id	sid	tid	token	lemma	pos
abrj-637	1	1	1	1	NUM
abrj-637	1	2	©	©	PROPN
abrj-637	1	3	2025	2025	NUM
abrj-637	1	4	by	by	ADP
abrj-637	1	5	the	the	DET
abrj-637	1	6	authors	author	NOUN
abrj-637	1	7	;	;	PUNCT
abrj-637	1	8	licensee	licensee	NOUN
abrj-637	1	9	eastern	eastern	ADJ
abrj-637	1	10	centre	centre	NOUN
abrj-637	1	11	of	of	ADP
abrj-637	1	12	science	science	NOUN
abrj-637	1	13	and	and	CCONJ
abrj-637	1	14	education	education	NOUN
abrj-637	1	15	,	,	PUNCT
abrj-637	1	16	usa	usa	PROPN
abrj-637	1	17	asian	asian	PROPN
abrj-637	1	18	business	business	PROPN
abrj-637	1	19	research	research	PROPN
abrj-637	1	20	journal	journal	PROPN
abrj-637	1	21	vol	vol	NOUN
abrj-637	1	22	.	.	PROPN
abrj-637	2	1	10	10	NUM
abrj-637	2	2	,	,	PUNCT
abrj-637	2	3	no	no	INTJ
abrj-637	2	4	.	.	NOUN
abrj-637	2	5	11	11	NUM
abrj-637	2	6	,	,	PUNCT
abrj-637	2	7	1	1	NUM
abrj-637	2	8	-	-	SYM
abrj-637	2	9	13	13	NUM
abrj-637	2	10	,	,	PUNCT
abrj-637	2	11	2025	2025	NUM
abrj-637	2	12	issn	issn	NOUN
abrj-637	2	13	:	:	PUNCT
abrj-637	2	14	2576	2576	NUM
abrj-637	2	15	-	-	SYM
abrj-637	2	16	6759	6759	NUM
abrj-637	2	17	doi	doi	NOUN
abrj-637	2	18	:	:	PUNCT
abrj-637	2	19	10.55220/2576	10.55220/2576	NUM
abrj-637	2	20	-	-	SYM
abrj-637	2	21	6759.637	6759.637	NUM
abrj-637	2	22	©	©	PROPN
abrj-637	2	23	2025	2025	NUM
abrj-637	2	24	by	by	ADP
abrj-637	2	25	the	the	DET
abrj-637	2	26	authors	author	NOUN
abrj-637	2	27	;	;	PUNCT
abrj-637	2	28	licensee	licensee	NOUN
abrj-637	2	29	eastern	eastern	ADJ
abrj-637	2	30	centre	centre	NOUN
abrj-637	2	31	of	of	ADP
abrj-637	2	32	science	science	NOUN
abrj-637	2	33	and	and	CCONJ
abrj-637	2	34	education	education	NOUN
abrj-637	2	35	,	,	PUNCT
abrj-637	2	36	usa	usa	NOUN
abrj-637	2	37	leveraging	leverage	VERB
abrj-637	2	38	machine	machine	NOUN
abrj-637	2	39	learning	learning	NOUN
abrj-637	2	40	for	for	ADP
abrj-637	2	41	tax	tax	NOUN
abrj-637	2	42	fraud	fraud	NOUN
abrj-637	2	43	detection	detection	NOUN
abrj-637	2	44	and	and	CCONJ
abrj-637	2	45	risk	risk	NOUN
abrj-637	2	46	scoring	scoring	NOUN
abrj-637	2	47	in	in	ADP
abrj-637	2	48	corporate	corporate	ADJ
abrj-637	2	49	filings	filing	NOUN
abrj-637	2	50	tiejiang	tiejiang	PROPN
abrj-637	2	51	sun1	sun1	PROPN
abrj-637	2	52	mengdie	mengdie	PROPN
abrj-637	2	53	wang2	wang2	PROPN
abrj-637	3	1			PROPN
abrj-637	3	2	jiaying	jiaying	PROPN
abrj-637	3	3	chen3	chen3	PROPN
abrj-637	3	4	1chang’an	1chang’an	NUM
abrj-637	3	5	university	university	NOUN
abrj-637	3	6	,	,	PUNCT
abrj-637	3	7	xi’an	xi’an	PROPN
abrj-637	3	8	710064	710064	NUM
abrj-637	3	9	,	,	PUNCT
abrj-637	3	10	china	china	PROPN
abrj-637	3	11	.	.	PUNCT
abrj-637	4	1	2shanghai	2shanghai	PROPN
abrj-637	4	2	lixin	lixin	PROPN
abrj-637	4	3	university	university	PROPN
abrj-637	4	4	of	of	ADP
abrj-637	4	5	accounting	accounting	NOUN
abrj-637	4	6	and	and	CCONJ
abrj-637	4	7	finance	finance	NOUN
abrj-637	4	8	,	,	PUNCT
abrj-637	4	9	shanghai	shanghai	PROPN
abrj-637	4	10	201620	201620	NUM
abrj-637	4	11	,	,	PUNCT
abrj-637	4	12	china	china	PROPN
abrj-637	4	13	.	.	PUNCT
abrj-637	5	1	3cornell	3cornell	NUM
abrj-637	5	2	university	university	PROPN
abrj-637	5	3	,	,	PUNCT
abrj-637	5	4	ithaca	ithaca	PROPN
abrj-637	5	5	,	,	PUNCT
abrj-637	5	6	ny	ny	PROPN
abrj-637	5	7	10022	10022	NUM
abrj-637	5	8	,	,	PUNCT
abrj-637	5	9	usa	usa	PROPN
abrj-637	5	10	.	.	PROPN
abrj-637	5	11	(	(	PUNCT
abrj-637	5	12			PROPN
abrj-637	5	13	corresponding	correspond	VERB
abrj-637	5	14	author	author	NOUN
abrj-637	5	15	)	)	PUNCT
abrj-637	5	16	abstract	abstract	ADJ
abrj-637	5	17	tax	tax	NOUN
abrj-637	5	18	fraud	fraud	NOUN
abrj-637	5	19	has	have	AUX
abrj-637	5	20	been	be	AUX
abrj-637	5	21	a	a	DET
abrj-637	5	22	thorn	thorn	NOUN
abrj-637	5	23	on	on	ADP
abrj-637	5	24	the	the	DET
abrj-637	5	25	flesh	flesh	NOUN
abrj-637	5	26	of	of	ADP
abrj-637	5	27	governments	government	NOUN
abrj-637	5	28	and	and	CCONJ
abrj-637	5	29	regulatory	regulatory	ADJ
abrj-637	5	30	bodies	body	NOUN
abrj-637	5	31	across	across	ADP
abrj-637	5	32	the	the	DET
abrj-637	5	33	globe	globe	NOUN
abrj-637	5	34	,	,	PUNCT
abrj-637	5	35	as	as	SCONJ
abrj-637	5	36	it	it	PRON
abrj-637	5	37	compromises	compromise	VERB
abrj-637	5	38	the	the	DET
abrj-637	5	39	financial	financial	ADJ
abrj-637	5	40	stability	stability	NOUN
abrj-637	5	41	and	and	CCONJ
abrj-637	5	42	confidence	confidence	NOUN
abrj-637	5	43	of	of	ADP
abrj-637	5	44	the	the	DET
abrj-637	5	45	citizens	citizen	NOUN
abrj-637	5	46	.	.	PUNCT
abrj-637	6	1	the	the	DET
abrj-637	6	2	conventional	conventional	ADJ
abrj-637	6	3	forms	form	NOUN
abrj-637	6	4	of	of	ADP
abrj-637	6	5	detection	detection	NOUN
abrj-637	6	6	,	,	PUNCT
abrj-637	6	7	which	which	PRON
abrj-637	6	8	are	be	AUX
abrj-637	6	9	mainly	mainly	ADV
abrj-637	6	10	rule	rule	NOUN
abrj-637	6	11	based	base	VERB
abrj-637	6	12	systems	system	NOUN
abrj-637	6	13	and	and	CCONJ
abrj-637	6	14	hand	hand	NOUN
abrj-637	6	15	audit	audit	NOUN
abrj-637	6	16	,	,	PUNCT
abrj-637	6	17	tend	tend	VERB
abrj-637	6	18	to	to	PART
abrj-637	6	19	be	be	AUX
abrj-637	6	20	lagging	lag	VERB
abrj-637	6	21	behind	behind	ADP
abrj-637	6	22	the	the	DET
abrj-637	6	23	intricacy	intricacy	NOUN
abrj-637	6	24	and	and	CCONJ
abrj-637	6	25	bulk	bulk	NOUN
abrj-637	6	26	of	of	ADP
abrj-637	6	27	the	the	DET
abrj-637	6	28	contemporary	contemporary	ADJ
abrj-637	6	29	corporate	corporate	ADJ
abrj-637	6	30	filings	filing	NOUN
abrj-637	6	31	.	.	PUNCT
abrj-637	7	1	this	this	DET
abrj-637	7	2	paper	paper	NOUN
abrj-637	7	3	will	will	AUX
abrj-637	7	4	discuss	discuss	VERB
abrj-637	7	5	the	the	DET
abrj-637	7	6	use	use	NOUN
abrj-637	7	7	of	of	ADP
abrj-637	7	8	machine	machine	NOUN
abrj-637	7	9	learning	learning	NOUN
abrj-637	7	10	(	(	PUNCT
abrj-637	7	11	ml	ml	NOUN
abrj-637	7	12	)	)	PUNCT
abrj-637	7	13	technologies	technology	NOUN
abrj-637	7	14	in	in	ADP
abrj-637	7	15	improving	improve	VERB
abrj-637	7	16	the	the	DET
abrj-637	7	17	process	process	NOUN
abrj-637	7	18	of	of	ADP
abrj-637	7	19	detecting	detect	VERB
abrj-637	7	20	tax	tax	NOUN
abrj-637	7	21	fraud	fraud	NOUN
abrj-637	7	22	and	and	CCONJ
abrj-637	7	23	risk	risk	NOUN
abrj-637	7	24	scoring	score	VERB
abrj-637	7	25	through	through	ADP
abrj-637	7	26	the	the	DET
abrj-637	7	27	use	use	NOUN
abrj-637	7	28	of	of	ADP
abrj-637	7	29	advanced	advanced	ADJ
abrj-637	7	30	data	datum	NOUN
abrj-637	7	31	analytics	analytic	NOUN
abrj-637	7	32	and	and	CCONJ
abrj-637	7	33	predictive	predictive	ADJ
abrj-637	7	34	models	model	NOUN
abrj-637	7	35	.	.	PUNCT
abrj-637	8	1	with	with	ADP
abrj-637	8	2	the	the	DET
abrj-637	8	3	help	help	NOUN
abrj-637	8	4	of	of	ADP
abrj-637	8	5	supervised	supervised	ADJ
abrj-637	8	6	,	,	PUNCT
abrj-637	8	7	unsupervised	unsupervised	ADJ
abrj-637	8	8	and	and	CCONJ
abrj-637	8	9	hybrid	hybrid	ADJ
abrj-637	8	10	learning	learning	NOUN
abrj-637	8	11	,	,	PUNCT
abrj-637	8	12	ml	ml	ADP
abrj-637	8	13	models	model	NOUN
abrj-637	8	14	are	be	AUX
abrj-637	8	15	able	able	ADJ
abrj-637	8	16	to	to	PART
abrj-637	8	17	discover	discover	VERB
abrj-637	8	18	the	the	DET
abrj-637	8	19	latent	latent	NOUN
abrj-637	8	20	patterns	pattern	NOUN
abrj-637	8	21	and	and	CCONJ
abrj-637	8	22	anomalies	anomaly	NOUN
abrj-637	8	23	and	and	CCONJ
abrj-637	8	24	come	come	VERB
abrj-637	8	25	up	up	ADP
abrj-637	8	26	with	with	ADP
abrj-637	8	27	risk	risk	NOUN
abrj-637	8	28	scores	score	NOUN
abrj-637	8	29	to	to	PART
abrj-637	8	30	determine	determine	VERB
abrj-637	8	31	the	the	DET
abrj-637	8	32	probability	probability	NOUN
abrj-637	8	33	of	of	ADP
abrj-637	8	34	fraud	fraud	NOUN
abrj-637	8	35	.	.	PUNCT
abrj-637	9	1	the	the	DET
abrj-637	9	2	paper	paper	NOUN
abrj-637	9	3	examines	examine	VERB
abrj-637	9	4	the	the	DET
abrj-637	9	5	current	current	ADJ
abrj-637	9	6	literature	literature	NOUN
abrj-637	9	7	on	on	ADP
abrj-637	9	8	the	the	DET
abrj-637	9	9	financial	financial	ADJ
abrj-637	9	10	and	and	CCONJ
abrj-637	9	11	tax	tax	NOUN
abrj-637	9	12	fraud	fraud	NOUN
abrj-637	9	13	detection	detection	NOUN
abrj-637	9	14	,	,	PUNCT
abrj-637	9	15	with	with	ADP
abrj-637	9	16	a	a	DET
abrj-637	9	17	specific	specific	ADJ
abrj-637	9	18	focus	focus	NOUN
abrj-637	9	19	on	on	ADP
abrj-637	9	20	how	how	SCONJ
abrj-637	9	21	these	these	DET
abrj-637	9	22	methods	method	NOUN
abrj-637	9	23	have	have	AUX
abrj-637	9	24	been	be	AUX
abrj-637	9	25	changing	change	VERB
abrj-637	9	26	towards	towards	ADP
abrj-637	9	27	adaptive	adaptive	ADJ
abrj-637	9	28	and	and	CCONJ
abrj-637	9	29	more	more	ADJ
abrj-637	9	30	data	data	NOUN
abrj-637	9	31	-	-	PUNCT
abrj-637	9	32	driven	drive	VERB
abrj-637	9	33	systems	system	NOUN
abrj-637	9	34	instead	instead	ADV
abrj-637	9	35	of	of	ADP
abrj-637	9	36	being	be	AUX
abrj-637	9	37	static	static	ADJ
abrj-637	9	38	and	and	CCONJ
abrj-637	9	39	rule	rule	NOUN
abrj-637	9	40	-	-	PUNCT
abrj-637	9	41	based	base	VERB
abrj-637	9	42	.	.	PUNCT
abrj-637	10	1	it	it	PRON
abrj-637	10	2	further	far	ADV
abrj-637	10	3	suggests	suggest	VERB
abrj-637	10	4	a	a	DET
abrj-637	10	5	structure	structure	NOUN
abrj-637	10	6	of	of	ADP
abrj-637	10	7	implementation	implementation	NOUN
abrj-637	10	8	which	which	PRON
abrj-637	10	9	takes	take	VERB
abrj-637	10	10	into	into	ADP
abrj-637	10	11	consideration	consideration	NOUN
abrj-637	10	12	data	datum	NOUN
abrj-637	10	13	preprocessing	preprocessing	NOUN
abrj-637	10	14	,	,	PUNCT
abrj-637	10	15	feature	feature	NOUN
abrj-637	10	16	engineering	engineering	NOUN
abrj-637	10	17	and	and	CCONJ
abrj-637	10	18	model	model	NOUN
abrj-637	10	19	evaluation	evaluation	NOUN
abrj-637	10	20	in	in	ADP
abrj-637	10	21	one	one	NUM
abrj-637	10	22	workflow	workflow	NOUN
abrj-637	10	23	that	that	PRON
abrj-637	10	24	is	be	AUX
abrj-637	10	25	fit	fit	ADJ
abrj-637	10	26	to	to	PART
abrj-637	10	27	be	be	AUX
abrj-637	10	28	used	use	VERB
abrj-637	10	29	by	by	ADP
abrj-637	10	30	tax	tax	NOUN
abrj-637	10	31	authorities	authority	NOUN
abrj-637	10	32	and	and	CCONJ
abrj-637	10	33	auditing	auditing	NOUN
abrj-637	10	34	firms	firm	NOUN
abrj-637	10	35	.	.	PUNCT
abrj-637	11	1	the	the	DET
abrj-637	11	2	proposed	propose	VERB
abrj-637	11	3	system	system	NOUN
abrj-637	11	4	makes	make	VERB
abrj-637	11	5	use	use	NOUN
abrj-637	11	6	of	of	ADP
abrj-637	11	7	algorithms	algorithm	NOUN
abrj-637	11	8	like	like	ADP
abrj-637	11	9	random	random	ADJ
abrj-637	11	10	forests	forest	NOUN
abrj-637	11	11	,	,	PUNCT
abrj-637	11	12	xgboost	xgboost	ADV
abrj-637	11	13	,	,	PUNCT
abrj-637	11	14	and	and	CCONJ
abrj-637	11	15	autoencoders	autoencoder	NOUN
abrj-637	11	16	to	to	PART
abrj-637	11	17	increase	increase	VERB
abrj-637	11	18	the	the	DET
abrj-637	11	19	accuracy	accuracy	NOUN
abrj-637	11	20	of	of	ADP
abrj-637	11	21	detection	detection	NOUN
abrj-637	11	22	and	and	CCONJ
abrj-637	11	23	minimize	minimize	VERB
abrj-637	11	24	the	the	DET
abrj-637	11	25	occurrence	occurrence	NOUN
abrj-637	11	26	of	of	ADP
abrj-637	11	27	false	false	ADJ
abrj-637	11	28	positives	positive	NOUN
abrj-637	11	29	.	.	PUNCT
abrj-637	12	1	moreover	moreover	ADV
abrj-637	12	2	,	,	PUNCT
abrj-637	12	3	the	the	DET
abrj-637	12	4	paper	paper	NOUN
abrj-637	12	5	emphasizes	emphasize	VERB
abrj-637	12	6	how	how	SCONJ
abrj-637	12	7	explainable	explainable	ADJ
abrj-637	12	8	ai	ai	NOUN
abrj-637	12	9	(	(	PUNCT
abrj-637	12	10	xai	xai	PROPN
abrj-637	12	11	)	)	PUNCT
abrj-637	12	12	can	can	AUX
abrj-637	12	13	be	be	AUX
abrj-637	12	14	important	important	ADJ
abrj-637	12	15	in	in	ADP
abrj-637	12	16	promoting	promote	VERB
abrj-637	12	17	transparency	transparency	NOUN
abrj-637	12	18	,	,	PUNCT
abrj-637	12	19	interpretability	interpretability	NOUN
abrj-637	12	20	,	,	PUNCT
abrj-637	12	21	and	and	CCONJ
abrj-637	12	22	adherence	adherence	NOUN
abrj-637	12	23	to	to	ADP
abrj-637	12	24	ethical	ethical	ADJ
abrj-637	12	25	and	and	CCONJ
abrj-637	12	26	legal	legal	ADJ
abrj-637	12	27	guidelines	guideline	NOUN
abrj-637	12	28	.	.	PUNCT
abrj-637	13	1	finally	finally	ADV
abrj-637	13	2	,	,	PUNCT
abrj-637	13	3	the	the	DET
abrj-637	13	4	study	study	NOUN
abrj-637	13	5	proves	prove	VERB
abrj-637	13	6	that	that	SCONJ
abrj-637	13	7	the	the	DET
abrj-637	13	8	application	application	NOUN
abrj-637	13	9	of	of	ADP
abrj-637	13	10	the	the	DET
abrj-637	13	11	ml	ml	ADV
abrj-637	13	12	-	-	PUNCT
abrj-637	13	13	based	base	VERB
abrj-637	13	14	fraud	fraud	NOUN
abrj-637	13	15	detection	detection	NOUN
abrj-637	13	16	and	and	CCONJ
abrj-637	13	17	risk	risk	NOUN
abrj-637	13	18	scoring	scoring	NOUN
abrj-637	13	19	can	can	AUX
abrj-637	13	20	become	become	VERB
abrj-637	13	21	a	a	DET
abrj-637	13	22	substantial	substantial	ADJ
abrj-637	13	23	enhancement	enhancement	NOUN
abrj-637	13	24	of	of	ADP
abrj-637	13	25	the	the	DET
abrj-637	13	26	effectiveness	effectiveness	NOUN
abrj-637	13	27	,	,	PUNCT
abrj-637	13	28	objectivity	objectivity	NOUN
abrj-637	13	29	,	,	PUNCT
abrj-637	13	30	and	and	CCONJ
abrj-637	13	31	scalability	scalability	NOUN
abrj-637	13	32	of	of	ADP
abrj-637	13	33	corporate	corporate	ADJ
abrj-637	13	34	tax	tax	NOUN
abrj-637	13	35	audits	audits	NOUN
abrj-637	13	36	.	.	PUNCT
abrj-637	14	1	the	the	DET
abrj-637	14	2	next	next	ADJ
abrj-637	14	3	step	step	NOUN
abrj-637	14	4	in	in	ADP
abrj-637	14	5	the	the	DET
abrj-637	14	6	work	work	NOUN
abrj-637	14	7	will	will	AUX
abrj-637	14	8	be	be	AUX
abrj-637	14	9	to	to	PART
abrj-637	14	10	incorporate	incorporate	VERB
abrj-637	14	11	deep	deep	ADJ
abrj-637	14	12	learning	learning	NOUN
abrj-637	14	13	,	,	PUNCT
abrj-637	14	14	natural	natural	ADJ
abrj-637	14	15	language	language	NOUN
abrj-637	14	16	processing	processing	NOUN
abrj-637	14	17	,	,	PUNCT
abrj-637	14	18	and	and	CCONJ
abrj-637	14	19	federated	federated	ADJ
abrj-637	14	20	systems	system	NOUN
abrj-637	14	21	to	to	PART
abrj-637	14	22	develop	develop	VERB
abrj-637	14	23	strong	strong	ADJ
abrj-637	14	24	,	,	PUNCT
abrj-637	14	25	privacy	privacy	NOUN
abrj-637	14	26	-	-	PUNCT
abrj-637	14	27	aware	aware	ADJ
abrj-637	14	28	frameworks	framework	NOUN
abrj-637	14	29	that	that	PRON
abrj-637	14	30	can	can	AUX
abrj-637	14	31	be	be	AUX
abrj-637	14	32	used	use	VERB
abrj-637	14	33	to	to	PART
abrj-637	14	34	detect	detect	VERB
abrj-637	14	35	fraud	fraud	NOUN
abrj-637	14	36	in	in	ADP
abrj-637	14	37	real	real	ADJ
abrj-637	14	38	-	-	PUNCT
abrj-637	14	39	time	time	NOUN
abrj-637	14	40	in	in	ADP
abrj-637	14	41	large	large	ADJ
abrj-637	14	42	-	-	PUNCT
abrj-637	14	43	scale	scale	NOUN
abrj-637	14	44	financial	financial	ADJ
abrj-637	14	45	ecosystems	ecosystem	NOUN
abrj-637	14	46	.	.	PUNCT
abrj-637	15	1	keywords	keyword	NOUN
abrj-637	15	2	:	:	PUNCT
abrj-637	15	3	corporate	corporate	ADJ
abrj-637	15	4	filings	filing	NOUN
abrj-637	15	5	,	,	PUNCT
abrj-637	15	6	financial	financial	ADJ
abrj-637	15	7	compliance	compliance	NOUN
abrj-637	15	8	,	,	PUNCT
abrj-637	15	9	machine	machine	NOUN
abrj-637	15	10	learning	learning	NOUN
abrj-637	15	11	,	,	PUNCT
abrj-637	15	12	risk	risk	NOUN
abrj-637	15	13	scoring	scoring	NOUN
abrj-637	15	14	,	,	PUNCT
abrj-637	15	15	tax	tax	NOUN
abrj-637	15	16	fraud	fraud	NOUN
abrj-637	15	17	detection	detection	NOUN
abrj-637	15	18	.	.	PUNCT
abrj-637	16	1	1	1	X
abrj-637	16	2	.	.	X
abrj-637	16	3	introduction	introduction	NOUN
abrj-637	16	4	tax	tax	NOUN
abrj-637	16	5	fraud	fraud	NOUN
abrj-637	16	6	is	be	AUX
abrj-637	16	7	one	one	NUM
abrj-637	16	8	of	of	ADP
abrj-637	16	9	the	the	DET
abrj-637	16	10	most	most	ADV
abrj-637	16	11	long	long	ADJ
abrj-637	16	12	running	run	VERB
abrj-637	16	13	and	and	CCONJ
abrj-637	16	14	intricate	intricate	ADJ
abrj-637	16	15	problems	problem	NOUN
abrj-637	16	16	of	of	ADP
abrj-637	16	17	governments	government	NOUN
abrj-637	16	18	,	,	PUNCT
abrj-637	16	19	regulators	regulator	NOUN
abrj-637	16	20	and	and	CCONJ
abrj-637	16	21	financial	financial	ADJ
abrj-637	16	22	institutions	institution	NOUN
abrj-637	16	23	around	around	ADP
abrj-637	16	24	the	the	DET
abrj-637	16	25	globe	globe	NOUN
abrj-637	16	26	.	.	PUNCT
abrj-637	17	1	with	with	ADP
abrj-637	17	2	the	the	DET
abrj-637	17	3	further	further	ADJ
abrj-637	17	4	globalization	globalization	NOUN
abrj-637	17	5	of	of	ADP
abrj-637	17	6	financial	financial	ADJ
abrj-637	17	7	systems	system	NOUN
abrj-637	17	8	and	and	CCONJ
abrj-637	17	9	their	their	PRON
abrj-637	17	10	interdependence	interdependence	NOUN
abrj-637	17	11	,	,	PUNCT
abrj-637	17	12	the	the	DET
abrj-637	17	13	scope	scope	NOUN
abrj-637	17	14	of	of	ADP
abrj-637	17	15	corporate	corporate	ADJ
abrj-637	17	16	submissions	submission	NOUN
abrj-637	17	17	and	and	CCONJ
abrj-637	17	18	tax	tax	NOUN
abrj-637	17	19	returns	return	NOUN
abrj-637	17	20	has	have	AUX
abrj-637	17	21	grown	grow	VERB
abrj-637	17	22	exponentially	exponentially	ADV
abrj-637	17	23	,	,	PUNCT
abrj-637	17	24	which	which	PRON
abrj-637	17	25	has	have	AUX
abrj-637	17	26	rendered	render	VERB
abrj-637	17	27	traditional	traditional	ADJ
abrj-637	17	28	detection	detection	NOUN
abrj-637	17	29	techniques	technique	NOUN
abrj-637	17	30	inefficient	inefficient	ADJ
abrj-637	17	31	and	and	CCONJ
abrj-637	17	32	insufficient	insufficient	ADJ
abrj-637	17	33	(	(	PUNCT
abrj-637	17	34	breslin	breslin	PROPN
abrj-637	17	35	,	,	PUNCT
abrj-637	17	36	2021	2021	NUM
abrj-637	17	37	;	;	PUNCT
abrj-637	17	38	wu	wu	PROPN
abrj-637	17	39	et	et	PROPN
abrj-637	17	40	al	al	PROPN
abrj-637	17	41	.	.	PROPN
abrj-637	17	42	,	,	PUNCT
abrj-637	17	43	2012	2012	NUM
abrj-637	17	44	)	)	PUNCT
abrj-637	17	45	.	.	PUNCT
abrj-637	18	1	the	the	DET
abrj-637	18	2	conventional	conventional	ADJ
abrj-637	18	3	audit	audit	NOUN
abrj-637	18	4	-	-	PUNCT
abrj-637	18	5	based	base	VERB
abrj-637	18	6	or	or	CCONJ
abrj-637	18	7	rule	rule	NOUN
abrj-637	18	8	-	-	PUNCT
abrj-637	18	9	driven	drive	VERB
abrj-637	18	10	methods	method	NOUN
abrj-637	18	11	are	be	AUX
abrj-637	18	12	mostly	mostly	ADV
abrj-637	18	13	dependent	dependent	ADJ
abrj-637	18	14	on	on	ADP
abrj-637	18	15	a	a	DET
abrj-637	18	16	priori	priori	ADJ
abrj-637	18	17	indicators	indicator	NOUN
abrj-637	18	18	and	and	CCONJ
abrj-637	18	19	human	human	ADJ
abrj-637	18	20	intelligence	intelligence	NOUN
abrj-637	18	21	,	,	PUNCT
abrj-637	18	22	which	which	PRON
abrj-637	18	23	do	do	AUX
abrj-637	18	24	not	not	PART
abrj-637	18	25	adequately	adequately	ADV
abrj-637	18	26	reflect	reflect	VERB
abrj-637	18	27	the	the	DET
abrj-637	18	28	nuanced	nuanced	ADJ
abrj-637	18	29	evolving	evolve	VERB
abrj-637	18	30	trends	trend	NOUN
abrj-637	18	31	of	of	ADP
abrj-637	18	32	fraudulent	fraudulent	ADJ
abrj-637	18	33	activity	activity	NOUN
abrj-637	18	34	in	in	ADP
abrj-637	18	35	a	a	DET
abrj-637	18	36	large	large	ADJ
abrj-637	18	37	-	-	PUNCT
abrj-637	18	38	scale	scale	NOUN
abrj-637	18	39	financial	financial	ADJ
abrj-637	18	40	data	datum	NOUN
abrj-637	18	41	(	(	PUNCT
abrj-637	18	42	ippolito	ippolito	PROPN
abrj-637	18	43	and	and	CCONJ
abrj-637	18	44	lozano	lozano	PROPN
abrj-637	18	45	,	,	PUNCT
abrj-637	18	46	2020	2020	NUM
abrj-637	18	47	;	;	PUNCT
abrj-637	18	48	tagbo	tagbo	NOUN
abrj-637	18	49	and	and	CCONJ
abrj-637	18	50	adekoya	adekoya	PROPN
abrj-637	18	51	,	,	PUNCT
abrj-637	18	52	2023	2023	NUM
abrj-637	18	53	)	)	PUNCT
abrj-637	18	54	.	.	PUNCT
abrj-637	19	1	the	the	DET
abrj-637	19	2	recent	recent	ADJ
abrj-637	19	3	innovations	innovation	NOUN
abrj-637	19	4	in	in	ADP
abrj-637	19	5	machine	machine	NOUN
abrj-637	19	6	learning	learning	NOUN
abrj-637	19	7	(	(	PUNCT
abrj-637	19	8	ml	ml	NOUN
abrj-637	19	9	)	)	PUNCT
abrj-637	19	10	have	have	AUX
abrj-637	19	11	presented	present	VERB
abrj-637	19	12	revolutionary	revolutionary	ADJ
abrj-637	19	13	possibilities	possibility	NOUN
abrj-637	19	14	in	in	ADP
abrj-637	19	15	the	the	DET
abrj-637	19	16	sphere	sphere	NOUN
abrj-637	19	17	of	of	ADP
abrj-637	19	18	tax	tax	NOUN
abrj-637	19	19	returns	return	NOUN
abrj-637	19	20	and	and	CCONJ
abrj-637	19	21	frauds	fraud	NOUN
abrj-637	19	22	.	.	PUNCT
abrj-637	20	1	ml	ml	NOUN
abrj-637	20	2	models	model	NOUN
abrj-637	20	3	can	can	AUX
abrj-637	20	4	be	be	AUX
abrj-637	20	5	trained	train	VERB
abrj-637	20	6	to	to	PART
abrj-637	20	7	process	process	VERB
abrj-637	20	8	high	high	ADJ
abrj-637	20	9	-	-	PUNCT
abrj-637	20	10	dimensional	dimensional	ADJ
abrj-637	20	11	data	datum	NOUN
abrj-637	20	12	that	that	PRON
abrj-637	20	13	is	be	AUX
abrj-637	20	14	complex	complex	ADJ
abrj-637	20	15	to	to	PART
abrj-637	20	16	detect	detect	VERB
abrj-637	20	17	anomalies	anomaly	NOUN
abrj-637	20	18	and	and	CCONJ
abrj-637	20	19	predict	predict	VERB
abrj-637	20	20	risk	risk	NOUN
abrj-637	20	21	scores	score	NOUN
abrj-637	20	22	as	as	ADV
abrj-637	20	23	well	well	ADV
abrj-637	20	24	as	as	ADP
abrj-637	20	25	perform	perform	VERB
abrj-637	20	26	fraud	fraud	NOUN
abrj-637	20	27	detection	detection	NOUN
abrj-637	20	28	processes	process	NOUN
abrj-637	20	29	in	in	ADP
abrj-637	20	30	an	an	DET
abrj-637	20	31	automated	automate	VERB
abrj-637	20	32	fashion	fashion	NOUN
abrj-637	20	33	(	(	PUNCT
abrj-637	20	34	acharya	acharya	NOUN
abrj-637	20	35	,	,	PUNCT
abrj-637	20	36	2025	2025	NUM
abrj-637	20	37	;	;	PUNCT
abrj-637	20	38	galla	galla	PROPN
abrj-637	20	39	,	,	PUNCT
abrj-637	20	40	2023	2023	NUM
abrj-637	20	41	)	)	PUNCT
abrj-637	20	42	.	.	PUNCT
abrj-637	21	1	machine	machine	NOUN
abrj-637	21	2	learning	learning	NOUN
abrj-637	21	3	models	model	NOUN
abrj-637	21	4	are	be	AUX
abrj-637	21	5	learning	learn	VERB
abrj-637	21	6	as	as	ADP
abrj-637	21	7	compared	compare	VERB
abrj-637	21	8	to	to	ADP
abrj-637	21	9	traditional	traditional	ADJ
abrj-637	21	10	systems	system	NOUN
abrj-637	21	11	,	,	PUNCT
abrj-637	21	12	which	which	PRON
abrj-637	21	13	rely	rely	VERB
abrj-637	21	14	on	on	ADP
abrj-637	21	15	fixed	fix	VERB
abrj-637	21	16	rules	rule	NOUN
abrj-637	21	17	and	and	CCONJ
abrj-637	21	18	constantly	constantly	ADV
abrj-637	21	19	change	change	VERB
abrj-637	21	20	to	to	PART
abrj-637	21	21	accommodate	accommodate	VERB
abrj-637	21	22	new	new	ADJ
abrj-637	21	23	types	type	NOUN
abrj-637	21	24	of	of	ADP
abrj-637	21	25	frauds	fraud	NOUN
abrj-637	21	26	(	(	PUNCT
abrj-637	21	27	nguyen	nguyen	NOUN
abrj-637	21	28	,	,	PUNCT
abrj-637	21	29	2025	2025	NUM
abrj-637	21	30	;	;	PUNCT
abrj-637	21	31	zhou	zhou	PROPN
abrj-637	21	32	et	et	PROPN
abrj-637	21	33	al	al	PROPN
abrj-637	21	34	.	.	PROPN
abrj-637	21	35	,	,	PUNCT
abrj-637	21	36	2024	2024	NUM
abrj-637	21	37	)	)	PUNCT
abrj-637	21	38	.	.	PUNCT
abrj-637	22	1	this	this	DET
abrj-637	22	2	deterministic	deterministic	ADJ
abrj-637	22	3	to	to	PART
abrj-637	22	4	probabilistic	probabilistic	VERB
abrj-637	22	5	analysis	analysis	NOUN
abrj-637	22	6	allows	allow	VERB
abrj-637	22	7	revealing	reveal	VERB
abrj-637	22	8	the	the	DET
abrj-637	22	9	concealed	conceal	VERB
abrj-637	22	10	correlation	correlation	NOUN
abrj-637	22	11	between	between	ADP
abrj-637	22	12	financial	financial	ADJ
abrj-637	22	13	attributes	attribute	NOUN
abrj-637	22	14	,	,	PUNCT
abrj-637	22	15	enhancing	enhance	VERB
abrj-637	22	16	the	the	DET
abrj-637	22	17	precision	precision	NOUN
abrj-637	22	18	and	and	CCONJ
abrj-637	22	19	speed	speed	NOUN
abrj-637	22	20	of	of	ADP
abrj-637	22	21	the	the	DET
abrj-637	22	22	fraud	fraud	NOUN
abrj-637	22	23	detection	detection	NOUN
abrj-637	22	24	process	process	NOUN
abrj-637	22	25	(	(	PUNCT
abrj-637	22	26	zhang	zhang	X
abrj-637	22	27	et	et	PROPN
abrj-637	22	28	al	al	PROPN
abrj-637	22	29	.	.	PROPN
abrj-637	22	30	,	,	PUNCT
abrj-637	22	31	2025	2025	NUM
abrj-637	22	32	;	;	PUNCT
abrj-637	22	33	mehta	mehta	PROPN
abrj-637	22	34	et	et	PROPN
abrj-637	22	35	al	al	PROPN
abrj-637	22	36	.	.	PROPN
abrj-637	22	37	,	,	PUNCT
abrj-637	22	38	2022	2022	NUM
abrj-637	22	39	)	)	PUNCT
abrj-637	22	40	.	.	PUNCT
abrj-637	23	1	the	the	DET
abrj-637	23	2	use	use	NOUN
abrj-637	23	3	of	of	ADP
abrj-637	23	4	supervised	supervised	ADJ
abrj-637	23	5	and	and	CCONJ
abrj-637	23	6	unsupervised	unsupervised	ADJ
abrj-637	23	7	learning	learning	NOUN
abrj-637	23	8	algorithms	algorithm	NOUN
abrj-637	23	9	,	,	PUNCT
abrj-637	23	10	including	include	VERB
abrj-637	23	11	random	random	ADJ
abrj-637	23	12	forest	forest	NOUN
abrj-637	23	13	,	,	PUNCT
abrj-637	23	14	gradient	gradient	ADJ
abrj-637	23	15	boosting	boosting	NOUN
abrj-637	23	16	and	and	CCONJ
abrj-637	23	17	autoencoders	autoencoder	NOUN
abrj-637	23	18	,	,	PUNCT
abrj-637	23	19	has	have	AUX
abrj-637	23	20	demonstrated	demonstrate	VERB
abrj-637	23	21	encouraging	encourage	VERB
abrj-637	23	22	outcomes	outcome	NOUN
abrj-637	23	23	in	in	ADP
abrj-637	23	24	the	the	DET
abrj-637	23	25	context	context	NOUN
abrj-637	23	26	of	of	ADP
abrj-637	23	27	corporate	corporate	ADJ
abrj-637	23	28	tax	tax	NOUN
abrj-637	23	29	administration	administration	NOUN
abrj-637	23	30	according	accord	VERB
abrj-637	23	31	to	to	ADP
abrj-637	23	32	differentiating	differentiate	VERB
abrj-637	23	33	between	between	ADP
abrj-637	23	34	legitimate	legitimate	ADJ
abrj-637	23	35	and	and	CCONJ
abrj-637	23	36	fraudulent	fraudulent	ADJ
abrj-637	23	37	filings	filing	NOUN
abrj-637	23	38	(	(	PUNCT
abrj-637	23	39	craja	craja	NOUN
abrj-637	23	40	et	et	PROPN
abrj-637	23	41	al	al	PROPN
abrj-637	23	42	.	.	PROPN
abrj-637	23	43	,	,	PUNCT
abrj-637	23	44	2020	2020	NUM
abrj-637	23	45	;	;	PUNCT
abrj-637	23	46	shujaaddeen	shujaaddeen	PROPN
abrj-637	23	47	et	et	PROPN
abrj-637	23	48	al	al	PROPN
abrj-637	23	49	.	.	PROPN
abrj-637	23	50	,	,	PUNCT
abrj-637	23	51	2024	2024	NUM
abrj-637	23	52	)	)	PUNCT
abrj-637	23	53	.	.	PUNCT
abrj-637	24	1	hybrid	hybrid	NOUN
abrj-637	24	2	and	and	CCONJ
abrj-637	24	3	ensemble	ensemble	ADJ
abrj-637	24	4	methods	method	NOUN
abrj-637	24	5	also	also	ADV
abrj-637	24	6	increase	increase	VERB
abrj-637	24	7	predictive	predictive	ADJ
abrj-637	24	8	reliability	reliability	NOUN
abrj-637	24	9	and	and	CCONJ
abrj-637	24	10	combine	combine	VERB
abrj-637	24	11	many	many	ADJ
abrj-637	24	12	algorithms	algorithm	NOUN
abrj-637	24	13	to	to	PART
abrj-637	24	14	decrease	decrease	VERB
abrj-637	24	15	overfitting	overfitte	VERB
abrj-637	24	16	and	and	CCONJ
abrj-637	24	17	future	future	ADJ
abrj-637	24	18	positives	positive	NOUN
abrj-637	24	19	(	(	PUNCT
abrj-637	24	20	choudhary	choudhary	PROPN
abrj-637	24	21	,	,	PUNCT
abrj-637	24	22	2025	2025	NUM
abrj-637	24	23	;	;	PUNCT
abrj-637	24	24	martínez	martínez	NOUN
abrj-637	24	25	,	,	PUNCT
abrj-637	24	26	2025	2025	NUM
abrj-637	24	27	)	)	PUNCT
abrj-637	24	28	.	.	PUNCT
abrj-637	25	1	an	an	DET
abrj-637	25	2	example	example	NOUN
abrj-637	25	3	is	be	AUX
abrj-637	25	4	that	that	SCONJ
abrj-637	25	5	ensemble	ensemble	ADJ
abrj-637	25	6	models	model	NOUN
abrj-637	25	7	implemented	implement	VERB
abrj-637	25	8	based	base	VERB
abrj-637	25	9	on	on	ADP
abrj-637	25	10	the	the	DET
abrj-637	25	11	soft	soft	ADJ
abrj-637	25	12	-	-	PUNCT
abrj-637	25	13	voting	voting	NOUN
abrj-637	25	14	and	and	CCONJ
abrj-637	25	15	stacking	stack	VERB
abrj-637	25	16	algorithms	algorithm	NOUN
abrj-637	25	17	have	have	AUX
abrj-637	25	18	been	be	AUX
abrj-637	25	19	useful	useful	ADJ
abrj-637	25	20	in	in	ADP
abrj-637	25	21	the	the	DET
abrj-637	25	22	development	development	NOUN
abrj-637	25	23	of	of	ADP
abrj-637	25	24	holistic	holistic	ADJ
abrj-637	25	25	fraud	fraud	NOUN
abrj-637	25	26	risk	risk	NOUN
abrj-637	25	27	scoring	scoring	NOUN
abrj-637	25	28	models	model	NOUN
abrj-637	25	29	(	(	PUNCT
abrj-637	25	30	zhou	zhou	X
abrj-637	25	31	et	et	PROPN
abrj-637	25	32	al	al	PROPN
abrj-637	25	33	.	.	PROPN
abrj-637	25	34	,	,	PUNCT
abrj-637	25	35	2024	2024	NUM
abrj-637	25	36	;	;	PUNCT
abrj-637	25	37	zhang	zhang	PROPN
abrj-637	25	38	et	et	PROPN
abrj-637	25	39	al	al	PROPN
abrj-637	25	40	.	.	PROPN
abrj-637	25	41	,	,	PUNCT
abrj-637	25	42	2025	2025	NUM
abrj-637	25	43	)	)	PUNCT
abrj-637	25	44	.	.	PUNCT
abrj-637	26	1	https://doi.org/10.55220/2576-6759.637	https://doi.org/10.55220/2576-6759.637	PROPN
abrj-637	26	2	asian	asian	PROPN
abrj-637	26	3	business	business	PROPN
abrj-637	26	4	research	research	NOUN
abrj-637	26	5	journal	journal	NOUN
abrj-637	26	6	,	,	PUNCT
abrj-637	26	7	2025	2025	NUM
abrj-637	26	8	,	,	PUNCT
abrj-637	26	9	10(11	10(11	NUM
abrj-637	26	10	):	):	PUNCT
abrj-637	26	11	1	1	NUM
abrj-637	26	12	-	-	SYM
abrj-637	26	13	13	13	NUM
abrj-637	26	14	2	2	NUM
abrj-637	26	15	©	©	PROPN
abrj-637	26	16	2025	2025	NUM
abrj-637	26	17	by	by	ADP
abrj-637	26	18	the	the	DET
abrj-637	26	19	authors	author	NOUN
abrj-637	26	20	;	;	PUNCT
abrj-637	26	21	licensee	licensee	NOUN
abrj-637	26	22	eastern	eastern	ADJ
abrj-637	26	23	centre	centre	NOUN
abrj-637	26	24	of	of	ADP
abrj-637	26	25	science	science	NOUN
abrj-637	26	26	and	and	CCONJ
