128	5	analysis	analysis	NOUN
cana-5485	128	6	issn	issn	NOUN
cana-5485	128	7	:	:	PUNCT
cana-5485	128	8	1074	1074	NUM
cana-5485	128	9	-	-	PUNCT
cana-5485	128	10	133x	133x	NUM
cana-5485	128	11	vol	vol	VERB
cana-5485	128	12	32	32	NUM
cana-5485	128	13	no	no	NOUN
cana-5485	128	14	.	.	PUNCT
cana-5485	129	1	10s	10	NOUN
cana-5485	129	2	(	(	PUNCT
cana-5485	129	3	2025	2025	NUM
cana-5485	129	4	)	)	PUNCT
cana-5485	129	5	2381	2381	NUM
cana-5485	129	6	https://internationalpubls.com	https://internationalpubls.com	X
cana-5485	130	1	[	[	X
cana-5485	130	2	7	7	NUM
cana-5485	130	3	]	]	X
cana-5485	130	4	lundberg	lundberg	PROPN
cana-5485	130	5	,	,	PUNCT
cana-5485	130	6	s.	s.	PROPN
cana-5485	130	7	m.	m.	PROPN
cana-5485	130	8	,	,	PUNCT
cana-5485	130	9	&	&	CCONJ
cana-5485	130	10	lee	lee	PROPN
cana-5485	130	11	,	,	PUNCT
cana-5485	130	12	s.-i	s.-i	PROPN
cana-5485	130	13	.	.	PUNCT
cana-5485	131	1	(	(	PUNCT
cana-5485	131	2	2017	2017	NUM
cana-5485	131	3	)	)	PUNCT
cana-5485	131	4	.	.	PUNCT
cana-5485	132	1	a	a	DET
cana-5485	132	2	unified	unified	ADJ
cana-5485	132	3	approach	approach	NOUN
cana-5485	132	4	to	to	ADP
cana-5485	132	5	interpreting	interpret	VERB
cana-5485	132	6	model	model	NOUN
cana-5485	132	7	predictions	prediction	NOUN
cana-5485	132	8	.	.	PUNCT
cana-5485	133	1	in	in	ADP
cana-5485	133	2	proceedings	proceeding	NOUN
cana-5485	133	3	of	of	ADP
cana-5485	133	4	the	the	DET
cana-5485	133	5	31st	31st	ADJ
cana-5485	133	6	international	international	ADJ
cana-5485	133	7	conference	conference	NOUN
cana-5485	133	8	on	on	ADP
cana-5485	133	9	neural	neural	ADJ
cana-5485	133	10	information	information	NOUN
cana-5485	133	11	processing	processing	NOUN
cana-5485	133	12	systems	system	NOUN
cana-5485	133	13	(	(	PUNCT
cana-5485	133	14	pp	pp	ADJ
cana-5485	133	15	.	.	PUNCT
cana-5485	133	16	4765–4774	4765–4774	NUM
cana-5485	133	17	)	)	PUNCT
cana-5485	133	18	.	.	PUNCT
cana-5485	134	1	[	[	X
cana-5485	134	2	8	8	X
cana-5485	134	3	]	]	X
cana-5485	134	4	rajpurkar	rajpurkar	NOUN
cana-5485	134	5	,	,	PUNCT
cana-5485	134	6	p.	p.	PROPN
cana-5485	134	7	,	,	PUNCT
cana-5485	134	8	&	&	CCONJ
cana-5485	134	9	zhang	zhang	PROPN
cana-5485	134	10	,	,	PUNCT
cana-5485	134	11	j.	j.	PROPN
cana-5485	134	12	(	(	PUNCT
cana-5485	134	13	2020	2020	NUM
cana-5485	134	14	)	)	PUNCT
cana-5485	134	15	.	.	PUNCT
cana-5485	135	1	"	"	PUNCT
cana-5485	135	2	ai2	ai2	NOUN
cana-5485	135	3	:	:	PUNCT
cana-5485	135	4	a	a	DET
cana-5485	135	5	novel	novel	NOUN
cana-5485	135	6	dataset	dataset	NOUN
cana-5485	135	7	for	for	ADP
cana-5485	135	8	detecting	detect	VERB
cana-5485	135	9	fake	fake	ADJ
cana-5485	135	10	news	news	NOUN
cana-5485	135	11	.	.	PUNCT
cana-5485	135	12	"	"	PUNCT
cana-5485	136	1	iclr	iclr	VERB
cana-5485	136	2	2020	2020	NUM
cana-5485	136	3	.	.	PUNCT
cana-5485	137	1	doi	doi	NOUN
cana-5485	137	2	:	:	PUNCT
cana-5485	137	3	10.5555/3451418.3451465	10.5555/3451418.3451465	NUM
cana-5485	137	4	.	.	PUNCT
cana-5485	138	1	[	[	X
cana-5485	138	2	9	9	NUM
cana-5485	138	3	]	]	X
cana-5485	138	4	bertolami	bertolami	NOUN
cana-5485	138	5	,	,	PUNCT
cana-5485	138	6	g.	g.	PROPN
cana-5485	138	7	,	,	PUNCT
cana-5485	138	8	&	&	CCONJ
cana-5485	138	9	veloso	veloso	PROPN
cana-5485	138	10	,	,	PUNCT
cana-5485	138	11	m.	m.	NOUN
cana-5485	138	12	(	(	PUNCT
cana-5485	138	13	2020	2020	NUM
cana-5485	138	14	)	)	PUNCT
cana-5485	138	15	.	.	PUNCT