abrj-637	26	27	education	education	NOUN
abrj-637	26	28	,	,	PUNCT
abrj-637	26	29	usa	usa	PROPN
abrj-637	26	30	the	the	DET
abrj-637	26	31	use	use	NOUN
abrj-637	26	32	of	of	ADP
abrj-637	26	33	natural	natural	ADJ
abrj-637	26	34	language	language	NOUN
abrj-637	26	35	processing	processing	NOUN
abrj-637	26	36	(	(	PUNCT
abrj-637	26	37	nlp	nlp	NOUN
abrj-637	26	38	)	)	PUNCT
abrj-637	26	39	and	and	CCONJ
abrj-637	26	40	textual	textual	ADJ
abrj-637	26	41	analysis	analysis	NOUN
abrj-637	26	42	in	in	ADP
abrj-637	26	43	fraud	fraud	NOUN
abrj-637	26	44	detection	detection	NOUN
abrj-637	26	45	models	model	NOUN
abrj-637	26	46	has	have	AUX
abrj-637	26	47	broadened	broaden	VERB
abrj-637	26	48	the	the	DET
abrj-637	26	49	analytical	analytical	ADJ
abrj-637	26	50	scope	scope	NOUN
abrj-637	26	51	of	of	ADP
abrj-637	26	52	fraud	fraud	NOUN
abrj-637	26	53	detection	detection	NOUN
abrj-637	26	54	to	to	PART
abrj-637	26	55	include	include	VERB
abrj-637	26	56	numerical	numerical	ADJ
abrj-637	26	57	indicators	indicator	NOUN
abrj-637	26	58	.	.	PUNCT
abrj-637	27	1	the	the	DET
abrj-637	27	2	research	research	NOUN
abrj-637	27	3	of	of	ADP
abrj-637	27	4	zhang	zhang	PROPN
abrj-637	27	5	et	et	PROPN
abrj-637	27	6	al	al	PROPN
abrj-637	27	7	.	.	PROPN
abrj-637	27	8	(	(	PUNCT
abrj-637	27	9	2024	2024	NUM
abrj-637	27	10	)	)	PUNCT
abrj-637	27	11	was	be	AUX
abrj-637	27	12	able	able	ADJ
abrj-637	27	13	to	to	PART
abrj-637	27	14	show	show	VERB
abrj-637	27	15	that	that	SCONJ
abrj-637	27	16	linguistic	linguistic	ADJ
abrj-637	27	17	readability	readability	NOUN
abrj-637	27	18	and	and	CCONJ
abrj-637	27	19	semantic	semantic	ADJ
abrj-637	27	20	aspects	aspect	NOUN
abrj-637	27	21	in	in	ADP
abrj-637	27	22	corporate	corporate	ADJ
abrj-637	27	23	filings	filing	NOUN
abrj-637	27	24	are	be	AUX
abrj-637	27	25	predictive	predictive	ADJ
abrj-637	27	26	of	of	ADP
abrj-637	27	27	fraudulent	fraudulent	ADJ
abrj-637	27	28	intent	intent	NOUN
abrj-637	27	29	.	.	PUNCT
abrj-637	28	1	likewise	likewise	ADV
abrj-637	28	2	,	,	PUNCT
abrj-637	28	3	unstructured	unstructured	ADJ
abrj-637	28	4	information	information	NOUN
abrj-637	28	5	,	,	PUNCT
abrj-637	28	6	including	include	VERB
abrj-637	28	7	annual	annual	ADJ
abrj-637	28	8	reports	report	NOUN
abrj-637	28	9	and	and	CCONJ
abrj-637	28	10	executive	executive	ADJ
abrj-637	28	11	declarations	declaration	NOUN
abrj-637	28	12	,	,	PUNCT
abrj-637	28	13	provides	provide	VERB
abrj-637	28	14	a	a	DET
abrj-637	28	15	valuable	valuable	ADJ
abrj-637	28	16	source	source	NOUN
abrj-637	28	17	of	of	ADP
abrj-637	28	18	information	information	NOUN
abrj-637	28	19	when	when	SCONJ
abrj-637	28	20	processed	process	VERB
abrj-637	28	21	with	with	ADP
abrj-637	28	22	the	the	DET
abrj-637	28	23	deep	deep	ADJ
abrj-637	28	24	learning	learning	NOUN
abrj-637	28	25	and	and	CCONJ
abrj-637	28	26	nlp	nlp	NOUN
abrj-637	28	27	-	-	PUNCT
abrj-637	28	28	based	base	VERB
abrj-637	28	29	algorithms	algorithm	NOUN
abrj-637	28	30	(	(	PUNCT
abrj-637	28	31	ji	ji	X
abrj-637	28	32	et	et	PROPN
abrj-637	28	33	al	al	PROPN
abrj-637	28	34	.	.	PROPN
abrj-637	28	35	,	,	PUNCT
abrj-637	28	36	2024	2024	NUM
abrj-637	28	37	;	;	PUNCT
abrj-637	28	38	zhang	zhang	PROPN
abrj-637	28	39	et	et	PROPN
abrj-637	28	40	al	al	PROPN
abrj-637	28	41	.	.	PROPN
abrj-637	28	42	,	,	PUNCT
abrj-637	28	43	2024	2024	NUM
abrj-637	28	44	)	)	PUNCT
abrj-637	28	45	.	.	PUNCT
abrj-637	29	1	in	in	ADP
abrj-637	29	2	spite	spite	NOUN
abrj-637	29	3	of	of	ADP
abrj-637	29	4	these	these	PRON
abrj-637	29	5	,	,	PUNCT
abrj-637	29	6	there	there	PRON
abrj-637	29	7	are	be	VERB
abrj-637	29	8	various	various	ADJ
abrj-637	29	9	hurdles	hurdle	NOUN
abrj-637	29	10	in	in	ADP
abrj-637	29	11	the	the	DET
abrj-637	29	12	application	application	NOUN
abrj-637	29	13	of	of	ADP
abrj-637	29	14	ml	ml	NOUN
abrj-637	29	15	systems	system	NOUN
abrj-637	29	16	to	to	PART
abrj-637	29	17	detect	detect	VERB
abrj-637	29	18	tax	tax	NOUN
abrj-637	29	19	frauds	fraud	NOUN
abrj-637	29	20	.	.	PUNCT
abrj-637	30	1	such	such	ADJ
abrj-637	30	2	problems	problem	NOUN
abrj-637	30	3	as	as	ADP
abrj-637	30	4	data	datum	NOUN
abrj-637	30	5	imbalance	imbalance	NOUN
abrj-637	30	6	,	,	PUNCT
abrj-637	30	7	restricted	restrict	VERB
abrj-637	30	8	access	access	NOUN
abrj-637	30	9	to	to	ADP
abrj-637	30	10	labeled	label	VERB
abrj-637	30	11	data	datum	NOUN
abrj-637	30	12	,	,	PUNCT
abrj-637	30	13	model	model	NOUN
abrj-637	30	14	explainability	explainability	NOUN
abrj-637	30	15	,	,	PUNCT
abrj-637	30	16	and	and	CCONJ
abrj-637	30	17	ethical	ethical	ADJ
abrj-637	30	18	aspects	aspect	NOUN
abrj-637	30	19	still	still	ADV
abrj-637	30	20	impede	impede	VERB
abrj-637	30	21	a	a	DET
abrj-637	30	22	mass	mass	ADJ
abrj-637	30	23	adoption	adoption	NOUN
abrj-637	30	24	(	(	PUNCT
abrj-637	30	25	tagbo	tagbo	NOUN
abrj-637	30	26	and	and	CCONJ
abrj-637	30	27	adekoya	adekoya	NOUN
abrj-637	30	28	,	,	PUNCT
abrj-637	30	29	2023	2023	NUM
abrj-637	30	30	;	;	PUNCT
abrj-637	30	31	wahyono	wahyono	PROPN
abrj-637	30	32	and	and	CCONJ
abrj-637	30	33	david	david	PROPN
abrj-637	30	34	,	,	PUNCT
abrj-637	30	35	2025	2025	NUM
abrj-637	30	36	)	)	PUNCT
abrj-637	30	37	.	.	PUNCT
abrj-637	31	1	the	the	DET
abrj-637	31	2	xai	xai	PROPN
abrj-637	31	3	frameworks	framework	NOUN
abrj-637	31	4	are	be	AUX
abrj-637	31	5	thus	thus	ADV
abrj-637	31	6	important	important	ADJ
abrj-637	31	7	towards	towards	ADP
abrj-637	31	8	the	the	DET
abrj-637	31	9	transparency	transparency	NOUN
abrj-637	31	10	and	and	CCONJ
abrj-637	31	11	accountability	accountability	NOUN
abrj-637	31	12	of	of	ADP
abrj-637	31	13	the	the	DET
abrj-637	31	14	model	model	NOUN
abrj-637	31	15	-	-	PUNCT
abrj-637	31	16	driven	drive	VERB
abrj-637	31	17	decision	decision	NOUN
abrj-637	31	18	-	-	PUNCT
abrj-637	31	19	making	make	VERB
abrj-637	31	20	process	process	NOUN
abrj-637	31	21	(	(	PUNCT
abrj-637	31	22	zhou	zhou	X
abrj-637	31	23	et	et	PROPN
abrj-637	31	24	al	al	PROPN
abrj-637	31	25	.	.	PROPN
abrj-637	31	26	,	,	PUNCT
abrj-637	31	27	2024	2024	NUM
abrj-637	31	28	)	)	PUNCT
abrj-637	31	29	.	.	PUNCT
abrj-637	32	1	moreover	moreover	ADV
abrj-637	32	2	,	,	PUNCT
abrj-637	32	3	the	the	DET
abrj-637	32	4	use	use	NOUN
abrj-637	32	5	of	of	ADP
abrj-637	32	6	ai	ai	NOUN
abrj-637	32	7	-	-	PUNCT
abrj-637	32	8	based	base	VERB
abrj-637	32	9	tools	tool	NOUN
abrj-637	32	10	and	and	CCONJ
abrj-637	32	11	human	human	ADJ
abrj-637	32	12	auditors	auditor	NOUN
abrj-637	32	13	is	be	AUX
abrj-637	32	14	crucial	crucial	ADJ
abrj-637	32	15	to	to	PART
abrj-637	32	16	ensure	ensure	VERB
abrj-637	32	17	that	that	SCONJ
abrj-637	32	18	there	there	PRON
abrj-637	32	19	is	be	VERB
abrj-637	32	20	no	no	DET
abrj-637	32	21	violation	violation	NOUN
abrj-637	32	22	of	of	ADP
abrj-637	32	23	the	the	DET
abrj-637	32	24	law	law	NOUN
abrj-637	32	25	and	and	CCONJ
abrj-637	32	26	people	people	NOUN
abrj-637	32	27	do	do	AUX
abrj-637	32	28	not	not	PART
abrj-637	32	29	lose	lose	VERB
abrj-637	32	30	their	their	PRON
abrj-637	32	31	faith	faith	NOUN
abrj-637	32	32	in	in	ADP
abrj-637	32	33	automated	automate	VERB
abrj-637	32	34	fraud	fraud	NOUN
abrj-637	32	35	-	-	PUNCT
abrj-637	32	36	detecting	detect	VERB
abrj-637	32	37	systems	system	NOUN
abrj-637	32	38	(	(	PUNCT
abrj-637	32	39	breslin	breslin	PROPN
abrj-637	32	40	,	,	PUNCT
abrj-637	32	41	2021	2021	NUM
abrj-637	32	42	;	;	PUNCT
abrj-637	32	43	acharya	acharya	NOUN
abrj-637	32	44	,	,	PUNCT
abrj-637	32	45	2025	2025	NUM
abrj-637	32	46	)	)	PUNCT
abrj-637	32	47	.	.	PUNCT
abrj-637	33	1	the	the	DET
abrj-637	33	2	current	current	ADJ
abrj-637	33	3	paper	paper	NOUN
abrj-637	33	4	will	will	AUX
abrj-637	33	5	discuss	discuss	VERB
abrj-637	33	6	the	the	DET
abrj-637	33	7	use	use	NOUN
abrj-637	33	8	of	of	ADP
abrj-637	33	9	machine	machine	NOUN
abrj-637	33	10	learning	learn	VERB
abrj-637	33	11	to	to	PART
abrj-637	33	12	improve	improve	VERB
abrj-637	33	13	tax	tax	NOUN
abrj-637	33	14	fraud	fraud	NOUN
abrj-637	33	15	detection	detection	NOUN
abrj-637	33	16	and	and	CCONJ
abrj-637	33	17	risk	risk	NOUN
abrj-637	33	18	scoring	scoring	NOUN
abrj-637	33	19	in	in	ADP
abrj-637	33	20	corporate	corporate	ADJ
abrj-637	33	21	filings	filing	NOUN
abrj-637	33	22	.	.	PUNCT
abrj-637	34	1	it	it	PRON
abrj-637	34	2	evaluates	evaluate	VERB
abrj-637	34	3	how	how	SCONJ
abrj-637	34	4	data	datum	NOUN
abrj-637	34	5	analytics	analytic	NOUN
abrj-637	34	6	,	,	PUNCT
abrj-637	34	7	model	model	NOUN
abrj-637	34	8	design	design	NOUN
abrj-637	34	9	,	,	PUNCT
abrj-637	34	10	and	and	CCONJ
abrj-637	34	11	risk	risk	NOUN
abrj-637	34	12	evaluation	evaluation	NOUN
abrj-637	34	13	can	can	AUX
abrj-637	34	14	be	be	AUX
abrj-637	34	15	combined	combine	VERB
abrj-637	34	16	into	into	ADP
abrj-637	34	17	a	a	DET
abrj-637	34	18	single	single	ADJ
abrj-637	34	19	,	,	PUNCT
abrj-637	34	20	scalable	scalable	ADJ
abrj-637	34	21	structure	structure	NOUN
abrj-637	34	22	by	by	ADP
abrj-637	34	23	analyzing	analyze	VERB
abrj-637	34	24	the	the	DET
abrj-637	34	25	literature	literature	NOUN
abrj-637	34	26	on	on	ADP
abrj-637	34	27	the	the	DET
abrj-637	34	28	topic	topic	NOUN
abrj-637	34	29	,	,	PUNCT
abrj-637	34	30	and	and	CCONJ
abrj-637	34	31	providing	provide	VERB
abrj-637	34	32	a	a	DET
abrj-637	34	33	framework	framework	NOUN
abrj-637	34	34	of	of	ADP
abrj-637	34	35	the	the	DET
abrj-637	34	36	related	relate	VERB
abrj-637	34	37	process	process	NOUN
abrj-637	34	38	.	.	PUNCT
abrj-637	35	1	the	the	DET
abrj-637	35	2	paper	paper	NOUN
abrj-637	35	3	is	be	AUX
abrj-637	35	4	a	a	DET
abrj-637	35	5	contribution	contribution	NOUN
abrj-637	35	6	to	to	ADP
abrj-637	35	7	the	the	DET
abrj-637	35	8	expanding	expand	VERB
abrj-637	35	9	domain	domain	NOUN
abrj-637	35	10	of	of	ADP
abrj-637	35	11	computational	computational	ADJ
abrj-637	35	12	tax	tax	NOUN
abrj-637	35	13	analytics	analytic	NOUN
abrj-637	35	14	,	,	PUNCT
abrj-637	35	15	as	as	SCONJ
abrj-637	35	16	it	it	PRON
abrj-637	35	17	suggests	suggest	VERB
abrj-637	35	18	that	that	SCONJ
abrj-637	35	19	ml	ml	NOUN
abrj-637	35	20	models	model	NOUN
abrj-637	35	21	can	can	AUX
abrj-637	35	22	enhance	enhance	VERB
abrj-637	35	23	efficiency	efficiency	NOUN
abrj-637	35	24	and	and	CCONJ
abrj-637	35	25	minimize	minimize	VERB
abrj-637	35	26	false	false	ADJ
abrj-637	35	27	alarms	alarm	NOUN
abrj-637	35	28	,	,	PUNCT
abrj-637	35	29	transparency	transparency	NOUN
abrj-637	35	30	in	in	ADP
abrj-637	35	31	the	the	DET
abrj-637	35	32	contemporary	contemporary	PROPN
abrj-637	35	33	tax	tax	PROPN
abrj-637	35	34	administration	administration	NOUN
abrj-637	35	35	(	(	PUNCT
abrj-637	35	36	wu	wu	PROPN
abrj-637	35	37	et	et	PROPN
abrj-637	35	38	al	al	PROPN
abrj-637	35	39	.	.	PROPN
abrj-637	35	40	,	,	PUNCT
abrj-637	35	41	2012	2012	NUM
abrj-637	35	42	;	;	PUNCT
abrj-637	35	43	zhang	zhang	PROPN
abrj-637	35	44	et	et	PROPN
abrj-637	35	45	al	al	PROPN
abrj-637	35	46	.	.	PROPN
abrj-637	35	47	,	,	PUNCT
abrj-637	35	48	2025	2025	NUM
abrj-637	35	49	)	)	PUNCT
abrj-637	35	50	.	.	PUNCT
abrj-637	36	1	2	2	X
abrj-637	36	2	.	.	X
abrj-637	36	3	literature	literature	NOUN
abrj-637	36	4	review	review	PROPN
abrj-637	36	5	tax	tax	NOUN
abrj-637	36	6	fraud	fraud	NOUN
abrj-637	36	7	is	be	AUX
abrj-637	36	8	an	an	DET
abrj-637	36	9	issue	issue	NOUN
abrj-637	36	10	that	that	PRON
abrj-637	36	11	has	have	AUX
abrj-637	36	12	been	be	AUX
abrj-637	36	13	very	very	ADV
abrj-637	36	14	difficult	difficult	ADJ
abrj-637	36	15	to	to	PART
abrj-637	36	16	detect	detect	VERB
abrj-637	36	17	by	by	ADP
abrj-637	36	18	the	the	DET
abrj-637	36	19	governments	government	NOUN
abrj-637	36	20	and	and	CCONJ
abrj-637	36	21	financial	financial	ADJ
abrj-637	36	22	regulators	regulator	NOUN
abrj-637	36	23	.	.	PUNCT
abrj-637	37	1	conventional	conventional	ADJ
abrj-637	37	2	auditing	auditing	NOUN
abrj-637	37	3	and	and	CCONJ
abrj-637	37	4	rule	rule	NOUN
abrj-637	37	5	based	base	VERB
abrj-637	37	6	analytics	analytic	NOUN
abrj-637	37	7	have	have	AUX
abrj-637	37	8	in	in	ADP
abrj-637	37	9	the	the	DET
abrj-637	37	10	past	past	NOUN
abrj-637	37	11	depended	depend	VERB
abrj-637	37	12	on	on	ADP
abrj-637	37	13	the	the	DET
abrj-637	37	14	use	use	NOUN
abrj-637	37	15	of	of	ADP
abrj-637	37	16	static	static	ADJ
abrj-637	37	17	criteria	criterion	NOUN
abrj-637	37	18	and	and	CCONJ
abrj-637	37	19	manual	manual	ADJ
abrj-637	37	20	evaluations	evaluation	NOUN
abrj-637	37	21	to	to	PART
abrj-637	37	22	detect	detect	VERB
abrj-637	37	23	anomalies	anomaly	NOUN
abrj-637	37	24	in	in	ADP
abrj-637	37	25	the	the	DET
abrj-637	37	26	corporate	corporate	ADJ
abrj-637	37	27	filings	filing	NOUN
abrj-637	37	28	.	.	PUNCT
abrj-637	38	1	nevertheless	nevertheless	ADV
abrj-637	38	2	,	,	PUNCT
abrj-637	38	3	the	the	DET
abrj-637	38	4	traditional	traditional	ADJ
abrj-637	38	5	techniques	technique	NOUN
abrj-637	38	6	have	have	AUX
abrj-637	38	7	not	not	PART
abrj-637	38	8	been	be	AUX
abrj-637	38	9	sufficient	sufficient	ADJ
abrj-637	38	10	to	to	PART
abrj-637	38	11	manage	manage	VERB
abrj-637	38	12	the	the	DET
abrj-637	38	13	constantly	constantly	ADV
abrj-637	38	14	growing	grow	VERB
abrj-637	38	15	amount	amount	NOUN
abrj-637	38	16	,	,	PUNCT
abrj-637	38	17	speed	speed	NOUN
abrj-637	38	18	,	,	PUNCT
abrj-637	38	19	and	and	CCONJ
abrj-637	38	20	diversity	diversity	NOUN
abrj-637	38	21	of	of	ADP
abrj-637	38	22	financial	financial	ADJ
abrj-637	38	23	information	information	NOUN
abrj-637	38	24	(	(	PUNCT
abrj-637	38	25	breslin	breslin	PROPN
abrj-637	38	26	,	,	PUNCT
abrj-637	38	27	2021	2021	NUM
abrj-637	38	28	;	;	PUNCT
abrj-637	38	29	wu	wu	PROPN
abrj-637	38	30	et	et	PROPN
abrj-637	38	31	al	al	PROPN
abrj-637	38	32	.	.	PROPN
abrj-637	38	33	,	,	PUNCT
abrj-637	38	34	2012	2012	NUM
abrj-637	38	35	)	)	PUNCT
abrj-637	38	36	.	.	PUNCT
abrj-637	39	1	with	with	SCONJ
abrj-637	39	2	more	more	ADV
abrj-637	39	3	complex	complex	ADJ
abrj-637	39	4	corporate	corporate	ADJ
abrj-637	39	5	financial	financial	ADJ
abrj-637	39	6	structures	structure	NOUN
abrj-637	39	7	,	,	PUNCT
abrj-637	39	8	automated	automate	VERB
abrj-637	39	9	,	,	PUNCT
abrj-637	39	10	data	data	NOUN
abrj-637	39	11	-	-	PUNCT
abrj-637	39	12	driven	drive	VERB
abrj-637	39	13	systems	system	NOUN
abrj-637	39	14	which	which	PRON
abrj-637	39	15	can	can	AUX
abrj-637	39	16	be	be	AUX
abrj-637	39	17	trained	train	VERB
abrj-637	39	18	to	to	PART
abrj-637	39	19	be	be	AUX
abrj-637	39	20	adaptive	adaptive	ADJ
abrj-637	39	21	learned	learn	VERB
abrj-637	39	22	have	have	AUX
abrj-637	39	23	risen	rise	VERB
abrj-637	39	24	to	to	PART
abrj-637	39	25	be	be	AUX
abrj-637	39	26	essential	essential	ADJ
abrj-637	39	27	in	in	ADP
abrj-637	39	28	fraud	fraud	NOUN
abrj-637	39	29	detection	detection	NOUN
abrj-637	39	30	and	and	CCONJ
abrj-637	39	31	tax	tax	NOUN
abrj-637	39	32	compliance	compliance	NOUN
abrj-637	39	33	(	(	PUNCT
abrj-637	39	34	acharya	acharya	NOUN
abrj-637	39	35	,	,	PUNCT
abrj-637	39	36	2025	2025	NUM
abrj-637	39	37	;	;	PUNCT
abrj-637	39	38	galla	galla	PROPN
abrj-637	39	39	,	,	PUNCT
abrj-637	39	40	2023	2023	NUM
abrj-637	39	41	)	)	PUNCT
abrj-637	39	42	.	.	PUNCT
abrj-637	40	1	in	in	ADP
abrj-637	40	2	the	the	DET
abrj-637	40	3	past	past	NOUN
abrj-637	40	4	,	,	PUNCT
abrj-637	40	5	detection	detection	NOUN
abrj-637	40	6	of	of	ADP
abrj-637	40	7	tax	tax	NOUN
abrj-637	40	8	fraud	fraud	NOUN
abrj-637	40	9	depended	depend	VERB
abrj-637	40	10	on	on	ADP
abrj-637	40	11	the	the	DET
abrj-637	40	12	deterministic	deterministic	ADJ
abrj-637	40	13	model	model	NOUN
abrj-637	40	14	,	,	PUNCT
abrj-637	40	15	like	like	ADP
abrj-637	40	16	rule	rule	NOUN
abrj-637	40	17	-	-	PUNCT
abrj-637	40	18	based	base	VERB
abrj-637	40	19	systems	system	NOUN
abrj-637	40	20	and	and	CCONJ
abrj-637	40	21	indicators	indicator	NOUN
abrj-637	40	22	that	that	PRON
abrj-637	40	23	are	be	AUX
abrj-637	40	24	defined	define	VERB
abrj-637	40	25	by	by	ADP
abrj-637	40	26	experts	expert	NOUN
abrj-637	40	27	and	and	CCONJ
abrj-637	40	28	that	that	PRON
abrj-637	40	29	are	be	AUX
abrj-637	40	30	applied	apply	VERB
abrj-637	40	31	by	by	ADP
abrj-637	40	32	auditors	auditor	NOUN
abrj-637	40	33	to	to	PART
abrj-637	40	34	identify	identify	VERB
abrj-637	40	35	suspicious	suspicious	ADJ
abrj-637	40	36	filings	filing	NOUN
abrj-637	40	37	(	(	PUNCT
abrj-637	40	38	wu	wu	PROPN
abrj-637	40	39	et	et	PROPN
abrj-637	40	40	al	al	PROPN
abrj-637	40	41	.	.	PROPN
abrj-637	40	42	,	,	PUNCT
abrj-637	40	43	2012	2012	NUM
abrj-637	40	44	)	)	PUNCT
abrj-637	40	45	.	.	PUNCT
abrj-637	41	1	although	although	SCONJ
abrj-637	41	2	good	good	ADJ
abrj-637	41	3	in	in	ADP
abrj-637	41	4	isolated	isolated	ADJ
abrj-637	41	5	and	and	CCONJ
abrj-637	41	6	repetitive	repetitive	ADJ
abrj-637	41	7	cases	case	NOUN
abrj-637	41	8	,	,	PUNCT
abrj-637	41	9	these	these	DET
abrj-637	41	10	models	model	NOUN
abrj-637	41	11	were	be	AUX
abrj-637	41	12	not	not	PART
abrj-637	41	13	scalable	scalable	ADJ
abrj-637	41	14	and	and	CCONJ
abrj-637	41	15	flexible	flexible	ADJ
abrj-637	41	16	.	.	PUNCT
abrj-637	42	1	the	the	DET
abrj-637	42	2	manual	manual	ADJ
abrj-637	42	3	audits	audits	NOUN
abrj-637	42	4	could	could	AUX
abrj-637	42	5	be	be	AUX
abrj-637	42	6	extremely	extremely	ADV
abrj-637	42	7	time	time	NOUN
abrj-637	42	8	-	-	PUNCT
abrj-637	42	9	consuming	consume	VERB
abrj-637	42	10	,	,	PUNCT
abrj-637	42	11	as	as	ADV
abrj-637	42	12	well	well	ADV
abrj-637	42	13	as	as	ADP
abrj-637	42	14	liable	liable	ADJ
abrj-637	42	15	to	to	ADP
abrj-637	42	16	human	human	ADJ
abrj-637	42	17	error	error	NOUN
abrj-637	42	18	,	,	PUNCT
abrj-637	42	19	which	which	PRON
abrj-637	42	20	caused	cause	VERB
abrj-637	42	21	inefficiencies	inefficiency	NOUN
abrj-637	42	22	and	and	CCONJ
abrj-637	42	23	discrepancies	discrepancy	NOUN
abrj-637	42	24	in	in	ADP
abrj-637	42	25	fraud	fraud	NOUN
abrj-637	42	26	detection	detection	NOUN
abrj-637	42	27	(	(	PUNCT
abrj-637	42	28	breslin	breslin	NOUN
abrj-637	42	29	,	,	PUNCT
abrj-637	42	30	2021	2021	NUM
abrj-637	42	31	)	)	PUNCT
abrj-637	42	32	.	.	PUNCT
abrj-637	43	1	rule	rule	NOUN
abrj-637	43	2	-	-	PUNCT
abrj-637	43	3	based	base	VERB
abrj-637	43	4	systems	system	NOUN
abrj-637	43	5	were	be	AUX
abrj-637	43	6	unable	unable	ADJ
abrj-637	43	7	to	to	PART
abrj-637	43	8	account	account	VERB
abrj-637	43	9	for	for	ADP
abrj-637	43	10	sophisticated	sophisticated	ADJ
abrj-637	43	11	trends	trend	NOUN
abrj-637	43	12	of	of	ADP
abrj-637	43	13	tax	tax	NOUN
abrj-637	43	14	evasion	evasion	NOUN
abrj-637	43	15	and	and	CCONJ
abrj-637	43	16	misreporting	misreporting	NOUN
abrj-637	43	17	as	as	ADP
abrj-637	43	18	tax	tax	NOUN
abrj-637	43	19	systems	system	NOUN
abrj-637	43	20	grew	grow	VERB
abrj-637	43	21	increasingly	increasingly	ADV
abrj-637	43	22	digital	digital	ADJ
abrj-637	43	23	.	.	PUNCT
abrj-637	44	1	the	the	DET
abrj-637	44	2	restrictions	restriction	NOUN
abrj-637	44	3	motivated	motivate	VERB
abrj-637	44	4	the	the	DET
abrj-637	44	5	creation	creation	NOUN
abrj-637	44	6	of	of	ADP
abrj-637	44	7	datadriven	datadriven	ADJ
abrj-637	44	8	approaches	approach	NOUN
abrj-637	44	9	that	that	PRON
abrj-637	44	10	can	can	AUX
abrj-637	44	11	learn	learn	VERB
abrj-637	44	12	large	large	ADJ
abrj-637	44	13	and	and	CCONJ
abrj-637	44	14	dynamic	dynamic	ADJ
abrj-637	44	15	datasets	dataset	NOUN
abrj-637	44	16	(	(	PUNCT
abrj-637	44	17	ippolito	ippolito	PROPN
abrj-637	44	18	and	and	CCONJ
abrj-637	44	19	lozano	lozano	PROPN
abrj-637	44	20	,	,	PUNCT
abrj-637	44	21	2020	2020	NUM
abrj-637	44	22	)	)	PUNCT
abrj-637	44	23	.	.	PUNCT
abrj-637	45	1	machine	machine	NOUN
abrj-637	45	2	learning	learn	VERB
abrj-637	45	3	as	as	ADP
abrj-637	45	4	a	a	DET
abrj-637	45	5	solution	solution	NOUN
abrj-637	45	6	to	to	PART
abrj-637	45	7	fraud	fraud	NOUN
abrj-637	45	8	detection	detection	NOUN
abrj-637	45	9	was	be	AUX
abrj-637	45	10	the	the	DET
abrj-637	45	11	answer	answer	NOUN
abrj-637	45	12	to	to	ADP
abrj-637	45	13	the	the	DET
abrj-637	45	14	limitations	limitation	NOUN
abrj-637	45	15	of	of	ADP
abrj-637	45	16	rule	rule	NOUN
abrj-637	45	17	-	-	PUNCT
abrj-637	45	18	based	base	VERB
abrj-637	45	19	systems	system	NOUN
abrj-637	45	20	.	.	PUNCT
abrj-637	46	1	machine	machine	NOUN
abrj-637	46	2	learning	learning	NOUN
abrj-637	46	3	-	-	PUNCT
abrj-637	46	4	based	base	VERB
abrj-637	46	5	models	model	NOUN
abrj-637	46	6	,	,	PUNCT
abrj-637	46	7	which	which	PRON
abrj-637	46	8	have	have	AUX
abrj-637	46	9	been	be	AUX
abrj-637	46	10	trained	train	VERB
abrj-637	46	11	on	on	ADP
abrj-637	46	12	historic	historic	ADJ
abrj-637	46	13	data	datum	NOUN
abrj-637	46	14	,	,	PUNCT
abrj-637	46	15	are	be	AUX
abrj-637	46	16	able	able	ADJ
abrj-637	46	17	to	to	PART
abrj-637	46	18	detect	detect	VERB
abrj-637	46	19	trends	trend	NOUN
abrj-637	46	20	,	,	PUNCT
abrj-637	46	21	deviations	deviation	NOUN
abrj-637	46	22	,	,	PUNCT
abrj-637	46	23	and	and	CCONJ
abrj-637	46	24	linkages	linkage	NOUN
abrj-637	46	25	that	that	PRON
abrj-637	46	26	could	could	AUX
abrj-637	46	27	be	be	AUX
abrj-637	46	28	indicative	indicative	ADJ
abrj-637	46	29	of	of	ADP
abrj-637	46	30	fraudulent	fraudulent	ADJ
abrj-637	46	31	behavior	behavior	NOUN
abrj-637	46	32	(	(	PUNCT
abrj-637	46	33	acharya	acharya	NOUN
abrj-637	46	34	,	,	PUNCT
abrj-637	46	35	2025	2025	NUM
abrj-637	46	36	;	;	PUNCT
abrj-637	46	37	choudhary	choudhary	PROPN
abrj-637	46	38	,	,	PUNCT
abrj-637	46	39	2025	2025	NUM
abrj-637	46	40	)	)	PUNCT
abrj-637	46	41	.	.	PUNCT
abrj-637	47	1	with	with	ADP
abrj-637	47	2	the	the	DET
abrj-637	47	3	help	help	NOUN
abrj-637	47	4	of	of	ADP
abrj-637	47	5	algorithms	algorithm	NOUN
abrj-637	47	6	,	,	PUNCT
abrj-637	47	7	including	include	VERB
abrj-637	47	8	random	random	ADJ
abrj-637	47	9	forests	forest	NOUN
abrj-637	47	10	and	and	CCONJ
abrj-637	47	11	gradient	gradient	ADJ
abrj-637	47	12	boosting	boost	VERB
abrj-637	47	13	machines	machine	NOUN
abrj-637	47	14	(	(	PUNCT
abrj-637	47	15	gbms	gbms	NOUN
abrj-637	47	16	)	)	PUNCT
abrj-637	47	17	and	and	CCONJ
abrj-637	47	18	neural	neural	ADJ
abrj-637	47	19	networks	network	NOUN
abrj-637	47	20	,	,	PUNCT
abrj-637	47	21	investigators	investigator	NOUN
abrj-637	47	22	have	have	AUX
abrj-637	47	23	already	already	ADV
abrj-637	47	24	shown	show	VERB
abrj-637	47	25	a	a	DET
abrj-637	47	26	high	high	ADJ
abrj-637	47	27	score	score	NOUN
abrj-637	47	28	in	in	ADP
abrj-637	47	29	classification	classification	NOUN
abrj-637	47	30	and	and	CCONJ
abrj-637	47	31	early	early	ADJ
abrj-637	47	32	detection	detection	NOUN
abrj-637	47	33	rates	rate	NOUN
abrj-637	47	34	(	(	PUNCT
abrj-637	47	35	galla	galla	PROPN
abrj-637	47	36	,	,	PUNCT
abrj-637	47	37	2023	2023	NUM
abrj-637	47	38	;	;	PUNCT
abrj-637	47	39	martínez	martínez	NOUN
abrj-637	47	40	,	,	PUNCT
abrj-637	47	41	2025	2025	NUM
abrj-637	47	42	)	)	PUNCT
abrj-637	47	43	.	.	PUNCT
abrj-637	48	1	craja	craja	PROPN
abrj-637	48	2	et	et	PROPN
abrj-637	48	3	al	al	PROPN
abrj-637	48	4	.	.	PROPN
abrj-637	48	5	(	(	PUNCT
abrj-637	48	6	2020	2020	NUM
abrj-637	48	7	)	)	PUNCT
abrj-637	48	8	emphasized	emphasize	VERB
abrj-637	48	9	the	the	DET
abrj-637	48	10	effectiveness	effectiveness	NOUN
abrj-637	48	11	of	of	ADP
abrj-637	48	12	deep	deep	ADJ
abrj-637	48	13	learning	learning	NOUN
abrj-637	48	14	models	model	NOUN
abrj-637	48	15	compared	compare	VERB
abrj-637	48	16	to	to	ADP
abrj-637	48	17	traditional	traditional	ADJ
abrj-637	48	18	ones	one	NOUN
abrj-637	48	19	in	in	ADP
abrj-637	48	20	identifying	identify	VERB
abrj-637	48	21	financial	financial	ADJ
abrj-637	48	22	statement	statement	NOUN
abrj-637	48	23	fraud	fraud	NOUN
abrj-637	48	24	especially	especially	ADV
abrj-637	48	25	when	when	SCONJ
abrj-637	48	26	datasets	dataset	NOUN
abrj-637	48	27	are	be	AUX
abrj-637	48	28	large	large	ADJ
abrj-637	48	29	and	and	CCONJ
abrj-637	48	30	the	the	DET
abrj-637	48	31	pattern	pattern	NOUN
abrj-637	48	32	is	be	AUX
abrj-637	48	33	non	non	ADJ
abrj-637	48	34	-	-	ADJ
abrj-637	48	35	linear	linear	ADJ
abrj-637	48	36	and	and	CCONJ
abrj-637	48	37	multidimensional	multidimensional	ADJ
abrj-637	48	38	.	.	PUNCT
abrj-637	49	1	in	in	ADP
abrj-637	49	2	a	a	DET
abrj-637	49	3	similar	similar	ADJ
abrj-637	49	4	way	way	NOUN
abrj-637	49	5	,	,	PUNCT
abrj-637	49	6	ippolito	ippolito	PROPN
abrj-637	49	7	and	and	CCONJ
abrj-637	49	8	lozano	lozano	PROPN
abrj-637	49	9	(	(	PUNCT
abrj-637	49	10	2020	2020	NUM
abrj-637	49	11	)	)	PUNCT
abrj-637	49	12	created	create	VERB
abrj-637	49	13	a	a	DET
abrj-637	49	14	model	model	NOUN
abrj-637	49	15	of	of	ADP
abrj-637	49	16	prediction	prediction	NOUN
abrj-637	49	17	of	of	ADP
abrj-637	49	18	tax	tax	NOUN
abrj-637	49	19	crimes	crime	NOUN
abrj-637	49	20	,	,	PUNCT
abrj-637	49	21	based	base	VERB
abrj-637	49	22	on	on	ADP
abrj-637	49	23	ml	ml	PROPN
abrj-637	49	24	techniques	technique	NOUN
abrj-637	49	25	,	,	PUNCT
abrj-637	49	26	which	which	PRON
abrj-637	49	27	demonstrated	demonstrate	VERB
abrj-637	49	28	an	an	DET
abrj-637	49	29	improvement	improvement	NOUN
abrj-637	49	30	in	in	ADP
abrj-637	49	31	the	the	DET
abrj-637	49	32	detection	detection	NOUN
abrj-637	49	33	accuracy	accuracy	NOUN
abrj-637	49	34	of	of	ADP
abrj-637	49	35	taxpayer	taxpayer	NOUN
abrj-637	49	36	behavior	behavior	NOUN
abrj-637	49	37	at	at	ADP
abrj-637	49	38	the	the	DET
abrj-637	49	39	municipal	municipal	ADJ
abrj-637	49	40	level	level	NOUN
abrj-637	49	41	because	because	SCONJ
abrj-637	49	42	of	of	ADP
abrj-637	49	43	the	the	DET
abrj-637	49	44	capture	capture	NOUN
abrj-637	49	45	of	of	ADP
abrj-637	49	46	hidden	hide	VERB
abrj-637	49	47	correlations	correlation	NOUN
abrj-637	49	48	of	of	ADP
abrj-637	49	49	taxation	taxation	NOUN
abrj-637	49	50	.	.	PUNCT
abrj-637	50	1	the	the	DET
abrj-637	50	2	developments	development	NOUN
abrj-637	50	3	are	be	AUX
abrj-637	50	4	a	a	DET
abrj-637	50	5	move	move	NOUN
abrj-637	50	6	towards	towards	ADP
abrj-637	50	7	probabilistic	probabilistic	ADJ
abrj-637	50	8	and	and	CCONJ
abrj-637	50	9	adaptive	adaptive	ADJ
abrj-637	50	10	risk	risk	NOUN
abrj-637	50	11	modeling	modeling	NOUN
abrj-637	50	12	,	,	PUNCT
abrj-637	50	13	no	no	ADV
abrj-637	50	14	longer	long	ADV
abrj-637	50	15	on	on	ADP
abrj-637	50	16	hard	hard	ADJ
abrj-637	50	17	-	-	PUNCT
abrj-637	50	18	and	and	CCONJ
abrj-637	50	19	-	-	PUNCT
abrj-637	50	20	fast	fast	ADJ
abrj-637	50	21	risk	risk	NOUN
abrj-637	50	22	thresholds	threshold	NOUN
abrj-637	50	23	but	but	CCONJ
abrj-637	50	24	toward	toward	ADP
abrj-637	50	25	a	a	DET
abrj-637	50	26	more	more	ADV
abrj-637	50	27	dynamic	dynamic	ADJ
abrj-637	50	28	scoring	scoring	NOUN
abrj-637	50	29	mechanism	mechanism	NOUN
abrj-637	50	30	.	.	PUNCT
abrj-637	51	1	supervised	supervised	ADJ
abrj-637	51	2	learning	learning	NOUN
abrj-637	51	3	methods	method	NOUN
abrj-637	51	4	are	be	AUX
abrj-637	51	5	based	base	VERB
abrj-637	51	6	on	on	ADP
abrj-637	51	7	the	the	DET
abrj-637	51	8	idea	idea	NOUN
abrj-637	51	9	that	that	SCONJ
abrj-637	51	10	an	an	DET
abrj-637	51	11	expert	expert	ADJ
abrj-637	51	12	coach	coach	NOUN
abrj-637	51	13	teaches	teach	VERB
abrj-637	51	14	the	the	DET
abrj-637	51	15	student.<|human|>2.3	student.<|human|>2.3	NUM
abrj-637	51	16	supervised	supervised	ADJ
abrj-637	51	17	learning	learning	NOUN
abrj-637	51	18	approaches	approach	NOUN
abrj-637	51	19	.	.	PUNCT
abrj-637	52	1	one	one	NUM
abrj-637	52	2	of	of	ADP
abrj-637	52	3	the	the	DET
abrj-637	52	4	most	most	ADV
abrj-637	52	5	common	common	ADJ
abrj-637	52	6	methods	method	NOUN
abrj-637	52	7	that	that	PRON
abrj-637	52	8	have	have	AUX
abrj-637	52	9	been	be	AUX
abrj-637	52	10	used	use	VERB
abrj-637	52	11	to	to	PART
abrj-637	52	12	detect	detect	VERB
abrj-637	52	13	tax	tax	NOUN
abrj-637	52	14	and	and	CCONJ
abrj-637	52	15	financial	financial	ADJ
abrj-637	52	16	fraud	fraud	NOUN
abrj-637	52	17	is	be	AUX
abrj-637	52	18	supervised	supervise	VERB
abrj-637	52	19	learning	learn	VERB
abrj-637	52	20	algorithms	algorithm	NOUN
abrj-637	52	21	.	.	PUNCT
abrj-637	53	1	the	the	DET
abrj-637	53	2	models	model	NOUN
abrj-637	53	3	are	be	AUX
abrj-637	53	4	based	base	VERB
abrj-637	53	5	on	on	ADP
abrj-637	53	6	labeled	label	VERB
abrj-637	53	7	data	datum	NOUN
abrj-637	53	8	,	,	PUNCT
abrj-637	53	9	so	so	SCONJ
abrj-637	53	10	every	every	DET
abrj-637	53	11	example	example	NOUN
abrj-637	53	12	of	of	ADP
abrj-637	53	13	financial	financial	ADJ
abrj-637	53	14	behavior	behavior	NOUN
abrj-637	53	15	(	(	PUNCT
abrj-637	53	16	fraudulent	fraudulent	ADJ
abrj-637	53	17	or	or	CCONJ
abrj-637	53	18	legitimate	legitimate	ADJ
abrj-637	53	19	)	)	PUNCT
abrj-637	53	20	is	be	AUX
abrj-637	53	21	known	know	VERB
abrj-637	53	22	and	and	CCONJ
abrj-637	53	23	the	the	DET
abrj-637	53	24	algorithm	algorithm	NOUN
abrj-637	53	25	can	can	AUX
abrj-637	53	26	learn	learn	VERB
abrj-637	53	27	discriminative	discriminative	NOUN
abrj-637	53	28	behavior	behavior	NOUN
abrj-637	53	29	.	.	PUNCT
abrj-637	54	1	random	random	ADJ
abrj-637	54	2	forests	forest	NOUN
abrj-637	54	3	,	,	PUNCT
abrj-637	54	4	decision	decision	NOUN
abrj-637	54	5	trees	tree	NOUN
abrj-637	54	6	,	,	PUNCT
abrj-637	54	7	and	and	CCONJ
abrj-637	54	8	support	support	VERB
abrj-637	54	9	vector	vector	NOUN
abrj-637	54	10	machines	machine	NOUN
abrj-637	54	11	(	(	PUNCT
abrj-637	54	12	svm	svm	PROPN
abrj-637	54	13	)	)	PUNCT
abrj-637	54	14	or	or	CCONJ
abrj-637	54	15	logistic	logistic	ADJ
abrj-637	54	16	regression	regression	NOUN
abrj-637	54	17	are	be	AUX
abrj-637	54	18	part	part	NOUN
abrj-637	54	19	of	of	ADP
abrj-637	54	20	the	the	DET
abrj-637	54	21	basics	basic	NOUN
abrj-637	54	22	of	of	ADP
abrj-637	54	23	fraud	fraud	NOUN
abrj-637	54	24	classification	classification	NOUN
abrj-637	54	25	works	work	NOUN
abrj-637	54	26	(	(	PUNCT
abrj-637	54	27	nguyen	nguyen	NOUN
abrj-637	54	28	,	,	PUNCT
abrj-637	54	29	2025	2025	NUM
abrj-637	54	30	;	;	PUNCT
abrj-637	54	31	acharya	acharya	NOUN
abrj-637	54	32	,	,	PUNCT
abrj-637	54	33	2025	2025	NUM
abrj-637	54	34	)	)	PUNCT
abrj-637	54	35	.	.	PUNCT
abrj-637	55	1	in	in	ADP
abrj-637	55	2	their	their	PRON
abrj-637	55	3	work	work	NOUN
abrj-637	55	4	,	,	PUNCT
abrj-637	55	5	zhou	zhou	PROPN
abrj-637	55	6	et	et	PROPN
abrj-637	55	7	al	al	PROPN
abrj-637	55	8	.	.	PROPN
abrj-637	56	1	(	(	PUNCT
abrj-637	56	2	2024	2024	NUM
abrj-637	56	3	)	)	PUNCT
abrj-637	56	4	proposed	propose	VERB
abrj-637	56	5	a	a	DET
abrj-637	56	6	soft	soft	ADJ
abrj-637	56	7	-	-	PUNCT
abrj-637	56	8	voting	voting	NOUN
abrj-637	56	9	ensemble	ensemble	ADJ
abrj-637	56	10	framework	framework	NOUN
abrj-637	56	11	,	,	PUNCT
abrj-637	56	12	which	which	PRON
abrj-637	56	13	is	be	AUX
abrj-637	56	14	a	a	DET
abrj-637	56	15	combination	combination	NOUN
abrj-637	56	16	of	of	ADP
abrj-637	56	17	multiple	multiple	ADJ
abrj-637	56	18	supervised	supervised	ADJ
abrj-637	56	19	predictors	predictor	NOUN
abrj-637	56	20	,	,	PUNCT
abrj-637	56	21	and	and	CCONJ
abrj-637	56	22	has	have	VERB
abrj-637	56	23	a	a	DET
abrj-637	56	24	better	well	ADJ
abrj-637	56	25	predictive	predictive	ADJ
abrj-637	56	26	accuracy	accuracy	NOUN
abrj-637	56	27	and	and	CCONJ
abrj-637	56	28	robustness	robustness	NOUN
abrj-637	56	29	.	.	PUNCT
abrj-637	57	1	in	in	ADP
abrj-637	57	2	the	the	DET
abrj-637	57	3	same	same	ADJ
abrj-637	57	4	fashion	fashion	NOUN
abrj-637	57	5	,	,	PUNCT
abrj-637	57	6	mehta	mehta	PROPN
abrj-637	57	7	et	et	PROPN
abrj-637	57	8	al	al	PROPN
abrj-637	57	9	.	.	PROPN
abrj-637	57	10	(	(	PUNCT
abrj-637	57	11	2022	2022	NUM
abrj-637	57	12	)	)	PUNCT
abrj-637	57	13	used	use	VERB
abrj-637	57	14	a	a	DET
abrj-637	57	15	bidirectional	bidirectional	ADJ
abrj-637	57	16	generative	generative	ADJ
abrj-637	57	17	adversarial	adversarial	ADJ
abrj-637	57	18	network	network	NOUN
abrj-637	57	19	(	(	PUNCT
abrj-637	57	20	gan	gan	PROPN
abrj-637	57	21	)	)	PUNCT
abrj-637	57	22	to	to	PART
abrj-637	57	23	generate	generate	VERB
abrj-637	57	24	artificial	artificial	ADJ
abrj-637	57	25	data	datum	NOUN
abrj-637	57	26	in	in	ADP
abrj-637	57	27	the	the	DET
abrj-637	57	28	field	field	NOUN
abrj-637	57	29	of	of	ADP
abrj-637	57	30	fraud	fraud	NOUN
abrj-637	57	31	in	in	ADP
abrj-637	57	32	taxation	taxation	NOUN
abrj-637	57	33	to	to	PART
abrj-637	57	34	enhance	enhance	VERB
abrj-637	57	35	the	the	DET
abrj-637	57	36	heterogeneity	heterogeneity	NOUN
abrj-637	57	37	and	and	CCONJ
abrj-637	57	38	applicability	applicability	NOUN
abrj-637	57	39	of	of	ADP
abrj-637	57	40	the	the	DET
abrj-637	57	41	training	training	NOUN
abrj-637	57	42	sample	sample	NOUN
abrj-637	57	43	.	.	PUNCT
abrj-637	58	1	these	these	DET
abrj-637	58	2	hybrid	hybrid	ADJ
abrj-637	58	3	approaches	approach	NOUN
abrj-637	58	4	are	be	AUX
abrj-637	58	5	a	a	DET
abrj-637	58	6	combination	combination	NOUN
abrj-637	58	7	of	of	ADP
abrj-637	58	8	predictive	predictive	ADJ
abrj-637	58	9	performance	performance	NOUN
abrj-637	58	10	and	and	CCONJ
abrj-637	58	11	increased	increase	VERB
abrj-637	58	12	sensitivity	sensitivity	NOUN
abrj-637	58	13	to	to	ADP
abrj-637	58	14	intricate	intricate	ADJ
abrj-637	58	15	and	and	CCONJ
abrj-637	58	16	obscure	obscure	ADJ
abrj-637	58	17	irregularities	irregularity	NOUN
abrj-637	58	18	in	in	ADP
abrj-637	58	19	tax	tax	NOUN
abrj-637	58	20	filings	filing	NOUN
abrj-637	58	21	.	.	PUNCT
abrj-637	59	1	although	although	SCONJ
abrj-637	59	2	supervised	supervised	ADJ
abrj-637	59	3	learning	learning	NOUN
abrj-637	59	4	relies	rely	VERB
abrj-637	59	5	on	on	ADP
abrj-637	59	6	the	the	DET
abrj-637	59	7	existence	existence	NOUN
abrj-637	59	8	of	of	ADP
abrj-637	59	9	labeled	label	VERB
abrj-637	59	10	data	datum	NOUN
abrj-637	59	11	,	,	PUNCT
abrj-637	59	12	acquiring	acquire	VERB
abrj-637	59	13	those	those	DET
abrj-637	59	14	datasets	dataset	NOUN
abrj-637	59	15	in	in	ADP
abrj-637	59	16	tax	tax	NOUN
abrj-637	59	17	fields	field	NOUN
abrj-637	59	18	is	be	AUX
abrj-637	59	19	frequently	frequently	ADV
abrj-637	59	20	difficult	difficult	ADJ
abrj-637	59	21	because	because	SCONJ
abrj-637	59	22	of	of	ADP
abrj-637	59	23	the	the	DET
abrj-637	59	24	confidentiality	confidentiality	NOUN
abrj-637	59	25	and	and	CCONJ
abrj-637	59	26	lack	lack	NOUN
abrj-637	59	27	of	of	ADP
abrj-637	59	28	data	datum	NOUN
abrj-637	59	29	(	(	PUNCT
abrj-637	59	30	tagbo	tagbo	NOUN
abrj-637	59	31	and	and	CCONJ
abrj-637	59	32	adekoya	adekoya	NOUN
abrj-637	59	33	,	,	PUNCT
abrj-637	59	34	2023	2023	NUM
abrj-637	59	35	;	;	PUNCT
abrj-637	59	36	wahyono	wahyono	PROPN
abrj-637	59	37	and	and	CCONJ
abrj-637	59	38	david	david	PROPN
abrj-637	59	39	,	,	PUNCT
abrj-637	59	40	2025	2025	NUM
abrj-637	59	41	)	)	PUNCT
abrj-637	59	42	.	.	PUNCT
abrj-637	60	1	this	this	PRON
abrj-637	60	2	in	in	ADP
abrj-637	60	3	turn	turn	NOUN
abrj-637	60	4	has	have	AUX
abrj-637	60	5	led	lead	VERB
abrj-637	60	6	to	to	ADP
abrj-637	60	7	the	the	DET
abrj-637	60	8	attention	attention	NOUN
abrj-637	60	9	of	of	ADP
abrj-637	60	10	unsupervised	unsupervised	ADJ
abrj-637	60	11	techniques	technique	NOUN
abrj-637	60	12	of	of	ADP
abrj-637	60	13	learning	learn	VERB
abrj-637	60	14	like	like	ADP
abrj-637	60	15	clustering	clustering	NOUN
abrj-637	60	16	and	and	CCONJ
abrj-637	60	17	anomaly	anomaly	NOUN
abrj-637	60	18	detection	detection	NOUN
abrj-637	60	19	.	.	PUNCT
abrj-637	61	1	methods	method	NOUN
abrj-637	61	2	such	such	ADJ
abrj-637	61	3	as	as	ADP
abrj-637	61	4	k	k	NOUN
abrj-637	61	5	-	-	PUNCT
abrj-637	61	6	means	mean	VERB
abrj-637	61	7	clustering	clustering	NOUN
abrj-637	61	8	,	,	PUNCT
abrj-637	61	9	isolation	isolation	NOUN
abrj-637	61	10	forests	forest	NOUN
abrj-637	61	11	,	,	PUNCT
abrj-637	61	12	and	and	CCONJ
abrj-637	61	13	autoencoders	autoencoder	NOUN
abrj-637	61	14	identify	identify	VERB
abrj-637	61	15	suspicious	suspicious	ADJ
abrj-637	61	16	transactions	transaction	NOUN
abrj-637	61	17	or	or	CCONJ
abrj-637	61	18	filing	file	VERB
abrj-637	61	19	patterns	pattern	NOUN
abrj-637	61	20	without	without	ADP
abrj-637	61	21	having	have	VERB
abrj-637	61	22	any	any	DET
abrj-637	61	23	idea	idea	NOUN
abrj-637	61	24	of	of	ADP
abrj-637	61	25	what	what	PRON
abrj-637	61	26	fraud	fraud	NOUN
abrj-637	61	27	is	be	AUX
abrj-637	61	28	(	(	PUNCT
abrj-637	61	29	wu	wu	PROPN
abrj-637	61	30	et	et	PROPN
abrj-637	61	31	al	al	PROPN
abrj-637	61	32	.	.	PROPN
abrj-637	61	33	,	,	PUNCT
abrj-637	61	34	2012	2012	NUM
abrj-637	61	35	;	;	PUNCT
abrj-637	61	36	zhou	zhou	PROPN
abrj-637	61	37	et	et	PROPN
abrj-637	61	38	al	al	PROPN
abrj-637	61	39	.	.	PROPN
abrj-637	61	40	,	,	PUNCT
abrj-637	61	41	2024	2024	NUM
abrj-637	61	42	)	)	PUNCT
abrj-637	61	43	.	.	PUNCT
abrj-637	62	1	hybrid	hybrid	ADJ
abrj-637	62	2	neural	neural	ADJ
abrj-637	62	3	network	network	NOUN
abrj-637	62	4	models	model	NOUN
abrj-637	62	5	were	be	AUX
abrj-637	62	6	investigated	investigate	VERB
abrj-637	62	7	by	by	ADP
abrj-637	62	8	choudhary	choudhary	PROPN
abrj-637	62	9	(	(	PUNCT
abrj-637	62	10	2025	2025	NUM
abrj-637	62	11	)	)	PUNCT
abrj-637	62	12	and	and	CCONJ
abrj-637	62	13	shujaaddeen	shujaaddeen	PROPN
abrj-637	62	14	et	et	PROPN
abrj-637	62	15	al	al	PROPN
abrj-637	62	16	.	.	PROPN
abrj-637	63	1	(	(	PUNCT
abrj-637	63	2	2024	2024	NUM
abrj-637	63	3	)	)	PUNCT
abrj-637	63	4	,	,	PUNCT
abrj-637	63	5	who	who	PRON
abrj-637	63	6	assumed	assume	VERB
abrj-637	63	7	that	that	SCONJ
abrj-637	63	8	supervised	supervised	ADJ
abrj-637	63	9	and	and	CCONJ
abrj-637	63	10	unsupervised	unsupervised	ADJ
abrj-637	63	11	architecture	architecture	NOUN
abrj-637	63	12	is	be	AUX
abrj-637	63	13	integrated	integrate	VERB
abrj-637	63	14	to	to	PART
abrj-637	63	15	improve	improve	VERB
abrj-637	63	16	the	the	DET
abrj-637	63	17	predictive	predictive	ADJ
abrj-637	63	18	capability	capability	NOUN
abrj-637	63	19	.	.	PUNCT
abrj-637	64	1	such	such	ADJ
abrj-637	64	2	systems	system	NOUN
abrj-637	64	3	ensure	ensure	VERB
abrj-637	64	4	that	that	SCONJ
abrj-637	64	5	tax	tax	NOUN
abrj-637	64	6	authorities	authority	NOUN
abrj-637	64	7	categorize	categorize	VERB
abrj-637	64	8	the	the	DET
abrj-637	64	9	suspicious	suspicious	ADJ
abrj-637	64	10	parties	party	NOUN
abrj-637	64	11	,	,	PUNCT
abrj-637	64	12	in	in	ADP
abrj-637	64	13	addition	addition	NOUN
abrj-637	64	14	to	to	ADP
abrj-637	64	15	identifying	identify	VERB
abrj-637	64	16	new	new	ADJ
abrj-637	64	17	trends	trend	NOUN
abrj-637	64	18	on	on	ADP
abrj-637	64	19	how	how	SCONJ
abrj-637	64	20	the	the	DET
abrj-637	64	21	fraudulent	fraudulent	ADJ
abrj-637	64	22	activities	activity	NOUN
abrj-637	64	23	can	can	AUX
abrj-637	64	24	be	be	AUX
abrj-637	64	25	carried	carry	VERB
abrj-637	64	26	out	out	ADP
abrj-637	64	27	.	.	PUNCT
abrj-637	65	1	one	one	NUM
abrj-637	65	2	example	example	NOUN
abrj-637	65	3	of	of	ADP
abrj-637	65	4	such	such	DET
abrj-637	65	5	a	a	DET
abrj-637	65	6	system	system	NOUN
abrj-637	65	7	is	be	AUX
abrj-637	65	8	autoencoders	autoencoder	NOUN
abrj-637	65	9	,	,	PUNCT
abrj-637	65	10	which	which	PRON
abrj-637	65	11	are	be	AUX
abrj-637	65	12	trained	train	VERB
abrj-637	65	13	to	to	PART
abrj-637	65	14	identify	identify	VERB
abrj-637	65	15	compressed	compress	VERB
abrj-637	65	16	instances	instance	NOUN
abrj-637	65	17	of	of	ADP
abrj-637	65	18	standard	standard	ADJ
abrj-637	65	19	financial	financial	ADJ
abrj-637	65	20	behavior	behavior	NOUN
abrj-637	65	21	;	;	PUNCT
abrj-637	65	22	exceptions	exception	NOUN
abrj-637	65	23	to	to	ADP
abrj-637	65	24	this	this	DET
abrj-637	65	25	behavior	behavior	NOUN
abrj-637	65	26	are	be	AUX
abrj-637	65	27	detected	detect	VERB
abrj-637	65	28	as	as	ADP
abrj-637	65	29	possible	possible	ADJ
abrj-637	65	30	anomalies	anomaly	NOUN
abrj-637	65	31	(	(	PUNCT
abrj-637	65	32	martínez	martínez	NOUN
abrj-637	65	33	,	,	PUNCT
abrj-637	65	34	2025	2025	NUM
abrj-637	65	35	)	)	PUNCT
abrj-637	65	36	.	.	PUNCT
abrj-637	66	1	asian	asian	ADJ
abrj-637	66	2	business	business	PROPN
abrj-637	66	3	research	research	NOUN
abrj-637	66	4	journal	journal	NOUN
abrj-637	66	5	,	,	PUNCT
abrj-637	66	6	2025	2025	NUM
abrj-637	66	7	,	,	PUNCT
abrj-637	66	8	10(11	10(11	NUM
abrj-637	66	9	):	):	PUNCT
abrj-637	66	10	1	1	NUM
abrj-637	66	11	-	-	SYM
abrj-637	66	12	13	13	NUM
abrj-637	66	13	3	3	NUM
abrj-637	66	14	©	©	PROPN
abrj-637	66	15	2025	2025	NUM
abrj-637	66	16	by	by	ADP
abrj-637	66	17	the	the	DET
abrj-637	66	18	authors	author	NOUN
abrj-637	66	19	;	;	PUNCT
abrj-637	66	20	licensee	licensee	NOUN
abrj-637	66	21	eastern	eastern	ADJ
abrj-637	66	22	centre	centre	NOUN
abrj-637	66	23	of	of	ADP
abrj-637	66	24	science	science	NOUN
abrj-637	66	25	and	and	CCONJ
abrj-637	66	26	education	education	NOUN
abrj-637	66	27	,	,	PUNCT
abrj-637	66	28	usa	usa	PROPN
abrj-637	66	29	recent	recent	ADJ
abrj-637	66	30	studies	study	NOUN
abrj-637	66	31	have	have	AUX
abrj-637	66	32	extended	extend	VERB
abrj-637	66	33	the	the	DET
abrj-637	66	34	definition	definition	NOUN
abrj-637	66	35	of	of	ADP
abrj-637	66	36	fraud	fraud	NOUN
abrj-637	66	37	detection	detection	NOUN
abrj-637	66	38	to	to	PART
abrj-637	66	39	include	include	VERB
abrj-637	66	40	natural	natural	ADJ
abrj-637	66	41	language	language	NOUN
abrj-637	66	42	processing	processing	NOUN
abrj-637	66	43	(	(	PUNCT
abrj-637	66	44	nlp	nlp	NOUN
abrj-637	66	45	)	)	PUNCT
abrj-637	66	46	methods	method	NOUN
abrj-637	66	47	as	as	SCONJ
abrj-637	66	48	opposed	oppose	VERB
abrj-637	66	49	to	to	ADP
abrj-637	66	50	numerical	numerical	ADJ
abrj-637	66	51	and	and	CCONJ
abrj-637	66	52	transactional	transactional	ADJ
abrj-637	66	53	data	datum	NOUN
abrj-637	66	54	.	.	PUNCT
abrj-637	67	1	linguistic	linguistic	ADJ
abrj-637	67	2	cues	cue	NOUN
abrj-637	67	3	of	of	ADP
abrj-637	67	4	deception	deception	NOUN
abrj-637	67	5	are	be	AUX
abrj-637	67	6	likely	likely	ADJ
abrj-637	67	7	to	to	PART
abrj-637	67	8	be	be	AUX
abrj-637	67	9	present	present	ADJ
abrj-637	67	10	in	in	ADP
abrj-637	67	11	textual	textual	ADJ
abrj-637	67	12	disclosures	disclosure	NOUN
abrj-637	67	13	in	in	ADP
abrj-637	67	14	corporate	corporate	ADJ
abrj-637	67	15	filings	filing	NOUN
abrj-637	67	16	,	,	PUNCT
abrj-637	67	17	management	management	NOUN
abrj-637	67	18	commentaries	commentary	NOUN
abrj-637	67	19	and	and	CCONJ
abrj-637	67	20	auditor	auditor	NOUN
abrj-637	67	21	statements	statement	NOUN
abrj-637	67	22	.	.	PUNCT
abrj-637	68	1	it	it	PRON
abrj-637	68	2	was	be	AUX
abrj-637	68	3	established	establish	VERB
abrj-637	68	4	by	by	ADP
abrj-637	68	5	zhang	zhang	PROPN
abrj-637	68	6	et	et	PROPN
abrj-637	68	7	al	al	PROPN
abrj-637	68	8	.	.	PROPN
abrj-637	69	1	(	(	PUNCT
abrj-637	69	2	2024	2024	NUM
abrj-637	69	3	)	)	PUNCT
abrj-637	69	4	that	that	PRON
abrj-637	69	5	readability	readability	NOUN
abrj-637	69	6	and	and	CCONJ
abrj-637	69	7	sentiment	sentiment	NOUN
abrj-637	69	8	characteristics	characteristic	NOUN
abrj-637	69	9	of	of	ADP
abrj-637	69	10	language	language	NOUN
abrj-637	69	11	use	use	NOUN
abrj-637	69	12	in	in	ADP
abrj-637	69	13	financial	financial	ADJ
abrj-637	69	14	documents	document	NOUN
abrj-637	69	15	can	can	AUX
abrj-637	69	16	be	be	AUX
abrj-637	69	17	regarded	regard	VERB
abrj-637	69	18	as	as	ADP
abrj-637	69	19	an	an	DET
abrj-637	69	20	indicator	indicator	NOUN
abrj-637	69	21	of	of	ADP
abrj-637	69	22	fraud	fraud	NOUN
abrj-637	69	23	.	.	PUNCT
abrj-637	70	1	the	the	DET
abrj-637	70	2	analysis	analysis	NOUN
abrj-637	70	3	of	of	ADP
abrj-637	70	4	their	their	PRON
abrj-637	70	5	study	study	NOUN
abrj-637	70	6	was	be	AUX
abrj-637	70	7	based	base	VERB
abrj-637	70	8	on	on	ADP
abrj-637	70	9	a	a	DET
abrj-637	70	10	combination	combination	NOUN
abrj-637	70	11	of	of	ADP
abrj-637	70	12	semantic	semantic	ADJ
abrj-637	70	13	and	and	CCONJ
abrj-637	70	14	syntactic	syntactic	ADJ
abrj-637	70	15	analysis	analysis	NOUN
abrj-637	70	16	with	with	ADP
abrj-637	70	17	ml	ml	NOUN
abrj-637	70	18	models	model	NOUN
abrj-637	70	19	to	to	PART
abrj-637	70	20	improve	improve	VERB
abrj-637	70	21	the	the	DET
abrj-637	70	22	accuracy	accuracy	NOUN
abrj-637	70	23	of	of	ADP
abrj-637	70	24	the	the	DET
abrj-637	70	25	classification	classification	NOUN
abrj-637	70	26	.	.	PUNCT
abrj-637	71	1	ji	ji	PROPN
abrj-637	71	2	et	et	PROPN
abrj-637	71	3	al	al	PROPN
abrj-637	71	4	.	.	PROPN
abrj-637	72	1	(	(	PUNCT
abrj-637	72	2	2024	2024	NUM
abrj-637	72	3	)	)	PUNCT
abrj-637	72	4	also	also	ADV
abrj-637	72	5	studied	study	VERB
abrj-637	72	6	the	the	DET
abrj-637	72	7	textual	textual	ADJ
abrj-637	72	8	characteristics	characteristic	NOUN
abrj-637	72	9	of	of	ADP
abrj-637	72	10	financial	financial	ADJ
abrj-637	72	11	anomalies	anomaly	NOUN
abrj-637	72	12	,	,	PUNCT
abrj-637	72	13	with	with	ADP
abrj-637	72	14	the	the	DET
abrj-637	72	15	authors	author	NOUN
abrj-637	72	16	discovering	discover	VERB
abrj-637	72	17	that	that	SCONJ
abrj-637	72	18	the	the	DET
abrj-637	72	19	use	use	NOUN
abrj-637	72	20	of	of	ADP
abrj-637	72	21	words	word	NOUN
abrj-637	72	22	,	,	PUNCT
abrj-637	72	23	tones	tone	NOUN
abrj-637	72	24	,	,	PUNCT
abrj-637	72	25	and	and	CCONJ
abrj-637	72	26	document	document	NOUN
abrj-637	72	27	complexity	complexity	NOUN
abrj-637	72	28	can	can	AUX
abrj-637	72	29	indicate	indicate	VERB
abrj-637	72	30	inconsistency	inconsistency	NOUN
abrj-637	72	31	in	in	ADP
abrj-637	72	32	the	the	DET
abrj-637	72	33	company	company	NOUN
abrj-637	72	34	descriptions	description	NOUN
abrj-637	72	35	.	.	PUNCT
abrj-637	73	1	the	the	DET
abrj-637	73	2	introduction	introduction	NOUN
abrj-637	73	3	of	of	ADP
abrj-637	73	4	nlp	nlp	NOUN
abrj-637	73	5	to	to	PART
abrj-637	73	6	ml	ml	VERB
abrj-637	73	7	pipelines	pipeline	NOUN
abrj-637	73	8	makes	make	VERB
abrj-637	73	9	it	it	PRON
abrj-637	73	10	possible	possible	ADJ
abrj-637	73	11	to	to	PART
abrj-637	73	12	carry	carry	VERB
abrj-637	73	13	out	out	ADP
abrj-637	73	14	a	a	DET
abrj-637	73	15	comprehensive	comprehensive	ADJ
abrj-637	73	16	approach	approach	NOUN
abrj-637	73	17	to	to	ADP
abrj-637	73	18	fraud	fraud	NOUN
abrj-637	73	19	detection	detection	NOUN
abrj-637	73	20	,	,	PUNCT
abrj-637	73	21	relying	rely	VERB
abrj-637	73	22	on	on	ADP
abrj-637	73	23	quantitative	quantitative	ADJ
abrj-637	73	24	and	and	CCONJ
abrj-637	73	25	qualitative	qualitative	ADJ
abrj-637	73	26	indicator	indicator	NOUN
abrj-637	73	27	(	(	PUNCT
abrj-637	73	28	acharya	acharya	PROPN
abrj-637	73	29	,	,	PUNCT
abrj-637	73	30	2025	2025	NUM
abrj-637	73	31	;	;	PUNCT
abrj-637	73	32	zhang	zhang	PROPN
abrj-637	73	33	et	et	PROPN
abrj-637	73	34	al	al	PROPN
abrj-637	73	35	.	.	PROPN
abrj-637	73	36	,	,	PUNCT
abrj-637	73	37	2024	2024	NUM
abrj-637	73	38	)	)	PUNCT
abrj-637	73	39	.	.	PUNCT
abrj-637	74	1	the	the	DET
abrj-637	74	2	use	use	NOUN
abrj-637	74	3	of	of	ADP
abrj-637	74	4	ensemble	ensemble	ADJ
abrj-637	74	5	learning	learning	NOUN
abrj-637	74	6	to	to	PART
abrj-637	74	7	unite	unite	VERB
abrj-637	74	8	several	several	ADJ
abrj-637	74	9	models	model	NOUN
abrj-637	74	10	to	to	PART
abrj-637	74	11	enhance	enhance	VERB
abrj-637	74	12	reliability	reliability	NOUN
abrj-637	74	13	has	have	AUX
abrj-637	74	14	taken	take	VERB
abrj-637	74	15	center	center	ADJ
abrj-637	74	16	stage	stage	NOUN
abrj-637	74	17	in	in	ADP
abrj-637	74	18	literature	literature	NOUN
abrj-637	74	19	as	as	ADP
abrj-637	74	20	a	a	DET
abrj-637	74	21	result	result	NOUN
abrj-637	74	22	of	of	ADP
abrj-637	74	23	its	its	PRON
abrj-637	74	24	capabilities	capability	NOUN
abrj-637	74	25	.	.	PUNCT
abrj-637	75	1	ensemble	ensemble	ADJ
abrj-637	75	2	-	-	PUNCT
abrj-637	75	3	based	base	VERB
abrj-637	75	4	models	model	NOUN
abrj-637	75	5	,	,	PUNCT
abrj-637	75	6	like	like	ADP
abrj-637	75	7	random	random	ADJ
abrj-637	75	8	forests	forest	NOUN
abrj-637	75	9	and	and	CCONJ
abrj-637	75	10	xgboost	xgboost	NOUN
abrj-637	75	11	are	be	AUX
abrj-637	75	12	also	also	ADV
abrj-637	75	13	built	build	VERB
abrj-637	75	14	on	on	ADP
abrj-637	75	15	the	the	DET
abrj-637	75	16	same	same	ADJ
abrj-637	75	17	principle	principle	NOUN
abrj-637	75	18	but	but	CCONJ
abrj-637	75	19	scholars	scholar	NOUN
abrj-637	75	20	have	have	AUX
abrj-637	75	21	developed	develop	VERB
abrj-637	75	22	other	other	ADJ
abrj-637	75	23	methods	method	NOUN
abrj-637	75	24	like	like	ADP
abrj-637	75	25	stacking	stack	VERB
abrj-637	75	26	and	and	CCONJ
abrj-637	75	27	voting	voting	NOUN
abrj-637	75	28	ensembles	ensemble	NOUN
abrj-637	75	29	that	that	PRON
abrj-637	75	30	combine	combine	VERB
abrj-637	75	31	different	different	ADJ
abrj-637	75	32	classifiers	classifier	NOUN
abrj-637	75	33	(	(	PUNCT
abrj-637	75	34	zhou	zhou	X
abrj-637	75	35	et	et	PROPN
abrj-637	75	36	al	al	PROPN
abrj-637	75	37	.	.	PROPN
abrj-637	75	38	,	,	PUNCT
abrj-637	75	39	2024	2024	NUM
abrj-637	75	40	;	;	PUNCT
abrj-637	76	1	zhang	zhang	PROPN
abrj-637	76	2	et	et	PROPN
abrj-637	76	3	al	al	PROPN
abrj-637	76	4	.	.	PROPN
abrj-637	76	5	,	,	PUNCT
abrj-637	76	6	2025	2025	NUM
abrj-637	76	7	)	)	PUNCT
abrj-637	76	8	.	.	PUNCT
abrj-637	77	1	as	as	ADP
abrj-637	77	2	a	a	DET
abrj-637	77	3	way	way	NOUN
abrj-637	77	4	to	to	PART
abrj-637	77	5	detect	detect	VERB
abrj-637	77	6	tax	tax	NOUN
abrj-637	77	7	fraud	fraud	NOUN
abrj-637	77	8	more	more	ADV
abrj-637	77	9	accurately	accurately	ADV
abrj-637	77	10	and	and	CCONJ
abrj-637	77	11	with	with	ADP
abrj-637	77	12	lower	low	ADJ
abrj-637	77	13	false	false	ADJ
abrj-637	77	14	-	-	PUNCT
abrj-637	77	15	positive	positive	ADJ
abrj-637	77	16	rates	rate	NOUN
abrj-637	77	17	,	,	PUNCT
abrj-637	77	18	zhou	zhou	PROPN
abrj-637	77	19	et	et	PROPN
abrj-637	77	20	al	al	PROPN
abrj-637	77	21	.	.	PROPN
abrj-637	77	22	(	(	PUNCT
abrj-637	77	23	2024	2024	NUM
abrj-637	77	24	)	)	PUNCT
abrj-637	77	25	made	make	VERB
abrj-637	77	26	a	a	DET
abrj-637	77	27	proposal	proposal	NOUN
abrj-637	77	28	of	of	ADP
abrj-637	77	29	a	a	DET
abrj-637	77	30	soft	soft	ADJ
abrj-637	77	31	-	-	PUNCT
abrj-637	77	32	voting	voting	NOUN
abrj-637	77	33	ensemble	ensemble	ADJ
abrj-637	77	34	,	,	PUNCT
abrj-637	77	35	which	which	PRON
abrj-637	77	36	involves	involve	VERB
abrj-637	77	37	encoder	encoder	NOUN
abrj-637	77	38	extraction	extraction	NOUN
abrj-637	77	39	methods	method	NOUN
abrj-637	77	40	to	to	PART
abrj-637	77	41	detect	detect	VERB
abrj-637	77	42	tax	tax	NOUN
abrj-637	77	43	fraud	fraud	NOUN
abrj-637	77	44	.	.	PUNCT
abrj-637	78	1	equally	equally	ADV
abrj-637	78	2	,	,	PUNCT
abrj-637	78	3	zhang	zhang	PROPN
abrj-637	78	4	et	et	PROPN
abrj-637	78	5	al	al	PROPN
abrj-637	78	6	.	.	PROPN
abrj-637	78	7	(	(	PUNCT
abrj-637	78	8	2025	2025	NUM
abrj-637	78	9	)	)	PUNCT
abrj-637	78	10	showed	show	VERB
abrj-637	78	11	that	that	SCONJ
abrj-637	78	12	stack	stack	NOUN
abrj-637	78	13	learning	learn	VERB
abrj-637	78	14	enhanced	enhance	VERB
abrj-637	78	15	detection	detection	NOUN
abrj-637	78	16	of	of	ADP
abrj-637	78	17	fraud	fraud	NOUN
abrj-637	78	18	in	in	ADP
abrj-637	78	19	financial	financial	ADJ
abrj-637	78	20	markets	market	NOUN
abrj-637	78	21	meaning	mean	VERB
abrj-637	78	22	that	that	SCONJ
abrj-637	78	23	it	it	PRON
abrj-637	78	24	could	could	AUX
abrj-637	78	25	be	be	AUX
abrj-637	78	26	used	use	VERB
abrj-637	78	27	in	in	ADP
abrj-637	78	28	corporate	corporate	ADJ
abrj-637	78	29	tax	tax	NOUN
abrj-637	78	30	analysis	analysis	NOUN
abrj-637	78	31	.	.	PUNCT
abrj-637	79	1	the	the	DET
abrj-637	79	2	mix	mix	NOUN
abrj-637	79	3	of	of	ADP
abrj-637	79	4	the	the	DET
abrj-637	79	5	linear	linear	ADJ
abrj-637	79	6	and	and	CCONJ
abrj-637	79	7	non	non	ADJ
abrj-637	79	8	-	-	ADJ
abrj-637	79	9	linear	linear	ADJ
abrj-637	79	10	learners	learner	NOUN
abrj-637	79	11	will	will	AUX
abrj-637	79	12	allow	allow	VERB
abrj-637	79	13	the	the	DET
abrj-637	79	14	system	system	NOUN
abrj-637	79	15	to	to	PART
abrj-637	79	16	seize	seize	VERB
abrj-637	79	17	macro	macro	PROPN
abrj-637	79	18	and	and	CCONJ
abrj-637	79	19	micro	micro	ADJ
abrj-637	79	20	-	-	NOUN
abrj-637	79	21	level	level	ADJ
abrj-637	79	22	fraud	fraud	NOUN
abrj-637	79	23	indicators	indicator	NOUN
abrj-637	79	24	,	,	PUNCT
abrj-637	79	25	which	which	PRON
abrj-637	79	26	will	will	AUX
abrj-637	79	27	be	be	AUX
abrj-637	79	28	more	more	ADV
abrj-637	79	29	interpretable	interpretable	ADJ
abrj-637	79	30	and	and	CCONJ
abrj-637	79	31	generalizable	generalizable	ADJ
abrj-637	79	32	.	.	PUNCT
abrj-637	80	1	although	although	SCONJ
abrj-637	80	2	there	there	PRON
abrj-637	80	3	is	be	VERB
abrj-637	80	4	increased	increase	VERB
abrj-637	80	5	technological	technological	ADJ
abrj-637	80	6	advances	advance	NOUN
abrj-637	80	7	,	,	PUNCT
abrj-637	80	8	issues	issue	NOUN
abrj-637	80	9	of	of	ADP
abrj-637	80	10	data	datum	NOUN
abrj-637	80	11	quality	quality	NOUN
abrj-637	80	12	,	,	PUNCT
abrj-637	80	13	bias	bias	NOUN
abrj-637	80	14	,	,	PUNCT
abrj-637	80	15	and	and	CCONJ
abrj-637	80	16	explainability	explainability	NOUN
abrj-637	80	17	continue	continue	VERB
abrj-637	80	18	to	to	PART
abrj-637	80	19	exist	exist	VERB
abrj-637	80	20	.	.	PUNCT
abrj-637	81	1	tax	tax	NOUN
abrj-637	81	2	datasets	dataset	NOUN
abrj-637	81	3	are	be	AUX
abrj-637	81	4	usually	usually	ADV
abrj-637	81	5	imbalanced	imbalanced	ADJ
abrj-637	81	6	(	(	PUNCT
abrj-637	81	7	there	there	PRON
abrj-637	81	8	are	be	VERB
abrj-637	81	9	few	few	ADJ
abrj-637	81	10	fraudulent	fraudulent	ADJ
abrj-637	81	11	cases	case	NOUN
abrj-637	81	12	compared	compare	VERB
abrj-637	81	13	to	to	ADP
abrj-637	81	14	total	total	ADJ
abrj-637	81	15	filings	filing	NOUN
abrj-637	81	16	)	)	PUNCT
abrj-637	81	17	,	,	PUNCT
abrj-637	81	18	resulting	result	VERB
abrj-637	81	19	in	in	ADP
abrj-637	81	20	bias	bias	NOUN
abrj-637	81	21	in	in	ADP
abrj-637	81	22	the	the	DET
abrj-637	81	23	learning	learning	NOUN
abrj-637	81	24	performance	performance	NOUN
abrj-637	81	25	unless	unless	SCONJ
abrj-637	81	26	it	it	PRON
abrj-637	81	27	is	be	AUX
abrj-637	81	28	managed	manage	VERB
abrj-637	81	29	correctly	correctly	ADV
abrj-637	81	30	(	(	PUNCT
abrj-637	81	31	tagbo	tagbo	NOUN
abrj-637	81	32	&	&	CCONJ
abrj-637	81	33	adekoya	adekoya	PROPN
abrj-637	81	34	,	,	PUNCT
abrj-637	81	35	2023	2023	NUM
abrj-637	81	36	;	;	PUNCT
abrj-637	81	37	wahyono	wahyono	PROPN
abrj-637	81	38	and	and	CCONJ
abrj-637	81	39	david	david	PROPN
abrj-637	81	40	,	,	PUNCT
abrj-637	81	41	2025	2025	NUM
abrj-637	81	42	)	)	PUNCT
abrj-637	81	43	.	.	PUNCT
abrj-637	82	1	further	far	ADV
abrj-637	82	2	,	,	PUNCT
abrj-637	82	3	the	the	DET
abrj-637	82	4	lack	lack	NOUN
abrj-637	82	5	of	of	ADP
abrj-637	82	6	accountability	accountability	NOUN
abrj-637	82	7	and	and	CCONJ
abrj-637	82	8	transparency	transparency	NOUN
abrj-637	82	9	is	be	AUX
abrj-637	82	10	of	of	ADP
abrj-637	82	11	concern	concern	NOUN
abrj-637	82	12	since	since	SCONJ
abrj-637	82	13	some	some	DET
abrj-637	82	14	ml	ml	NOUN
abrj-637	82	15	algorithms	algorithm	NOUN
abrj-637	82	16	,	,	PUNCT
abrj-637	82	17	especially	especially	ADV
abrj-637	82	18	deep	deep	ADJ
abrj-637	82	19	learning	learning	NOUN
abrj-637	82	20	models	model	NOUN
abrj-637	82	21	,	,	PUNCT
abrj-637	82	22	are	be	AUX
abrj-637	82	23	opaque	opaque	ADJ
abrj-637	82	24	by	by	ADP
abrj-637	82	25	nature	nature	NOUN
abrj-637	82	26	(	(	PUNCT
abrj-637	82	27	acharya	acharya	PROPN
abrj-637	82	28	,	,	PUNCT
abrj-637	82	29	2025	2025	NUM
abrj-637	82	30	)	)	PUNCT
abrj-637	82	31	.	.	PUNCT
abrj-637	83	1	the	the	DET
abrj-637	83	2	use	use	NOUN
abrj-637	83	3	of	of	ADP
abrj-637	83	4	explainable	explainable	ADJ
abrj-637	83	5	artificial	artificial	ADJ
abrj-637	83	6	intelligence	intelligence	NOUN
abrj-637	83	7	(	(	PUNCT
abrj-637	83	8	xai	xai	PROPN
abrj-637	83	9	)	)	PUNCT
abrj-637	83	10	methods	method	NOUN
abrj-637	83	11	,	,	PUNCT
abrj-637	83	12	including	include	VERB
abrj-637	83	13	shap	shap	NOUN
abrj-637	83	14	(	(	PUNCT
abrj-637	83	15	shapley	shapley	ADJ
abrj-637	83	16	additive	additive	ADJ
abrj-637	83	17	explanations	explanation	NOUN
abrj-637	83	18	)	)	PUNCT
abrj-637	83	19	and	and	CCONJ
abrj-637	83	20	lime	lime	NOUN
abrj-637	83	21	(	(	PUNCT
abrj-637	83	22	local	local	ADJ
abrj-637	83	23	interpretable	interpretable	ADJ
abrj-637	83	24	model	model	ADJ
abrj-637	83	25	-	-	ADJ
abrj-637	83	26	agnostic	agnostic	ADJ
abrj-637	83	27	explanations	explanation	NOUN
abrj-637	83	28	)	)	PUNCT
abrj-637	83	29	,	,	PUNCT
abrj-637	83	30	is	be	AUX
abrj-637	83	31	becoming	become	VERB
abrj-637	83	32	more	more	ADV
abrj-637	83	33	widespread	widespread	ADJ
abrj-637	83	34	to	to	PART
abrj-637	83	35	explain	explain	VERB
abrj-637	83	36	the	the	DET
abrj-637	83	37	predictions	prediction	NOUN
abrj-637	83	38	of	of	ADP
abrj-637	83	39	the	the	DET
abrj-637	83	40	model	model	NOUN
abrj-637	83	41	,	,	PUNCT
abrj-637	83	42	so	so	SCONJ
abrj-637	83	43	that	that	SCONJ
abrj-637	83	44	they	they	PRON
abrj-637	83	45	can	can	AUX
abrj-637	83	46	be	be	AUX
abrj-637	83	47	regulated	regulate	VERB
abrj-637	83	48	and	and	CCONJ
abrj-637	83	49	trusted	trust	VERB
abrj-637	83	50	by	by	ADP
abrj-637	83	51	the	the	DET
abrj-637	83	52	auditors	auditor	NOUN
abrj-637	83	53	(	(	PUNCT
abrj-637	83	54	zhou	zhou	X
abrj-637	83	55	et	et	PROPN
abrj-637	83	56	al	al	PROPN
abrj-637	83	57	.	.	PROPN
abrj-637	83	58	,	,	PUNCT
abrj-637	83	59	2024	2024	NUM
abrj-637	83	60	)	)	PUNCT
abrj-637	83	61	.	.	PUNCT
abrj-637	84	1	there	there	PRON
abrj-637	84	2	are	be	VERB
abrj-637	84	3	other	other	ADJ
abrj-637	84	4	ethical	ethical	ADJ
abrj-637	84	5	issues	issue	NOUN
abrj-637	84	6	,	,	PUNCT
abrj-637	84	7	such	such	ADJ