cana-5485	139	1	"	"	PUNCT
cana-5485	139	2	a	a	DET
cana-5485	139	3	hybrid	hybrid	ADJ
cana-5485	139	4	approach	approach	NOUN
cana-5485	139	5	for	for	ADP
cana-5485	139	6	fake	fake	ADJ
cana-5485	139	7	news	news	NOUN
cana-5485	139	8	detection	detection	NOUN
cana-5485	139	9	utilizing	utilize	VERB
cana-5485	139	10	attention	attention	NOUN
cana-5485	139	11	mechanisms	mechanism	NOUN
cana-5485	139	12	.	.	PUNCT
cana-5485	139	13	"	"	PUNCT
cana-5485	140	1	journal	journal	NOUN
cana-5485	140	2	of	of	ADP
cana-5485	140	3	computer	computer	NOUN
cana-5485	140	4	science	science	NOUN
cana-5485	140	5	and	and	CCONJ
cana-5485	140	6	technology	technology	NOUN
cana-5485	140	7	,	,	PUNCT
cana-5485	140	8	35(5	35(5	NUM
cana-5485	140	9	)	)	PUNCT
cana-5485	140	10	,	,	PUNCT
cana-5485	140	11	951	951	NUM
cana-5485	140	12	-	-	SYM
cana-5485	140	13	965	965	NUM
cana-5485	140	14	.	.	PUNCT
cana-5485	141	1	doi	doi	NOUN
cana-5485	141	2	:	:	PUNCT
cana-5485	141	3	10.1007	10.1007	NUM
cana-5485	141	4	/	/	SYM
cana-5485	141	5	s11390	s11390	NOUN
cana-5485	141	6	-	-	PUNCT
cana-5485	141	7	020	020	NUM
cana-5485	141	8	-	-	PUNCT
cana-5485	141	9	0992	0992	NUM
cana-5485	141	10	-	-	PUNCT
cana-5485	141	11	4	4	NUM
cana-5485	141	12	.	.	PUNCT
cana-5485	142	1	[	[	X
cana-5485	142	2	10	10	NUM
cana-5485	142	3	]	]	X
cana-5485	142	4	zhang	zhang	PROPN
cana-5485	142	5	,	,	PUNCT
cana-5485	142	6	y.	y.	PROPN
cana-5485	142	7	,	,	PUNCT
cana-5485	142	8	&	&	CCONJ
cana-5485	142	9	wang	wang	PROPN
cana-5485	142	10	,	,	PUNCT
cana-5485	142	11	s.	s.	PROPN
cana-5485	142	12	(	(	PUNCT
cana-5485	142	13	2020	2020	NUM
cana-5485	142	14	)	)	PUNCT
cana-5485	142	15	.	.	PUNCT
cana-5485	143	1	"	"	PUNCT
cana-5485	143	2	case	case	NOUN
cana-5485	143	3	study	study	NOUN
cana-5485	143	4	on	on	ADP
cana-5485	143	5	fake	fake	ADJ
cana-5485	143	6	news	news	NOUN
cana-5485	143	7	detection	detection	NOUN
cana-5485	143	8	in	in	ADP
cana-5485	143	9	chinese	chinese	ADJ
cana-5485	143	10	social	social	ADJ
cana-5485	143	11	media	medium	NOUN
cana-5485	143	12	.	.	PUNCT
cana-5485	143	13	"	"	PUNCT
cana-5485	144	1	ieee	ieee	NOUN
cana-5485	144	2	transactions	transaction	NOUN
cana-5485	144	3	on	on	ADP
cana-5485	144	4	knowledge	knowledge	NOUN
cana-5485	144	5	and	and	CCONJ
cana-5485	144	6	data	datum	NOUN
cana-5485	144	7	engineering	engineering	NOUN
cana-5485	144	8	,	,	PUNCT
cana-5485	144	9	32(12	32(12	NUM
cana-5485	144	10	)	)	PUNCT
cana-5485	144	11	,	,	PUNCT
cana-5485	144	12	2347	2347	NUM
cana-5485	144	13	-	-	SYM
cana-5485	144	14	2360	2360	NUM
cana-5485	144	15	.	.	PUNCT
cana-5485	145	1	doi	doi	NOUN
cana-5485	145	2	:	:	PUNCT
cana-5485	145	3	10.1109	10.1109	NUM
cana-5485	145	4	/	/	SYM
cana-5485	145	5	tkde.2020.3027248	tkde.2020.3027248	PROPN
cana-5485	145	6	.	.	PUNCT
cana-5485	146	1	[	[	X
cana-5485	146	2	11	11	NUM
cana-5485	146	3	]	]	X
cana-5485	146	4	joulin	joulin	PROPN
cana-5485	146	5	,	,	PUNCT
cana-5485	146	6	a.	a.	NOUN
cana-5485	146	7	,	,	PUNCT
cana-5485	146	8	grave	grave	NOUN
cana-5485	146	9	,	,	PUNCT
cana-5485	146	10	e.	e.	PROPN
cana-5485	146	11	,	,	PUNCT
cana-5485	146	12	bojanowski	bojanowski	VERB
cana-5485	146	13	,	,	PUNCT
cana-5485	146	14	p.	p.	NOUN
cana-5485	146	15	,	,	PUNCT
cana-5485	146	16	&	&	CCONJ
cana-5485	146	17	mikolov	mikolov	PROPN
cana-5485	146	18	,	,	PUNCT
cana-5485	146	19	t.	t.	PROPN
cana-5485	146	20	(	(	PUNCT