abrj-637	84	8	as	as	ADP
abrj-637	84	9	data	data	NOUN
abrj-637	84	10	privacy	privacy	NOUN
abrj-637	84	11	,	,	PUNCT
abrj-637	84	12	equity	equity	NOUN
abrj-637	84	13	,	,	PUNCT
abrj-637	84	14	and	and	CCONJ
abrj-637	84	15	over	over	ADP
abrj-637	84	16	-	-	PUNCT
abrj-637	84	17	automation	automation	NOUN
abrj-637	84	18	,	,	PUNCT
abrj-637	84	19	which	which	PRON
abrj-637	84	20	have	have	VERB
abrj-637	84	21	to	to	PART
abrj-637	84	22	be	be	AUX
abrj-637	84	23	taken	take	VERB
abrj-637	84	24	into	into	ADP
abrj-637	84	25	account	account	NOUN
abrj-637	84	26	prior	prior	ADV
abrj-637	84	27	to	to	ADP
abrj-637	84	28	adopting	adopt	VERB
abrj-637	84	29	ml	ml	ADP
abrj-637	84	30	systems	system	NOUN
abrj-637	84	31	into	into	ADP
abrj-637	84	32	tax	tax	NOUN
abrj-637	84	33	governance	governance	NOUN
abrj-637	84	34	systems	system	NOUN
abrj-637	84	35	(	(	PUNCT
abrj-637	84	36	breslin	breslin	PROPN
abrj-637	84	37	,	,	PUNCT
abrj-637	84	38	2021	2021	NUM
abrj-637	84	39	)	)	PUNCT
abrj-637	84	40	.	.	PUNCT
abrj-637	85	1	literature	literature	NOUN
abrj-637	85	2	as	as	ADP
abrj-637	85	3	a	a	DET
abrj-637	85	4	whole	whole	NOUN
abrj-637	85	5	is	be	AUX
abrj-637	85	6	in	in	ADP
abrj-637	85	7	favor	favor	NOUN
abrj-637	85	8	of	of	ADP
abrj-637	85	9	the	the	DET
abrj-637	85	10	transformative	transformative	ADJ
abrj-637	85	11	value	value	NOUN
abrj-637	85	12	of	of	ADP
abrj-637	85	13	machine	machine	NOUN
abrj-637	85	14	learning	learning	NOUN
abrj-637	85	15	in	in	ADP
abrj-637	85	16	the	the	DET
abrj-637	85	17	detection	detection	NOUN
abrj-637	85	18	of	of	ADP
abrj-637	85	19	tax	tax	NOUN
abrj-637	85	20	fraud	fraud	NOUN
abrj-637	85	21	and	and	CCONJ
abrj-637	85	22	corporate	corporate	ADJ
abrj-637	85	23	risk	risk	NOUN
abrj-637	85	24	assessment	assessment	NOUN
abrj-637	85	25	.	.	PUNCT
abrj-637	86	1	hybrid	hybrid	ADJ
abrj-637	86	2	systems	system	NOUN
abrj-637	86	3	,	,	PUNCT
abrj-637	86	4	ensemble	ensemble	ADJ
abrj-637	86	5	and	and	CCONJ
abrj-637	86	6	deep	deep	ADJ
abrj-637	86	7	learning	learning	NOUN
abrj-637	86	8	systems	system	NOUN
abrj-637	86	9	are	be	AUX
abrj-637	86	10	slowly	slowly	ADV
abrj-637	86	11	replacing	replace	VERB
abrj-637	86	12	tablebased	tablebase	VERB
abrj-637	86	13	systems	system	NOUN
abrj-637	86	14	and	and	CCONJ
abrj-637	86	15	shallow	shallow	ADJ
abrj-637	86	16	classifiers	classifier	NOUN
abrj-637	86	17	in	in	ADP
abrj-637	86	18	response	response	NOUN
abrj-637	86	19	to	to	ADP
abrj-637	86	20	structured	structured	ADJ
abrj-637	86	21	and	and	CCONJ
abrj-637	86	22	unstructured	unstructured	ADJ
abrj-637	86	23	data	datum	NOUN
abrj-637	86	24	(	(	PUNCT
abrj-637	86	25	acharya	acharya	NOUN
abrj-637	86	26	,	,	PUNCT
abrj-637	86	27	2025	2025	NUM
abrj-637	86	28	;	;	PUNCT
abrj-637	86	29	zhang	zhang	PROPN
abrj-637	86	30	et	et	PROPN
abrj-637	86	31	al	al	PROPN
abrj-637	86	32	.	.	PROPN
abrj-637	86	33	,	,	PUNCT
abrj-637	86	34	2025	2025	NUM
abrj-637	86	35	)	)	PUNCT
abrj-637	86	36	.	.	PUNCT
abrj-637	87	1	research	research	NOUN
abrj-637	87	2	always	always	ADV
abrj-637	87	3	records	record	VERB
abrj-637	87	4	an	an	DET
abrj-637	87	5	increase	increase	NOUN
abrj-637	87	6	in	in	ADP
abrj-637	87	7	detection	detection	NOUN
abrj-637	87	8	accuracy	accuracy	NOUN
abrj-637	87	9	,	,	PUNCT
abrj-637	87	10	scalability	scalability	NOUN
abrj-637	87	11	,	,	PUNCT
abrj-637	87	12	and	and	CCONJ
abrj-637	87	13	adaptability	adaptability	NOUN
abrj-637	87	14	in	in	ADP
abrj-637	87	15	the	the	DET
abrj-637	87	16	case	case	NOUN
abrj-637	87	17	of	of	ADP
abrj-637	87	18	applying	apply	VERB
abrj-637	87	19	ml	ml	NOUN
abrj-637	87	20	models	model	NOUN
abrj-637	87	21	to	to	ADP
abrj-637	87	22	tax	tax	NOUN
abrj-637	87	23	and	and	CCONJ
abrj-637	87	24	corporate	corporate	ADJ
abrj-637	87	25	data	datum	NOUN
abrj-637	87	26	.	.	PUNCT
abrj-637	88	1	however	however	ADV
abrj-637	88	2	,	,	PUNCT
abrj-637	88	3	gaps	gap	NOUN
abrj-637	88	4	remain	remain	VERB
abrj-637	88	5	.	.	PUNCT
abrj-637	89	1	the	the	DET
abrj-637	89	2	existing	exist	VERB
abrj-637	89	3	studies	study	NOUN
abrj-637	89	4	are	be	AUX
abrj-637	89	5	characterized	characterize	VERB
abrj-637	89	6	by	by	ADP
abrj-637	89	7	the	the	DET
abrj-637	89	8	focus	focus	NOUN
abrj-637	89	9	on	on	ADP
abrj-637	89	10	technical	technical	ADJ
abrj-637	89	11	performance	performance	NOUN
abrj-637	89	12	that	that	PRON
abrj-637	89	13	frequently	frequently	ADV
abrj-637	89	14	ignores	ignore	VERB
abrj-637	89	15	operational	operational	ADJ
abrj-637	89	16	issues	issue	NOUN
abrj-637	89	17	,	,	PUNCT
abrj-637	89	18	including	include	VERB
abrj-637	89	19	data	data	NOUN
abrj-637	89	20	access	access	NOUN
abrj-637	89	21	,	,	PUNCT
abrj-637	89	22	interpretability	interpretability	NOUN
abrj-637	89	23	,	,	PUNCT
abrj-637	89	24	and	and	CCONJ
abrj-637	89	25	operational	operational	ADJ
abrj-637	89	26	control	control	NOUN
abrj-637	89	27	.	.	PUNCT
abrj-637	90	1	there	there	PRON
abrj-637	90	2	are	be	VERB
abrj-637	90	3	no	no	DET
abrj-637	90	4	studies	study	NOUN
abrj-637	90	5	that	that	PRON
abrj-637	90	6	touch	touch	VERB
abrj-637	90	7	on	on	ADP
abrj-637	90	8	the	the	DET
abrj-637	90	9	incorporation	incorporation	NOUN
abrj-637	90	10	of	of	ADP
abrj-637	90	11	ml	ml	NOUN
abrj-637	90	12	systems	system	NOUN
abrj-637	90	13	in	in	ADP
abrj-637	90	14	practical	practical	ADJ
abrj-637	90	15	tax	tax	NOUN
abrj-637	90	16	audit	audit	NOUN
abrj-637	90	17	process	process	NOUN
abrj-637	90	18	or	or	CCONJ
abrj-637	90	19	their	their	PRON
abrj-637	90	20	suitability	suitability	NOUN
abrj-637	90	21	with	with	ADP
abrj-637	90	22	regulatory	regulatory	ADJ
abrj-637	90	23	systems	system	NOUN
abrj-637	90	24	.	.	PUNCT
abrj-637	91	1	responsible	responsible	ADJ
abrj-637	91	2	and	and	CCONJ
abrj-637	91	3	transparent	transparent	ADJ
abrj-637	91	4	deployment	deployment	NOUN
abrj-637	91	5	of	of	ADP
abrj-637	91	6	ml	ml	PRON
abrj-637	91	7	then	then	ADV
abrj-637	91	8	needs	need	VERB
abrj-637	91	9	to	to	PART
abrj-637	91	10	be	be	AUX
abrj-637	91	11	the	the	DET
abrj-637	91	12	focus	focus	NOUN
abrj-637	91	13	of	of	ADP
abrj-637	91	14	future	future	ADJ
abrj-637	91	15	research	research	NOUN
abrj-637	91	16	as	as	SCONJ
abrj-637	91	17	it	it	PRON
abrj-637	91	18	can	can	AUX
abrj-637	91	19	be	be	AUX
abrj-637	91	20	done	do	VERB
abrj-637	91	21	by	by	ADP
abrj-637	91	22	developing	develop	VERB
abrj-637	91	23	explainable	explainable	ADJ
abrj-637	91	24	,	,	PUNCT
abrj-637	91	25	auditor	auditor	NOUN
abrj-637	91	26	-	-	PUNCT
abrj-637	91	27	assistive	assistive	NOUN
abrj-637	91	28	,	,	PUNCT
abrj-637	91	29	and	and	CCONJ
abrj-637	91	30	privacy	privacy	NOUN
abrj-637	91	31	-	-	PUNCT
abrj-637	91	32	preserving	preserve	VERB
abrj-637	91	33	architectures	architecture	NOUN
abrj-637	91	34	(	(	PUNCT
abrj-637	91	35	wahyono	wahyono	PROPN
abrj-637	91	36	and	and	CCONJ
abrj-637	91	37	david	david	PROPN
abrj-637	91	38	,	,	PUNCT
abrj-637	91	39	2025	2025	NUM
abrj-637	91	40	;	;	PUNCT
abrj-637	91	41	zhou	zhou	PROPN
abrj-637	91	42	et	et	PROPN
abrj-637	91	43	al	al	PROPN
abrj-637	91	44	.	.	PROPN
abrj-637	91	45	,	,	PUNCT
abrj-637	91	46	2024	2024	NUM
abrj-637	91	47	)	)	PUNCT
abrj-637	91	48	.	.	PUNCT
abrj-637	92	1	table	table	NOUN
abrj-637	93	1	1	1	NUM
abrj-637	93	2	.	.	PUNCT
abrj-637	94	1	summary	summary	NOUN
abrj-637	94	2	of	of	ADP
abrj-637	94	3	key	key	ADJ
abrj-637	94	4	studies	study	NOUN
abrj-637	94	5	on	on	ADP
abrj-637	94	6	machine	machine	NOUN
abrj-637	94	7	learning	learn	VERB
abrj-637	94	8	for	for	ADP
abrj-637	94	9	tax	tax	NOUN
abrj-637	94	10	fraud	fraud	NOUN
abrj-637	94	11	detection	detection	NOUN
abrj-637	94	12	.	.	PUNCT
abrj-637	95	1	author(s	author(s	NOUN
abrj-637	95	2	)	)	PUNCT
abrj-637	95	3	&	&	CCONJ
abrj-637	95	4	year	year	NOUN
abrj-637	95	5	focus	focus	NOUN
abrj-637	95	6	area	area	NOUN
abrj-637	95	7	method	method	VERB
abrj-637	95	8	key	key	ADJ
abrj-637	95	9	findings	finding	NOUN
abrj-637	95	10	breslin	breslin	PROPN
abrj-637	95	11	(	(	PUNCT
abrj-637	95	12	2021	2021	NUM
abrj-637	95	13	)	)	PUNCT
abrj-637	95	14	tax	tax	NOUN
abrj-637	95	15	audit	audit	NOUN
abrj-637	95	16	efficiency	efficiency	NOUN
abrj-637	95	17	ml	ml	VERB
abrj-637	95	18	-	-	PUNCT
abrj-637	95	19	based	base	VERB
abrj-637	95	20	auditing	auditing	NOUN
abrj-637	95	21	improves	improve	VERB
abrj-637	95	22	speed	speed	NOUN
abrj-637	95	23	and	and	CCONJ
abrj-637	95	24	accuracy	accuracy	NOUN
abrj-637	95	25	wu	wu	PROPN
abrj-637	95	26	et	et	PROPN
abrj-637	95	27	al	al	PROPN
abrj-637	95	28	.	.	PROPN
abrj-637	96	1	(	(	PUNCT
abrj-637	96	2	2012	2012	NUM
abrj-637	96	3	)	)	PUNCT
abrj-637	96	4	tax	tax	NOUN
abrj-637	96	5	evasion	evasion	NOUN
abrj-637	96	6	detection	detection	NOUN
abrj-637	96	7	data	datum	NOUN
abrj-637	96	8	mining	mining	NOUN
abrj-637	96	9	enhances	enhance	VERB
abrj-637	96	10	fraud	fraud	NOUN
abrj-637	96	11	identification	identification	NOUN
abrj-637	96	12	acharya	acharya	NOUN
abrj-637	96	13	(	(	PUNCT
abrj-637	96	14	2025	2025	NUM
abrj-637	96	15	)	)	PUNCT
abrj-637	96	16	corporate	corporate	ADJ
abrj-637	96	17	fraud	fraud	NOUN
abrj-637	96	18	detection	detection	NOUN
abrj-637	96	19	ml	ml	NOUN
abrj-637	96	20	models	model	NOUN
abrj-637	96	21	boosts	boost	VERB
abrj-637	96	22	reliability	reliability	NOUN
abrj-637	96	23	in	in	ADP
abrj-637	96	24	filings	filing	NOUN
abrj-637	96	25	ippolito	ippolito	PROPN
abrj-637	96	26	&	&	CCONJ
abrj-637	96	27	lozano	lozano	PROPN
abrj-637	96	28	(	(	PUNCT
abrj-637	96	29	2020	2020	NUM
abrj-637	96	30	)	)	PUNCT
abrj-637	96	31	tax	tax	NOUN
abrj-637	96	32	crime	crime	NOUN
abrj-637	96	33	prediction	prediction	NOUN
abrj-637	96	34	predictive	predictive	ADJ
abrj-637	96	35	ml	ml	ADP
abrj-637	96	36	outperforms	outperform	VERB
abrj-637	96	37	manual	manual	ADJ
abrj-637	96	38	audits	audits	NOUN
abrj-637	96	39	craja	craja	NOUN
abrj-637	96	40	et	et	PROPN
abrj-637	96	41	al	al	PROPN
abrj-637	96	42	.	.	PROPN
abrj-637	97	1	(	(	PUNCT
abrj-637	97	2	2020	2020	NUM
abrj-637	97	3	)	)	PUNCT
abrj-637	97	4	financial	financial	ADJ
abrj-637	97	5	statement	statement	NOUN
abrj-637	97	6	fraud	fraud	NOUN
abrj-637	97	7	deep	deep	ADJ
abrj-637	97	8	learning	learning	NOUN
abrj-637	97	9	captures	capture	VERB
abrj-637	97	10	complex	complex	ADJ
abrj-637	97	11	fraud	fraud	NOUN
abrj-637	97	12	patterns	pattern	NOUN
abrj-637	97	13	mehta	mehta	PROPN
abrj-637	97	14	et	et	PROPN
abrj-637	97	15	al	al	PROPN
abrj-637	97	16	.	.	PROPN
abrj-637	98	1	(	(	PUNCT
abrj-637	98	2	2022	2022	NUM
abrj-637	98	3	)	)	PUNCT
abrj-637	98	4	tax	tax	NOUN
abrj-637	98	5	fraud	fraud	NOUN
abrj-637	98	6	simulation	simulation	NOUN
abrj-637	98	7	gan	gan	PROPN
abrj-637	98	8	improves	improve	VERB
abrj-637	98	9	model	model	NOUN
abrj-637	98	10	training	training	NOUN
abrj-637	98	11	data	datum	NOUN
abrj-637	98	12	shujaaddeen	shujaaddeen	PROPN
abrj-637	98	13	et	et	PROPN
abrj-637	98	14	al	al	PROPN
abrj-637	98	15	.	.	PROPN
abrj-637	99	1	(	(	PUNCT
abrj-637	99	2	2024	2024	NUM
abrj-637	99	3	)	)	PUNCT
abrj-637	99	4	tax	tax	NOUN
abrj-637	99	5	evasion	evasion	NOUN
abrj-637	99	6	levels	level	NOUN
abrj-637	99	7	hybrid	hybrid	ADJ
abrj-637	99	8	neural	neural	ADJ
abrj-637	99	9	net	net	ADJ
abrj-637	99	10	detects	detect	NOUN
abrj-637	99	11	multi	multi	ADJ
abrj-637	99	12	-	-	ADJ
abrj-637	99	13	level	level	ADJ
abrj-637	99	14	evasion	evasion	NOUN
abrj-637	99	15	martínez	martínez	NOUN
abrj-637	99	16	(	(	PUNCT
abrj-637	99	17	2025	2025	NUM
abrj-637	99	18	)	)	PUNCT
abrj-637	99	19	model	model	NOUN
abrj-637	99	20	comparison	comparison	NOUN
abrj-637	99	21	ensemble	ensemble	ADJ
abrj-637	99	22	ml	ml	ADP
abrj-637	99	23	hybrid	hybrid	ADJ
abrj-637	99	24	models	model	NOUN
abrj-637	99	25	yield	yield	VERB
abrj-637	99	26	higher	high	ADJ
abrj-637	99	27	accuracy	accuracy	NOUN
abrj-637	99	28	zhang	zhang	PROPN
abrj-637	99	29	et	et	PROPN
abrj-637	99	30	al	al	PROPN
abrj-637	99	31	.	.	PROPN
abrj-637	100	1	(	(	PUNCT
abrj-637	100	2	2024	2024	NUM
abrj-637	100	3	)	)	PUNCT
abrj-637	100	4	textual	textual	ADJ
abrj-637	100	5	fraud	fraud	NOUN
abrj-637	100	6	signals	signal	VERB
abrj-637	100	7	nlp	nlp	NOUN
abrj-637	100	8	+	+	CCONJ
abrj-637	100	9	ml	ml	X
abrj-637	100	10	linguistic	linguistic	ADJ
abrj-637	100	11	cues	cue	NOUN
abrj-637	100	12	predict	predict	VERB
abrj-637	100	13	fraud	fraud	NOUN
abrj-637	100	14	zhou	zhou	PROPN
abrj-637	100	15	et	et	PROPN
abrj-637	100	16	al	al	PROPN
abrj-637	100	17	.	.	PROPN
abrj-637	101	1	(	(	PUNCT
abrj-637	101	2	2024	2024	NUM
abrj-637	101	3	)	)	PUNCT
abrj-637	101	4	tax	tax	NOUN
abrj-637	101	5	fraud	fraud	NOUN
abrj-637	101	6	scoring	score	VERB
abrj-637	101	7	ensemble	ensemble	ADJ
abrj-637	101	8	learning	learning	NOUN
abrj-637	101	9	reduces	reduce	VERB
abrj-637	101	10	false	false	ADJ
abrj-637	101	11	positives	positive	NOUN
abrj-637	101	12	3	3	NUM
abrj-637	101	13	.	.	X
abrj-637	101	14	methodology	methodology	NOUN
abrj-637	101	15	the	the	DET
abrj-637	101	16	proposed	propose	VERB
abrj-637	101	17	study	study	NOUN
abrj-637	101	18	will	will	AUX
abrj-637	101	19	take	take	VERB
abrj-637	101	20	a	a	DET
abrj-637	101	21	quantitative	quantitative	ADJ
abrj-637	101	22	,	,	PUNCT
abrj-637	101	23	evidence	evidence	NOUN
abrj-637	101	24	-	-	PUNCT
abrj-637	101	25	based	base	VERB
abrj-637	101	26	methodology	methodology	NOUN
abrj-637	101	27	and	and	CCONJ
abrj-637	101	28	combine	combine	VERB
abrj-637	101	29	supervised	supervise	VERB
abrj-637	101	30	,	,	PUNCT
abrj-637	101	31	unsupervised	unsupervised	ADJ
abrj-637	101	32	,	,	PUNCT
abrj-637	101	33	and	and	CCONJ
abrj-637	101	34	hybrid	hybrid	ADJ
abrj-637	101	35	machine	machine	NOUN
abrj-637	101	36	learning	learning	NOUN
abrj-637	101	37	(	(	PUNCT
abrj-637	101	38	ml	ml	NOUN
abrj-637	101	39	)	)	PUNCT
abrj-637	101	40	to	to	PART
abrj-637	101	41	identify	identify	VERB
abrj-637	101	42	fraud	fraud	NOUN
abrj-637	101	43	in	in	ADP
abrj-637	101	44	corporate	corporate	ADJ
abrj-637	101	45	tax	tax	NOUN
abrj-637	101	46	filings	filing	NOUN
abrj-637	101	47	.	.	PUNCT
abrj-637	102	1	the	the	DET
abrj-637	102	2	workflow	workflow	NOUN
abrj-637	102	3	of	of	ADP
abrj-637	102	4	the	the	DET
abrj-637	102	5	methodology	methodology	NOUN
abrj-637	102	6	has	have	VERB
abrj-637	102	7	seven	seven	NUM
abrj-637	102	8	steps	step	NOUN
abrj-637	102	9	that	that	PRON
abrj-637	102	10	include	include	VERB
abrj-637	102	11	data	datum	NOUN
abrj-637	102	12	acquisition	acquisition	NOUN
abrj-637	102	13	,	,	PUNCT
abrj-637	102	14	preprocessing	preprocessing	NOUN
abrj-637	102	15	,	,	PUNCT
abrj-637	102	16	feature	feature	NOUN
abrj-637	102	17	engineering	engineering	NOUN
abrj-637	102	18	,	,	PUNCT
abrj-637	102	19	model	model	NOUN
abrj-637	102	20	training	training	NOUN
abrj-637	102	21	,	,	PUNCT
abrj-637	102	22	evaluation	evaluation	NOUN
abrj-637	102	23	,	,	PUNCT
abrj-637	102	24	and	and	CCONJ
abrj-637	102	25	interpretability	interpretability	NOUN
abrj-637	102	26	.	.	PUNCT
abrj-637	103	1	it	it	PRON
abrj-637	103	2	will	will	AUX
abrj-637	103	3	seek	seek	VERB
abrj-637	103	4	to	to	PART
abrj-637	103	5	establish	establish	VERB
abrj-637	103	6	a	a	DET
abrj-637	103	7	prediction	prediction	NOUN
abrj-637	103	8	system	system	NOUN
abrj-637	103	9	that	that	PRON
abrj-637	103	10	will	will	AUX
abrj-637	103	11	be	be	AUX
abrj-637	103	12	able	able	ADJ
abrj-637	103	13	to	to	PART
abrj-637	103	14	detect	detect	VERB
abrj-637	103	15	high	high	ADJ
abrj-637	103	16	-	-	PUNCT
abrj-637	103	17	risk	risk	NOUN
abrj-637	103	18	corporate	corporate	ADJ
abrj-637	103	19	filings	filing	NOUN
abrj-637	103	20	based	base	VERB
abrj-637	103	21	on	on	ADP
abrj-637	103	22	the	the	DET
abrj-637	103	23	past	past	NOUN
abrj-637	103	24	and	and	CCONJ
abrj-637	103	25	current	current	ADJ
abrj-637	103	26	financial	financial	ADJ
abrj-637	103	27	data	datum	NOUN
abrj-637	103	28	.	.	PUNCT
abrj-637	104	1	asian	asian	ADJ
abrj-637	104	2	business	business	PROPN
abrj-637	104	3	research	research	NOUN
abrj-637	104	4	journal	journal	NOUN
abrj-637	104	5	,	,	PUNCT
abrj-637	104	6	2025	2025	NUM
abrj-637	104	7	,	,	PUNCT
abrj-637	104	8	10(11	10(11	NUM
abrj-637	104	9	):	):	PUNCT
abrj-637	104	10	1	1	NUM
abrj-637	104	11	-	-	SYM
abrj-637	104	12	13	13	NUM
abrj-637	104	13	4	4	NUM
abrj-637	104	14	©	©	PROPN
abrj-637	104	15	2025	2025	NUM
abrj-637	104	16	by	by	ADP
abrj-637	104	17	the	the	DET
abrj-637	104	18	authors	author	NOUN
abrj-637	104	19	;	;	PUNCT
abrj-637	104	20	licensee	licensee	NOUN
abrj-637	104	21	eastern	eastern	ADJ
abrj-637	104	22	centre	centre	NOUN
abrj-637	104	23	of	of	ADP
abrj-637	104	24	science	science	NOUN
abrj-637	104	25	and	and	CCONJ
abrj-637	104	26	education	education	NOUN
abrj-637	104	27	,	,	PUNCT
abrj-637	104	28	usa	usa	PROPN
abrj-637	104	29	figure	figure	NOUN
abrj-637	104	30	1	1	NUM
abrj-637	104	31	.	.	PUNCT
abrj-637	104	32	general	general	ADJ
abrj-637	104	33	research	research	NOUN
abrj-637	104	34	model	model	NOUN
abrj-637	104	35	of	of	ADP
abrj-637	104	36	proposed	propose	VERB
abrj-637	104	37	machine	machine	NOUN
abrj-637	104	38	learning	learning	NOUN
abrj-637	104	39	-	-	PUNCT
abrj-637	104	40	based	base	VERB
abrj-637	104	41	tax	tax	NOUN
abrj-637	104	42	fraud	fraud	NOUN
abrj-637	104	43	detection	detection	NOUN
abrj-637	104	44	.	.	PUNCT
abrj-637	105	1	this	this	DET
abrj-637	105	2	theoretical	theoretical	ADJ
abrj-637	105	3	character	character	NOUN
abrj-637	105	4	shows	show	VERB
abrj-637	105	5	the	the	DET
abrj-637	105	6	workflow	workflow	NOUN
abrj-637	105	7	use	use	NOUN
abrj-637	105	8	chronologically	chronologically	ADV
abrj-637	105	9	in	in	ADP
abrj-637	105	10	this	this	DET
abrj-637	105	11	research	research	NOUN
abrj-637	105	12	the	the	DET
abrj-637	105	13	collection	collection	NOUN
abrj-637	105	14	and	and	CCONJ
abrj-637	105	15	preprocessing	preprocessing	NOUN
abrj-637	105	16	of	of	ADP
abrj-637	105	17	data	datum	NOUN
abrj-637	105	18	to	to	PART
abrj-637	105	19	feature	feature	VERB
abrj-637	105	20	engineering	engineering	NOUN
abrj-637	105	21	,	,	PUNCT
abrj-637	105	22	training	training	NOUN
abrj-637	105	23	of	of	ADP
abrj-637	105	24	a	a	DET
abrj-637	105	25	model	model	NOUN
abrj-637	105	26	,	,	PUNCT
abrj-637	105	27	risk	risk	NOUN
abrj-637	105	28	scoring	scoring	NOUN
abrj-637	105	29	,	,	PUNCT
abrj-637	105	30	and	and	CCONJ
abrj-637	105	31	explainability	explainability	NOUN
abrj-637	105	32	.	.	PUNCT
abrj-637	106	1	it	it	PRON
abrj-637	106	2	also	also	ADV
abrj-637	106	3	focuses	focus	VERB
abrj-637	106	4	on	on	ADP
abrj-637	106	5	the	the	DET
abrj-637	106	6	combination	combination	NOUN
abrj-637	106	7	of	of	ADP
abrj-637	106	8	structured	structured	ADJ
abrj-637	106	9	,	,	PUNCT
abrj-637	106	10	unstructured	unstructured	ADJ
abrj-637	106	11	and	and	CCONJ
abrj-637	106	12	external	external	ADJ
abrj-637	106	13	data	datum	NOUN
abrj-637	106	14	streams	stream	NOUN
abrj-637	106	15	,	,	PUNCT
abrj-637	106	16	serving	serve	VERB
abrj-637	106	17	into	into	ADP
abrj-637	106	18	supervised	supervised	ADJ
abrj-637	106	19	,	,	PUNCT
abrj-637	106	20	unsupervised	unsupervised	ADJ
abrj-637	106	21	and	and	CCONJ
abrj-637	106	22	hybrid	hybrid	ADJ
abrj-637	106	23	machine	machine	NOUN
abrj-637	106	24	-	-	PUNCT
abrj-637	106	25	learning	learn	VERB
abrj-637	106	26	predictions	prediction	NOUN
abrj-637	106	27	,	,	PUNCT
abrj-637	106	28	to	to	PART
abrj-637	106	29	generate	generate	VERB
abrj-637	106	30	readable	readable	ADJ
abrj-637	106	31	scores	score	NOUN
abrj-637	106	32	of	of	ADP
abrj-637	106	33	fraud	fraud	NOUN
abrj-637	106	34	-	-	PUNCT
abrj-637	106	35	risk	risk	NOUN
abrj-637	106	36	to	to	ADP
abrj-637	106	37	tax	tax	NOUN
abrj-637	106	38	authorities	authority	NOUN
abrj-637	106	39	.	.	PUNCT
abrj-637	107	1	3.1	3.1	NUM
abrj-637	107	2	.	.	PUNCT
abrj-637	107	3	research	research	NOUN
abrj-637	107	4	design	design	NOUN
abrj-637	107	5	the	the	DET
abrj-637	107	6	study	study	NOUN
abrj-637	107	7	plan	plan	NOUN
abrj-637	107	8	is	be	AUX
abrj-637	107	9	an	an	DET
abrj-637	107	10	iterative	iterative	NOUN
abrj-637	107	11	ml	ml	ADP
abrj-637	107	12	pipeline	pipeline	NOUN
abrj-637	107	13	,	,	PUNCT
abrj-637	107	14	which	which	PRON
abrj-637	107	15	will	will	AUX
abrj-637	107	16	start	start	VERB
abrj-637	107	17	with	with	ADP
abrj-637	107	18	data	datum	NOUN
abrj-637	107	19	collection	collection	NOUN
abrj-637	107	20	and	and	CCONJ
abrj-637	107	21	preprocessing	preprocessing	NOUN
abrj-637	107	22	,	,	PUNCT
abrj-637	107	23	feature	feature	NOUN
abrj-637	107	24	engineering	engineering	NOUN
abrj-637	107	25	,	,	PUNCT
abrj-637	107	26	training	training	NOUN
abrj-637	107	27	and	and	CCONJ
abrj-637	107	28	validation	validation	NOUN
abrj-637	107	29	of	of	ADP
abrj-637	107	30	the	the	DET
abrj-637	107	31	model	model	NOUN
abrj-637	107	32	,	,	PUNCT
abrj-637	107	33	and	and	CCONJ
abrj-637	107	34	interpretation	interpretation	NOUN
abrj-637	107	35	.	.	PUNCT
abrj-637	108	1	all	all	DET
abrj-637	108	2	phases	phase	NOUN
abrj-637	108	3	are	be	AUX
abrj-637	108	4	interrelated	interrelate	VERB
abrj-637	108	5	in	in	ADP
abrj-637	108	6	order	order	NOUN
abrj-637	108	7	to	to	PART
abrj-637	108	8	guarantee	guarantee	VERB
abrj-637	108	9	that	that	SCONJ
abrj-637	108	10	the	the	DET
abrj-637	108	11	integrity	integrity	NOUN
abrj-637	108	12	and	and	CCONJ
abrj-637	108	13	explainability	explainability	NOUN
abrj-637	108	14	of	of	ADP
abrj-637	108	15	data	datum	NOUN
abrj-637	108	16	are	be	AUX
abrj-637	108	17	preserved	preserve	VERB
abrj-637	108	18	throughout	throughout	ADP
abrj-637	108	19	the	the	DET
abrj-637	108	20	process	process	NOUN
abrj-637	108	21	.	.	PUNCT
abrj-637	109	1	the	the	DET
abrj-637	109	2	design	design	NOUN
abrj-637	109	3	is	be	AUX
abrj-637	109	4	inspired	inspire	VERB
abrj-637	109	5	by	by	ADP
abrj-637	109	6	the	the	DET
abrj-637	109	7	available	available	ADJ
abrj-637	109	8	literature	literature	NOUN
abrj-637	109	9	regarding	regard	VERB
abrj-637	109	10	hybrid	hybrid	NOUN
abrj-637	109	11	ml	ml	NOUN
abrj-637	109	12	systems	system	NOUN
abrj-637	109	13	used	use	VERB
abrj-637	109	14	to	to	PART
abrj-637	109	15	detect	detect	VERB
abrj-637	109	16	financial	financial	ADJ
abrj-637	109	17	anomalies	anomaly	NOUN
abrj-637	109	18	(	(	PUNCT
abrj-637	109	19	acharya	acharya	PROPN
abrj-637	109	20	,	,	PUNCT
abrj-637	109	21	2025	2025	NUM
abrj-637	109	22	;	;	PUNCT
abrj-637	109	23	zhou	zhou	PROPN
abrj-637	109	24	et	et	PROPN
abrj-637	109	25	al	al	PROPN
abrj-637	109	26	.	.	PROPN
abrj-637	109	27	,	,	PUNCT
abrj-637	109	28	2024	2024	NUM
abrj-637	109	29	;	;	PUNCT
abrj-637	109	30	martinez	martinez	PROPN
abrj-637	109	31	,	,	PUNCT
abrj-637	109	32	2025	2025	NUM
abrj-637	109	33	)	)	PUNCT
abrj-637	109	34	.	.	PUNCT
abrj-637	110	1	asian	asian	ADJ
abrj-637	110	2	business	business	PROPN
abrj-637	110	3	research	research	NOUN
abrj-637	110	4	journal	journal	NOUN
abrj-637	110	5	,	,	PUNCT
abrj-637	110	6	2025	2025	NUM
abrj-637	110	7	,	,	PUNCT
abrj-637	110	8	10(11	10(11	NUM
abrj-637	110	9	):	):	PUNCT
abrj-637	110	10	1	1	NUM
abrj-637	110	11	-	-	SYM
abrj-637	110	12	13	13	NUM
abrj-637	110	13	5	5	NUM
abrj-637	110	14	©	©	PROPN
abrj-637	110	15	2025	2025	NUM
abrj-637	110	16	by	by	ADP
abrj-637	110	17	the	the	DET
abrj-637	110	18	authors	author	NOUN
abrj-637	110	19	;	;	PUNCT
abrj-637	110	20	licensee	licensee	NOUN
abrj-637	110	21	eastern	eastern	ADJ
abrj-637	110	22	centre	centre	NOUN
abrj-637	110	23	of	of	ADP
abrj-637	110	24	science	science	NOUN
abrj-637	110	25	and	and	CCONJ
abrj-637	110	26	education	education	NOUN
abrj-637	110	27	,	,	PUNCT
abrj-637	110	28	usa	usa	PROPN
abrj-637	110	29	figure	figure	NOUN
abrj-637	110	30	2	2	NUM
abrj-637	110	31	.	.	PUNCT
abrj-637	110	32	fraud	fraud	NOUN
abrj-637	110	33	detection	detection	NOUN
abrj-637	110	34	pipeline	pipeline	NOUN
abrj-637	110	35	design	design	NOUN
abrj-637	110	36	step	step	NOUN
abrj-637	110	37	by	by	ADP
abrj-637	110	38	step	step	NOUN
abrj-637	110	39	.	.	PUNCT
abrj-637	111	1	this	this	DET
abrj-637	111	2	value	value	NOUN
abrj-637	111	3	describes	describe	VERB
abrj-637	111	4	the	the	DET
abrj-637	111	5	elaborate	elaborate	ADJ
abrj-637	111	6	fraud	fraud	NOUN
abrj-637	111	7	detection	detection	NOUN
abrj-637	111	8	pipeline	pipeline	NOUN
abrj-637	111	9	that	that	PRON
abrj-637	111	10	is	be	AUX
abrj-637	111	11	used	use	VERB
abrj-637	111	12	to	to	PART
abrj-637	111	13	identify	identify	VERB
abrj-637	111	14	fraud	fraud	NOUN
abrj-637	111	15	in	in	ADP
abrj-637	111	16	corporate	corporate	ADJ
abrj-637	111	17	filings	filing	NOUN
abrj-637	111	18	based	base	VERB
abrj-637	111	19	on	on	ADP
abrj-637	111	20	machine	machine	NOUN
abrj-637	111	21	learning	learning	NOUN
abrj-637	111	22	.	.	PUNCT
abrj-637	112	1	the	the	DET
abrj-637	112	2	pipeline	pipeline	NOUN
abrj-637	112	3	is	be	AUX
abrj-637	112	4	a	a	DET
abrj-637	112	5	multi	multi	ADJ
abrj-637	112	6	-	-	ADJ
abrj-637	112	7	stage	stage	ADJ
abrj-637	112	8	structure	structure	NOUN
abrj-637	112	9	,	,	PUNCT
abrj-637	112	10	which	which	PRON
abrj-637	112	11	is	be	AUX
abrj-637	112	12	an	an	DET
abrj-637	112	13	integration	integration	NOUN
abrj-637	112	14	of	of	ADP
abrj-637	112	15	various	various	ADJ
abrj-637	112	16	data	datum	NOUN
abrj-637	112	17	sources	source	NOUN
abrj-637	112	18	and	and	CCONJ
abrj-637	112	19	is	be	AUX
abrj-637	112	20	then	then	ADV
abrj-637	112	21	subjected	subject	VERB
abrj-637	112	22	to	to	ADP
abrj-637	112	23	intensive	intensive	ADJ
abrj-637	112	24	data	datum	NOUN
abrj-637	112	25	preprocessing	preprocessing	NOUN
abrj-637	112	26	and	and	CCONJ
abrj-637	112	27	model	model	NOUN
abrj-637	112	28	training	training	NOUN
abrj-637	112	29	.	.	PUNCT
abrj-637	113	1	3.2	3.2	NUM
abrj-637	113	2	.	.	PUNCT
abrj-637	114	1	data	datum	NOUN
abrj-637	114	2	sources	source	NOUN
abrj-637	114	3	the	the	DET
abrj-637	114	4	proposed	propose	VERB
abrj-637	114	5	framework	framework	NOUN
abrj-637	114	6	uses	use	VERB
abrj-637	114	7	three	three	NUM
abrj-637	114	8	significant	significant	ADJ
abrj-637	114	9	types	type	NOUN
abrj-637	114	10	of	of	ADP
abrj-637	114	11	data	datum	NOUN
abrj-637	114	12	.	.	PUNCT
abrj-637	115	1	the	the	DET
abrj-637	115	2	structured	structured	ADJ
abrj-637	115	3	data	data	NOUN
abrj-637	115	4	consists	consist	VERB
abrj-637	115	5	of	of	ADP
abrj-637	115	6	numerical	numerical	ADJ
abrj-637	115	7	corporate	corporate	ADJ
abrj-637	115	8	information	information	NOUN
abrj-637	115	9	like	like	ADP
abrj-637	115	10	balance	balance	NOUN
abrj-637	115	11	sheets	sheet	NOUN
abrj-637	115	12	,	,	PUNCT
abrj-637	115	13	income	income	NOUN
abrj-637	115	14	statements	statement	NOUN
abrj-637	115	15	,	,	PUNCT
abrj-637	115	16	and	and	CCONJ
abrj-637	115	17	tax	tax	NOUN
abrj-637	115	18	filing	filing	NOUN
abrj-637	115	19	(	(	PUNCT
abrj-637	115	20	wu	wu	PROPN
abrj-637	115	21	et	et	PROPN
abrj-637	115	22	al	al	PROPN
abrj-637	115	23	.	.	PROPN
abrj-637	115	24	,	,	PUNCT
abrj-637	115	25	2012	2012	NUM
abrj-637	115	26	;	;	PUNCT
abrj-637	115	27	acharya	acharya	NOUN
abrj-637	115	28	,	,	PUNCT
abrj-637	115	29	2025	2025	NUM
abrj-637	115	30	)	)	PUNCT
abrj-637	115	31	.	.	PUNCT
abrj-637	116	1	unformatted	unformatted	ADJ
abrj-637	116	2	data	datum	NOUN
abrj-637	116	3	include	include	VERB
abrj-637	116	4	the	the	DET
abrj-637	116	5	textual	textual	ADJ
abrj-637	116	6	disclosures	disclosure	NOUN
abrj-637	116	7	,	,	PUNCT
abrj-637	116	8	including	include	VERB
abrj-637	116	9	management	management	NOUN
abrj-637	116	10	commentaries	commentary	NOUN
abrj-637	116	11	and	and	CCONJ
abrj-637	116	12	auditor	auditor	NOUN
abrj-637	116	13	notes	note	NOUN
abrj-637	116	14	,	,	PUNCT
abrj-637	116	15	which	which	PRON
abrj-637	116	16	can	can	AUX
abrj-637	116	17	be	be	AUX
abrj-637	116	18	useful	useful	ADJ
abrj-637	116	19	in	in	ADP
abrj-637	116	20	terms	term	NOUN
abrj-637	116	21	of	of	ADP
abrj-637	116	22	linguistic	linguistic	ADJ
abrj-637	116	23	indicators	indicator	NOUN
abrj-637	116	24	of	of	ADP
abrj-637	116	25	fraud	fraud	NOUN
abrj-637	116	26	(	(	PUNCT
abrj-637	116	27	zhang	zhang	PROPN
abrj-637	116	28	et	et	PROPN
abrj-637	116	29	al	al	PROPN
abrj-637	116	30	.	.	PROPN
abrj-637	116	31	,	,	PUNCT
abrj-637	116	32	2024	2024	NUM
abrj-637	116	33	;	;	PUNCT
abrj-637	116	34	ji	ji	PROPN
abrj-637	116	35	et	et	PROPN
abrj-637	116	36	al	al	PROPN
abrj-637	116	37	.	.	PROPN
abrj-637	116	38	,	,	PUNCT
abrj-637	116	39	2024	2024	NUM
abrj-637	116	40	)	)	PUNCT
abrj-637	116	41	.	.	PUNCT
abrj-637	117	1	contextual	contextual	ADJ
abrj-637	117	2	information	information	NOUN
abrj-637	117	3	is	be	AUX
abrj-637	117	4	added	add	VERB
abrj-637	117	5	with	with	ADP
abrj-637	117	6	external	external	ADJ