cana-5485	146	21	2017	2017	NUM
cana-5485	146	22	)	)	PUNCT
cana-5485	146	23	.	.	PUNCT
cana-5485	147	1	"	"	PUNCT
cana-5485	147	2	bag	bag	NOUN
cana-5485	147	3	of	of	ADP
cana-5485	147	4	tricks	trick	NOUN
cana-5485	147	5	for	for	ADP
cana-5485	147	6	efficient	efficient	ADJ
cana-5485	147	7	text	text	NOUN
cana-5485	147	8	classification	classification	NOUN
cana-5485	147	9	.	.	PUNCT
cana-5485	147	10	"	"	PUNCT
cana-5485	148	1	proceedings	proceeding	NOUN
cana-5485	148	2	of	of	ADP
cana-5485	148	3	the	the	DET
cana-5485	148	4	15th	15th	ADJ
cana-5485	148	5	conference	conference	NOUN
cana-5485	148	6	of	of	ADP
cana-5485	148	7	the	the	DET
cana-5485	148	8	european	european	ADJ
cana-5485	148	9	chapter	chapter	NOUN
cana-5485	148	10	of	of	ADP
cana-5485	148	11	the	the	DET
cana-5485	148	12	association	association	NOUN
cana-5485	148	13	for	for	ADP
cana-5485	148	14	computational	computational	ADJ
cana-5485	148	15	linguistics	linguistic	NOUN
cana-5485	148	16	(	(	PUNCT
cana-5485	148	17	eacl	eacl	PROPN
cana-5485	148	18	)	)	PUNCT
cana-5485	148	19	,	,	PUNCT
cana-5485	148	20	427–431	427–431	NUM
cana-5485	148	21	.	.	PUNCT
cana-5485	149	1	doi	doi	NOUN
cana-5485	149	2	:	:	PUNCT
cana-5485	149	3	10.18653	10.18653	NUM
cana-5485	149	4	/	/	SYM
cana-5485	149	5	v1	v1	NOUN
cana-5485	149	6	/	/	SYM
cana-5485	149	7	e17	e17	NOUN
cana-5485	149	8	-	-	PUNCT
cana-5485	149	9	2068	2068	NUM
cana-5485	149	10	.	.	PUNCT
cana-5485	150	1	12	12	NUM
cana-5485	150	2	]	]	PUNCT
cana-5485	150	3	rajpurkar	rajpurkar	NOUN
cana-5485	150	4	,	,	PUNCT
cana-5485	150	5	p.	p.	PROPN
cana-5485	150	6	,	,	PUNCT
cana-5485	150	7	&	&	CCONJ
cana-5485	150	8	zhang	zhang	PROPN
cana-5485	150	9	,	,	PUNCT
cana-5485	150	10	j.	j.	PROPN
cana-5485	150	11	(	(	PUNCT
cana-5485	150	12	2020	2020	NUM
cana-5485	150	13	)	)	PUNCT
cana-5485	150	14	.	.	PUNCT
cana-5485	151	1	ai2	ai2	NOUN
cana-5485	151	2	:	:	PUNCT
cana-5485	151	3	a	a	DET
cana-5485	151	4	novel	novel	NOUN
cana-5485	151	5	dataset	dataset	NOUN
cana-5485	151	6	for	for	ADP
cana-5485	151	7	detecting	detect	VERB
cana-5485	151	8	fake	fake	ADJ
cana-5485	151	9	news	news	NOUN
cana-5485	151	10	.	.	PUNCT
cana-5485	152	1	iclr	iclr	NOUN
cana-5485	152	2	2020	2020	NUM
cana-5485	152	3	.	.	PUNCT
cana-5485	153	1	doi	doi	NOUN
cana-5485	153	2	:	:	PUNCT
cana-5485	153	3	10.5555/3451418.3451465	10.5555/3451418.3451465	NUM
cana-5485	153	4	.	.	PUNCT
cana-5485	154	1	[	[	X
cana-5485	154	2	13	13	NUM
cana-5485	154	3	]	]	X
cana-5485	154	4	ribeiro	ribeiro	PROPN
cana-5485	154	5	,	,	PUNCT
cana-5485	154	6	m.	m.	NOUN
cana-5485	154	7	t.	t.	PROPN
cana-5485	154	8	,	,	PUNCT
cana-5485	154	9	singh	singh	PROPN
cana-5485	154	10	,	,	PUNCT
cana-5485	154	11	s.	s.	PROPN
cana-5485	154	12	,	,	PUNCT
cana-5485	154	13	&	&	CCONJ
cana-5485	154	14	guestrin	guestrin	PROPN
cana-5485	154	15	,	,	PUNCT
cana-5485	154	16	c.	c.	PROPN
cana-5485	154	17	(	(	PUNCT
cana-5485	154	18	2016	2016	NUM
cana-5485	154	19	)	)	PUNCT
cana-5485	154	20	.	.	PUNCT
cana-5485	155	1	why	why	SCONJ
cana-5485	155	2	should	should	AUX
cana-5485	155	3	i	i	PRON
cana-5485	155	4	trust	trust	VERB
cana-5485	155	5	you	you	PRON
cana-5485	155	6	?	?	PUNCT
cana-5485	156	1	explaining	explain	VERB
cana-5485	156	2	the	the	DET
cana-5485	156	3	predictions	prediction	NOUN
cana-5485	156	4	of	of	ADP
cana-5485	156	5	any	any	DET
cana-5485	156	6	classifier	classifier	NOUN
cana-5485	156	7	.	.	PUNCT
cana-5485	157	1	proceedings	proceeding	NOUN
cana-5485	157	2	of	of	ADP
cana-5485	157	3	the	the	DET
cana-5485	157	4	22nd	22nd	PROPN
cana-5485	157	5	acm	acm	PROPN
cana-5485	157	6	sigkdd	sigkdd	PROPN
cana-5485	157	7	international	international	ADJ
cana-5485	157	8	conference	conference	NOUN
cana-5485	157	9	on	on	ADP
cana-5485	157	10	knowledge	knowledge	NOUN
cana-5485	157	11	discovery	discovery	PROPN
cana-5485	157	12	and	and	CCONJ
cana-5485	157	13	data	datum	NOUN
cana-5485	157	14	mining	mining	NOUN
cana-5485	157	15	(	(	PUNCT
cana-5485	157	16	kdd	kdd	PROPN
cana-5485	157	17	)	)	PUNCT
cana-5485	157	18	,	,	PUNCT
cana-5485	157	19	1135–1144	1135–1144	NUM
cana-5485	157	20	.	.	PUNCT
cana-5485	158	1	doi	doi	NOUN
cana-5485	158	2	:	:	PUNCT
cana-5485	158	3	10.1145/2939672.2939778	10.1145/2939672.2939778	NUM
cana-5485	158	4	.	.	PUNCT
cana-5485	159	1	[	[	X
cana-5485	159	2	14	14	NUM
cana-5485	159	3	]	]	X
cana-5485	159	4	lundberg	lundberg	PROPN
cana-5485	159	5	,	,	PUNCT
cana-5485	159	6	s.	s.	PROPN
cana-5485	159	7	m.	m.	PROPN
cana-5485	159	8	,	,	PUNCT
cana-5485	159	9	&	&	CCONJ
cana-5485	159	10	lee	lee	PROPN
cana-5485	159	11	,	,	PUNCT
cana-5485	159	12	s.	s.	PROPN
cana-5485	159	13	i.	i.	PROPN
cana-5485	159	14	(	(	PUNCT
cana-5485	159	15	2017	2017	NUM
cana-5485	159	16	)	)	PUNCT
cana-5485	159	17	.	.	PUNCT
cana-5485	160	1	a	a	DET
cana-5485	160	2	unified	unified	ADJ
cana-5485	160	3	approach	approach	NOUN
cana-5485	160	4	to	to	ADP
cana-5485	160	5	interpreting	interpret	VERB
cana-5485	160	6	model	model	NOUN
cana-5485	160	7	predictions	prediction	NOUN
cana-5485	160	8	.	.	PUNCT
cana-5485	161	1	advances	advance	NOUN
cana-5485	161	2	in	in	ADP
cana-5485	161	3	neural	neural	ADJ
cana-5485	161	4	information	information	NOUN
cana-5485	161	5	processing	processing	NOUN
cana-5485	161	6	systems	system	NOUN
cana-5485	161	7	(	(	PUNCT
cana-5485	161	8	neurips	neurip	NOUN
cana-5485	161	9	)	)	PUNCT
cana-5485	161	10	,	,	PUNCT
cana-5485	161	11	30	30	NUM
cana-5485	161	12	.	.	PUNCT
cana-5485	162	1	[	[	X
cana-5485	162	2	15	15	NUM
cana-5485	162	3	]	]	X
cana-5485	162	4	li	li	PROPN
cana-5485	162	5	,	,	PUNCT
cana-5485	162	6	r.	r.	PROPN
cana-5485	162	7	,	,	PUNCT
cana-5485	162	8	&	&	CCONJ
cana-5485	162	9	chang	chang	PROPN
cana-5485	162	10	,	,	PUNCT
cana-5485	162	11	k.	k.	PROPN
cana-5485	162	12	w.	w.	PROPN
cana-5485	162	13	(	(	PUNCT
cana-5485	162	14	2022	2022	NUM
cana-5485	162	15	)	)	PUNCT
cana-5485	162	16	.	.	PUNCT
cana-5485	163	1	roberta	roberta	PROPN
cana-5485	163	2	-	-	PUNCT
cana-5485	163	3	based	base	VERB
cana-5485	163	4	approaches	approach	NOUN
cana-5485	163	5	for	for	ADP
cana-5485	163	6	detecting	detect	VERB
cana-5485	163	7	fake	fake	ADJ
cana-5485	163	8	news	news	NOUN
cana-5485	163	9	.	.	PUNCT
cana-5485	164	1	computational	computational	ADJ
cana-5485	164	2	linguistics	linguistic	NOUN
cana-5485	164	3	,	,	PUNCT
cana-5485	164	4	48(2	48(2	NUM
cana-5485	164	5	)	)	PUNCT
cana-5485	164	6	,	,	PUNCT
cana-5485	164	7	321–335	321–335	NUM
cana-5485	164	8	.	.	PUNCT
cana-5485	165	1	doi	doi	NOUN
cana-5485	165	2	:	:	PUNCT
cana-5485	165	3	10.1162	10.1162	NUM
cana-5485	165	4	/	/	SYM
cana-5485	165	5	coli_a_00441	coli_a_00441	NOUN
cana-5485	165	6	.	.	PUNCT
cana-5485	166	1	[	[	X
cana-5485	166	2	16	16	NUM
cana-5485	166	3	]	]	X
cana-5485	166	4	caruana	caruana	PROPN
cana-5485	166	5	,	,	PUNCT