abrj-637	117	7	data	datum	NOUN
abrj-637	117	8	such	such	ADJ
abrj-637	117	9	as	as	ADP
abrj-637	117	10	macroeconomic	macroeconomic	ADJ
abrj-637	117	11	data	datum	NOUN
abrj-637	117	12	,	,	PUNCT
abrj-637	117	13	previous	previous	ADJ
abrj-637	117	14	compliance	compliance	NOUN
abrj-637	117	15	records	record	NOUN
abrj-637	117	16	,	,	PUNCT
abrj-637	117	17	and	and	CCONJ
abrj-637	117	18	transaction	transaction	NOUN
abrj-637	117	19	history	history	NOUN
abrj-637	117	20	(	(	PUNCT
abrj-637	117	21	ippolito	ippolito	PROPN
abrj-637	117	22	and	and	CCONJ
abrj-637	117	23	lozano	lozano	PROPN
abrj-637	117	24	,	,	PUNCT
abrj-637	117	25	2020	2020	NUM
abrj-637	117	26	;	;	PUNCT
abrj-637	117	27	galla	galla	PROPN
abrj-637	117	28	,	,	PUNCT
abrj-637	117	29	2023	2023	NUM
abrj-637	117	30	)	)	PUNCT
abrj-637	117	31	.	.	PUNCT
abrj-637	118	1	every	every	DET
abrj-637	118	2	dataset	dataset	NOUN
abrj-637	118	3	is	be	AUX
abrj-637	118	4	normalized	normalize	VERB
abrj-637	118	5	to	to	PART
abrj-637	118	6	bring	bring	VERB
abrj-637	118	7	about	about	ADP
abrj-637	118	8	consistency	consistency	NOUN
abrj-637	118	9	and	and	CCONJ
abrj-637	118	10	compatibility	compatibility	NOUN
abrj-637	118	11	with	with	ADP
abrj-637	118	12	ml	ml	NOUN
abrj-637	118	13	models	model	NOUN
abrj-637	118	14	.	.	PUNCT
abrj-637	119	1	table	table	NOUN
abrj-637	119	2	1	1	NUM
abrj-637	119	3	.	.	PUNCT
abrj-637	120	1	data	datum	NOUN
abrj-637	120	2	and	and	CCONJ
abrj-637	120	3	sources	source	NOUN
abrj-637	120	4	of	of	ADP
abrj-637	120	5	data	datum	NOUN
abrj-637	120	6	to	to	PART
abrj-637	120	7	use	use	VERB
abrj-637	120	8	in	in	ADP
abrj-637	120	9	the	the	DET
abrj-637	120	10	study	study	NOUN
abrj-637	120	11	.	.	PUNCT
abrj-637	121	1	data	datum	NOUN
abrj-637	121	2	type	type	NOUN
abrj-637	121	3	source	source	NOUN
abrj-637	121	4	description	description	NOUN
abrj-637	121	5	structured	structure	VERB
abrj-637	121	6	data	datum	NOUN
abrj-637	121	7	corporate	corporate	ADJ
abrj-637	121	8	financial	financial	ADJ
abrj-637	121	9	records	record	NOUN
abrj-637	121	10	includes	include	VERB
abrj-637	121	11	balance	balance	NOUN
abrj-637	121	12	sheets	sheet	NOUN
abrj-637	121	13	,	,	PUNCT
abrj-637	121	14	income	income	NOUN
abrj-637	121	15	statements	statement	NOUN
abrj-637	121	16	,	,	PUNCT
abrj-637	121	17	and	and	CCONJ
abrj-637	121	18	tax	tax	NOUN
abrj-637	121	19	filings	filing	NOUN
abrj-637	121	20	,	,	PUNCT
abrj-637	121	21	which	which	PRON
abrj-637	121	22	provide	provide	VERB
abrj-637	121	23	quantitative	quantitative	ADJ
abrj-637	121	24	financial	financial	ADJ
abrj-637	121	25	data	datum	NOUN
abrj-637	121	26	.	.	PUNCT
abrj-637	122	1	structured	structured	ADJ
abrj-637	122	2	data	datum	NOUN
abrj-637	122	3	compliance	compliance	NOUN
abrj-637	122	4	reports	report	NOUN
abrj-637	122	5	contains	contain	VERB
abrj-637	122	6	records	record	NOUN
abrj-637	122	7	of	of	ADP
abrj-637	122	8	corporate	corporate	ADJ
abrj-637	122	9	compliance	compliance	NOUN
abrj-637	122	10	with	with	ADP
abrj-637	122	11	tax	tax	NOUN
abrj-637	122	12	regulations	regulation	NOUN
abrj-637	122	13	,	,	PUNCT
abrj-637	122	14	which	which	PRON
abrj-637	122	15	can	can	AUX
abrj-637	122	16	help	help	VERB
abrj-637	122	17	in	in	ADP
abrj-637	122	18	detecting	detect	VERB
abrj-637	122	19	discrepancies	discrepancy	NOUN
abrj-637	122	20	.	.	PUNCT
abrj-637	123	1	structured	structured	ADJ
abrj-637	123	2	data	datum	NOUN
abrj-637	123	3	external	external	ADJ
abrj-637	123	4	databases	database	NOUN
abrj-637	123	5	(	(	PUNCT
abrj-637	123	6	e.g.	e.g.	ADV
abrj-637	123	7	,	,	PUNCT
abrj-637	123	8	economic	economic	ADJ
abrj-637	123	9	indicators	indicator	NOUN
abrj-637	123	10	,	,	PUNCT
abrj-637	123	11	transaction	transaction	NOUN
abrj-637	123	12	history	history	NOUN
abrj-637	123	13	)	)	PUNCT
abrj-637	123	14	encompasses	encompass	VERB
abrj-637	123	15	external	external	ADJ
abrj-637	123	16	financial	financial	ADJ
abrj-637	123	17	data	datum	NOUN
abrj-637	123	18	that	that	PRON
abrj-637	123	19	provides	provide	VERB
abrj-637	123	20	contextual	contextual	ADJ
abrj-637	123	21	information	information	NOUN
abrj-637	123	22	relevant	relevant	ADJ
abrj-637	123	23	to	to	ADP
abrj-637	123	24	corporate	corporate	ADJ
abrj-637	123	25	filings	filing	NOUN
abrj-637	123	26	.	.	PUNCT
abrj-637	124	1	unstructured	unstructured	ADJ
abrj-637	124	2	data	data	PROPN
abrj-637	124	3	management	management	NOUN
abrj-637	124	4	commentaries	commentary	NOUN
abrj-637	124	5	and	and	CCONJ
abrj-637	124	6	reports	report	VERB
abrj-637	124	7	textual	textual	ADJ
abrj-637	124	8	data	datum	NOUN
abrj-637	124	9	from	from	ADP
abrj-637	124	10	company	company	NOUN
abrj-637	124	11	reports	report	NOUN
abrj-637	124	12	and	and	CCONJ
abrj-637	124	13	management	management	NOUN
abrj-637	124	14	discussions	discussion	NOUN
abrj-637	124	15	that	that	PRON
abrj-637	124	16	may	may	AUX
abrj-637	124	17	contain	contain	VERB
abrj-637	124	18	linguistic	linguistic	ADJ
abrj-637	124	19	cues	cue	NOUN
abrj-637	124	20	of	of	ADP
abrj-637	124	21	fraud	fraud	NOUN
abrj-637	124	22	.	.	PUNCT
abrj-637	125	1	unstructured	unstructured	ADJ
abrj-637	125	2	data	data	PROPN
abrj-637	125	3	auditor	auditor	NOUN
abrj-637	125	4	notes	note	NOUN
abrj-637	125	5	and	and	CCONJ
abrj-637	125	6	external	external	ADJ
abrj-637	125	7	evaluations	evaluation	NOUN
abrj-637	125	8	includes	include	VERB
abrj-637	125	9	qualitative	qualitative	ADJ
abrj-637	125	10	insights	insight	NOUN
abrj-637	125	11	from	from	ADP
abrj-637	125	12	auditors	auditor	NOUN
abrj-637	125	13	that	that	PRON
abrj-637	125	14	may	may	AUX
abrj-637	125	15	help	help	AUX
abrj-637	125	16	reveal	reveal	VERB
abrj-637	125	17	fraudulent	fraudulent	ADJ
abrj-637	125	18	behavior	behavior	NOUN
abrj-637	125	19	not	not	PART
abrj-637	125	20	captured	capture	VERB
abrj-637	125	21	in	in	ADP
abrj-637	125	22	numerical	numerical	ADJ
abrj-637	125	23	data	data	PROPN
abrj-637	125	24	.	.	PUNCT
abrj-637	126	1	external	external	ADJ
abrj-637	126	2	data	datum	NOUN
abrj-637	126	3	transaction	transaction	NOUN
abrj-637	126	4	histories	history	NOUN
abrj-637	126	5	,	,	PUNCT
abrj-637	126	6	historical	historical	ADJ
abrj-637	126	7	compliance	compliance	NOUN
abrj-637	126	8	behavior	behavior	NOUN
abrj-637	126	9	provides	provide	VERB
abrj-637	126	10	context	context	NOUN
abrj-637	126	11	on	on	ADP
abrj-637	126	12	previous	previous	ADJ
abrj-637	126	13	behavior	behavior	NOUN
abrj-637	126	14	,	,	PUNCT
abrj-637	126	15	helping	helping	AUX
abrj-637	126	16	assess	assess	VERB
abrj-637	126	17	the	the	DET
abrj-637	126	18	likelihood	likelihood	NOUN
abrj-637	126	19	of	of	ADP
abrj-637	126	20	fraud	fraud	NOUN
abrj-637	126	21	based	base	VERB
abrj-637	126	22	on	on	ADP
abrj-637	126	23	past	past	ADJ
abrj-637	126	24	trends	trend	NOUN
abrj-637	126	25	.	.	PUNCT
abrj-637	127	1	asian	asian	ADJ
abrj-637	127	2	business	business	PROPN
abrj-637	127	3	research	research	NOUN
abrj-637	127	4	journal	journal	NOUN
abrj-637	127	5	,	,	PUNCT
abrj-637	127	6	2025	2025	NUM
abrj-637	127	7	,	,	PUNCT
abrj-637	127	8	10(11	10(11	NUM
abrj-637	127	9	):	):	PUNCT
abrj-637	127	10	1	1	NUM
abrj-637	127	11	-	-	SYM
abrj-637	127	12	13	13	NUM
abrj-637	127	13	6	6	NUM
abrj-637	127	14	©	©	ADP
abrj-637	127	15	2025	2025	NUM
abrj-637	127	16	by	by	ADP
abrj-637	127	17	the	the	DET
abrj-637	127	18	authors	author	NOUN
abrj-637	127	19	;	;	PUNCT
abrj-637	127	20	licensee	licensee	NOUN
abrj-637	127	21	eastern	eastern	ADJ
abrj-637	127	22	centre	centre	NOUN
abrj-637	127	23	of	of	ADP
abrj-637	127	24	science	science	NOUN
abrj-637	127	25	and	and	CCONJ
abrj-637	127	26	education	education	NOUN
abrj-637	127	27	,	,	PUNCT
abrj-637	127	28	usa	usa	PROPN
abrj-637	127	29	this	this	DET
abrj-637	127	30	table	table	NOUN
abrj-637	127	31	summarizes	summarize	VERB
abrj-637	127	32	the	the	DET
abrj-637	127	33	different	different	ADJ
abrj-637	127	34	data	datum	NOUN
abrj-637	127	35	types	type	NOUN
abrj-637	127	36	and	and	CCONJ
abrj-637	127	37	sources	source	NOUN
abrj-637	127	38	used	use	VERB
abrj-637	127	39	in	in	ADP
abrj-637	127	40	the	the	DET
abrj-637	127	41	research	research	NOUN
abrj-637	127	42	towards	towards	ADP
abrj-637	127	43	identifying	identify	VERB
abrj-637	127	44	tax	tax	NOUN
abrj-637	127	45	fraud	fraud	NOUN
abrj-637	127	46	in	in	ADP
abrj-637	127	47	corporate	corporate	ADJ
abrj-637	127	48	filings	filing	NOUN
abrj-637	127	49	with	with	ADP
abrj-637	127	50	the	the	DET
abrj-637	127	51	help	help	NOUN
abrj-637	127	52	of	of	ADP
abrj-637	127	53	machine	machine	NOUN
abrj-637	127	54	learning	learning	NOUN
abrj-637	127	55	.	.	PUNCT
abrj-637	128	1	these	these	DET
abrj-637	128	2	sources	source	NOUN
abrj-637	128	3	offer	offer	VERB
abrj-637	128	4	a	a	DET
abrj-637	128	5	multi	multi	ADJ
abrj-637	128	6	-	-	ADJ
abrj-637	128	7	dimensional	dimensional	ADJ
abrj-637	128	8	data	datum	NOUN
abrj-637	128	9	,	,	PUNCT
abrj-637	128	10	which	which	PRON
abrj-637	128	11	is	be	AUX
abrj-637	128	12	broad	broad	ADJ
abrj-637	128	13	,	,	PUNCT
abrj-637	128	14	covering	cover	VERB
abrj-637	128	15	both	both	CCONJ
abrj-637	128	16	structured	structured	ADJ
abrj-637	128	17	and	and	CCONJ
abrj-637	128	18	unstructured	unstructured	ADJ
abrj-637	128	19	data	datum	NOUN
abrj-637	128	20	.	.	PUNCT
abrj-637	129	1	the	the	DET
abrj-637	129	2	various	various	ADJ
abrj-637	129	3	types	type	NOUN
abrj-637	129	4	of	of	ADP
abrj-637	129	5	data	datum	NOUN
abrj-637	129	6	are	be	AUX
abrj-637	129	7	important	important	ADJ
abrj-637	129	8	into	into	ADP
abrj-637	129	9	the	the	DET
abrj-637	129	10	fraud	fraud	NOUN
abrj-637	129	11	detection	detection	NOUN
abrj-637	129	12	pipeline	pipeline	NOUN
abrj-637	129	13	because	because	SCONJ
abrj-637	129	14	they	they	PRON
abrj-637	129	15	provide	provide	VERB
abrj-637	129	16	both	both	CCONJ
abrj-637	129	17	qualitative	qualitative	ADJ
abrj-637	129	18	and	and	CCONJ
abrj-637	129	19	financial	financial	ADJ
abrj-637	129	20	measures	measure	NOUN
abrj-637	129	21	of	of	ADP
abrj-637	129	22	data	datum	NOUN
abrj-637	129	23	.	.	PUNCT
abrj-637	130	1	3.3	3.3	NUM
abrj-637	130	2	.	.	PUNCT
abrj-637	131	1	data	datum	NOUN
abrj-637	131	2	preprocessing	preprocesse	VERB
abrj-637	131	3	preprocessing	preprocesse	VERB
abrj-637	131	4	the	the	DET
abrj-637	131	5	data	data	NOUN
abrj-637	131	6	is	be	AUX
abrj-637	131	7	on	on	ADP
abrj-637	131	8	the	the	DET
abrj-637	131	9	basis	basis	NOUN
abrj-637	131	10	of	of	ADP
abrj-637	131	11	reliability	reliability	NOUN
abrj-637	131	12	and	and	CCONJ
abrj-637	131	13	accuracy	accuracy	NOUN
abrj-637	131	14	before	before	ADP
abrj-637	131	15	modeling	modeling	NOUN
abrj-637	131	16	.	.	PUNCT
abrj-637	132	1	this	this	PRON
abrj-637	132	2	is	be	AUX
abrj-637	132	3	done	do	VERB
abrj-637	132	4	by	by	ADP
abrj-637	132	5	cleaning	clean	VERB
abrj-637	132	6	to	to	PART
abrj-637	132	7	remove	remove	VERB
abrj-637	132	8	duplicates	duplicate	NOUN
abrj-637	132	9	and	and	CCONJ
abrj-637	132	10	missing	miss	VERB
abrj-637	132	11	values	value	NOUN
abrj-637	132	12	(	(	PUNCT
abrj-637	132	13	breslin	breslin	NOUN
abrj-637	132	14	,	,	PUNCT
abrj-637	132	15	2021	2021	NUM
abrj-637	132	16	)	)	PUNCT
abrj-637	132	17	,	,	PUNCT
abrj-637	132	18	normalization	normalization	NOUN
abrj-637	132	19	to	to	PART
abrj-637	132	20	put	put	VERB
abrj-637	132	21	the	the	DET
abrj-637	132	22	numerical	numerical	ADJ
abrj-637	132	23	scales	scale	NOUN
abrj-637	132	24	on	on	ADP
abrj-637	132	25	the	the	DET
abrj-637	132	26	same	same	ADJ
abrj-637	132	27	level	level	NOUN
abrj-637	132	28	,	,	PUNCT
abrj-637	132	29	and	and	CCONJ
abrj-637	132	30	encoding	encode	VERB
abrj-637	132	31	the	the	DET
abrj-637	132	32	categorical	categorical	ADJ
abrj-637	132	33	variables	variable	NOUN
abrj-637	132	34	with	with	ADP
abrj-637	132	35	one	one	NUM
abrj-637	132	36	-	-	PUNCT
abrj-637	132	37	hot	hot	ADJ
abrj-637	132	38	or	or	CCONJ
abrj-637	132	39	label	label	NOUN
abrj-637	132	40	encoding	encoding	NOUN
abrj-637	132	41	.	.	PUNCT
abrj-637	133	1	in	in	ADP
abrj-637	133	2	the	the	DET
abrj-637	133	3	case	case	NOUN
abrj-637	133	4	of	of	ADP
abrj-637	133	5	unstructured	unstructured	ADJ
abrj-637	133	6	textual	textual	ADJ
abrj-637	133	7	data	datum	NOUN
abrj-637	133	8	,	,	PUNCT
abrj-637	133	9	preprocessing	preprocesse	VERB
abrj-637	133	10	requires	require	VERB
abrj-637	133	11	tokenization	tokenization	NOUN
abrj-637	133	12	of	of	ADP
abrj-637	133	13	data	datum	NOUN
abrj-637	133	14	,	,	PUNCT
abrj-637	133	15	removal	removal	NOUN
abrj-637	133	16	of	of	ADP
abrj-637	133	17	stop	stop	NOUN
abrj-637	133	18	-	-	PUNCT
abrj-637	133	19	words	word	NOUN
abrj-637	133	20	,	,	PUNCT
abrj-637	133	21	and	and	CCONJ
abrj-637	133	22	sentiment	sentiment	NOUN
abrj-637	133	23	analysis	analysis	NOUN
abrj-637	133	24	(	(	PUNCT
abrj-637	133	25	zhang	zhang	X
abrj-637	133	26	et	et	PROPN
abrj-637	133	27	al	al	PROPN
abrj-637	133	28	.	.	PROPN
abrj-637	133	29	,	,	PUNCT
abrj-637	133	30	2024	2024	NUM
abrj-637	133	31	)	)	PUNCT
abrj-637	133	32	.	.	PUNCT
abrj-637	134	1	synthetic	synthetic	ADJ
abrj-637	134	2	minority	minority	NOUN
abrj-637	134	3	oversampling	oversample	VERB
abrj-637	134	4	technique	technique	NOUN
abrj-637	134	5	(	(	PUNCT
abrj-637	134	6	smote	smote	NOUN
abrj-637	134	7	)	)	PUNCT
abrj-637	134	8	and	and	CCONJ
abrj-637	134	9	undersampling	undersample	VERB
abrj-637	134	10	are	be	AUX
abrj-637	134	11	used	use	VERB
abrj-637	134	12	as	as	ADP
abrj-637	134	13	methods	method	NOUN
abrj-637	134	14	of	of	ADP
abrj-637	134	15	reducing	reduce	VERB
abrj-637	134	16	the	the	DET
abrj-637	134	17	issue	issue	NOUN
abrj-637	134	18	of	of	ADP
abrj-637	134	19	class	class	NOUN
abrj-637	134	20	imbalance	imbalance	NOUN
abrj-637	134	21	(	(	PUNCT
abrj-637	134	22	tagbo	tagbo	NOUN
abrj-637	134	23	&	&	CCONJ
abrj-637	134	24	adekoya	adekoya	PROPN
abrj-637	134	25	,	,	PUNCT
abrj-637	134	26	2023	2023	NUM
abrj-637	134	27	;	;	PUNCT
abrj-637	134	28	wahyono	wahyono	PROPN
abrj-637	134	29	and	and	CCONJ
abrj-637	134	30	david	david	PROPN
abrj-637	134	31	,	,	PUNCT
abrj-637	134	32	2025	2025	NUM
abrj-637	134	33	)	)	PUNCT
abrj-637	134	34	.	.	PUNCT
abrj-637	135	1	figure	figure	VERB
abrj-637	135	2	3	3	NUM
abrj-637	135	3	.	.	PUNCT
abrj-637	136	1	preprocessing	preprocessing	NOUN
abrj-637	136	2	and	and	CCONJ
abrj-637	136	3	transformation	transformation	NOUN
abrj-637	136	4	of	of	ADP
abrj-637	136	5	data	datum	NOUN
abrj-637	136	6	.	.	PUNCT
abrj-637	137	1	3.4	3.4	NUM
abrj-637	137	2	.	.	PUNCT
abrj-637	137	3	feature	feature	NOUN
abrj-637	137	4	engineering	engineer	VERB
abrj-637	137	5	the	the	DET
abrj-637	137	6	feature	feature	NOUN
abrj-637	137	7	engineering	engineering	NOUN
abrj-637	137	8	increases	increase	VERB
abrj-637	137	9	the	the	DET
abrj-637	137	10	accuracy	accuracy	NOUN
abrj-637	137	11	and	and	CCONJ
abrj-637	137	12	interpretability	interpretability	NOUN
abrj-637	137	13	of	of	ADP
abrj-637	137	14	the	the	DET
abrj-637	137	15	model	model	NOUN
abrj-637	137	16	.	.	PUNCT
abrj-637	138	1	ratios	ratio	NOUN
abrj-637	138	2	(	(	PUNCT
abrj-637	138	3	profit	profit	NOUN
abrj-637	138	4	margins	margin	NOUN
abrj-637	138	5	,	,	PUNCT
abrj-637	138	6	debtto	debtto	NOUN
abrj-637	138	7	-	-	PUNCT
abrj-637	138	8	equity	equity	NOUN
abrj-637	138	9	,	,	PUNCT
abrj-637	138	10	and	and	CCONJ
abrj-637	138	11	effective	effective	ADJ
abrj-637	138	12	tax	tax	NOUN
abrj-637	138	13	rate	rate	NOUN
abrj-637	138	14	)	)	PUNCT
abrj-637	138	15	and	and	CCONJ
abrj-637	138	16	such	such	ADJ
abrj-637	138	17	aspects	aspect	NOUN
abrj-637	138	18	of	of	ADP
abrj-637	138	19	transactions	transaction	NOUN
abrj-637	138	20	as	as	ADP
abrj-637	138	21	frequency	frequency	NOUN
abrj-637	138	22	and	and	CCONJ
abrj-637	138	23	large	large	ADJ
abrj-637	138	24	transaction	transaction	NOUN
abrj-637	138	25	volume	volume	NOUN
abrj-637	138	26	are	be	AUX
abrj-637	138	27	calculated	calculate	VERB
abrj-637	138	28	(	(	PUNCT
abrj-637	138	29	craja	craja	NOUN
abrj-637	138	30	et	et	PROPN
abrj-637	138	31	al	al	PROPN
abrj-637	138	32	.	.	PROPN
abrj-637	138	33	,	,	PUNCT
abrj-637	138	34	2020	2020	NUM
abrj-637	138	35	)	)	PUNCT
abrj-637	138	36	.	.	PUNCT
abrj-637	139	1	this	this	PRON
abrj-637	139	2	is	be	AUX
abrj-637	139	3	based	base	VERB
abrj-637	139	4	on	on	ADP
abrj-637	139	5	behavioral	behavioral	ADJ
abrj-637	139	6	indicators	indicator	NOUN
abrj-637	139	7	,	,	PUNCT
abrj-637	139	8	that	that	ADV
abrj-637	139	9	is	is	ADV
abrj-637	139	10	,	,	PUNCT
abrj-637	139	11	the	the	DET
abrj-637	139	12	late	late	ADJ
abrj-637	139	13	filings	filing	NOUN
abrj-637	139	14	and	and	CCONJ
abrj-637	139	15	revenue	revenue	NOUN
abrj-637	139	16	restatements	restatement	NOUN
abrj-637	139	17	,	,	PUNCT
abrj-637	139	18	which	which	PRON
abrj-637	139	19	reflect	reflect	VERB
abrj-637	139	20	risky	risky	ADJ
abrj-637	139	21	corporate	corporate	ADJ
abrj-637	139	22	behavior	behavior	NOUN
abrj-637	139	23	(	(	PUNCT
abrj-637	139	24	breslin	breslin	PROPN
abrj-637	139	25	,	,	PUNCT
abrj-637	139	26	2021	2021	NUM
abrj-637	139	27	)	)	PUNCT
abrj-637	139	28	.	.	PUNCT
abrj-637	140	1	textual	textual	ADJ
abrj-637	140	2	characteristics	characteristic	NOUN
abrj-637	140	3	are	be	AUX
abrj-637	140	4	sentiment	sentiment	NOUN
abrj-637	140	5	polarity	polarity	NOUN
abrj-637	140	6	,	,	PUNCT
abrj-637	140	7	readability	readability	NOUN
abrj-637	140	8	and	and	CCONJ
abrj-637	140	9	linguistic	linguistic	ADJ
abrj-637	140	10	ambiguity	ambiguity	NOUN
abrj-637	140	11	(	(	PUNCT
abrj-637	140	12	zhang	zhang	PROPN
abrj-637	140	13	et	et	PROPN
abrj-637	140	14	al	al	PROPN
abrj-637	140	15	.	.	PROPN
abrj-637	140	16	,	,	PUNCT
abrj-637	140	17	2024	2024	NUM
abrj-637	140	18	;	;	PUNCT
abrj-637	140	19	ji	ji	PROPN
abrj-637	140	20	et	et	PROPN
abrj-637	140	21	al	al	PROPN
abrj-637	140	22	.	.	PROPN
abrj-637	140	23	,	,	PUNCT
abrj-637	140	24	2024	2024	NUM
abrj-637	140	25	)	)	PUNCT
abrj-637	140	26	.	.	PUNCT
abrj-637	141	1	table	table	NOUN
abrj-637	141	2	2	2	NUM
abrj-637	141	3	.	.	X
abrj-637	141	4	extraction	extraction	NOUN
abrj-637	141	5	of	of	ADP
abrj-637	141	6	quantitative	quantitative	ADJ
abrj-637	141	7	and	and	CCONJ
abrj-637	141	8	textual	textual	ADJ
abrj-637	141	9	analysis	analysis	NOUN
abrj-637	141	10	.	.	PUNCT
abrj-637	142	1	feature	feature	NOUN
abrj-637	142	2	type	type	NOUN
abrj-637	142	3	data	datum	NOUN
abrj-637	142	4	source	source	NOUN
abrj-637	142	5	extraction	extraction	NOUN
abrj-637	142	6	method	method	NOUN
abrj-637	142	7	description	description	NOUN
abrj-637	142	8	quantitative	quantitative	ADJ
abrj-637	142	9	corporate	corporate	ADJ
abrj-637	142	10	financial	financial	ADJ
abrj-637	142	11	statements	statement	NOUN
abrj-637	142	12	financial	financial	ADJ
abrj-637	142	13	ratio	ratio	NOUN
abrj-637	142	14	analysis	analysis	NOUN
abrj-637	142	15	,	,	PUNCT
abrj-637	142	16	including	include	VERB
abrj-637	142	17	profit	profit	NOUN
abrj-637	142	18	margins	margin	NOUN
abrj-637	142	19	,	,	PUNCT
abrj-637	142	20	debt	debt	NOUN
abrj-637	142	21	-	-	PUNCT
abrj-637	142	22	to	to	ADP
abrj-637	142	23	-	-	PUNCT
abrj-637	142	24	equity	equity	NOUN
abrj-637	142	25	,	,	PUNCT
abrj-637	142	26	effective	effective	ADJ
abrj-637	142	27	tax	tax	NOUN
abrj-637	142	28	rates	rate	NOUN
abrj-637	142	29	key	key	ADJ
abrj-637	142	30	financial	financial	ADJ
abrj-637	142	31	ratios	ratio	NOUN
abrj-637	142	32	used	use	VERB
abrj-637	142	33	to	to	PART
abrj-637	142	34	assess	assess	VERB
abrj-637	142	35	the	the	DET
abrj-637	142	36	health	health	NOUN
abrj-637	142	37	and	and	CCONJ
abrj-637	142	38	potential	potential	ADJ
abrj-637	142	39	risks	risk	NOUN
abrj-637	142	40	of	of	ADP
abrj-637	142	41	a	a	DET
abrj-637	142	42	company	company	NOUN
abrj-637	142	43	.	.	PUNCT
abrj-637	143	1	quantitative	quantitative	ADJ
abrj-637	143	2	transaction	transaction	NOUN
abrj-637	143	3	data	datum	NOUN
abrj-637	143	4	frequency	frequency	NOUN
abrj-637	143	5	,	,	PUNCT
abrj-637	143	6	magnitude	magnitude	NOUN
abrj-637	143	7	,	,	PUNCT
abrj-637	143	8	and	and	CCONJ
abrj-637	143	9	timing	timing	NOUN
abrj-637	143	10	of	of	ADP
abrj-637	143	11	large	large	ADJ
abrj-637	143	12	transactions	transaction	NOUN
abrj-637	143	13	identifies	identify	VERB
abrj-637	143	14	outliers	outlier	NOUN
abrj-637	143	15	and	and	CCONJ
abrj-637	143	16	unusual	unusual	ADJ
abrj-637	143	17	activity	activity	NOUN
abrj-637	143	18	in	in	ADP
abrj-637	143	19	financial	financial	ADJ
abrj-637	143	20	transactions	transaction	NOUN
abrj-637	143	21	.	.	PUNCT
abrj-637	144	1	quantitative	quantitative	ADJ
abrj-637	144	2	compliance	compliance	NOUN
abrj-637	144	3	reports	report	VERB
abrj-637	144	4	history	history	NOUN
abrj-637	144	5	of	of	ADP
abrj-637	144	6	compliance	compliance	NOUN
abrj-637	144	7	behavior	behavior	NOUN
abrj-637	144	8	,	,	PUNCT
abrj-637	144	9	tax	tax	NOUN
abrj-637	144	10	filings	filing	NOUN
abrj-637	144	11	,	,	PUNCT
abrj-637	144	12	and	and	CCONJ
abrj-637	144	13	amendments	amendment	NOUN
abrj-637	144	14	tracks	track	VERB
abrj-637	144	15	deviations	deviation	NOUN
abrj-637	144	16	in	in	ADP
abrj-637	144	17	compliance	compliance	NOUN
abrj-637	144	18	over	over	ADP
abrj-637	144	19	time	time	NOUN
abrj-637	144	20	to	to	PART
abrj-637	144	21	flag	flag	VERB
abrj-637	144	22	possible	possible	ADJ
abrj-637	144	23	fraudulent	fraudulent	ADJ
abrj-637	144	24	activity	activity	NOUN
abrj-637	144	25	.	.	PUNCT
abrj-637	145	1	textual	textual	ADJ
abrj-637	145	2	management	management	NOUN
abrj-637	145	3	commentaries	commentary	NOUN
abrj-637	145	4	and	and	CCONJ
abrj-637	145	5	reports	report	VERB
abrj-637	145	6	natural	natural	ADJ
abrj-637	145	7	language	language	NOUN
abrj-637	145	8	processing	processing	NOUN
abrj-637	145	9	(	(	PUNCT
abrj-637	145	10	nlp	nlp	NOUN
abrj-637	145	11	)	)	PUNCT
abrj-637	145	12	for	for	ADP
abrj-637	145	13	sentiment	sentiment	NOUN
abrj-637	145	14	analysis	analysis	NOUN
abrj-637	145	15	and	and	CCONJ
abrj-637	145	16	keyword	keyword	NOUN
abrj-637	145	17	extraction	extraction	NOUN
abrj-637	145	18	identifies	identify	VERB
abrj-637	145	19	linguistic	linguistic	ADJ
abrj-637	145	20	cues	cue	NOUN
abrj-637	145	21	indicating	indicate	VERB
abrj-637	145	22	deception	deception	NOUN
abrj-637	145	23	or	or	CCONJ
abrj-637	145	24	inconsistency	inconsistency	NOUN
abrj-637	145	25	in	in	ADP
abrj-637	145	26	corporate	corporate	ADJ
abrj-637	145	27	reports	report	NOUN
abrj-637	145	28	.	.	PUNCT
abrj-637	146	1	textual	textual	ADJ
abrj-637	146	2	auditor	auditor	NOUN
abrj-637	146	3	notes	note	VERB
abrj-637	146	4	sentiment	sentiment	NOUN
abrj-637	146	5	analysis	analysis	NOUN
abrj-637	146	6	,	,	PUNCT
abrj-637	146	7	tone	tone	NOUN
abrj-637	146	8	detection	detection	NOUN
abrj-637	146	9	,	,	PUNCT
abrj-637	146	10	frequency	frequency	NOUN
abrj-637	146	11	of	of	ADP
abrj-637	146	12	conflict	conflict	NOUN
abrj-637	146	13	-	-	PUNCT
abrj-637	146	14	related	relate	VERB
abrj-637	146	15	terms	term	NOUN
abrj-637	146	16	detects	detect	NOUN
abrj-637	146	17	inconsistencies	inconsistency	NOUN
abrj-637	146	18	and	and	CCONJ
abrj-637	146	19	potential	potential	ADJ
abrj-637	146	20	fraud	fraud	NOUN
abrj-637	146	21	based	base	VERB
abrj-637	146	22	on	on	ADP
abrj-637	146	23	the	the	DET
abrj-637	146	24	tone	tone	NOUN
abrj-637	146	25	and	and	CCONJ
abrj-637	146	26	language	language	NOUN
abrj-637	146	27	used	use	VERB
abrj-637	146	28	in	in	ADP
abrj-637	146	29	auditor	auditor	NOUN
abrj-637	146	30	reports	report	NOUN
abrj-637	146	31	.	.	PUNCT
abrj-637	147	1	textual	textual	ADJ
abrj-637	147	2	executive	executive	ADJ
abrj-637	147	3	statements	statement	NOUN
abrj-637	147	4	and	and	CCONJ
abrj-637	147	5	annual	annual	ADJ
abrj-637	147	6	reports	report	NOUN
abrj-637	147	7	textual	textual	ADJ
abrj-637	147	8	feature	feature	NOUN
abrj-637	147	9	extraction	extraction	NOUN
abrj-637	147	10	using	use	VERB
abrj-637	147	11	tf	tf	PROPN
abrj-637	147	12	-	-	PUNCT
abrj-637	147	13	idf	idf	PROPN
abrj-637	147	14	and	and	CCONJ
abrj-637	147	15	syntactic	syntactic	ADJ
abrj-637	147	16	analysis	analysis	NOUN
abrj-637	147	17	analyzes	analyze	VERB
abrj-637	147	18	text	text	NOUN
abrj-637	147	19	complexity	complexity	NOUN
abrj-637	147	20	and	and	CCONJ
abrj-637	147	21	semantic	semantic	ADJ
abrj-637	147	22	structure	structure	NOUN
abrj-637	147	23	to	to	PART
abrj-637	147	24	detect	detect	VERB
abrj-637	147	25	fraud	fraud	NOUN
abrj-637	147	26	-	-	PUNCT
abrj-637	147	27	related	relate	VERB
abrj-637	147	28	signals	signal	NOUN
abrj-637	147	29	.	.	PUNCT
abrj-637	148	1	asian	asian	ADJ
abrj-637	148	2	business	business	PROPN
abrj-637	148	3	research	research	NOUN
abrj-637	148	4	journal	journal	NOUN
abrj-637	148	5	,	,	PUNCT
abrj-637	148	6	2025	2025	NUM
abrj-637	148	7	,	,	PUNCT
abrj-637	148	8	10(11	10(11	NUM
abrj-637	148	9	):	):	PUNCT
abrj-637	148	10	1	1	NUM
abrj-637	148	11	-	-	SYM
abrj-637	148	12	13	13	NUM
abrj-637	148	13	7	7	NUM
abrj-637	148	14	©	©	ADP
abrj-637	148	15	2025	2025	NUM
abrj-637	148	16	by	by	ADP
abrj-637	148	17	the	the	DET
abrj-637	148	18	authors	author	NOUN
abrj-637	148	19	;	;	PUNCT
abrj-637	148	20	licensee	licensee	NOUN
abrj-637	148	21	eastern	eastern	ADJ
abrj-637	148	22	centre	centre	NOUN
abrj-637	148	23	of	of	ADP
abrj-637	148	24	science	science	NOUN
abrj-637	148	25	and	and	CCONJ
abrj-637	148	26	education	education	NOUN
abrj-637	148	27	,	,	PUNCT
abrj-637	148	28	usa	usa	PROPN
abrj-637	148	29	the	the	DET
abrj-637	148	30	following	follow	VERB
abrj-637	148	31	table	table	NOUN
abrj-637	148	32	defines	define	VERB
abrj-637	148	33	the	the	DET
abrj-637	148	34	quantitative	quantitative	ADJ
abrj-637	148	35	and	and	CCONJ
abrj-637	148	36	textual	textual	ADJ
abrj-637	148	37	properties	property	NOUN
abrj-637	148	38	that	that	PRON
abrj-637	148	39	were	be	AUX
abrj-637	148	40	obtained	obtain	VERB
abrj-637	148	41	in	in	ADP
abrj-637	148	42	the	the	DET
abrj-637	148	43	process	process	NOUN
abrj-637	148	44	of	of	ADP
abrj-637	148	45	data	datum	NOUN
abrj-637	148	46	preprocessing	preprocessing	NOUN
abrj-637	148	47	and	and	CCONJ
abrj-637	148	48	feature	feature	NOUN
abrj-637	148	49	engineering	engineering	NOUN
abrj-637	148	50	in	in	ADP
abrj-637	148	51	the	the	DET
abrj-637	148	52	fraud	fraud	NOUN
abrj-637	148	53	detection	detection	NOUN
abrj-637	148	54	pipeline	pipeline	NOUN
abrj-637	148	55	.	.	PUNCT
abrj-637	149	1	these	these	DET
abrj-637	149	2	characteristics	characteristic	NOUN
abrj-637	149	3	play	play	VERB
abrj-637	149	4	an	an	DET
abrj-637	149	5	important	important	ADJ
abrj-637	149	6	role	role	NOUN
abrj-637	149	7	in	in	ADP
abrj-637	149	8	improving	improve	VERB
abrj-637	149	9	predictive	predictive	ADJ
abrj-637	149	10	power	power	NOUN
abrj-637	149	11	of	of	ADP
abrj-637	149	12	machine	machine	NOUN
abrj-637	149	13	learning	learning	NOUN
abrj-637	149	14	models	model	NOUN
abrj-637	149	15	,	,	PUNCT
abrj-637	149	16	as	as	SCONJ
abrj-637	149	17	it	it	PRON
abrj-637	149	18	offers	offer	VERB
abrj-637	149	19	both	both	DET
abrj-637	149	20	numerical	numerical	ADJ
abrj-637	149	21	and	and	CCONJ
abrj-637	149	22	textual	textual	ADJ
abrj-637	149	23	information	information	NOUN
abrj-637	149	24	.	.	PUNCT
abrj-637	150	1	3.5	3.5	NUM
abrj-637	150	2	.	.	PUNCT
abrj-637	150	3	model	model	PROPN
abrj-637	150	4	development	development	PROPN
abrj-637	150	5	the	the	DET
abrj-637	150	6	model	model	NOUN
abrj-637	150	7	development	development	NOUN
abrj-637	150	8	is	be	AUX
abrj-637	150	9	a	a	DET
abrj-637	150	10	blend	blend	NOUN
abrj-637	150	11	of	of	ADP
abrj-637	150	12	the	the	DET
abrj-637	150	13	supervised	supervised	ADJ
abrj-637	150	14	and	and	CCONJ
abrj-637	150	15	unsupervised	unsupervised	ADJ
abrj-637	150	16	and	and	CCONJ
abrj-637	150	17	ensemble	ensemble	ADJ
abrj-637	150	18	methods	method	NOUN
abrj-637	150	19	.	.	PUNCT
abrj-637	151	1	model	model	NOUN
abrj-637	151	2	(	(	PUNCT
abrj-637	151	3	e.g.	e.g.	ADV
abrj-637	151	4	logistic	logistic	ADJ
abrj-637	151	5	regression	regression	NOUN
abrj-637	151	6	,	,	PUNCT
abrj-637	151	7	random	random	ADJ
abrj-637	151	8	forest	forest	NOUN
abrj-637	151	9	,	,	PUNCT
abrj-637	151	10	xgboost	xgboost	PROPN
abrj-637	151	11	)	)	PUNCT
abrj-637	151	12	,	,	PUNCT
abrj-637	151	13	which	which	PRON
abrj-637	151	14	is	be	AUX
abrj-637	151	15	monitored	monitor	VERB
abrj-637	151	16	,	,	PUNCT
abrj-637	151	17	has	have	VERB
abrj-637	151	18	an	an	DET
abrj-637	151	19	interpretation	interpretation	NOUN
abrj-637	151	20	ability	ability	NOUN
abrj-637	151	21	and	and	CCONJ
abrj-637	151	22	provides	provide	VERB
abrj-637	151	23	a	a	DET
abrj-637	151	24	strong	strong	ADJ