cana-5485	166	6	r.	r.	PROPN
cana-5485	166	7	,	,	PUNCT
cana-5485	166	8	lou	lou	PROPN
cana-5485	166	9	,	,	PUNCT
cana-5485	166	10	y.	y.	PROPN
cana-5485	166	11	,	,	PUNCT
cana-5485	166	12	gehrke	gehrke	PROPN
cana-5485	166	13	,	,	PUNCT
cana-5485	166	14	j.	j.	PROPN
cana-5485	166	15	,	,	PUNCT
cana-5485	166	16	koch	koch	PROPN
cana-5485	166	17	,	,	PUNCT
cana-5485	166	18	p.	p.	PROPN
cana-5485	166	19	,	,	PUNCT
cana-5485	166	20	sturm	sturm	PROPN
cana-5485	166	21	,	,	PUNCT
cana-5485	166	22	m.	m.	NOUN
cana-5485	166	23	,	,	PUNCT
cana-5485	166	24	&	&	CCONJ
cana-5485	166	25	elhadad	elhadad	PROPN
cana-5485	166	26	,	,	PUNCT
cana-5485	166	27	n.	n.	NOUN
cana-5485	166	28	(	(	PUNCT
cana-5485	166	29	2015	2015	NUM
cana-5485	166	30	)	)	PUNCT
cana-5485	166	31	.	.	PUNCT
cana-5485	167	1	intelligible	intelligible	ADJ
cana-5485	167	2	models	model	NOUN
cana-5485	167	3	for	for	ADP
cana-5485	167	4	healthcare	healthcare	NOUN
cana-5485	167	5	.	.	PUNCT
cana-5485	168	1	proceedings	proceeding	NOUN
cana-5485	168	2	of	of	ADP
cana-5485	168	3	the	the	DET
cana-5485	168	4	21st	21st	ADJ
cana-5485	168	5	acm	acm	NOUN
cana-5485	168	6	sigkdd	sigkdd	NOUN
cana-5485	168	7	conference	conference	NOUN
cana-5485	168	8	on	on	ADP
cana-5485	168	9	knowledge	knowledge	NOUN
cana-5485	168	10	discovery	discovery	PROPN
cana-5485	168	11	and	and	CCONJ
cana-5485	168	12	data	datum	NOUN
cana-5485	168	13	mining	mining	NOUN
cana-5485	168	14	(	(	PUNCT
cana-5485	168	15	kdd	kdd	PROPN
cana-5485	168	16	)	)	PUNCT
cana-5485	168	17	,	,	PUNCT
cana-5485	168	18	1721–1730	1721–1730	NUM
cana-5485	168	19	.	.	PUNCT
cana-5485	169	1	[	[	X
cana-5485	169	2	17	17	NUM
cana-5485	169	3	]	]	X
cana-5485	169	4	zhang	zhang	PROPN
cana-5485	169	5	,	,	PUNCT
cana-5485	169	6	x.	x.	PROPN
cana-5485	169	7	,	,	PUNCT
cana-5485	169	8	zhao	zhao	PROPN
cana-5485	169	9	,	,	PUNCT
cana-5485	169	10	j.	j.	PROPN
cana-5485	169	11	,	,	PUNCT
cana-5485	169	12	&	&	CCONJ
cana-5485	169	13	lecun	lecun	PROPN
cana-5485	169	14	,	,	PUNCT
cana-5485	169	15	y.	y.	PROPN
cana-5485	169	16	(	(	PUNCT
cana-5485	169	17	2015	2015	NUM
cana-5485	169	18	)	)	PUNCT
cana-5485	169	19	.	.	PUNCT
cana-5485	170	1	character	character	NOUN
cana-5485	170	2	-	-	PUNCT
cana-5485	170	3	level	level	NOUN
cana-5485	170	4	convolutional	convolutional	ADJ
cana-5485	170	5	networks	network	NOUN
cana-5485	170	6	for	for	ADP
cana-5485	170	7	text	text	NOUN
cana-5485	170	8	classification	classification	NOUN
cana-5485	170	9	.	.	PUNCT
cana-5485	171	1	advances	advance	NOUN
cana-5485	171	2	in	in	ADP
cana-5485	171	3	neural	neural	ADJ
cana-5485	171	4	information	information	NOUN
cana-5485	171	5	processing	processing	NOUN
cana-5485	171	6	systems	system	NOUN
cana-5485	171	7	(	(	PUNCT
cana-5485	171	8	neurips	neurip	NOUN
cana-5485	171	9	)	)	PUNCT
cana-5485	171	10	,	,	PUNCT
cana-5485	171	11	28	28	NUM
cana-5485	171	12	,	,	PUNCT
cana-5485	171	13	649–657	649–657	NUM
cana-5485	171	14	.	.	PUNCT
cana-5485	172	1	[	[	X
cana-5485	172	2	18	18	NUM
cana-5485	172	3	]	]	X
cana-5485	172	4	kleinberg	kleinberg	PROPN
cana-5485	172	5	,	,	PUNCT
cana-5485	172	6	b.	b.	PROPN
cana-5485	172	7	,	,	PUNCT
cana-5485	172	8	mozes	mozes	PROPN
cana-5485	172	9	,	,	PUNCT
cana-5485	172	10	m.	m.	NOUN
cana-5485	172	11	,	,	PUNCT
cana-5485	172	12	&	&	CCONJ
cana-5485	172	13	arntz	arntz	PROPN
cana-5485	172	14	,	,	PUNCT
cana-5485	172	15	a.	a.	NOUN
cana-5485	172	16	(	(	PUNCT
cana-5485	172	17	2020	2020	NUM
cana-5485	172	18	)	)	PUNCT
cana-5485	172	19	.	.	PUNCT
cana-5485	173	1	automatic	automatic	ADJ
cana-5485	173	2	fake	fake	ADJ
cana-5485	173	3	news	news	NOUN
cana-5485	173	4	detection	detection	NOUN
cana-5485	173	5	using	use	VERB
cana-5485	173	6	linguistic	linguistic	ADJ
cana-5485	173	7	features	feature	NOUN
cana-5485	173	8	and	and	CCONJ
cana-5485	173	9	deep	deep	ADJ
cana-5485	173	10	learning	learning	NOUN
cana-5485	173	11	.	.	PUNCT
cana-5485	174	1	computational	computational	ADJ
cana-5485	174	2	linguistics	linguistic	NOUN
cana-5485	174	3	,	,	PUNCT
cana-5485	174	4	46(2	46(2	NOUN
cana-5485	174	5	)	)	PUNCT
cana-5485	174	6	,	,	PUNCT
cana-5485	174	7	1–19	1–19	NOUN
cana-5485	174	8	.	.	PUNCT
cana-5485	175	1	[	[	X
cana-5485	175	2	19	19	NUM
cana-5485	175	3	]	]	X
cana-5485	175	4	xie	xie	PROPN
cana-5485	175	5	,	,	PUNCT
cana-5485	175	6	y.	y.	PROPN
cana-5485	175	7	,	,	PUNCT
cana-5485	175	8	&	&	CCONJ
cana-5485	175	9	xu	xu	PROPN
cana-5485	175	10	,	,	PUNCT
cana-5485	175	11	x.	x.	NOUN
cana-5485	175	12	(	(	PUNCT
cana-5485	175	13	2020	2020	NUM
cana-5485	175	14	)	)	PUNCT
cana-5485	175	15	.	.	PUNCT
cana-5485	176	1	a	a	DET
cana-5485	176	2	hybrid	hybrid	ADJ
cana-5485	176	3	deep	deep	ADJ
cana-5485	176	4	learning	learning	NOUN
cana-5485	176	5	framework	framework	NOUN
cana-5485	176	6	for	for	ADP
cana-5485	176	7	fake	fake	ADJ
cana-5485	176	8	news	news	NOUN
cana-5485	176	9	detection	detection	NOUN
cana-5485	176	10	.	.	PUNCT
cana-5485	177	1	applied	apply	VERB
cana-5485	177	2	intelligence	intelligence	NOUN
cana-5485	177	3	,	,	PUNCT
cana-5485	177	4	50(10	50(10	NOUN
cana-5485	177	5	)	)	PUNCT
cana-5485	177	6	,	,	PUNCT
cana-5485	177	7	3464–3475	3464–3475	NUM
cana-5485	177	8	.	.	PUNCT
cana-5485	178	1	[	[	X
cana-5485	178	2	20	20	NUM
cana-5485	178	3	]	]	X
cana-5485	178	4	chen	chen	PROPN
cana-5485	178	5	,	,	PUNCT
cana-5485	178	6	y.	y.	PROPN
cana-5485	178	7	,	,	PUNCT
cana-5485	178	8	&	&	CCONJ
cana-5485	178	9	li	li	PROPN
cana-5485	178	10	,	,	PUNCT
cana-5485	178	11	j.	j.	PROPN
cana-5485	178	12	(	(	PUNCT
cana-5485	178	13	2021	2021	NUM
cana-5485	178	14	)	)	PUNCT
cana-5485	178	15	.	.	PUNCT
cana-5485	179	1	incorporating	incorporate	VERB
cana-5485	179	2	explainable	explainable	ADJ
cana-5485	179	3	ai	ai	NOUN
cana-5485	179	4	for	for	ADP
cana-5485	179	5	fake	fake	ADJ
cana-5485	179	6	news	news	NOUN
cana-5485	179	7	detection	detection	NOUN
cana-5485	179	8	.	.	PUNCT
cana-5485	180	1	journal	journal	NOUN
cana-5485	180	2	of	of	ADP
cana-5485	180	3	intelligent	intelligent	ADJ
cana-5485	180	4	information	information	NOUN
cana-5485	180	5	systems	system	NOUN
cana-5485	180	6	,	,	PUNCT
cana-5485	180	7	56(2	56(2	ADJ
cana-5485	180	8	)	)	PUNCT
cana-5485	180	9	,	,	PUNCT
cana-5485	180	10	239–257	239–257	NUM
cana-5485	180	11	.	.	PUNCT
cana-5485	181	1	[	[	X
cana-5485	181	2	21	21	NUM
cana-5485	181	3	]	]	X
cana-5485	181	4	goodfellow	goodfellow	PROPN
cana-5485	181	5	,	,	PUNCT
cana-5485	181	6	i.	i.	PROPN
cana-5485	181	7	,	,	PUNCT
cana-5485	181	8	bengio	bengio	PROPN
cana-5485	181	9	,	,	PUNCT
cana-5485	181	10	y.	y.	PROPN
cana-5485	181	11	,	,	PUNCT
cana-5485	181	12	&	&	CCONJ
cana-5485	181	13	courville	courville	PROPN
cana-5485	181	14	,	,	PUNCT
cana-5485	181	15	a.	a.	NOUN
cana-5485	181	16	(	(	PUNCT
cana-5485	181	17	2016	2016	NUM