abrj-637	151	25	classification	classification	NOUN
abrj-637	151	26	(	(	PUNCT
abrj-637	151	27	nguyen	nguyen	NOUN
abrj-637	151	28	,	,	PUNCT
abrj-637	151	29	2025	2025	NUM
abrj-637	151	30	;	;	PUNCT
abrj-637	151	31	acharya	acharya	NOUN
abrj-637	151	32	,	,	PUNCT
abrj-637	151	33	2025	2025	NUM
abrj-637	151	34	)	)	PUNCT
abrj-637	151	35	.	.	PUNCT
abrj-637	152	1	unsupervised	unsupervised	ADJ
abrj-637	152	2	approaches	approach	NOUN
abrj-637	152	3	(	(	PUNCT
abrj-637	152	4	e.g.	e.g.	ADV
abrj-637	152	5	,	,	PUNCT
abrj-637	152	6	autoencoders	autoencoder	NOUN
abrj-637	152	7	,	,	PUNCT
abrj-637	152	8	isolation	isolation	NOUN
abrj-637	152	9	forests	forest	NOUN
abrj-637	152	10	)	)	PUNCT
abrj-637	152	11	identify	identify	VERB
abrj-637	152	12	new	new	ADJ
abrj-637	152	13	malpractices	malpractice	NOUN
abrj-637	152	14	in	in	ADP
abrj-637	152	15	unlabeled	unlabeled	ADJ
abrj-637	152	16	data	datum	NOUN
abrj-637	152	17	(	(	PUNCT
abrj-637	152	18	zhou	zhou	X
abrj-637	152	19	et	et	PROPN
abrj-637	152	20	al	al	PROPN
abrj-637	152	21	.	.	PROPN
abrj-637	152	22	,	,	PUNCT
abrj-637	152	23	2024	2024	NUM
abrj-637	152	24	)	)	PUNCT
abrj-637	152	25	.	.	PUNCT
abrj-637	153	1	hybrid	hybrid	ADJ
abrj-637	153	2	ones	one	NOUN
abrj-637	153	3	combine	combine	VERB
abrj-637	153	4	the	the	DET
abrj-637	153	5	paradigms	paradigm	NOUN
abrj-637	153	6	to	to	PART
abrj-637	153	7	enhance	enhance	VERB
abrj-637	153	8	the	the	DET
abrj-637	153	9	process	process	NOUN
abrj-637	153	10	of	of	ADP
abrj-637	153	11	generalization	generalization	NOUN
abrj-637	153	12	(	(	PUNCT
abrj-637	153	13	shujaaddeen	shujaaddeen	PROPN
abrj-637	153	14	et	et	PROPN
abrj-637	153	15	al	al	PROPN
abrj-637	153	16	.	.	PROPN
abrj-637	153	17	,	,	PUNCT
abrj-637	153	18	2024	2024	NUM
abrj-637	153	19	;	;	PUNCT
abrj-637	153	20	mehta	mehta	PROPN
abrj-637	153	21	et	et	PROPN
abrj-637	153	22	al	al	PROPN
abrj-637	153	23	.	.	PROPN
abrj-637	153	24	,	,	PUNCT
abrj-637	153	25	2022	2022	NUM
abrj-637	153	26	)	)	PUNCT
abrj-637	153	27	.	.	PUNCT
abrj-637	154	1	accuracy	accuracy	NOUN
abrj-637	154	2	is	be	AUX
abrj-637	154	3	further	far	ADV
abrj-637	154	4	improved	improve	VERB
abrj-637	154	5	with	with	ADP
abrj-637	154	6	ensemble	ensemble	ADJ
abrj-637	154	7	models	model	NOUN
abrj-637	154	8	which	which	PRON
abrj-637	154	9	utilize	utilize	VERB
abrj-637	154	10	the	the	DET
abrj-637	154	11	soft	soft	ADJ
abrj-637	154	12	-	-	PUNCT
abrj-637	154	13	voting	voting	NOUN
abrj-637	154	14	and	and	CCONJ
abrj-637	154	15	stacking	stacking	NOUN
abrj-637	154	16	(	(	PUNCT
abrj-637	154	17	zhang	zhang	PROPN
abrj-637	154	18	et	et	PROPN
abrj-637	154	19	al	al	PROPN
abrj-637	154	20	.	.	PROPN
abrj-637	154	21	,	,	PUNCT
abrj-637	154	22	2025	2025	NUM
abrj-637	154	23	;	;	PUNCT
abrj-637	154	24	zhou	zhou	PROPN
abrj-637	154	25	et	et	PROPN
abrj-637	154	26	al	al	PROPN
abrj-637	154	27	.	.	PROPN
abrj-637	154	28	,	,	PUNCT
abrj-637	154	29	2024	2024	NUM
abrj-637	154	30	)	)	PUNCT
abrj-637	154	31	.	.	PUNCT
abrj-637	155	1	kfold	kfold	PROPN
abrj-637	155	2	cross	cross	NOUN
abrj-637	155	3	-	-	NOUN
abrj-637	155	4	validation	validation	NOUN
abrj-637	155	5	is	be	AUX
abrj-637	155	6	applied	apply	VERB
abrj-637	155	7	on	on	ADP
abrj-637	155	8	each	each	DET
abrj-637	155	9	model	model	NOUN
abrj-637	155	10	to	to	PART
abrj-637	155	11	make	make	VERB
abrj-637	155	12	them	they	PRON
abrj-637	155	13	robust	robust	ADJ
abrj-637	155	14	and	and	CCONJ
abrj-637	155	15	avoid	avoid	VERB
abrj-637	155	16	overfitting	overfitte	VERB
abrj-637	155	17	.	.	PUNCT
abrj-637	156	1	3.6	3.6	NUM
abrj-637	156	2	.	.	PUNCT
abrj-637	156	3	risk	risk	NOUN
abrj-637	156	4	scoring	scoring	NOUN
abrj-637	156	5	framework	framework	NOUN
abrj-637	156	6	a	a	DET
abrj-637	156	7	risk	risk	NOUN
abrj-637	156	8	scoring	scoring	NOUN
abrj-637	156	9	mechanism	mechanism	NOUN
abrj-637	156	10	is	be	AUX
abrj-637	156	11	a	a	DET
abrj-637	156	12	probabilistic	probabilistic	ADJ
abrj-637	156	13	model	model	NOUN
abrj-637	156	14	that	that	PRON
abrj-637	156	15	transforms	transform	VERB
abrj-637	156	16	model	model	NOUN
abrj-637	156	17	outputs	output	NOUN
abrj-637	156	18	to	to	PART
abrj-637	156	19	interpretable	interpretable	ADJ
abrj-637	156	20	signs	sign	NOUN
abrj-637	156	21	to	to	ADP
abrj-637	156	22	the	the	DET
abrj-637	156	23	auditors	auditor	NOUN
abrj-637	156	24	.	.	PUNCT
abrj-637	157	1	the	the	DET
abrj-637	157	2	risks	risk	NOUN
abrj-637	157	3	of	of	ADP
abrj-637	157	4	fraud	fraud	NOUN
abrj-637	157	5	are	be	AUX
abrj-637	157	6	plotted	plot	VERB
abrj-637	157	7	on	on	ADP
abrj-637	157	8	a	a	DET
abrj-637	157	9	0	0	NUM
abrj-637	157	10	-	-	SYM
abrj-637	157	11	1	1	NUM
abrj-637	157	12	scale	scale	NOUN
abrj-637	157	13	:	:	PUNCT
abrj-637	157	14	0.00–0.30	0.00–0.30	NUM
abrj-637	157	15	(	(	PUNCT
abrj-637	157	16	low	low	ADJ
abrj-637	157	17	risk	risk	NOUN
abrj-637	157	18	)	)	PUNCT
abrj-637	157	19	0.31–0.70	0.31–0.70	NOUN
abrj-637	158	1	(	(	PUNCT
abrj-637	158	2	medium	medium	ADJ
abrj-637	158	3	risk	risk	NOUN
abrj-637	158	4	)	)	PUNCT
abrj-637	158	5	0.71–1.00	0.71–1.00	NUM
abrj-637	159	1	(	(	PUNCT
abrj-637	159	2	high	high	ADJ
abrj-637	159	3	risk	risk	NOUN
abrj-637	159	4	)	)	PUNCT
abrj-637	159	5	the	the	DET
abrj-637	159	6	system	system	NOUN
abrj-637	159	7	takes	take	VERB
abrj-637	159	8	the	the	DET
abrj-637	159	9	results	result	NOUN
abrj-637	159	10	of	of	ADP
abrj-637	159	11	the	the	DET
abrj-637	159	12	ensemble	ensemble	NOUN
abrj-637	159	13	of	of	ADP
abrj-637	159	14	classifiers	classifier	NOUN
abrj-637	159	15	and	and	CCONJ
abrj-637	159	16	anomaly	anomaly	NOUN
abrj-637	159	17	detectors	detector	NOUN
abrj-637	159	18	and	and	CCONJ
abrj-637	159	19	weighs	weigh	NOUN
abrj-637	159	20	by	by	ADP
abrj-637	159	21	the	the	DET
abrj-637	159	22	confidence	confidence	NOUN
abrj-637	159	23	level	level	NOUN
abrj-637	159	24	(	(	PUNCT
abrj-637	159	25	choudhary	choudhary	PROPN
abrj-637	159	26	,	,	PUNCT
abrj-637	159	27	2025	2025	NUM
abrj-637	159	28	;	;	PUNCT
abrj-637	159	29	zhou	zhou	PROPN
abrj-637	159	30	et	et	PROPN
abrj-637	159	31	al	al	PROPN
abrj-637	159	32	.	.	PROPN
abrj-637	159	33	,	,	PUNCT
abrj-637	159	34	2024	2024	NUM
abrj-637	159	35	)	)	PUNCT
abrj-637	159	36	.	.	PUNCT
abrj-637	160	1	this	this	DET
abrj-637	160	2	aids	aid	NOUN
abrj-637	160	3	in	in	ADP
abrj-637	160	4	prioritization	prioritization	NOUN
abrj-637	160	5	of	of	ADP
abrj-637	160	6	high	high	ADJ
abrj-637	160	7	-	-	PUNCT
abrj-637	160	8	risk	risk	NOUN
abrj-637	160	9	filings	filing	NOUN
abrj-637	160	10	to	to	ADP
abrj-637	160	11	the	the	DET
abrj-637	160	12	further	further	ADJ
abrj-637	160	13	audit	audit	NOUN
abrj-637	160	14	.	.	PUNCT
abrj-637	161	1	figure	figure	VERB
abrj-637	161	2	4	4	NUM
abrj-637	161	3	.	.	PUNCT
abrj-637	161	4	decision	decision	NOUN
abrj-637	161	5	-	-	PUNCT
abrj-637	161	6	making	make	VERB
abrj-637	161	7	framework	framework	NOUN
abrj-637	161	8	and	and	CCONJ
abrj-637	161	9	risk	risk	NOUN
abrj-637	161	10	scoring	scoring	NOUN
abrj-637	161	11	.	.	PUNCT
abrj-637	162	1	3.7	3.7	NUM
abrj-637	162	2	.	.	PUNCT
abrj-637	162	3	model	model	NOUN
abrj-637	162	4	evaluation	evaluation	NOUN
abrj-637	162	5	metrics	metric	NOUN
abrj-637	162	6	the	the	DET
abrj-637	162	7	measures	measure	NOUN
abrj-637	162	8	of	of	ADP
abrj-637	162	9	evaluation	evaluation	NOUN
abrj-637	162	10	are	be	AUX
abrj-637	162	11	precision	precision	NOUN
abrj-637	162	12	,	,	PUNCT
abrj-637	162	13	recall	recall	NOUN
abrj-637	162	14	,	,	PUNCT
abrj-637	162	15	f1	f1	NOUN
abrj-637	162	16	-	-	PUNCT
abrj-637	162	17	score	score	NOUN
abrj-637	162	18	,	,	PUNCT
abrj-637	162	19	and	and	CCONJ
abrj-637	162	20	roc	roc	NOUN
abrj-637	162	21	-	-	PUNCT
abrj-637	162	22	auc	auc	NOUN
abrj-637	162	23	which	which	PRON
abrj-637	162	24	are	be	AUX
abrj-637	162	25	used	use	VERB
abrj-637	162	26	to	to	PART
abrj-637	162	27	measure	measure	VERB
abrj-637	162	28	performance	performance	NOUN
abrj-637	162	29	of	of	ADP
abrj-637	162	30	classification	classification	NOUN
abrj-637	162	31	.	.	PUNCT
abrj-637	163	1	the	the	DET
abrj-637	163	2	confusion	confusion	NOUN
abrj-637	163	3	matrices	matrix	NOUN
abrj-637	163	4	are	be	AUX
abrj-637	163	5	used	use	VERB
abrj-637	163	6	to	to	PART
abrj-637	163	7	identify	identify	VERB
abrj-637	163	8	misclassifications	misclassification	NOUN
abrj-637	163	9	and	and	CCONJ
abrj-637	163	10	scores	score	NOUN
abrj-637	163	11	of	of	ADP
abrj-637	163	12	crossvalidation	crossvalidation	NOUN
abrj-637	163	13	are	be	AUX
abrj-637	163	14	used	use	VERB
abrj-637	163	15	to	to	PART
abrj-637	163	16	verify	verify	VERB
abrj-637	163	17	that	that	SCONJ
abrj-637	163	18	the	the	DET
abrj-637	163	19	model	model	NOUN
abrj-637	163	20	is	be	AUX
abrj-637	163	21	generalizable	generalizable	ADJ
abrj-637	163	22	(	(	PUNCT
abrj-637	163	23	martínez	martínez	NOUN
abrj-637	163	24	,	,	PUNCT
abrj-637	163	25	2025	2025	NUM
abrj-637	163	26	;	;	PUNCT
abrj-637	163	27	wahyono	wahyono	PROPN
abrj-637	163	28	and	and	CCONJ
abrj-637	163	29	david	david	PROPN
abrj-637	163	30	,	,	PUNCT
abrj-637	163	31	2025	2025	NUM
abrj-637	163	32	)	)	PUNCT
abrj-637	163	33	.	.	PUNCT
abrj-637	164	1	to	to	PART
abrj-637	164	2	apply	apply	VERB
abrj-637	164	3	the	the	DET
abrj-637	164	4	model	model	NOUN
abrj-637	164	5	predictions	prediction	NOUN
abrj-637	164	6	,	,	PUNCT
abrj-637	164	7	shap	shap	PROPN
abrj-637	164	8	values	value	NOUN
abrj-637	164	9	are	be	AUX
abrj-637	164	10	used	use	VERB
abrj-637	164	11	,	,	PUNCT
abrj-637	164	12	and	and	CCONJ
abrj-637	164	13	this	this	PRON
abrj-637	164	14	is	be	AUX
abrj-637	164	15	to	to	PART
abrj-637	164	16	ensure	ensure	VERB
abrj-637	164	17	that	that	SCONJ
abrj-637	164	18	the	the	DET
abrj-637	164	19	predictions	prediction	NOUN
abrj-637	164	20	remain	remain	VERB
abrj-637	164	21	within	within	ADP
abrj-637	164	22	the	the	DET
abrj-637	164	23	scope	scope	NOUN
abrj-637	164	24	of	of	ADP
abrj-637	164	25	the	the	DET
abrj-637	164	26	ethical	ethical	ADJ
abrj-637	164	27	standards	standard	NOUN
abrj-637	164	28	(	(	PUNCT
abrj-637	164	29	zhou	zhou	X
abrj-637	164	30	et	et	PROPN
abrj-637	164	31	al	al	PROPN
abrj-637	164	32	.	.	PROPN
abrj-637	164	33	,	,	PUNCT
abrj-637	164	34	2024	2024	NUM
abrj-637	164	35	;	;	PUNCT
abrj-637	164	36	tagbo	tagbo	NOUN
abrj-637	164	37	and	and	CCONJ
abrj-637	164	38	adekoya	adekoya	NOUN
abrj-637	164	39	,	,	PUNCT
abrj-637	164	40	2023	2023	NUM
abrj-637	164	41	)	)	PUNCT
abrj-637	164	42	.	.	PUNCT
abrj-637	165	1	3.8	3.8	NUM
abrj-637	165	2	.	.	PUNCT
abrj-637	166	1	legal	legal	ADJ
abrj-637	166	2	and	and	CCONJ
abrj-637	166	3	ethical	ethical	ADJ
abrj-637	166	4	concerns	concern	NOUN
abrj-637	166	5	.	.	PUNCT
abrj-637	167	1	the	the	DET
abrj-637	167	2	model	model	NOUN
abrj-637	167	3	follows	follow	VERB
abrj-637	167	4	the	the	DET
abrj-637	167	5	principles	principle	NOUN
abrj-637	167	6	of	of	ADP
abrj-637	167	7	privacy	privacy	NOUN
abrj-637	167	8	,	,	PUNCT
abrj-637	167	9	fairness	fairness	NOUN
abrj-637	167	10	,	,	PUNCT
abrj-637	167	11	and	and	CCONJ
abrj-637	167	12	transparency	transparency	NOUN
abrj-637	167	13	.	.	PUNCT
abrj-637	168	1	data	datum	NOUN
abrj-637	168	2	on	on	ADP
abrj-637	168	3	taxpayers	taxpayer	NOUN
abrj-637	168	4	is	be	AUX
abrj-637	168	5	made	make	VERB
abrj-637	168	6	anonymous	anonymous	ADJ
abrj-637	168	7	and	and	CCONJ
abrj-637	168	8	can	can	AUX
abrj-637	168	9	be	be	AUX
abrj-637	168	10	processed	process	VERB
abrj-637	168	11	according	accord	VERB
abrj-637	168	12	to	to	ADP
abrj-637	168	13	the	the	DET
abrj-637	168	14	gdpr	gdpr	NOUN
abrj-637	168	15	and	and	CCONJ
abrj-637	168	16	other	other	ADJ
abrj-637	168	17	laws	law	NOUN
abrj-637	168	18	on	on	ADP
abrj-637	168	19	data	datum	NOUN
abrj-637	168	20	protection	protection	NOUN
abrj-637	168	21	.	.	PUNCT
abrj-637	169	1	discrimination	discrimination	NOUN
abrj-637	169	2	against	against	ADP
abrj-637	169	3	the	the	DET
abrj-637	169	4	results	result	NOUN
abrj-637	169	5	is	be	AUX
abrj-637	169	6	avoided	avoid	VERB
abrj-637	169	7	by	by	ADP
abrj-637	169	8	integrating	integrate	VERB
abrj-637	169	9	bias	bias	NOUN
abrj-637	169	10	detection	detection	NOUN
abrj-637	169	11	systems	system	NOUN
abrj-637	169	12	(	(	PUNCT
abrj-637	169	13	wahyono	wahyono	PROPN
abrj-637	169	14	and	and	CCONJ
abrj-637	169	15	david	david	PROPN
abrj-637	169	16	,	,	PUNCT
abrj-637	169	17	2025	2025	NUM
abrj-637	169	18	;	;	PUNCT
abrj-637	169	19	breslin	breslin	PROPN
abrj-637	169	20	,	,	PUNCT
abrj-637	169	21	2021	2021	NUM
abrj-637	169	22	)	)	PUNCT
abrj-637	169	23	.	.	PUNCT
abrj-637	170	1	explainable	explainable	ADJ
abrj-637	170	2	ai	ai	PROPN
abrj-637	170	3	(	(	PUNCT
abrj-637	170	4	xai	xai	PROPN
abrj-637	170	5	)	)	PUNCT
abrj-637	170	6	should	should	AUX
abrj-637	170	7	also	also	ADV
abrj-637	170	8	be	be	AUX
abrj-637	170	9	integrated	integrate	VERB
abrj-637	170	10	to	to	PART
abrj-637	170	11	enable	enable	VERB
abrj-637	170	12	accountability	accountability	NOUN
abrj-637	170	13	and	and	CCONJ
abrj-637	170	14	trust	trust	NOUN
abrj-637	170	15	as	as	SCONJ
abrj-637	170	16	the	the	DET
abrj-637	170	17	auditors	auditor	NOUN
abrj-637	170	18	can	can	AUX
abrj-637	170	19	audit	audit	VERB
abrj-637	170	20	the	the	DET
abrj-637	170	21	model	model	NOUN
abrj-637	170	22	decision	decision	NOUN
abrj-637	170	23	(	(	PUNCT
abrj-637	170	24	zhou	zhou	X
abrj-637	170	25	et	et	PROPN
abrj-637	170	26	al	al	PROPN
abrj-637	170	27	.	.	PROPN
abrj-637	170	28	,	,	PUNCT
abrj-637	170	29	2024	2024	NUM
abrj-637	170	30	)	)	PUNCT
abrj-637	170	31	.	.	PUNCT
abrj-637	171	1	3.9	3.9	NUM
abrj-637	171	2	.	.	PUNCT
abrj-637	171	3	summary	summary	VERB
abrj-637	171	4	the	the	DET
abrj-637	171	5	approach	approach	NOUN
abrj-637	171	6	combines	combine	VERB
abrj-637	171	7	the	the	DET
abrj-637	171	8	data	data	NOUN
abrj-637	171	9	-	-	PUNCT
abrj-637	171	10	driven	drive	VERB
abrj-637	171	11	innovation	innovation	NOUN
abrj-637	171	12	with	with	ADP
abrj-637	171	13	ethical	ethical	ADJ
abrj-637	171	14	governance	governance	NOUN
abrj-637	171	15	in	in	ADP
abrj-637	171	16	an	an	DET
abrj-637	171	17	attempt	attempt	NOUN
abrj-637	171	18	to	to	PART
abrj-637	171	19	deliver	deliver	VERB
abrj-637	171	20	a	a	DET
abrj-637	171	21	scalable	scalable	ADJ
abrj-637	171	22	,	,	PUNCT
abrj-637	171	23	transparent	transparent	ADJ
abrj-637	171	24	,	,	PUNCT
abrj-637	171	25	and	and	CCONJ
abrj-637	171	26	accurate	accurate	ADJ
abrj-637	171	27	tax	tax	NOUN
abrj-637	171	28	fraud	fraud	NOUN
abrj-637	171	29	detection	detection	NOUN
abrj-637	171	30	system	system	NOUN
abrj-637	171	31	.	.	PUNCT
abrj-637	172	1	the	the	DET
abrj-637	172	2	framework	framework	NOUN
abrj-637	172	3	advances	advance	VERB
abrj-637	172	4	the	the	DET
abrj-637	172	5	existing	exist	VERB
abrj-637	172	6	literature	literature	NOUN
abrj-637	172	7	on	on	ADP
abrj-637	172	8	the	the	DET
abrj-637	172	9	topic	topic	NOUN
abrj-637	172	10	asian	asian	PROPN
abrj-637	172	11	business	business	PROPN
abrj-637	172	12	research	research	NOUN
abrj-637	172	13	journal	journal	NOUN
abrj-637	172	14	,	,	PUNCT
abrj-637	172	15	2025	2025	NUM
abrj-637	172	16	,	,	PUNCT
abrj-637	172	17	10(11	10(11	NUM
abrj-637	172	18	):	):	PUNCT
abrj-637	172	19	1	1	NUM
abrj-637	172	20	-	-	SYM
abrj-637	172	21	13	13	NUM
abrj-637	172	22	8	8	NUM
abrj-637	172	23	©	©	ADP
abrj-637	172	24	2025	2025	NUM
abrj-637	172	25	by	by	ADP
abrj-637	172	26	the	the	DET
abrj-637	172	27	authors	author	NOUN
abrj-637	172	28	;	;	PUNCT
abrj-637	172	29	licensee	licensee	NOUN
abrj-637	172	30	eastern	eastern	ADJ
abrj-637	172	31	centre	centre	NOUN
abrj-637	172	32	of	of	ADP
abrj-637	172	33	science	science	NOUN
abrj-637	172	34	and	and	CCONJ
abrj-637	172	35	education	education	NOUN
abrj-637	172	36	,	,	PUNCT
abrj-637	172	37	usa	usa	PROPN
abrj-637	172	38	(	(	PUNCT
abrj-637	172	39	acharya	acharya	PROPN
abrj-637	172	40	,	,	PUNCT
abrj-637	172	41	2025	2025	NUM
abrj-637	172	42	;	;	PUNCT
abrj-637	172	43	zhang	zhang	PROPN
abrj-637	172	44	et	et	PROPN
abrj-637	172	45	al	al	PROPN
abrj-637	172	46	.	.	PROPN
abrj-637	172	47	,	,	PUNCT
abrj-637	172	48	2025	2025	NUM
abrj-637	172	49	;	;	PUNCT
abrj-637	172	50	zhou	zhou	PROPN
abrj-637	172	51	et	et	PROPN
abrj-637	172	52	al	al	PROPN
abrj-637	172	53	.	.	PROPN
abrj-637	172	54	,	,	PUNCT
abrj-637	172	55	2024	2024	NUM
abrj-637	172	56	)	)	PUNCT
abrj-637	172	57	by	by	ADP
abrj-637	172	58	integrating	integrate	VERB
abrj-637	172	59	ensemble	ensemble	ADJ
abrj-637	172	60	learning	learning	NOUN
abrj-637	172	61	,	,	PUNCT
abrj-637	172	62	text	text	NOUN
abrj-637	172	63	analytics	analytic	NOUN
abrj-637	172	64	,	,	PUNCT
abrj-637	172	65	and	and	CCONJ
abrj-637	172	66	risk	risk	NOUN
abrj-637	172	67	scoring	scoring	NOUN
abrj-637	172	68	and	and	CCONJ
abrj-637	172	69	focusing	focus	VERB
abrj-637	172	70	on	on	ADP
abrj-637	172	71	the	the	DET
abrj-637	172	72	applicability	applicability	NOUN
abrj-637	172	73	and	and	CCONJ
abrj-637	172	74	interpretability	interpretability	NOUN
abrj-637	172	75	of	of	ADP
abrj-637	172	76	the	the	DET
abrj-637	172	77	practices	practice	NOUN
abrj-637	172	78	to	to	ADP
abrj-637	172	79	the	the	DET
abrj-637	172	80	real	real	ADJ
abrj-637	172	81	-	-	PUNCT
abrj-637	172	82	world	world	NOUN
abrj-637	172	83	regulators	regulator	NOUN
abrj-637	172	84	.	.	PUNCT
abrj-637	173	1	system	system	NOUN
abrj-637	173	2	architecture	architecture	NOUN
abrj-637	173	3	is	be	AUX
abrj-637	173	4	an	an	DET
abrj-637	173	5	overview	overview	NOUN
abrj-637	173	6	of	of	ADP
abrj-637	173	7	the	the	DET
abrj-637	173	8	entire	entire	ADJ
abrj-637	173	9	system	system	NOUN
abrj-637	173	10	,	,	PUNCT
abrj-637	173	11	including	include	VERB
abrj-637	173	12	its	its	PRON
abrj-637	173	13	design	design	NOUN
abrj-637	173	14	,	,	PUNCT
abrj-637	173	15	implementation	implementation	NOUN
abrj-637	173	16	,	,	PUNCT
abrj-637	173	17	and	and	CCONJ
abrj-637	173	18	testing	testing	NOUN
abrj-637	173	19	processes	process	NOUN
abrj-637	173	20	.	.	PUNCT
abrj-637	174	1	3.10	3.10	NUM
abrj-637	174	2	.	.	PUNCT
abrj-637	174	3	system	system	NOUN
abrj-637	174	4	architecture	architecture	NOUN
abrj-637	174	5	overview	overview	NOUN
abrj-637	174	6	system	system	NOUN
abrj-637	174	7	architecture	architecture	NOUN
abrj-637	174	8	is	be	AUX
abrj-637	174	9	a	a	DET
abrj-637	174	10	description	description	NOUN
abrj-637	174	11	of	of	ADP
abrj-637	174	12	the	the	DET
abrj-637	174	13	whole	whole	ADJ
abrj-637	174	14	system	system	NOUN
abrj-637	174	15	,	,	PUNCT
abrj-637	174	16	both	both	CCONJ
abrj-637	174	17	in	in	ADP
abrj-637	174	18	terms	term	NOUN
abrj-637	174	19	of	of	ADP
abrj-637	174	20	design	design	NOUN
abrj-637	174	21	,	,	PUNCT
abrj-637	174	22	implementation	implementation	NOUN
abrj-637	174	23	and	and	CCONJ
abrj-637	174	24	testing	test	VERB
abrj-637	174	25	the	the	DET
abrj-637	174	26	suggested	suggest	VERB
abrj-637	174	27	system	system	NOUN
abrj-637	174	28	design	design	NOUN
abrj-637	174	29	is	be	AUX
abrj-637	174	30	modular	modular	ADJ
abrj-637	174	31	as	as	SCONJ
abrj-637	174	32	it	it	PRON
abrj-637	174	33	consists	consist	VERB
abrj-637	174	34	of	of	ADP
abrj-637	174	35	four	four	NUM
abrj-637	174	36	layers	layer	NOUN
abrj-637	174	37	of	of	ADP
abrj-637	174	38	significance	significance	NOUN
abrj-637	174	39	:	:	PUNCT
abrj-637	174	40	data	datum	NOUN
abrj-637	174	41	ingestion	ingestion	NOUN
abrj-637	174	42	,	,	PUNCT
abrj-637	174	43	data	datum	NOUN
abrj-637	174	44	processing	processing	NOUN
abrj-637	174	45	and	and	CCONJ
abrj-637	174	46	analytics	analytic	NOUN
abrj-637	174	47	,	,	PUNCT
abrj-637	174	48	risk	risk	NOUN
abrj-637	174	49	scoring	scoring	NOUN
abrj-637	174	50	and	and	CCONJ
abrj-637	174	51	interpretation	interpretation	NOUN
abrj-637	174	52	,	,	PUNCT
abrj-637	174	53	and	and	CCONJ
abrj-637	174	54	visualization	visualization	NOUN
abrj-637	174	55	and	and	CCONJ
abrj-637	174	56	reporting	reporting	NOUN
abrj-637	174	57	.	.	PUNCT
abrj-637	175	1	the	the	DET
abrj-637	175	2	former	former	ADJ
abrj-637	175	3	layer	layer	NOUN
abrj-637	175	4	,	,	PUNCT
abrj-637	175	5	data	data	NOUN
abrj-637	175	6	ingestion	ingestion	NOUN
abrj-637	175	7	,	,	PUNCT
abrj-637	175	8	gathers	gather	VERB
abrj-637	175	9	structured	structure	VERB
abrj-637	175	10	and	and	CCONJ
abrj-637	175	11	unstructured	unstructured	ADJ
abrj-637	175	12	data	datum	NOUN
abrj-637	175	13	of	of	ADP
abrj-637	175	14	various	various	ADJ
abrj-637	175	15	sources	source	NOUN
abrj-637	175	16	including	include	VERB
abrj-637	175	17	corporation	corporation	NOUN
abrj-637	175	18	financial	financial	ADJ
abrj-637	175	19	reports	report	NOUN
abrj-637	175	20	and	and	CCONJ
abrj-637	175	21	compliance	compliance	NOUN
abrj-637	175	22	reports	report	NOUN
abrj-637	175	23	,	,	PUNCT
abrj-637	175	24	as	as	ADV
abrj-637	175	25	well	well	ADV
abrj-637	175	26	as	as	ADP
abrj-637	175	27	external	external	ADJ
abrj-637	175	28	databases	database	NOUN
abrj-637	175	29	.	.	PUNCT
abrj-637	176	1	these	these	DET
abrj-637	176	2	data	datum	NOUN
abrj-637	176	3	are	be	AUX
abrj-637	176	4	then	then	ADV
abrj-637	176	5	subjected	subject	VERB
abrj-637	176	6	to	to	ADP
abrj-637	176	7	a	a	DET
abrj-637	176	8	processing	processing	NOUN
abrj-637	176	9	and	and	CCONJ
abrj-637	176	10	analytics	analytic	NOUN
abrj-637	176	11	layer	layer	NOUN
abrj-637	176	12	where	where	SCONJ
abrj-637	176	13	they	they	PRON
abrj-637	176	14	are	be	AUX
abrj-637	176	15	preprocessed	preprocesse	VERB
abrj-637	176	16	,	,	PUNCT
abrj-637	176	17	features	feature	NOUN
abrj-637	176	18	extracted	extract	VERB
abrj-637	176	19	and	and	CCONJ
abrj-637	176	20	model	model	NOUN
abrj-637	176	21	trained	train	VERB
abrj-637	176	22	.	.	PUNCT
abrj-637	177	1	the	the	DET
abrj-637	177	2	risk	risk	NOUN
abrj-637	177	3	scoring	scoring	NOUN
abrj-637	177	4	and	and	CCONJ
abrj-637	177	5	interpretation	interpretation	NOUN
abrj-637	177	6	layer	layer	NOUN
abrj-637	177	7	uses	use	VERB
abrj-637	177	8	the	the	DET
abrj-637	177	9	trained	train	VERB
abrj-637	177	10	models	model	NOUN
abrj-637	177	11	on	on	ADP
abrj-637	177	12	the	the	DET
abrj-637	177	13	data	datum	NOUN
abrj-637	177	14	to	to	PART
abrj-637	177	15	create	create	VERB
abrj-637	177	16	the	the	DET
abrj-637	177	17	risk	risk	NOUN
abrj-637	177	18	score	score	NOUN
abrj-637	177	19	of	of	ADP
abrj-637	177	20	fraud	fraud	NOUN
abrj-637	177	21	along	along	ADP
abrj-637	177	22	with	with	ADP
abrj-637	177	23	giving	give	VERB
abrj-637	177	24	actionable	actionable	ADJ
abrj-637	177	25	data	datum	NOUN
abrj-637	177	26	to	to	ADP
abrj-637	177	27	the	the	DET
abrj-637	177	28	auditors	auditor	NOUN
abrj-637	177	29	.	.	PUNCT
abrj-637	178	1	lastly	lastly	ADV
abrj-637	178	2	,	,	PUNCT
abrj-637	178	3	the	the	DET
abrj-637	178	4	visualization	visualization	NOUN
abrj-637	178	5	and	and	CCONJ
abrj-637	178	6	reporting	reporting	NOUN
abrj-637	178	7	layer	layer	NOUN
abrj-637	178	8	of	of	ADP
abrj-637	178	9	the	the	DET
abrj-637	178	10	system	system	NOUN
abrj-637	178	11	provides	provide	VERB
abrj-637	178	12	information	information	NOUN
abrj-637	178	13	to	to	ADP
abrj-637	178	14	the	the	DET
abrj-637	178	15	auditors	auditor	NOUN
abrj-637	178	16	,	,	PUNCT
abrj-637	178	17	who	who	PRON
abrj-637	178	18	can	can	AUX
abrj-637	178	19	evaluate	evaluate	VERB
abrj-637	178	20	high	high	ADJ
abrj-637	178	21	-	-	PUNCT
abrj-637	178	22	risk	risk	NOUN
abrj-637	178	23	cases	case	NOUN
abrj-637	178	24	and	and	CCONJ
abrj-637	178	25	take	take	VERB
abrj-637	178	26	relevant	relevant	ADJ
abrj-637	178	27	actions	action	NOUN
abrj-637	178	28	with	with	ADP
abrj-637	178	29	the	the	DET
abrj-637	178	30	help	help	NOUN
abrj-637	178	31	of	of	ADP
abrj-637	178	32	interactive	interactive	ADJ
abrj-637	178	33	dashboards	dashboard	NOUN
abrj-637	178	34	.	.	PUNCT
abrj-637	179	1	3.11	3.11	NUM
abrj-637	179	2	.	.	PUNCT
abrj-637	179	3	data	datum	NOUN
abrj-637	179	4	pipeline	pipeline	NOUN
abrj-637	179	5	and	and	CCONJ
abrj-637	179	6	integration	integration	NOUN
abrj-637	179	7	data	datum	NOUN
abrj-637	179	8	pipeline	pipeline	NOUN
abrj-637	179	9	is	be	AUX
abrj-637	179	10	a	a	DET
abrj-637	179	11	very	very	ADV
abrj-637	179	12	important	important	ADJ
abrj-637	179	13	component	component	NOUN
abrj-637	179	14	of	of	ADP
abrj-637	179	15	integrity	integrity	NOUN
abrj-637	179	16	and	and	CCONJ
abrj-637	179	17	consistency	consistency	NOUN
abrj-637	179	18	of	of	ADP
abrj-637	179	19	the	the	DET
abrj-637	179	20	data	datum	NOUN
abrj-637	179	21	that	that	PRON
abrj-637	179	22	is	be	AUX
abrj-637	179	23	being	be	AUX
abrj-637	179	24	fed	feed	VERB
abrj-637	179	25	into	into	ADP
abrj-637	179	26	the	the	DET
abrj-637	179	27	system	system	NOUN
abrj-637	179	28	.	.	PUNCT
abrj-637	180	1	it	it	PRON
abrj-637	180	2	automates	automate	VERB
abrj-637	180	3	the	the	DET
abrj-637	180	4	ingestion	ingestion	NOUN
abrj-637	180	5	,	,	PUNCT
abrj-637	180	6	preprocessing	preprocessing	NOUN
abrj-637	180	7	and	and	CCONJ
abrj-637	180	8	transformation	transformation	NOUN
abrj-637	180	9	of	of	ADP
abrj-637	180	10	data	datum	NOUN
abrj-637	180	11	in	in	ADP
abrj-637	180	12	order	order	NOUN
abrj-637	180	13	to	to	PART
abrj-637	180	14	perform	perform	VERB
abrj-637	180	15	real	real	ADJ
abrj-637	180	16	time	time	NOUN
abrj-637	180	17	analytics	analytic	NOUN
abrj-637	180	18	and	and	CCONJ
abrj-637	180	19	decision	decision	NOUN
abrj-637	180	20	making	making	NOUN
abrj-637	180	21	.	.	PUNCT
abrj-637	181	1	there	there	PRON
abrj-637	181	2	are	be	VERB
abrj-637	181	3	numerous	numerous	ADJ
abrj-637	181	4	sources	source	NOUN
abrj-637	181	5	of	of	ADP
abrj-637	181	6	data	datum	NOUN
abrj-637	181	7	tapped	tap	VERB
abrj-637	181	8	by	by	ADP
abrj-637	181	9	the	the	DET
abrj-637	181	10	pipeline	pipeline	NOUN
abrj-637	181	11	,	,	PUNCT
abrj-637	181	12	which	which	PRON
abrj-637	181	13	include	include	VERB
abrj-637	181	14	tax	tax	NOUN
abrj-637	181	15	returns	return	NOUN
abrj-637	181	16	,	,	PUNCT
abrj-637	181	17	the	the	DET
abrj-637	181	18	financial	financial	ADJ
abrj-637	181	19	reports	report	NOUN
abrj-637	181	20	of	of	ADP
abrj-637	181	21	the	the	DET
abrj-637	181	22	companies	company	NOUN
abrj-637	181	23	and	and	CCONJ
abrj-637	181	24	external	external	ADJ
abrj-637	181	25	economic	economic	ADJ
abrj-637	181	26	indicators	indicator	NOUN
abrj-637	181	27	.	.	PUNCT
abrj-637	182	1	data	datum	NOUN
abrj-637	182	2	fusion	fusion	NOUN
abrj-637	182	3	is	be	AUX
abrj-637	182	4	being	be	AUX
abrj-637	182	5	controlled	control	VERB
abrj-637	182	6	with	with	ADP
abrj-637	182	7	the	the	DET
abrj-637	182	8	help	help	NOUN
abrj-637	182	9	of	of	ADP
abrj-637	182	10	automated	automate	VERB
abrj-637	182	11	etl	etl	NOUN
abrj-637	182	12	(	(	PUNCT
abrj-637	182	13	extract	extract	NOUN
abrj-637	182	14	,	,	PUNCT
abrj-637	182	15	transform	transform	NOUN
abrj-637	182	16	,	,	PUNCT
abrj-637	182	17	load	load	NOUN
abrj-637	182	18	)	)	PUNCT
abrj-637	182	19	operations	operation	NOUN
abrj-637	182	20	that	that	PRON
abrj-637	182	21	normalize	normalize	VERB
abrj-637	182	22	data	datum	NOUN
abrj-637	182	23	and	and	CCONJ
abrj-637	182	24	make	make	VERB
abrj-637	182	25	it	it	PRON
abrj-637	182	26	compatible	compatible	ADJ
abrj-637	182	27	and	and	CCONJ
abrj-637	182	28	consistent	consistent	ADJ
abrj-637	182	29	with	with	ADP
abrj-637	182	30	machine	machine	NOUN
abrj-637	182	31	learning	learning	NOUN
abrj-637	182	32	models	model	NOUN
abrj-637	182	33	.	.	PUNCT
abrj-637	183	1	the	the	DET
abrj-637	183	2	system	system	NOUN
abrj-637	183	3	uses	use	VERB
abrj-637	183	4	api	api	NOUN
abrj-637	183	5	integrations	integration	NOUN
abrj-637	183	6	to	to	PART
abrj-637	183	7	guarantee	guarantee	VERB
abrj-637	183	8	a	a	DET
abrj-637	183	9	smooth	smooth	ADJ
abrj-637	183	10	exchange	exchange	NOUN
abrj-637	183	11	between	between	ADP
abrj-637	183	12	the	the	DET
abrj-637	183	13	fraud	fraud	NOUN
abrj-637	183	14	detection	detection	NOUN
abrj-637	183	15	model	model	NOUN
abrj-637	183	16	and	and	CCONJ
abrj-637	183	17	the	the	DET
abrj-637	183	18	external	external	ADJ
abrj-637	183	19	tax	tax	NOUN
abrj-637	183	20	authority	authority	NOUN
abrj-637	183	21	databases	database	NOUN
abrj-637	183	22	.	.	PUNCT
abrj-637	184	1	the	the	DET
abrj-637	184	2	system	system	NOUN
abrj-637	184	3	data	data	NOUN
abrj-637	184	4	storage	storage	NOUN
abrj-637	184	5	solution	solution	NOUN
abrj-637	184	6	is	be	AUX
abrj-637	184	7	built	build	VERB
abrj-637	184	8	on	on	ADP
abrj-637	184	9	scalable	scalable	ADJ
abrj-637	184	10	and	and	CCONJ
abrj-637	184	11	secure	secure	ADJ
abrj-637	184	12	cloud	cloud	NOUN
abrj-637	184	13	storage	storage	NOUN
abrj-637	184	14	or	or	CCONJ
abrj-637	184	15	data	data	NOUN
abrj-637	184	16	warehouses	warehouse	NOUN