cana-5485	181	18	)	)	PUNCT
cana-5485	181	19	.	.	PUNCT
cana-5485	182	1	"	"	PUNCT
cana-5485	182	2	deep	deep	ADJ
cana-5485	182	3	learning	learning	NOUN
cana-5485	182	4	.	.	PUNCT
cana-5485	182	5	"	"	PUNCT
cana-5485	183	1	mit	mit	PROPN
cana-5485	183	2	press	press	NOUN
cana-5485	183	3	.	.	PUNCT
cana-5485	184	1	communications	communication	NOUN
cana-5485	184	2	on	on	ADP
cana-5485	184	3	applied	apply	VERB
cana-5485	184	4	nonlinear	nonlinear	ADJ
cana-5485	184	5	analysis	analysis	NOUN
cana-5485	184	6	issn	issn	NOUN
cana-5485	184	7	:	:	PUNCT
cana-5485	184	8	1074	1074	NUM
cana-5485	184	9	-	-	PUNCT
cana-5485	184	10	133x	133x	NUM
cana-5485	184	11	vol	vol	VERB
cana-5485	184	12	32	32	NUM
cana-5485	184	13	no	no	NOUN
cana-5485	184	14	.	.	PUNCT
cana-5485	185	1	10s	10	NOUN
cana-5485	185	2	(	(	PUNCT
cana-5485	185	3	2025	2025	NUM
cana-5485	185	4	)	)	PUNCT
cana-5485	185	5	2382	2382	NUM
cana-5485	185	6	https://internationalpubls.com	https://internationalpubls.com	X
cana-5485	186	1	[	[	X
cana-5485	186	2	22	22	NUM
cana-5485	186	3	]	]	X
cana-5485	186	4	zhang	zhang	PROPN
cana-5485	186	5	,	,	PUNCT
cana-5485	186	6	h.	h.	PROPN
cana-5485	186	7	,	,	PUNCT
cana-5485	186	8	xu	xu	PROPN
cana-5485	186	9	,	,	PUNCT
cana-5485	186	10	l.	l.	PROPN
cana-5485	186	11	,	,	PUNCT
cana-5485	186	12	&	&	CCONJ
cana-5485	186	13	liu	liu	PROPN
cana-5485	186	14	,	,	PUNCT
cana-5485	186	15	s.	s.	PROPN
cana-5485	186	16	(	(	PUNCT
cana-5485	186	17	2019	2019	NUM
cana-5485	186	18	)	)	PUNCT
cana-5485	186	19	.	.	PUNCT
cana-5485	187	1	"	"	PUNCT
cana-5485	187	2	multi	multi	ADJ
cana-5485	187	3	-	-	ADJ
cana-5485	187	4	modal	modal	ADJ
cana-5485	187	5	fake	fake	ADJ
cana-5485	187	6	news	news	NOUN
cana-5485	187	7	detection	detection	NOUN
cana-5485	187	8	with	with	ADP
cana-5485	187	9	deep	deep	ADJ
cana-5485	187	10	learning	learning	NOUN
cana-5485	187	11	.	.	PUNCT
cana-5485	187	12	"	"	PUNCT
cana-5485	188	1	proceedings	proceeding	NOUN
cana-5485	188	2	of	of	ADP
cana-5485	188	3	the	the	DET
cana-5485	188	4	2019	2019	NUM
cana-5485	188	5	conference	conference	NOUN
cana-5485	188	6	on	on	ADP
cana-5485	188	7	empirical	empirical	ADJ
cana-5485	188	8	methods	method	NOUN
cana-5485	188	9	in	in	ADP
cana-5485	188	10	natural	natural	ADJ
cana-5485	188	11	language	language	NOUN
cana-5485	188	12	processing	processing	NOUN
cana-5485	188	13	(	(	PUNCT
cana-5485	188	14	emnlp	emnlp	ADJ
cana-5485	188	15	)	)	PUNCT
cana-5485	188	16	,	,	PUNCT
cana-5485	188	17	113	113	NUM
cana-5485	188	18	-	-	SYM
cana-5485	188	19	122	122	NUM
cana-5485	188	20	.	.	PUNCT
cana-5485	189	1	[	[	X
cana-5485	189	2	23	23	NUM
cana-5485	189	3	]	]	X
cana-5485	189	4	liao	liao	PROPN
cana-5485	189	5	,	,	PUNCT
cana-5485	189	6	q.	q.	PROPN
cana-5485	189	7	v.	v.	PROPN
cana-5485	189	8	,	,	PUNCT
cana-5485	189	9	gruen	gruen	PROPN
cana-5485	189	10	,	,	PUNCT
cana-5485	189	11	d.	d.	PROPN
cana-5485	189	12	,	,	PUNCT
cana-5485	189	13	&	&	CCONJ
cana-5485	189	14	miller	miller	PROPN
cana-5485	189	15	,	,	PUNCT
cana-5485	189	16	s.	s.	PROPN
cana-5485	189	17	(	(	PUNCT
cana-5485	189	18	2020	2020	NUM
cana-5485	189	19	)	)	PUNCT
cana-5485	189	20	.	.	PUNCT
cana-5485	190	1	"	"	PUNCT
cana-5485	190	2	questioning	question	VERB
cana-5485	190	3	the	the	DET
cana-5485	190	4	ai	ai	NOUN
cana-5485	190	5	:	:	PUNCT
cana-5485	190	6	informing	inform	VERB
cana-5485	190	7	design	design	NOUN
cana-5485	190	8	practices	practice	NOUN
cana-5485	190	9	for	for	ADP
cana-5485	190	10	explainable	explainable	ADJ
cana-5485	190	11	ai	ai	NOUN
cana-5485	190	12	user	user	NOUN
cana-5485	190	13	experiences	experience	NOUN
cana-5485	190	14	.	.	PUNCT
cana-5485	190	15	"	"	PUNCT
cana-5485	190	16	proceedings	proceeding	NOUN
cana-5485	190	17	of	of	ADP
cana-5485	190	18	chi	chi	PROPN
cana-5485	190	19	2020	2020	NUM
cana-5485	190	20	.	.	PUNCT
cana-5485	191	1	[	[	X
cana-5485	191	2	24	24	NUM
cana-5485	191	3	]	]	X
cana-5485	191	4	mikolov	mikolov	NOUN
cana-5485	191	5	,	,	PUNCT
cana-5485	191	6	t.	t.	PROPN
cana-5485	191	7	,	,	PUNCT
cana-5485	191	8	sutskever	sutskever	PROPN
cana-5485	191	9	,	,	PUNCT
cana-5485	191	10	i.	i.	PROPN
cana-5485	191	11	,	,	PUNCT
cana-5485	191	12	chen	chen	PROPN
cana-5485	191	13	,	,	PUNCT
cana-5485	191	14	k.	k.	PROPN
cana-5485	191	15	,	,	PUNCT
cana-5485	191	16	corrado	corrado	PROPN
cana-5485	191	17	,	,	PUNCT
cana-5485	191	18	g.	g.	PROPN
cana-5485	191	19	s.	s.	PROPN
cana-5485	191	20	,	,	PUNCT
cana-5485	191	21	&	&	CCONJ
cana-5485	191	22	dean	dean	PROPN
cana-5485	191	23	,	,	PUNCT
cana-5485	191	24	j.	j.	PROPN
cana-5485	191	25	(	(	PUNCT
cana-5485	191	26	2013	2013	NUM
cana-5485	191	27	)	)	PUNCT
cana-5485	191	28	.	.	PUNCT
cana-5485	192	1	"	"	PUNCT
cana-5485	192	2	distributed	distribute	VERB
cana-5485	192	3	representations	representation	NOUN
cana-5485	192	4	of	of	ADP
cana-5485	192	5	words	word	NOUN
cana-5485	192	6	and	and	CCONJ
cana-5485	192	7	phrases	phrase	NOUN
cana-5485	192	8	and	and	CCONJ
cana-5485	192	9	their	their	PRON
cana-5485	192	10	compositionality	compositionality	NOUN
cana-5485	192	11	.	.	PUNCT
cana-5485	192	12	"	"	PUNCT
cana-5485	193	1	advances	advance	NOUN
cana-5485	193	2	in	in	ADP
cana-5485	193	3	neural	neural	ADJ
cana-5485	193	4	information	information	NOUN
cana-5485	193	5	processing	processing	NOUN
cana-5485	193	6	systems	system	NOUN
cana-5485	193	7	(	(	PUNCT
cana-5485	193	8	neurips	neurip	NOUN
cana-5485	193	9	)	)	PUNCT
cana-5485	193	10	,	,	PUNCT
cana-5485	193	11	26	26	NUM
cana-5485	193	12	.	.	PUNCT
cana-5485	194	1	[	[	X
cana-5485	194	2	25	25	NUM
cana-5485	194	3	]	]	X
cana-5485	194	4	kim	kim	PROPN
cana-5485	194	5	,	,	PUNCT
cana-5485	194	6	y.	y.	PROPN
cana-5485	194	7	(	(	PUNCT
cana-5485	194	8	2014	2014	NUM
cana-5485	194	9	)	)	PUNCT
cana-5485	194	10	.	.	PUNCT
cana-5485	195	1	"	"	PUNCT
cana-5485	195	2	convolutional	convolutional	ADJ
cana-5485	195	3	neural	neural	ADJ
cana-5485	195	4	networks	network	NOUN
cana-5485	195	5	for	for	ADP
cana-5485	195	6	sentence	sentence	NOUN
cana-5485	195	7	classification	classification	NOUN
cana-5485	195	8	.	.	PUNCT
cana-5485	195	9	"	"	PUNCT
cana-5485	196	1	proceedings	proceeding	NOUN
cana-5485	196	2	of	of	ADP
cana-5485	196	3	the	the	DET
cana-5485	196	4	2014	2014	NUM
cana-5485	196	5	conference	conference	NOUN
cana-5485	196	6	on	on	ADP
cana-5485	196	7	empirical	empirical	ADJ
cana-5485	196	8	methods	method	NOUN
cana-5485	196	9	in	in	ADP
cana-5485	196	10	natural	natural	ADJ
cana-5485	196	11	language	language	NOUN
cana-5485	196	12	processing	processing	NOUN
cana-5485	196	13	(	(	PUNCT
cana-5485	196	14	emnlp	emnlp	ADJ
cana-5485	196	15	)	)	PUNCT
cana-5485	196	16	,	,	PUNCT
cana-5485	196	17	1746	1746	NUM
cana-5485	196	18	-	-	SYM
cana-5485	196	19	1751	1751	NUM
cana-5485	196	20	.	.	PUNCT