abrj-637	184	17	on	on	ADP
abrj-637	184	18	which	which	PRON
abrj-637	184	19	all	all	DET
abrj-637	184	20	the	the	DET
abrj-637	184	21	processed	process	VERB
abrj-637	184	22	data	data	NOUN
abrj-637	184	23	is	be	AUX
abrj-637	184	24	saved	save	VERB
abrj-637	184	25	in	in	ADP
abrj-637	184	26	an	an	DET
abrj-637	184	27	encrypted	encrypt	VERB
abrj-637	184	28	form	form	NOUN
abrj-637	184	29	so	so	SCONJ
abrj-637	184	30	that	that	SCONJ
abrj-637	184	31	they	they	PRON
abrj-637	184	32	satisfy	satisfy	VERB
abrj-637	184	33	the	the	DET
abrj-637	184	34	regulatory	regulatory	ADJ
abrj-637	184	35	requirements	requirement	NOUN
abrj-637	184	36	.	.	PUNCT
abrj-637	185	1	this	this	DET
abrj-637	185	2	table	table	NOUN
abrj-637	185	3	will	will	AUX
abrj-637	185	4	describe	describe	VERB
abrj-637	185	5	different	different	ADJ
abrj-637	185	6	sources	source	NOUN
abrj-637	185	7	of	of	ADP
abrj-637	185	8	data	datum	NOUN
abrj-637	185	9	and	and	CCONJ
abrj-637	185	10	integration	integration	NOUN
abrj-637	185	11	tools	tool	NOUN
abrj-637	185	12	that	that	PRON
abrj-637	185	13	will	will	AUX
abrj-637	185	14	be	be	AUX
abrj-637	185	15	used	use	VERB
abrj-637	185	16	in	in	ADP
abrj-637	185	17	the	the	DET
abrj-637	185	18	fraud	fraud	NOUN
abrj-637	185	19	detection	detection	NOUN
abrj-637	185	20	system	system	NOUN
abrj-637	185	21	.	.	PUNCT
abrj-637	186	1	these	these	DET
abrj-637	186	2	various	various	ADJ
abrj-637	186	3	sources	source	NOUN
abrj-637	186	4	of	of	ADP
abrj-637	186	5	data	datum	NOUN
abrj-637	186	6	are	be	AUX
abrj-637	186	7	integrated	integrate	VERB
abrj-637	186	8	making	make	VERB
abrj-637	186	9	the	the	DET
abrj-637	186	10	system	system	NOUN
abrj-637	186	11	provide	provide	VERB
abrj-637	186	12	a	a	DET
abrj-637	186	13	complete	complete	ADJ
abrj-637	186	14	and	and	CCONJ
abrj-637	186	15	sound	sound	ADJ
abrj-637	186	16	analysis	analysis	NOUN
abrj-637	186	17	.	.	PUNCT
abrj-637	187	1	the	the	DET
abrj-637	187	2	tools	tool	NOUN
abrj-637	187	3	play	play	VERB
abrj-637	187	4	an	an	DET
abrj-637	187	5	important	important	ADJ
abrj-637	187	6	role	role	NOUN
abrj-637	187	7	in	in	ADP
abrj-637	187	8	the	the	DET
abrj-637	187	9	smooth	smooth	ADJ
abrj-637	187	10	movement	movement	NOUN
abrj-637	187	11	of	of	ADP
abrj-637	187	12	data	datum	NOUN
abrj-637	187	13	between	between	ADP
abrj-637	187	14	external	external	ADJ
abrj-637	187	15	databases	database	NOUN
abrj-637	187	16	into	into	ADP
abrj-637	187	17	the	the	DET
abrj-637	187	18	fraud	fraud	NOUN
abrj-637	187	19	detection	detection	NOUN
abrj-637	187	20	pipeline	pipeline	NOUN
abrj-637	187	21	with	with	ADP
abrj-637	187	22	retaining	retain	VERB
abrj-637	187	23	data	datum	NOUN
abrj-637	187	24	integrity	integrity	NOUN
abrj-637	187	25	,	,	PUNCT
abrj-637	187	26	scalability	scalability	NOUN
abrj-637	187	27	,	,	PUNCT
abrj-637	187	28	and	and	CCONJ
abrj-637	187	29	compatibility	compatibility	NOUN
abrj-637	187	30	with	with	ADP
abrj-637	187	31	machine	machine	NOUN
abrj-637	187	32	learning	learn	VERB
abrj-637	187	33	algorithms	algorithm	NOUN
abrj-637	187	34	.	.	PUNCT
abrj-637	188	1	table	table	NOUN
abrj-637	188	2	3	3	NUM
abrj-637	188	3	.	.	PUNCT
abrj-637	188	4	data	datum	NOUN
abrj-637	188	5	sources	source	NOUN
abrj-637	188	6	and	and	CCONJ
abrj-637	188	7	integration	integration	NOUN
abrj-637	188	8	tools	tool	NOUN
abrj-637	188	9	to	to	PART
abrj-637	188	10	be	be	AUX
abrj-637	188	11	used	use	VERB
abrj-637	188	12	in	in	ADP
abrj-637	188	13	the	the	DET
abrj-637	188	14	system	system	NOUN
abrj-637	188	15	.	.	PUNCT
abrj-637	189	1	data	datum	NOUN
abrj-637	189	2	source	source	NOUN
abrj-637	189	3	description	description	NOUN
abrj-637	189	4	integration	integration	NOUN
abrj-637	189	5	tools	tool	NOUN
abrj-637	189	6	purpose	purpose	VERB
abrj-637	189	7	corporate	corporate	ADJ
abrj-637	189	8	financial	financial	ADJ
abrj-637	189	9	data	datum	NOUN
abrj-637	189	10	includes	include	VERB
abrj-637	189	11	balance	balance	NOUN
abrj-637	189	12	sheets	sheet	NOUN
abrj-637	189	13	,	,	PUNCT
abrj-637	189	14	income	income	NOUN
abrj-637	189	15	statements	statement	NOUN
abrj-637	189	16	,	,	PUNCT
abrj-637	189	17	and	and	CCONJ
abrj-637	189	18	tax	tax	NOUN
abrj-637	189	19	filings	filing	NOUN
abrj-637	189	20	.	.	PUNCT
abrj-637	190	1	api	api	NOUN
abrj-637	190	2	connections	connection	NOUN
abrj-637	190	3	to	to	ADP
abrj-637	190	4	corporate	corporate	ADJ
abrj-637	190	5	databases	database	NOUN
abrj-637	190	6	provides	provide	VERB
abrj-637	190	7	structured	structured	ADJ
abrj-637	190	8	financial	financial	ADJ
abrj-637	190	9	data	datum	NOUN
abrj-637	190	10	used	use	VERB
abrj-637	190	11	to	to	PART
abrj-637	190	12	evaluate	evaluate	VERB
abrj-637	190	13	the	the	DET
abrj-637	190	14	company	company	NOUN
abrj-637	190	15	’s	’s	PART
abrj-637	190	16	financial	financial	ADJ
abrj-637	190	17	health	health	NOUN
abrj-637	190	18	.	.	PUNCT
abrj-637	191	1	compliance	compliance	NOUN
abrj-637	191	2	reports	report	NOUN
abrj-637	191	3	contains	contain	VERB
abrj-637	191	4	records	record	NOUN
abrj-637	191	5	of	of	ADP
abrj-637	191	6	company	company	NOUN
abrj-637	191	7	compliance	compliance	NOUN
abrj-637	191	8	with	with	ADP
abrj-637	191	9	tax	tax	NOUN
abrj-637	191	10	regulations	regulation	NOUN
abrj-637	191	11	.	.	PUNCT
abrj-637	192	1	etl	etl	NOUN
abrj-637	192	2	(	(	PUNCT
abrj-637	192	3	extract	extract	NOUN
abrj-637	192	4	,	,	PUNCT
abrj-637	192	5	transform	transform	NOUN
abrj-637	192	6	,	,	PUNCT
abrj-637	192	7	load	load	NOUN
abrj-637	192	8	)	)	PUNCT
abrj-637	192	9	pipeline	pipeline	NOUN
abrj-637	192	10	,	,	PUNCT
abrj-637	192	11	cloud	cloud	NOUN
abrj-637	192	12	storage	storage	NOUN
abrj-637	192	13	tracks	track	VERB
abrj-637	192	14	deviations	deviation	NOUN
abrj-637	192	15	in	in	ADP
abrj-637	192	16	tax	tax	NOUN
abrj-637	192	17	compliance	compliance	NOUN
abrj-637	192	18	,	,	PUNCT
abrj-637	192	19	helping	helping	AUX
abrj-637	192	20	identify	identify	VERB
abrj-637	192	21	filings	filing	NOUN
abrj-637	192	22	with	with	ADP
abrj-637	192	23	irregularities	irregularity	NOUN
abrj-637	192	24	.	.	PUNCT
abrj-637	193	1	external	external	ADJ
abrj-637	193	2	databases	database	NOUN
abrj-637	193	3	includes	include	VERB
abrj-637	193	4	macroeconomic	macroeconomic	ADJ
abrj-637	193	5	indicators	indicator	NOUN
abrj-637	193	6	and	and	CCONJ
abrj-637	193	7	transaction	transaction	NOUN
abrj-637	193	8	histories	history	NOUN
abrj-637	193	9	.	.	PUNCT
abrj-637	194	1	cloud	cloud	NOUN
abrj-637	194	2	storage	storage	NOUN
abrj-637	194	3	,	,	PUNCT
abrj-637	194	4	external	external	ADJ
abrj-637	194	5	api	api	NOUN
abrj-637	194	6	connectors	connector	NOUN
abrj-637	194	7	provides	provide	VERB
abrj-637	194	8	contextual	contextual	ADJ
abrj-637	194	9	financial	financial	ADJ
abrj-637	194	10	data	datum	NOUN
abrj-637	194	11	that	that	PRON
abrj-637	194	12	helps	help	VERB
abrj-637	194	13	assess	assess	VERB
abrj-637	194	14	external	external	ADJ
abrj-637	194	15	factors	factor	NOUN
abrj-637	194	16	influencing	influence	VERB
abrj-637	194	17	compliance	compliance	NOUN
abrj-637	194	18	.	.	PUNCT
abrj-637	195	1	management	management	NOUN
abrj-637	195	2	reports	report	NOUN
abrj-637	195	3	includes	include	VERB
abrj-637	195	4	management	management	NOUN
abrj-637	195	5	commentaries	commentary	NOUN
abrj-637	195	6	and	and	CCONJ
abrj-637	195	7	strategic	strategic	ADJ
abrj-637	195	8	reports	report	NOUN
abrj-637	195	9	.	.	PUNCT
abrj-637	196	1	text	text	NOUN
abrj-637	196	2	extraction	extraction	NOUN
abrj-637	196	3	,	,	PUNCT
abrj-637	196	4	nlp	nlp	ADJ
abrj-637	196	5	processing	processing	NOUN
abrj-637	196	6	tools	tool	NOUN
abrj-637	196	7	provides	provide	VERB
abrj-637	196	8	textual	textual	ADJ
abrj-637	196	9	data	datum	NOUN
abrj-637	196	10	for	for	ADP
abrj-637	196	11	linguistic	linguistic	ADJ
abrj-637	196	12	analysis	analysis	NOUN
abrj-637	196	13	,	,	PUNCT
abrj-637	196	14	revealing	reveal	VERB
abrj-637	196	15	potential	potential	ADJ
abrj-637	196	16	signs	sign	NOUN
abrj-637	196	17	of	of	ADP
abrj-637	196	18	fraudulent	fraudulent	ADJ
abrj-637	196	19	behavior	behavior	NOUN
abrj-637	196	20	.	.	PUNCT
abrj-637	197	1	auditor	auditor	NOUN
abrj-637	197	2	notes	note	NOUN
abrj-637	197	3	includes	include	VERB
abrj-637	197	4	notes	note	NOUN
abrj-637	197	5	from	from	ADP
abrj-637	197	6	tax	tax	NOUN
abrj-637	197	7	auditors	auditor	NOUN
abrj-637	197	8	regarding	regard	VERB
abrj-637	197	9	company	company	NOUN
abrj-637	197	10	filings	filing	NOUN
abrj-637	197	11	and	and	CCONJ
abrj-637	197	12	behavior	behavior	NOUN
abrj-637	197	13	.	.	PUNCT
abrj-637	198	1	cloud	cloud	NOUN
abrj-637	198	2	storage	storage	NOUN
abrj-637	198	3	,	,	PUNCT
abrj-637	198	4	api	api	NOUN
abrj-637	198	5	connections	connection	NOUN
abrj-637	198	6	offers	offer	VERB
abrj-637	198	7	qualitative	qualitative	ADJ
abrj-637	198	8	insights	insight	NOUN
abrj-637	198	9	into	into	ADP
abrj-637	198	10	company	company	NOUN
abrj-637	198	11	operations	operation	NOUN
abrj-637	198	12	,	,	PUNCT
abrj-637	198	13	assisting	assist	VERB
abrj-637	198	14	in	in	ADP
abrj-637	198	15	fraud	fraud	NOUN
abrj-637	198	16	detection	detection	NOUN
abrj-637	198	17	.	.	PUNCT
abrj-637	199	1	transaction	transaction	NOUN
abrj-637	199	2	data	datum	NOUN
abrj-637	199	3	provides	provide	VERB
abrj-637	199	4	detailed	detailed	ADJ
abrj-637	199	5	records	record	NOUN
abrj-637	199	6	of	of	ADP
abrj-637	199	7	transactions	transaction	NOUN
abrj-637	199	8	within	within	ADP
abrj-637	199	9	the	the	DET
abrj-637	199	10	company	company	NOUN
abrj-637	199	11	.	.	PUNCT
abrj-637	200	1	etl	etl	NOUN
abrj-637	200	2	pipeline	pipeline	NOUN
abrj-637	200	3	,	,	PUNCT
abrj-637	200	4	data	datum	NOUN
abrj-637	200	5	aggregation	aggregation	NOUN
abrj-637	200	6	tools	tool	NOUN
abrj-637	200	7	identifies	identify	VERB
abrj-637	200	8	unusual	unusual	ADJ
abrj-637	200	9	or	or	CCONJ
abrj-637	200	10	large	large	ADJ
abrj-637	200	11	transactions	transaction	NOUN
abrj-637	200	12	that	that	PRON
abrj-637	200	13	could	could	AUX
abrj-637	200	14	be	be	AUX
abrj-637	200	15	indicative	indicative	ADJ
abrj-637	200	16	of	of	ADP
abrj-637	200	17	fraud	fraud	NOUN
abrj-637	200	18	.	.	PUNCT
abrj-637	201	1	3.12	3.12	NUM
abrj-637	201	2	.	.	PUNCT
abrj-637	201	3	model	model	NOUN
abrj-637	201	4	deployment	deployment	ADJ
abrj-637	201	5	workflow	workflow	NOUN
abrj-637	201	6	the	the	DET
abrj-637	201	7	machine	machine	NOUN
abrj-637	201	8	learning	learn	VERB
abrj-637	201	9	model	model	NOUN
abrj-637	201	10	deployment	deployment	NOUN
abrj-637	201	11	process	process	NOUN
abrj-637	201	12	of	of	ADP
abrj-637	201	13	the	the	DET
abrj-637	201	14	fraud	fraud	NOUN
abrj-637	201	15	detection	detection	NOUN
abrj-637	201	16	system	system	NOUN
abrj-637	201	17	is	be	AUX
abrj-637	201	18	described	describe	VERB
abrj-637	201	19	in	in	ADP
abrj-637	201	20	three	three	NUM
abrj-637	201	21	significant	significant	ADJ
abrj-637	201	22	steps	step	NOUN
abrj-637	201	23	,	,	PUNCT
abrj-637	201	24	including	include	VERB
abrj-637	201	25	training	training	NOUN
abrj-637	201	26	,	,	PUNCT
abrj-637	201	27	validation	validation	NOUN
abrj-637	201	28	,	,	PUNCT
abrj-637	201	29	and	and	CCONJ
abrj-637	201	30	deployment	deployment	NOUN
abrj-637	201	31	.	.	PUNCT
abrj-637	202	1	the	the	DET
abrj-637	202	2	training	training	NOUN
abrj-637	202	3	of	of	ADP
abrj-637	202	4	both	both	CCONJ
abrj-637	202	5	supervised	supervised	ADJ
abrj-637	202	6	and	and	CCONJ
abrj-637	202	7	unsupervised	unsupervised	ADJ
abrj-637	202	8	models	model	NOUN
abrj-637	202	9	such	such	ADJ
abrj-637	202	10	as	as	ADP
abrj-637	202	11	random	random	ADJ
abrj-637	202	12	forest	forest	NOUN
abrj-637	202	13	,	,	PUNCT
abrj-637	202	14	xgboost	xgboost	PROPN
abrj-637	202	15	,	,	PUNCT
abrj-637	202	16	autoencoders	autoencoder	NOUN
abrj-637	202	17	,	,	PUNCT
abrj-637	202	18	and	and	CCONJ
abrj-637	202	19	isolation	isolation	NOUN
abrj-637	202	20	forests	forest	NOUN
abrj-637	202	21	is	be	AUX
abrj-637	202	22	trained	train	VERB
abrj-637	202	23	using	use	VERB
abrj-637	202	24	historical	historical	ADJ
abrj-637	202	25	data	datum	NOUN
abrj-637	202	26	during	during	ADP
abrj-637	202	27	the	the	DET
abrj-637	202	28	training	training	NOUN
abrj-637	202	29	stage	stage	NOUN
abrj-637	202	30	.	.	PUNCT
abrj-637	203	1	these	these	DET
abrj-637	203	2	models	model	NOUN
abrj-637	203	3	are	be	AUX
abrj-637	203	4	further	far	ADV
abrj-637	203	5	cross	cross	ADJ
abrj-637	203	6	-	-	VERB
abrj-637	203	7	validated	validated	ADJ
abrj-637	203	8	with	with	ADP
abrj-637	203	9	the	the	DET
abrj-637	203	10	help	help	NOUN
abrj-637	203	11	of	of	ADP
abrj-637	203	12	k	k	ADJ
abrj-637	203	13	-	-	ADJ
abrj-637	203	14	fold	fold	ADJ
abrj-637	203	15	cross	cross	NOUN
abrj-637	203	16	-	-	NOUN
abrj-637	203	17	validation	validation	NOUN
abrj-637	203	18	which	which	PRON
abrj-637	203	19	is	be	AUX
abrj-637	203	20	used	use	VERB
abrj-637	203	21	to	to	PART
abrj-637	203	22	check	check	VERB
abrj-637	203	23	the	the	DET
abrj-637	203	24	robustness	robustness	NOUN
abrj-637	203	25	of	of	ADP
abrj-637	203	26	the	the	DET
abrj-637	203	27	models	model	NOUN
abrj-637	203	28	and	and	CCONJ
abrj-637	203	29	prevent	prevent	VERB
abrj-637	203	30	overfitting	overfitting	NOUN
abrj-637	203	31	by	by	ADP
abrj-637	203	32	trying	try	VERB
abrj-637	203	33	them	they	PRON
abrj-637	203	34	on	on	ADP
abrj-637	203	35	various	various	ADJ
abrj-637	203	36	subsets	subset	NOUN
abrj-637	203	37	of	of	ADP
abrj-637	203	38	the	the	DET
abrj-637	203	39	data	datum	NOUN
abrj-637	203	40	.	.	PUNCT
abrj-637	204	1	when	when	SCONJ
abrj-637	204	2	the	the	DET
abrj-637	204	3	models	model	NOUN
abrj-637	204	4	are	be	AUX
abrj-637	204	5	validated	validate	VERB
abrj-637	204	6	,	,	PUNCT
abrj-637	204	7	the	the	DET
abrj-637	204	8	most	most	ADV
abrj-637	204	9	effective	effective	ADJ
abrj-637	204	10	ones	one	NOUN
abrj-637	204	11	are	be	AUX
abrj-637	204	12	implemented	implement	VERB
abrj-637	204	13	into	into	ADP
abrj-637	204	14	the	the	DET
abrj-637	204	15	production	production	NOUN
abrj-637	204	16	system	system	NOUN
abrj-637	204	17	where	where	SCONJ
abrj-637	204	18	they	they	PRON
abrj-637	204	19	are	be	AUX
abrj-637	204	20	capable	capable	ADJ
abrj-637	204	21	of	of	ADP
abrj-637	204	22	processing	process	VERB
abrj-637	204	23	new	new	ADJ
abrj-637	204	24	incoming	incoming	ADJ
abrj-637	204	25	data	datum	NOUN
abrj-637	204	26	and	and	CCONJ
abrj-637	204	27	making	make	VERB
abrj-637	204	28	it	it	PRON
abrj-637	204	29	available	available	ADJ
abrj-637	204	30	in	in	ADP
abrj-637	204	31	real	real	ADJ
abrj-637	204	32	-	-	PUNCT
abrj-637	204	33	time	time	NOUN
abrj-637	204	34	to	to	PART
abrj-637	204	35	predict	predict	VERB
abrj-637	204	36	the	the	DET
abrj-637	204	37	risk	risk	NOUN
abrj-637	204	38	of	of	ADP
abrj-637	204	39	fraud	fraud	NOUN
abrj-637	204	40	.	.	PUNCT
abrj-637	205	1	the	the	DET
abrj-637	205	2	continuous	continuous	ADJ
abrj-637	205	3	learning	learning	NOUN
abrj-637	205	4	is	be	AUX
abrj-637	205	5	possible	possible	ADJ
abrj-637	205	6	in	in	ADP
abrj-637	205	7	this	this	DET
abrj-637	205	8	stage	stage	NOUN
abrj-637	205	9	of	of	ADP
abrj-637	205	10	deployment	deployment	NOUN
abrj-637	205	11	since	since	SCONJ
abrj-637	205	12	the	the	DET
abrj-637	205	13	models	model	NOUN
abrj-637	205	14	can	can	AUX
abrj-637	205	15	be	be	AUX
abrj-637	205	16	regularly	regularly	ADV
abrj-637	205	17	updated	update	VERB
abrj-637	205	18	,	,	PUNCT
abrj-637	205	19	according	accord	VERB
abrj-637	205	20	to	to	ADP
abrj-637	205	21	new	new	ADJ
abrj-637	205	22	data	datum	NOUN
abrj-637	205	23	,	,	PUNCT
abrj-637	205	24	so	so	SCONJ
abrj-637	205	25	that	that	SCONJ
abrj-637	205	26	they	they	PRON
abrj-637	205	27	are	be	AUX
abrj-637	205	28	effective	effective	ADJ
abrj-637	205	29	as	as	SCONJ
abrj-637	205	30	the	the	DET
abrj-637	205	31	fraud	fraud	NOUN
abrj-637	205	32	detection	detection	NOUN
abrj-637	205	33	patterns	pattern	NOUN
abrj-637	205	34	will	will	AUX
abrj-637	205	35	constantly	constantly	ADV
abrj-637	205	36	change	change	VERB
abrj-637	205	37	.	.	PUNCT
abrj-637	206	1	asian	asian	ADJ
abrj-637	206	2	business	business	PROPN
abrj-637	206	3	research	research	NOUN
abrj-637	206	4	journal	journal	NOUN
abrj-637	206	5	,	,	PUNCT
abrj-637	206	6	2025	2025	NUM
abrj-637	206	7	,	,	PUNCT
abrj-637	206	8	10(11	10(11	NUM
abrj-637	206	9	):	):	PUNCT
abrj-637	206	10	1	1	NUM
abrj-637	206	11	-	-	SYM
abrj-637	206	12	13	13	NUM
abrj-637	206	13	9	9	NUM
abrj-637	206	14	©	©	PROPN
abrj-637	206	15	2025	2025	NUM
abrj-637	206	16	by	by	ADP
abrj-637	206	17	the	the	DET
abrj-637	206	18	authors	author	NOUN
abrj-637	206	19	;	;	PUNCT
abrj-637	206	20	licensee	licensee	NOUN
abrj-637	206	21	eastern	eastern	ADJ
abrj-637	206	22	centre	centre	NOUN
abrj-637	206	23	of	of	ADP
abrj-637	206	24	science	science	NOUN
abrj-637	206	25	and	and	CCONJ
abrj-637	206	26	education	education	NOUN
abrj-637	206	27	,	,	PUNCT
abrj-637	206	28	usa	usa	PROPN
abrj-637	206	29	figure	figure	NOUN
abrj-637	206	30	5	5	NUM
abrj-637	206	31	.	.	PUNCT
abrj-637	206	32	training	training	NOUN
abrj-637	206	33	and	and	CCONJ
abrj-637	206	34	deployment	deployment	NOUN
abrj-637	206	35	workflow	workflow	NOUN
abrj-637	206	36	model	model	NOUN
abrj-637	206	37	.	.	PUNCT
abrj-637	207	1	3.13	3.13	NUM
abrj-637	207	2	.	.	PUNCT
abrj-637	208	1	risk	risk	NOUN
abrj-637	208	2	scoring	scoring	NOUN
abrj-637	208	3	mechanism	mechanism	NOUN
abrj-637	208	4	the	the	DET
abrj-637	208	5	risk	risk	NOUN
abrj-637	208	6	scoring	scoring	NOUN
abrj-637	208	7	scheme	scheme	NOUN
abrj-637	208	8	translates	translate	VERB
abrj-637	208	9	the	the	DET
abrj-637	208	10	results	result	NOUN
abrj-637	208	11	of	of	ADP
abrj-637	208	12	machine	machine	NOUN
abrj-637	208	13	learning	learning	NOUN
abrj-637	208	14	models	model	NOUN
abrj-637	208	15	into	into	ADP
abrj-637	208	16	a	a	DET
abrj-637	208	17	simplified	simplified	ADJ
abrj-637	208	18	and	and	CCONJ
abrj-637	208	19	highly	highly	ADV
abrj-637	208	20	interpretable	interpretable	ADJ
abrj-637	208	21	framework	framework	NOUN
abrj-637	208	22	that	that	PRON
abrj-637	208	23	are	be	AUX
abrj-637	208	24	easily	easily	ADV
abrj-637	208	25	comprehensible	comprehensible	ADJ
abrj-637	208	26	to	to	ADP
abrj-637	208	27	the	the	DET
abrj-637	208	28	auditors	auditor	NOUN
abrj-637	208	29	.	.	PUNCT
abrj-637	209	1	the	the	DET
abrj-637	209	2	probability	probability	NOUN
abrj-637	209	3	of	of	ADP
abrj-637	209	4	fraud	fraud	NOUN
abrj-637	209	5	is	be	AUX
abrj-637	209	6	used	use	VERB
abrj-637	209	7	to	to	PART
abrj-637	209	8	rate	rate	VERB
abrj-637	209	9	the	the	DET
abrj-637	209	10	risk	risk	NOUN
abrj-637	209	11	of	of	ADP
abrj-637	209	12	fraud	fraud	NOUN
abrj-637	209	13	in	in	ADP
abrj-637	209	14	each	each	DET
abrj-637	209	15	corporate	corporate	ADJ
abrj-637	209	16	filing	filing	NOUN
abrj-637	209	17	with	with	ADP
abrj-637	209	18	a	a	DET
abrj-637	209	19	range	range	NOUN
abrj-637	209	20	of	of	ADP
abrj-637	209	21	score	score	NOUN
abrj-637	209	22	between	between	ADP
abrj-637	209	23	0.00	0.00	NUM
abrj-637	209	24	and	and	CCONJ
abrj-637	209	25	1.00	1.00	NUM
abrj-637	209	26	.	.	PUNCT
abrj-637	210	1	the	the	DET
abrj-637	210	2	score	score	NOUN
abrj-637	210	3	range	range	NOUN
abrj-637	210	4	is	be	AUX
abrj-637	210	5	broken	break	VERB
abrj-637	210	6	down	down	ADP
abrj-637	210	7	into	into	ADP
abrj-637	210	8	three	three	NUM
abrj-637	210	9	categories	category	NOUN
abrj-637	210	10	,	,	PUNCT
abrj-637	210	11	which	which	PRON
abrj-637	210	12	include	include	VERB
abrj-637	210	13	low	low	ADJ
abrj-637	210	14	risk	risk	NOUN
abrj-637	210	15	(	(	PUNCT
abrj-637	210	16	0.00	0.00	NUM
abrj-637	210	17	-	-	NUM
abrj-637	210	18	0.30	0.30	NUM
abrj-637	210	19	)	)	PUNCT
abrj-637	210	20	,	,	PUNCT
abrj-637	210	21	medium	medium	ADJ
abrj-637	210	22	risk	risk	NOUN
abrj-637	210	23	(	(	PUNCT
abrj-637	210	24	0.31	0.31	NUM
abrj-637	210	25	-	-	SYM
abrj-637	210	26	0.70	0.70	NUM
abrj-637	210	27	)	)	PUNCT
abrj-637	210	28	,	,	PUNCT
abrj-637	210	29	and	and	CCONJ
abrj-637	210	30	high	high	ADJ
abrj-637	210	31	risk	risk	NOUN
abrj-637	210	32	(	(	PUNCT
abrj-637	210	33	0.71	0.71	NUM
abrj-637	210	34	-	-	SYM
abrj-637	210	35	1.00	1.00	NUM
abrj-637	210	36	)	)	PUNCT
abrj-637	210	37	.	.	PUNCT
abrj-637	211	1	such	such	ADJ
abrj-637	211	2	scores	score	NOUN
abrj-637	211	3	are	be	AUX
abrj-637	211	4	then	then	ADV
abrj-637	211	5	employed	employ	VERB
abrj-637	211	6	to	to	PART
abrj-637	211	7	give	give	VERB
abrj-637	211	8	turnover	turnover	NOUN
abrj-637	211	9	to	to	ADP
abrj-637	211	10	further	further	ADJ
abrj-637	211	11	audits	audits	NOUN
abrj-637	211	12	,	,	PUNCT
abrj-637	211	13	with	with	ADP
abrj-637	211	14	the	the	DET
abrj-637	211	15	high	high	ADJ
abrj-637	211	16	-	-	PUNCT
abrj-637	211	17	risk	risk	NOUN
abrj-637	211	18	ones	one	NOUN
abrj-637	211	19	being	be	AUX
abrj-637	211	20	followed	follow	VERB
abrj-637	211	21	up	up	ADP
abrj-637	211	22	by	by	ADP
abrj-637	211	23	the	the	DET
abrj-637	211	24	auditors	auditor	NOUN
abrj-637	211	25	instantly	instantly	ADV
abrj-637	211	26	.	.	PUNCT
abrj-637	212	1	the	the	DET
abrj-637	212	2	risk	risk	NOUN
abrj-637	212	3	scoring	scoring	NOUN
abrj-637	212	4	system	system	NOUN
abrj-637	212	5	takes	take	VERB
abrj-637	212	6	the	the	DET
abrj-637	212	7	outputs	output	NOUN
abrj-637	212	8	of	of	ADP
abrj-637	212	9	several	several	ADJ
abrj-637	212	10	different	different	ADJ
abrj-637	212	11	models	model	NOUN
abrj-637	212	12	,	,	PUNCT
abrj-637	212	13	e.g.	e.g.	ADV
abrj-637	212	14	,	,	PUNCT
abrj-637	212	15	ensemble	ensemble	ADJ
abrj-637	212	16	classifiers	classifier	NOUN
abrj-637	212	17	and	and	CCONJ
abrj-637	212	18	anomaly	anomaly	NOUN
abrj-637	212	19	detectors	detector	NOUN
abrj-637	212	20	and	and	CCONJ
abrj-637	212	21	weighs	weigh	VERB
abrj-637	212	22	them	they	PRON
abrj-637	212	23	by	by	ADP
abrj-637	212	24	their	their	PRON
abrj-637	212	25	confidence	confidence	NOUN
abrj-637	212	26	.	.	PUNCT
abrj-637	213	1	this	this	DET
abrj-637	213	2	approach	approach	NOUN
abrj-637	213	3	can	can	AUX
abrj-637	213	4	guarantee	guarantee	VERB
abrj-637	213	5	that	that	SCONJ
abrj-637	213	6	the	the	DET
abrj-637	213	7	end	end	NOUN
abrj-637	213	8	-	-	PUNCT
abrj-637	213	9	result	result	NOUN
abrj-637	213	10	risk	risk	NOUN
abrj-637	213	11	score	score	NOUN
abrj-637	213	12	will	will	AUX
abrj-637	213	13	be	be	AUX
abrj-637	213	14	as	as	ADV
abrj-637	213	15	precise	precise	ADJ
abrj-637	213	16	and	and	CCONJ
abrj-637	213	17	dependable	dependable	ADJ
abrj-637	213	18	as	as	ADP
abrj-637	213	19	possible	possible	ADJ
abrj-637	213	20	,	,	PUNCT
abrj-637	213	21	which	which	PRON
abrj-637	213	22	will	will	AUX
abrj-637	213	23	allow	allow	VERB
abrj-637	213	24	the	the	DET
abrj-637	213	25	tax	tax	NOUN
abrj-637	213	26	authorities	authority	NOUN
abrj-637	213	27	to	to	PART
abrj-637	213	28	have	have	VERB
abrj-637	213	29	a	a	DET
abrj-637	213	30	clear	clear	ADJ
abrj-637	213	31	list	list	NOUN
abrj-637	213	32	of	of	ADP
abrj-637	213	33	priorities	priority	NOUN
abrj-637	213	34	when	when	SCONJ
abrj-637	213	35	it	it	PRON
abrj-637	213	36	comes	come	VERB
abrj-637	213	37	to	to	ADP
abrj-637	213	38	conducting	conduct	VERB
abrj-637	213	39	the	the	DET
abrj-637	213	40	audit	audit	NOUN
abrj-637	213	41	.	.	PUNCT
abrj-637	214	1	3.14	3.14	NUM
abrj-637	214	2	.	.	PUNCT
abrj-637	215	1	explainable	explainable	ADJ
abrj-637	215	2	artificial	artificial	ADJ
abrj-637	215	3	intelligence	intelligence	NOUN
abrj-637	215	4	(	(	PUNCT
abrj-637	215	5	xai	xai	PROPN
abrj-637	215	6	)	)	PUNCT
abrj-637	215	7	layer	layer	NOUN
abrj-637	215	8	explainable	explainable	ADJ
abrj-637	215	9	artificial	artificial	ADJ
abrj-637	215	10	intelligence	intelligence	NOUN
abrj-637	215	11	(	(	PUNCT
abrj-637	215	12	xai	xai	PROPN
abrj-637	215	13	)	)	PUNCT
abrj-637	215	14	is	be	AUX
abrj-637	215	15	among	among	ADP
abrj-637	215	16	the	the	DET
abrj-637	215	17	most	most	ADV
abrj-637	215	18	important	important	ADJ
abrj-637	215	19	elements	element	NOUN
abrj-637	215	20	of	of	ADP
abrj-637	215	21	the	the	DET
abrj-637	215	22	proposed	propose	VERB
abrj-637	215	23	system	system	NOUN
abrj-637	215	24	.	.	PUNCT
abrj-637	216	1	this	this	DET
abrj-637	216	2	layer	layer	NOUN
abrj-637	216	3	promotes	promote	VERB
abrj-637	216	4	transparency	transparency	NOUN
abrj-637	216	5	and	and	CCONJ
abrj-637	216	6	accountability	accountability	NOUN
abrj-637	216	7	in	in	ADP
abrj-637	216	8	the	the	DET
abrj-637	216	9	decision	decision	NOUN
abrj-637	216	10	-	-	PUNCT
abrj-637	216	11	making	make	VERB
abrj-637	216	12	process	process	NOUN
abrj-637	216	13	of	of	ADP
abrj-637	216	14	the	the	DET
abrj-637	216	15	model	model	NOUN
abrj-637	216	16	,	,	PUNCT
abrj-637	216	17	which	which	PRON
abrj-637	216	18	is	be	AUX
abrj-637	216	19	very	very	ADV
abrj-637	216	20	important	important	ADJ
abrj-637	216	21	in	in	ADP
abrj-637	216	22	ensuring	ensure	VERB
abrj-637	216	23	that	that	SCONJ
abrj-637	216	24	people	people	NOUN
abrj-637	216	25	have	have	VERB
abrj-637	216	26	trust	trust	NOUN
abrj-637	216	27	in	in	ADP
abrj-637	216	28	automated	automate	VERB
abrj-637	216	29	fraud	fraud	NOUN
abrj-637	216	30	detection	detection	NOUN
abrj-637	216	31	systems	system	NOUN
abrj-637	216	32	.	.	PUNCT
abrj-637	217	1	shap	shap	PROPN
abrj-637	217	2	(	(	PUNCT
abrj-637	217	3	shapley	shapley	NOUN
abrj-637	217	4	additive	additive	ADJ
abrj-637	217	5	explainations	explaination	NOUN
abrj-637	217	6	)	)	PUNCT
abrj-637	217	7	and	and	CCONJ
abrj-637	217	8	lime	lime	NOUN
abrj-637	217	9	(	(	PUNCT
abrj-637	217	10	local	local	ADJ
abrj-637	217	11	interpretable	interpretable	ADJ
abrj-637	217	12	model	model	ADJ
abrj-637	217	13	-	-	PUNCT
abrj-637	217	14	agnostic	agnostic	ADJ
abrj-637	217	15	explainations	explaination	NOUN
abrj-637	217	16	)	)	PUNCT
abrj-637	217	17	are	be	AUX
abrj-637	217	18	provided	provide	VERB
abrj-637	217	19	to	to	PART
abrj-637	217	20	explain	explain	VERB
abrj-637	217	21	the	the	DET
abrj-637	217	22	model	model	NOUN
abrj-637	217	23	predictions	prediction	NOUN
abrj-637	217	24	and	and	CCONJ
abrj-637	217	25	provide	provide	VERB
abrj-637	217	26	auditors	auditor	NOUN
abrj-637	217	27	with	with	ADP
abrj-637	217	28	knowledge	knowledge	NOUN
abrj-637	217	29	of	of	ADP
abrj-637	217	30	the	the	DET
abrj-637	217	31	particular	particular	ADJ
abrj-637	217	32	characteristics	characteristic	NOUN
abrj-637	217	33	that	that	PRON
abrj-637	217	34	led	lead	VERB
abrj-637	217	35	to	to	ADP
abrj-637	217	36	the	the	DET
abrj-637	217	37	classification	classification	NOUN
abrj-637	217	38	of	of	ADP
abrj-637	217	39	a	a	DET
abrj-637	217	40	filing	filing	NOUN
abrj-637	217	41	as	as	ADP
abrj-637	217	42	a	a	DET
abrj-637	217	43	high	high	ADJ
abrj-637	217	44	or	or	CCONJ
abrj-637	217	45	low	low	ADJ
abrj-637	217	46	risk	risk	NOUN
abrj-637	217	47	.	.	PUNCT
abrj-637	218	1	decision	decision	NOUN
abrj-637	218	2	-	-	PUNCT
abrj-637	218	3	making	making	NOUN
abrj-637	218	4	can	can	AUX
abrj-637	218	5	also	also	ADV
abrj-637	218	6	be	be	AUX
abrj-637	218	7	supported	support	VERB
abrj-637	218	8	by	by	ADP
abrj-637	218	9	the	the	DET
abrj-637	218	10	xai	xai	PROPN
abrj-637	218	11	layer	layer	NOUN
abrj-637	218	12	,	,	PUNCT
abrj-637	218	13	which	which	PRON
abrj-637	218	14	determines	determine	VERB
abrj-637	218	15	the	the	DET
abrj-637	218	16	most	most	ADV
abrj-637	218	17	significant	significant	ADJ
abrj-637	218	18	variables	variable	NOUN
abrj-637	218	19	,	,	PUNCT
abrj-637	218	20	including	include	VERB
abrj-637	218	21	unusual	unusual	ADJ
abrj-637	218	22	revenue	revenue	NOUN
abrj-637	218	23	trends	trend	NOUN
abrj-637	218	24	or	or	CCONJ
abrj-637	218	25	the	the	DET
abrj-637	218	26	adverse	adverse	ADJ
abrj-637	218	27	sentiments	sentiment	NOUN
abrj-637	218	28	in	in	ADP
abrj-637	218	29	the	the	DET
abrj-637	218	30	textual	textual	ADJ
abrj-637	218	31	reports	report	NOUN
abrj-637	218	32	,	,	PUNCT
abrj-637	218	33	that	that	PRON
abrj-637	218	34	resulted	result	VERB
abrj-637	218	35	in	in	ADP
abrj-637	218	36	a	a	DET
abrj-637	218	37	certain	certain	ADJ
abrj-637	218	38	fraud	fraud	NOUN
abrj-637	218	39	risk	risk	NOUN
abrj-637	218	40	score	score	NOUN
abrj-637	218	41	.	.	PUNCT
abrj-637	219	1	with	with	ADP
abrj-637	219	2	such	such	ADJ
abrj-637	219	3	explanations	explanation	NOUN
abrj-637	219	4	,	,	PUNCT
abrj-637	219	5	the	the	DET
abrj-637	219	6	system	system	NOUN
abrj-637	219	7	enables	enable	VERB
abrj-637	219	8	the	the	DET
abrj-637	219	9	auditors	auditor	NOUN
abrj-637	219	10	to	to	PART
abrj-637	219	11	have	have	VERB
abrj-637	219	12	a	a	DET
abrj-637	219	13	more	more	ADJ
abrj-637	219	14	insight	insight	NOUN
abrj-637	219	15	into	into	ADP
abrj-637	219	16	the	the	DET
abrj-637	219	17	behavior	behavior	NOUN
abrj-637	219	18	of	of	ADP
abrj-637	219	19	the	the	DET
abrj-637	219	20	model	model	NOUN
abrj-637	219	21	,	,	PUNCT
abrj-637	219	22	and	and	CCONJ
abrj-637	219	23	any	any	DET
abrj-637	219	24	filings	filing	NOUN
abrj-637	219	25	flagged	flag	VERB
abrj-637	219	26	will	will	AUX
abrj-637	219	27	be	be	AUX
abrj-637	219	28	looked	look	VERB
abrj-637	219	29	into	into	ADP
abrj-637	219	30	with	with	ADP
abrj-637	219	31	the	the	DET
abrj-637	219	32	context	context	NOUN
abrj-637	219	33	of	of	ADP
abrj-637	219	34	the	the	DET
abrj-637	219	35	model	model	NOUN
abrj-637	219	36	.	.	PUNCT
abrj-637	220	1	3.15	3.15	NUM
abrj-637	220	2	.	.	PUNCT
abrj-637	220	3	interrelation	interrelation	NOUN
abrj-637	220	4	of	of	ADP
abrj-637	220	5	audit	audit	NOUN
abrj-637	220	6	and	and	CCONJ
abrj-637	220	7	compliance	compliance	NOUN
abrj-637	220	8	system	system	NOUN
abrj-637	220	9	.	.	PUNCT
abrj-637	221	1	to	to	PART
abrj-637	221	2	be	be	AUX
abrj-637	221	3	useful	useful	ADJ
abrj-637	221	4	in	in	ADP
abrj-637	221	5	a	a	DET
abrj-637	221	6	real	real	ADJ
abrj-637	221	7	-	-	PUNCT
abrj-637	221	8	world	world	NOUN
abrj-637	221	9	environment	environment	NOUN
abrj-637	221	10	,	,	PUNCT
abrj-637	221	11	the	the	DET
abrj-637	221	12	machine	machine	NOUN
abrj-637	221	13	learning	learning	NOUN
abrj-637	221	14	-	-	PUNCT
abrj-637	221	15	based	base	VERB
abrj-637	221	16	fraud	fraud	NOUN
abrj-637	221	17	detection	detection	NOUN
abrj-637	221	18	system	system	NOUN
abrj-637	221	19	should	should	AUX
abrj-637	221	20	be	be	AUX
abrj-637	221	21	connected	connect	VERB
abrj-637	221	22	with	with	ADP
abrj-637	221	23	the	the	DET
abrj-637	221	24	existing	exist	VERB
abrj-637	221	25	enterprise	enterprise	NOUN
abrj-637	221	26	audit	audit	NOUN
abrj-637	221	27	systems	system	NOUN
abrj-637	221	28	and	and	CCONJ
abrj-637	221	29	government	government	NOUN
abrj-637	221	30	tax	tax	NOUN
abrj-637	221	31	portals	portal	NOUN
abrj-637	221	32	.	.	PUNCT
abrj-637	222	1	this	this	PRON
abrj-637	222	2	can	can	AUX
abrj-637	222	3	be	be	AUX
abrj-637	222	4	integrated	integrate	VERB
abrj-637	222	5	by	by	ADP
abrj-637	222	6	use	use	NOUN
abrj-637	222	7	of	of	ADP
abrj-637	222	8	secure	secure	ADJ
abrj-637	222	9	api	api	NOUN
abrj-637	222	10	connections	connection	NOUN
abrj-637	222	11	,	,	PUNCT
abrj-637	222	12	where	where	SCONJ
abrj-637	222	13	fraud	fraud	NOUN
abrj-637	222	14	detecting	detect	VERB
abrj-637	222	15	system	system	NOUN
abrj-637	222	16	would	would	AUX
abrj-637	222	17	be	be	AUX
abrj-637	222	18	able	able	ADJ
abrj-637	222	19	to	to	PART
abrj-637	222	20	tap	tap	VERB
abrj-637	222	21	into	into	ADP
abrj-637	222	22	real	real	ADJ
abrj-637	222	23	time	time	NOUN
abrj-637	222	24	data	datum	NOUN
abrj-637	222	25	of	of	ADP
abrj-637	222	26	the	the	DET
abrj-637	222	27	tax	tax	NOUN
abrj-637	222	28	authorities	authority	NOUN
abrj-637	222	29	databases	database	NOUN
abrj-637	222	30	to	to	PART
abrj-637	222	31	communicate	communicate	VERB
abrj-637	222	32	smoothly	smoothly	ADV
abrj-637	222	33	with	with	ADP
abrj-637	222	34	other	other	ADJ
abrj-637	222	35	auditing	auditing	NOUN
abrj-637	222	36	systems	system	NOUN
abrj-637	222	37	.	.	PUNCT
abrj-637	223	1	in	in	ADP
abrj-637	223	2	addition	addition	NOUN
abrj-637	223	3	,	,	PUNCT
abrj-637	223	4	the	the	DET
abrj-637	223	5	system	system	NOUN
abrj-637	223	6	also	also	ADV
abrj-637	223	7	promotes	promote	VERB
abrj-637	223	8	dashboard	dashboard	NOUN
abrj-637	223	9	analytics	analytic	NOUN
abrj-637	223	10	,	,	PUNCT
abrj-637	223	11	so	so	SCONJ
abrj-637	223	12	the	the	DET
abrj-637	223	13	auditors	auditor	NOUN
abrj-637	223	14	are	be	AUX
abrj-637	223	15	able	able	ADJ
abrj-637	223	16	to	to	PART
abrj-637	223	17	visualize	visualize	VERB
abrj-637	223	18	risk	risk	NOUN
abrj-637	223	19	scores	score	NOUN
abrj-637	223	20	,	,	PUNCT
abrj-637	223	21	monitor	monitor	VERB
abrj-637	223	22	trends	trend	NOUN
abrj-637	223	23	of	of	ADP
abrj-637	223	24	frauds	fraud	NOUN
abrj-637	223	25	and	and	CCONJ
abrj-637	223	26	get	get	VERB
abrj-637	223	27	alerts	alert	NOUN
abrj-637	223	28	on	on	ADP
abrj-637	223	29	high	high	ADJ
abrj-637	223	30	-	-	PUNCT
abrj-637	223	31	risk	risk	NOUN
abrj-637	223	32	filing	filing	NOUN
abrj-637	223	33	.	.	PUNCT
abrj-637	224	1	this	this	PRON
abrj-637	224	2	is	be	AUX
abrj-637	224	3	an	an	DET
abrj-637	224	4	automated	automate	VERB
abrj-637	224	5	workflow	workflow	NOUN
abrj-637	224	6	which	which	PRON
abrj-637	224	7	improves	improve	VERB
abrj-637	224	8	efficiency	efficiency	NOUN
abrj-637	224	9	of	of	ADP
abrj-637	224	10	the	the	DET
abrj-637	224	11	auditors	auditor	NOUN
abrj-637	224	12	since	since	SCONJ
abrj-637	224	13	they	they	PRON
abrj-637	224	14	do	do	AUX
abrj-637	224	15	not	not	PART
abrj-637	224	16	have	have	VERB
abrj-637	224	17	to	to	PART
abrj-637	224	18	be	be	AUX
abrj-637	224	19	occupied	occupy	VERB
abrj-637	224	20	with	with	ADP
abrj-637	224	21	routine	routine	ADJ
abrj-637	224	22	checks	check	NOUN
abrj-637	224	23	but	but	CCONJ
abrj-637	224	24	concentrate	concentrate	VERB
abrj-637	224	25	on	on	ADP
abrj-637	224	26	the	the	DET
abrj-637	224	27	most	most	ADV
abrj-637	224	28	important	important	ADJ
abrj-637	224	29	cases	case	NOUN
abrj-637	224	30	.	.	PUNCT
abrj-637	225	1	3.16	3.16	NUM
abrj-637	225	2	.	.	PUNCT
abrj-637	226	1	performance	performance	NOUN
abrj-637	226	2	optimization	optimization	NOUN
abrj-637	226	3	a	a	DET
abrj-637	226	4	number	number	NOUN
abrj-637	226	5	of	of	ADP
abrj-637	226	6	methods	method	NOUN
abrj-637	226	7	are	be	AUX
abrj-637	226	8	used	use	VERB
abrj-637	226	9	to	to	PART
abrj-637	226	10	guarantee	guarantee	VERB
abrj-637	226	11	that	that	SCONJ
abrj-637	226	12	the	the	DET
abrj-637	226	13	performance	performance	NOUN
abrj-637	226	14	of	of	ADP
abrj-637	226	15	the	the	DET
abrj-637	226	16	developed	develop	VERB
abrj-637	226	17	fraud	fraud	NOUN
abrj-637	226	18	detection	detection	NOUN
abrj-637	226	19	system	system	NOUN
abrj-637	226	20	can	can	AUX
abrj-637	226	21	be	be	AUX
abrj-637	226	22	optimized	optimize	VERB
abrj-637	226	23	,	,	PUNCT
abrj-637	226	24	among	among	ADP
abrj-637	226	25	them	they	PRON
abrj-637	226	26	being	be	AUX
abrj-637	226	27	parallel	parallel	ADJ
abrj-637	226	28	processing	processing	NOUN
abrj-637	226	29	in	in	ADP
abrj-637	226	30	the	the	DET
abrj-637	226	31	gpus	gpus	PROPN
abrj-637	226	32	or	or	CCONJ
abrj-637	226	33	cloud	cloud	NOUN
abrj-637	226	34	clusters	cluster	NOUN
abrj-637	226	35	to	to	PART
abrj-637	226	36	train	train	VERB
abrj-637	226	37	models	model	NOUN
abrj-637	226	38	faster	fast	ADV
abrj-637	226	39	and	and	CCONJ
abrj-637	226	40	also	also	ADV
abrj-637	226	41	model	model	VERB
abrj-637	226	42	pruning	prune	VERB
abrj-637	226	43	to	to	PART
abrj-637	226	44	eliminate	eliminate	VERB
abrj-637	226	45	irrelevant	irrelevant	ADJ
abrj-637	226	46	parameters	parameter	NOUN
abrj-637	226	47	and	and	CCONJ
abrj-637	226	48	speed	speed	VERB
abrj-637	226	49	up	up	ADP
abrj-637	226	50	the	the	DET
abrj-637	226	51	process	process	NOUN
abrj-637	226	52	of	of	ADP
abrj-637	226	53	inference	inference	NOUN
abrj-637	226	54	.	.	PUNCT
abrj-637	227	1	it	it	PRON
abrj-637	227	2	applies	apply	VERB
abrj-637	227	3	real	real	ADJ
abrj-637	227	4	-	-	PUNCT
abrj-637	227	5	time	time	NOUN
abrj-637	227	6	processing	processing	NOUN
abrj-637	227	7	,	,	PUNCT
abrj-637	227	8	whereby	whereby	SCONJ
abrj-637	227	9	the	the	DET
abrj-637	227	10	system	system	NOUN
abrj-637	227	11	tracks	track	VERB
abrj-637	227	12	the	the	DET
abrj-637	227	13	corporate	corporate	ADJ
abrj-637	227	14	filings	filing	NOUN
abrj-637	227	15	as	as	SCONJ
abrj-637	227	16	they	they	PRON
abrj-637	227	17	are	be	AUX
abrj-637	227	18	received	receive	VERB
abrj-637	227	19	and	and	CCONJ
abrj-637	227	20	gives	give	VERB
abrj-637	227	21	instant	instant	ADJ
abrj-637	227	22	fraud	fraud	NOUN
abrj-637	227	23	risk	risk	NOUN
abrj-637	227	24	analysis	analysis	NOUN
abrj-637	227	25	.	.	PUNCT
abrj-637	228	1	moreover	moreover	ADV
abrj-637	228	2	,	,	PUNCT
abrj-637	228	3	the	the	DET
abrj-637	228	4	system	system	NOUN
abrj-637	228	5	will	will	AUX
abrj-637	228	6	be	be	AUX
abrj-637	228	7	developed	develop	VERB
abrj-637	228	8	in	in	ADP
abrj-637	228	9	a	a	DET
abrj-637	228	10	way	way	NOUN
abrj-637	228	11	that	that	PRON
abrj-637	228	12	it	it	PRON
abrj-637	228	13	will	will	AUX
abrj-637	228	14	constantly	constantly	ADV
abrj-637	228	15	become	become	VERB
abrj-637	228	16	better	well	ADJ
abrj-637	228	17	.	.	PUNCT
abrj-637	229	1	the	the	DET
abrj-637	229	2	models	model	NOUN
abrj-637	229	3	can	can	AUX
abrj-637	229	4	be	be	AUX
abrj-637	229	5	retrained	retrain	VERB
abrj-637	229	6	and	and	CCONJ
abrj-637	229	7	refined	refine	VERB
abrj-637	229	8	as	as	SCONJ
abrj-637	229	9	additional	additional	ADJ
abrj-637	229	10	data	datum	NOUN
abrj-637	229	11	is	be	AUX
abrj-637	229	12	made	make	VERB
abrj-637	229	13	available	available	ADJ
abrj-637	229	14	and	and	CCONJ
abrj-637	229	15	so	so	ADV
abrj-637	229	16	the	the	DET
abrj-637	229	17	system	system	NOUN
abrj-637	229	18	will	will	AUX
abrj-637	229	19	be	be	AUX
abrj-637	229	20	effective	effective	ADJ
abrj-637	229	21	in	in	ADP
abrj-637	229	22	detecting	detect	VERB
abrj-637	229	23	new	new	ADJ
abrj-637	229	24	types	type	NOUN
abrj-637	229	25	of	of	ADP
abrj-637	229	26	tax	tax	NOUN
abrj-637	229	27	fraud	fraud	NOUN
abrj-637	229	28	.	.	PUNCT
abrj-637	230	1	asian	asian	ADJ
abrj-637	230	2	business	business	PROPN
abrj-637	230	3	research	research	NOUN
abrj-637	230	4	journal	journal	NOUN
abrj-637	230	5	,	,	PUNCT
abrj-637	230	6	2025	2025	NUM
abrj-637	230	7	,	,	PUNCT
abrj-637	230	8	10(11	10(11	NUM
abrj-637	230	9	):	):	PUNCT
abrj-637	230	10	1	1	NUM
abrj-637	230	11	-	-	SYM
abrj-637	230	12	13	13	NUM
abrj-637	230	13	10	10	NUM
abrj-637	230	14	©	©	PROPN
abrj-637	230	15	2025	2025	NUM
abrj-637	230	16	by	by	ADP
abrj-637	230	17	the	the	DET
abrj-637	230	18	authors	author	NOUN
abrj-637	230	19	;	;	PUNCT
abrj-637	230	20	licensee	licensee	NOUN
abrj-637	230	21	eastern	eastern	ADJ
abrj-637	230	22	centre	centre	NOUN
abrj-637	230	23	of	of	ADP
abrj-637	230	24	science	science	NOUN
abrj-637	230	25	and	and	CCONJ
abrj-637	230	26	education	education	NOUN
abrj-637	230	27	,	,	PUNCT
abrj-637	230	28	usa	usa	PROPN
abrj-637	230	29	3.17	3.17	NUM
abrj-637	230	30	.	.	PUNCT
abrj-637	231	1	security	security	NOUN
abrj-637	231	2	,	,	PUNCT
abrj-637	231	3	privacy	privacy	NOUN
abrj-637	231	4	and	and	CCONJ
abrj-637	231	5	governance	governance	NOUN
abrj-637	231	6	in	in	ADP
abrj-637	231	7	the	the	DET
abrj-637	231	8	design	design	NOUN
abrj-637	231	9	of	of	ADP
abrj-637	231	10	the	the	DET
abrj-637	231	11	fraud	fraud	NOUN
abrj-637	231	12	detection	detection	NOUN
abrj-637	231	13	system	system	NOUN
abrj-637	231	14	due	due	ADP
abrj-637	231	15	to	to	ADP
abrj-637	231	16	the	the	DET
abrj-637	231	17	sensitivity	sensitivity	NOUN
abrj-637	231	18	of	of	ADP
abrj-637	231	19	tax	tax	NOUN
abrj-637	231	20	data	datum	NOUN
abrj-637	231	21	,	,	PUNCT
abrj-637	231	22	security	security	NOUN
abrj-637	231	23	,	,	PUNCT
abrj-637	231	24	privacy	privacy	NOUN
abrj-637	231	25	and	and	CCONJ
abrj-637	231	26	governance	governance	NOUN
abrj-637	231	27	are	be	AUX
abrj-637	231	28	of	of	ADP
abrj-637	231	29	top	top	ADJ
abrj-637	231	30	priority	priority	NOUN
abrj-637	231	31	.	.	PUNCT
abrj-637	232	1	the	the	DET
abrj-637	232	2	information	information	NOUN
abrj-637	232	3	is	be	AUX
abrj-637	232	4	encrypted	encrypt	VERB
abrj-637	232	5	when	when	SCONJ
abrj-637	232	6	it	it	PRON
abrj-637	232	7	is	be	AUX
abrj-637	232	8	being	be	AUX
abrj-637	232	9	transferred	transfer	VERB
abrj-637	232	10	and	and	CCONJ
abrj-637	232	11	when	when	SCONJ
abrj-637	232	12	it	it	PRON
abrj-637	232	13	is	be	AUX
abrj-637	232	14	stored	store	VERB
abrj-637	232	15	so	so	SCONJ
abrj-637	232	16	that	that	SCONJ
abrj-637	232	17	it	it	PRON
abrj-637	232	18	is	be	AUX
abrj-637	232	19	not	not	PART
abrj-637	232	20	accessed	access	VERB
abrj-637	232	21	by	by	ADP
abrj-637	232	22	unauthorized	unauthorized	ADJ
abrj-637	232	23	parties	party	NOUN
abrj-637	232	24	.	.	PUNCT
abrj-637	233	1	role	role	NOUN
abrj-637	233	2	-	-	PUNCT
abrj-637	233	3	based	base	VERB
abrj-637	233	4	access	access	NOUN
abrj-637	233	5	control	control	NOUN
abrj-637	233	6	is	be	AUX
abrj-637	233	7	installed	instal	VERB
abrj-637	233	8	to	to	PART
abrj-637	233	9	make	make	VERB
abrj-637	233	10	sure	sure	ADJ
abrj-637	233	11	that	that	SCONJ
abrj-637	233	12	only	only	ADV
abrj-637	233	13	the	the	DET
abrj-637	233	14	authorized	authorized	ADJ
abrj-637	233	15	people	people	NOUN
abrj-637	233	16	will	will	AUX
abrj-637	233	17	have	have	VERB
abrj-637	233	18	access	access	NOUN
abrj-637	233	19	to	to	ADP
abrj-637	233	20	sensitive	sensitive	ADJ
abrj-637	233	21	data	datum	NOUN
abrj-637	233	22	.	.	PUNCT
abrj-637	234	1	moreover	moreover	ADV
abrj-637	234	2	,	,	PUNCT
abrj-637	234	3	the	the	DET
abrj-637	234	4	system	system	NOUN
abrj-637	234	5	is	be	AUX
abrj-637	234	6	in	in	ADP
abrj-637	234	7	line	line	NOUN
abrj-637	234	8	with	with	ADP
abrj-637	234	9	data	datum	NOUN
abrj-637	234	10	protection	protection	NOUN
abrj-637	234	11	policies	policy	NOUN
abrj-637	234	12	,	,	PUNCT
abrj-637	234	13	including	include	VERB
abrj-637	234	14	gdpr	gdpr	NOUN
abrj-637	234	15	and	and	CCONJ
abrj-637	234	16	ccpa	ccpa	VERB
abrj-637	234	17	,	,	PUNCT
abrj-637	234	18	and	and	CCONJ
abrj-637	234	19	the	the	DET
abrj-637	234	20	information	information	NOUN
abrj-637	234	21	of	of	ADP
abrj-637	234	22	the	the	DET
abrj-637	234	23	taxpayer	taxpayer	NOUN
abrj-637	234	24	is	be	AUX
abrj-637	234	25	processed	process	VERB
abrj-637	234	26	with	with	ADP
abrj-637	234	27	the	the	DET
abrj-637	234	28	utmost	utmost	ADJ
abrj-637	234	29	degree	degree	NOUN
abrj-637	234	30	of	of	ADP
abrj-637	234	31	privacy	privacy	NOUN
abrj-637	234	32	.	.	PUNCT
abrj-637	235	1	there	there	PRON
abrj-637	235	2	is	be	VERB
abrj-637	235	3	also	also	ADV
abrj-637	235	4	an	an	DET
abrj-637	235	5	audit	audit	NOUN
abrj-637	235	6	logging	log	VERB
abrj-637	235	7	option	option	NOUN
abrj-637	235	8	that	that	PRON
abrj-637	235	9	can	can	AUX
abrj-637	235	10	be	be	AUX
abrj-637	235	11	used	use	VERB
abrj-637	235	12	to	to	PART
abrj-637	235	13	monitor	monitor	VERB
abrj-637	235	14	the	the	DET
abrj-637	235	15	activities	activity	NOUN
abrj-637	235	16	of	of	ADP
abrj-637	235	17	the	the	DET
abrj-637	235	18	system	system	NOUN
abrj-637	235	19	and	and	CCONJ
abrj-637	235	20	be	be	AUX
abrj-637	235	21	able	able	ADJ
abrj-637	235	22	to	to	PART
abrj-637	235	23	hold	hold	VERB
abrj-637	235	24	the	the	DET
abrj-637	235	25	accountability	accountability	NOUN
abrj-637	235	26	and	and	CCONJ
abrj-637	235	27	traceability	traceability	NOUN
abrj-637	235	28	of	of	ADP
abrj-637	235	29	decisions	decision	NOUN
abrj-637	235	30	the	the	DET
abrj-637	235	31	machine	machine	NOUN
abrj-637	235	32	learning	learning	NOUN
abrj-637	235	33	models	model	NOUN
abrj-637	235	34	make	make	VERB
abrj-637	235	35	.	.	PUNCT
abrj-637	236	1	4	4	X
abrj-637	236	2	.	.	X
abrj-637	236	3	case	case	NOUN
abrj-637	236	4	study	study	NOUN
abrj-637	236	5	/example	/example	NOUN
abrj-637	236	6	application	application	NOUN
abrj-637	236	7	in	in	ADP
abrj-637	236	8	order	order	NOUN
abrj-637	236	9	to	to	PART
abrj-637	236	10	show	show	VERB
abrj-637	236	11	the	the	DET
abrj-637	236	12	relevance	relevance	NOUN
abrj-637	236	13	and	and	CCONJ
abrj-637	236	14	usefulness	usefulness	NOUN
abrj-637	236	15	of	of	ADP
abrj-637	236	16	the	the	DET
abrj-637	236	17	proposed	propose	VERB
abrj-637	236	18	machine	machine	NOUN
abrj-637	236	19	learning	learning	NOUN
abrj-637	236	20	-	-	PUNCT
abrj-637	236	21	based	base	VERB
abrj-637	236	22	scheme	scheme	NOUN
abrj-637	236	23	to	to	PART
abrj-637	236	24	detect	detect	VERB
abrj-637	236	25	tax	tax	NOUN
abrj-637	236	26	frauds	fraud	NOUN
abrj-637	236	27	and	and	CCONJ
abrj-637	236	28	risk	risk	NOUN
abrj-637	236	29	score	score	NOUN
abrj-637	236	30	,	,	PUNCT
abrj-637	236	31	the	the	DET
abrj-637	236	32	following	follow	VERB
abrj-637	236	33	section	section	NOUN
abrj-637	236	34	provides	provide	VERB
abrj-637	236	35	a	a	DET
abrj-637	236	36	hypothetical	hypothetical	ADJ
abrj-637	236	37	case	case	NOUN
abrj-637	236	38	study	study	NOUN
abrj-637	236	39	that	that	PRON
abrj-637	236	40	simulates	simulate	VERB
abrj-637	236	41	real	real	ADJ
abrj-637	236	42	-	-	PUNCT
abrj-637	236	43	world	world	NOUN
abrj-637	236	44	corporate	corporate	ADJ
abrj-637	236	45	tax	tax	NOUN
abrj-637	236	46	filings	filing	NOUN
abrj-637	236	47	.	.	PUNCT
abrj-637	237	1	the	the	DET
abrj-637	237	2	case	case	NOUN
abrj-637	237	3	exemplifies	exemplify	VERB
abrj-637	237	4	how	how	SCONJ
abrj-637	237	5	the	the	DET
abrj-637	237	6	supervised	supervised	ADJ
abrj-637	237	7	and	and	CCONJ
abrj-637	237	8	the	the	DET
abrj-637	237	9	unsupervised	unsupervised	ADJ
abrj-637	237	10	learning	learning	NOUN
abrj-637	237	11	models	model	NOUN
abrj-637	237	12	would	would	AUX
abrj-637	237	13	be	be	AUX
abrj-637	237	14	useful	useful	ADJ
abrj-637	237	15	in	in	ADP
abrj-637	237	16	the	the	DET
abrj-637	237	17	identification	identification	NOUN
abrj-637	237	18	of	of	ADP
abrj-637	237	19	the	the	DET
abrj-637	237	20	possible	possible	ADJ
abrj-637	237	21	fraudulent	fraudulent	ADJ
abrj-637	237	22	activity	activity	NOUN
abrj-637	237	23	and	and	CCONJ
abrj-637	237	24	the	the	DET
abrj-637	237	25	allocation	allocation	NOUN
abrj-637	237	26	of	of	ADP
abrj-637	237	27	the	the	DET
abrj-637	237	28	quantitative	quantitative	ADJ
abrj-637	237	29	risk	risk	NOUN
abrj-637	237	30	scores	score	NOUN
abrj-637	237	31	that	that	PRON
abrj-637	237	32	would	would	AUX
abrj-637	237	33	serve	serve	VERB
abrj-637	237	34	as	as	ADP
abrj-637	237	35	the	the	DET
abrj-637	237	36	basis	basis	NOUN
abrj-637	237	37	of	of	ADP
abrj-637	237	38	targeted	target	VERB
abrj-637	237	39	audits	audit	NOUN
abrj-637	237	40	.	.	PUNCT
abrj-637	238	1	4.1	4.1	NUM
abrj-637	238	2	.	.	PUNCT
abrj-637	238	3	case	case	NOUN
abrj-637	238	4	study	study	NOUN
abrj-637	238	5	overview	overview	VERB
abrj-637	238	6	the	the	DET
abrj-637	238	7	simulated	simulate	VERB
abrj-637	238	8	data	data	NOUN
abrj-637	238	9	consists	consist	VERB
abrj-637	238	10	of	of	ADP
abrj-637	238	11	5,000	5,000	NUM
abrj-637	238	12	corporate	corporate	ADJ
abrj-637	238	13	filing	filing	NOUN
abrj-637	238	14	records	record	NOUN
abrj-637	238	15	that	that	PRON
abrj-637	238	16	belong	belong	VERB
abrj-637	238	17	to	to	ADP
abrj-637	238	18	various	various	ADJ
abrj-637	238	19	industries	industry	NOUN
abrj-637	238	20	,	,	PUNCT
abrj-637	238	21	such	such	ADJ
abrj-637	238	22	as	as	ADP
abrj-637	238	23	manufacturing	manufacturing	NOUN
abrj-637	238	24	,	,	PUNCT
abrj-637	238	25	services	service	NOUN
abrj-637	238	26	,	,	PUNCT
abrj-637	238	27	and	and	CCONJ
abrj-637	238	28	technology	technology	NOUN
abrj-637	238	29	industries	industry	NOUN
abrj-637	238	30	.	.	PUNCT
abrj-637	239	1	they	they	PRON
abrj-637	239	2	consist	consist	VERB
abrj-637	239	3	of	of	ADP
abrj-637	239	4	structured	structured	ADJ
abrj-637	239	5	financial	financial	ADJ
abrj-637	239	6	information	information	NOUN
abrj-637	239	7	(	(	PUNCT
abrj-637	239	8	e.g.	e.g.	ADV
abrj-637	239	9	revenue	revenue	NOUN
abrj-637	239	10	,	,	PUNCT
abrj-637	239	11	assets	asset	NOUN
abrj-637	239	12	,	,	PUNCT
abrj-637	239	13	tax	tax	NOUN
abrj-637	239	14	payable	payable	NOUN
abrj-637	239	15	)	)	PUNCT
abrj-637	239	16	and	and	CCONJ
abrj-637	239	17	unstructured	unstructured	ADJ
abrj-637	239	18	text	text	NOUN
abrj-637	239	19	information	information	NOUN
abrj-637	239	20	(	(	PUNCT
abrj-637	239	21	e.g.	e.g.	ADV
abrj-637	239	22	management	management	NOUN
abrj-637	239	23	commentary	commentary	NOUN
abrj-637	239	24	,	,	PUNCT
abrj-637	239	25	auditor	auditor	NOUN
abrj-637	239	26	notes	note	NOUN
abrj-637	239	27	)	)	PUNCT
abrj-637	239	28	.	.	PUNCT
abrj-637	240	1	the	the	DET
abrj-637	240	2	percentage	percentage	NOUN
abrj-637	240	3	of	of	ADP
abrj-637	240	4	these	these	DET
abrj-637	240	5	filings	filing	NOUN
abrj-637	240	6	is	be	AUX
abrj-637	240	7	around	around	ADV
abrj-637	240	8	8	8	NUM
abrj-637	240	9	percent	percent	NOUN
abrj-637	240	10	,	,	PUNCT
abrj-637	240	11	which	which	PRON
abrj-637	240	12	is	be	AUX
abrj-637	240	13	labeled	label	VERB
abrj-637	240	14	as	as	ADP
abrj-637	240	15	the	the	DET
abrj-637	240	16	fraudulent	fraudulent	ADJ
abrj-637	240	17	ones	one	NOUN
abrj-637	240	18	according	accord	VERB
abrj-637	240	19	to	to	ADP
abrj-637	240	20	the	the	DET
abrj-637	240	21	past	past	ADJ
abrj-637	240	22	audit	audit	NOUN
abrj-637	240	23	results	result	NOUN
abrj-637	240	24	,	,	PUNCT
abrj-637	240	25	which	which	PRON
abrj-637	240	26	is	be	AUX
abrj-637	240	27	a	a	DET
abrj-637	240	28	moderately	moderately	ADV
abrj-637	240	29	unbalanced	unbalanced	ADJ
abrj-637	240	30	data	datum	NOUN
abrj-637	240	31	(	(	PUNCT
abrj-637	240	32	wu	wu	PROPN
abrj-637	240	33	et	et	PROPN
abrj-637	240	34	al	al	PROPN
abrj-637	240	35	.	.	PROPN
abrj-637	240	36	,	,	PUNCT
abrj-637	240	37	2012	2012	NUM
abrj-637	240	38	;	;	PUNCT
abrj-637	240	39	martínez	martínez	NOUN
abrj-637	240	40	,	,	PUNCT
abrj-637	240	41	2025	2025	NUM
abrj-637	240	42	)	)	PUNCT
abrj-637	240	43	.	.	PUNCT
abrj-637	241	1	this	this	DET
abrj-637	241	2	case	case	NOUN
abrj-637	241	3	study	study	NOUN
abrj-637	241	4	aims	aim	VERB
abrj-637	241	5	to	to	PART
abrj-637	241	6	emphasize	emphasize	VERB
abrj-637	241	7	how	how	SCONJ
abrj-637	241	8	machine	machine	NOUN
abrj-637	241	9	learning	learning	NOUN
abrj-637	241	10	models	model	NOUN
abrj-637	241	11	,	,	PUNCT
abrj-637	241	12	which	which	PRON
abrj-637	241	13	are	be	AUX
abrj-637	241	14	random	random	ADJ
abrj-637	241	15	forest	forest	NOUN
abrj-637	241	16	,	,	PUNCT
abrj-637	241	17	xgboost	xgboost	ADV
abrj-637	241	18	,	,	PUNCT
abrj-637	241	19	and	and	CCONJ
abrj-637	241	20	autoencoder	autoencoder	NOUN
abrj-637	241	21	structures	structure	NOUN
abrj-637	241	22	,	,	PUNCT
abrj-637	241	23	can	can	AUX
abrj-637	241	24	be	be	AUX
abrj-637	241	25	used	use	VERB
abrj-637	241	26	to	to	PART
abrj-637	241	27	identify	identify	VERB
abrj-637	241	28	fraudulent	fraudulent	ADJ
abrj-637	241	29	behavior	behavior	NOUN
abrj-637	241	30	and	and	CCONJ
abrj-637	241	31	give	give	VERB
abrj-637	241	32	meaningful	meaningful	ADJ
abrj-637	241	33	risk	risk	NOUN
abrj-637	241	34	scores	score	NOUN
abrj-637	241	35	.	.	PUNCT
abrj-637	242	1	4.2	4.2	NUM
abrj-637	242	2	.	.	PUNCT
abrj-637	242	3	preparation	preparation	NOUN
abrj-637	242	4	of	of	ADP
abrj-637	242	5	data	datum	NOUN
abrj-637	242	6	and	and	CCONJ
abrj-637	242	7	extraction	extraction	NOUN
abrj-637	242	8	of	of	ADP
abrj-637	242	9	features	feature	NOUN
abrj-637	242	10	.	.	PUNCT
abrj-637	243	1	the	the	DET
abrj-637	243	2	pipeline	pipeline	NOUN
abrj-637	243	3	followed	follow	VERB
abrj-637	243	4	in	in	ADP
abrj-637	243	5	the	the	DET
abrj-637	243	6	preprocessing	preprocessing	NOUN
abrj-637	243	7	of	of	ADP
abrj-637	243	8	data	datum	NOUN
abrj-637	243	9	involved	involve	VERB
abrj-637	243	10	the	the	DET
abrj-637	243	11	methodology	methodology	NOUN
abrj-637	243	12	section	section	NOUN
abrj-637	243	13	.	.	PUNCT
abrj-637	244	1	numerical	numerical	PROPN
abrj-637	244	2	characteristics	characteristics	PROPN
abrj-637	244	3	were	be	AUX
abrj-637	244	4	normalized	normalize	VERB
abrj-637	244	5	,	,	PUNCT
abrj-637	244	6	categorical	categorical	ADJ
abrj-637	244	7	ones	one	NOUN
abrj-637	244	8	coded	code	VERB
abrj-637	244	9	,	,	PUNCT
abrj-637	244	10	and	and	CCONJ
abrj-637	244	11	missing	miss	VERB
abrj-637	244	12	values	value	NOUN
abrj-637	244	13	were	be	AUX
abrj-637	244	14	addressed	address	VERB
abrj-637	244	15	with	with	ADP
abrj-637	244	16	the	the	DET
abrj-637	244	17	help	help	NOUN
abrj-637	244	18	of	of	ADP
abrj-637	244	19	interpolation	interpolation	NOUN
abrj-637	244	20	.	.	PUNCT
abrj-637	245	1	in	in	ADP
abrj-637	245	2	the	the	DET
abrj-637	245	3	case	case	NOUN
abrj-637	245	4	of	of	ADP
abrj-637	245	5	text	text	NOUN
abrj-637	245	6	data	datum	NOUN
abrj-637	245	7	,	,	PUNCT
abrj-637	245	8	the	the	DET
abrj-637	245	9	linguistic	linguistic	ADJ
abrj-637	245	10	data	datum	NOUN
abrj-637	245	11	mining	mining	NOUN
abrj-637	245	12	(	(	PUNCT
abrj-637	245	13	tf	tf	PROPN
abrj-637	245	14	-	-	PUNCT
abrj-637	245	15	idf	idf	NOUN
abrj-637	245	16	)	)	PUNCT
abrj-637	245	17	and	and	CCONJ
abrj-637	245	18	sentiment	sentiment	NOUN
abrj-637	245	19	analysis	analysis	NOUN
abrj-637	245	20	methods	method	NOUN
abrj-637	245	21	were	be	AUX
abrj-637	245	22	employed	employ	VERB
abrj-637	245	23	to	to	PART
abrj-637	245	24	extract	extract	VERB
abrj-637	245	25	linguistic	linguistic	ADJ
abrj-637	245	26	evidence	evidence	NOUN
abrj-637	245	27	of	of	ADP
abrj-637	245	28	deceit	deceit	NOUN
abrj-637	245	29	(	(	PUNCT
abrj-637	245	30	zhang	zhang	PROPN
abrj-637	245	31	et	et	PROPN
abrj-637	245	32	al	al	PROPN
abrj-637	245	33	.	.	PROPN
abrj-637	245	34	,	,	PUNCT
abrj-637	245	35	2024	2024	NUM
abrj-637	245	36	;	;	PUNCT
abrj-637	245	37	ji	ji	PROPN
abrj-637	245	38	et	et	PROPN
abrj-637	245	39	al	al	PROPN
abrj-637	245	40	.	.	PROPN
abrj-637	245	41	,	,	PUNCT
abrj-637	245	42	2024	2024	NUM
abrj-637	245	43	)	)	PUNCT
abrj-637	245	44	.	.	PUNCT
abrj-637	246	1	4.2.1	4.2.1	X
abrj-637	246	2	.	.	PUNCT
abrj-637	247	1	key	key	ADJ
abrj-637	247	2	features	feature	NOUN
abrj-637	247	3	included	include	VERB
abrj-637	247	4	•	•	NUM
abrj-637	247	5	financial	financial	ADJ
abrj-637	247	6	ratios	ratio	NOUN
abrj-637	247	7	(	(	PUNCT
abrj-637	247	8	e.g.	e.g.	ADV
abrj-637	247	9	effective	effective	ADJ
abrj-637	247	10	tax	tax	NOUN
abrj-637	247	11	rate	rate	NOUN
abrj-637	247	12	,	,	PUNCT
abrj-637	247	13	debt	debt	NOUN
abrj-637	247	14	-	-	PUNCT
abrj-637	247	15	to	to	ADP
abrj-637	247	16	-	-	PUNCT
abrj-637	247	17	equity	equity	NOUN
abrj-637	247	18	,	,	PUNCT
abrj-637	247	19	profit	profit	NOUN
abrj-637	247	20	margin	margin	NOUN
abrj-637	247	21	)	)	PUNCT
abrj-637	247	22	•	•	NUM
abrj-637	247	23	late	late	ADJ
abrj-637	247	24	filings	filing	NOUN
abrj-637	247	25	,	,	PUNCT
abrj-637	247	26	restatements	restatement	NOUN
abrj-637	247	27	,	,	PUNCT
abrj-637	247	28	and	and	CCONJ
abrj-637	247	29	amendments	amendment	NOUN
abrj-637	247	30	(	(	PUNCT
abrj-637	247	31	behavior	behavior	NOUN
abrj-637	247	32	)	)	PUNCT
abrj-637	247	33	.	.	PUNCT
abrj-637	248	1	•	•	NUM
abrj-637	248	2	text	text	NOUN
abrj-637	248	3	sentiment	sentiment	NOUN
abrj-637	248	4	(	(	PUNCT
abrj-637	248	5	e.g.	e.g.	ADV
abrj-637	248	6	use	use	NOUN
abrj-637	248	7	of	of	ADP
abrj-637	248	8	too	too	ADV
abrj-637	248	9	much	much	ADJ
abrj-637	248	10	positive	positive	ADJ
abrj-637	248	11	tone	tone	NOUN
abrj-637	248	12	or	or	CCONJ
abrj-637	248	13	evasive	evasive	ADJ
abrj-637	248	14	language	language	NOUN
abrj-637	248	15	)	)	PUNCT
abrj-637	248	16	the	the	DET
abrj-637	248	17	data	datum	NOUN
abrj-637	248	18	was	be	AUX
abrj-637	248	19	subsequently	subsequently	ADV
abrj-637	248	20	divided	divide	VERB
abrj-637	248	21	into	into	ADP
abrj-637	248	22	70	70	NUM
abrj-637	248	23	percent	percent	NOUN
abrj-637	248	24	training	training	NOUN
abrj-637	248	25	and	and	CCONJ
abrj-637	248	26	30	30	NUM
abrj-637	248	27	percent	percent	NOUN
abrj-637	248	28	-	-	PUNCT
abrj-637	248	29	testing	test	VERB
abrj-637	248	30	data	datum	NOUN
abrj-637	248	31	where	where	SCONJ
abrj-637	248	32	stratification	stratification	NOUN
abrj-637	248	33	of	of	ADP
abrj-637	248	34	classes	class	NOUN
abrj-637	248	35	was	be	AUX
abrj-637	248	36	done	do	VERB
abrj-637	248	37	to	to	PART
abrj-637	248	38	maintain	maintain	VERB
abrj-637	248	39	the	the	DET
abrj-637	248	40	distribution	distribution	NOUN
abrj-637	248	41	of	of	ADP
abrj-637	248	42	fraud	fraud	NOUN
abrj-637	248	43	(	(	PUNCT
abrj-637	248	44	tagbo	tagbo	NOUN
abrj-637	248	45	&	&	CCONJ
abrj-637	248	46	adekoya	adekoya	PROPN
abrj-637	248	47	,	,	PUNCT
abrj-637	248	48	2023	2023	NUM
abrj-637	248	49	)	)	PUNCT
abrj-637	248	50	.	.	PUNCT
abrj-637	249	1	4.3	4.3	NUM
abrj-637	249	2	.	.	PUNCT
abrj-637	249	3	model	model	NOUN
abrj-637	249	4	implementation	implementation	NOUN
abrj-637	249	5	three	three	NUM
abrj-637	249	6	models	model	NOUN
abrj-637	249	7	were	be	AUX
abrj-637	249	8	put	put	VERB
abrj-637	249	9	in	in	ADP
abrj-637	249	10	place	place	NOUN
abrj-637	249	11	in	in	ADP
abrj-637	249	12	order	order	NOUN
abrj-637	249	13	to	to	PART
abrj-637	249	14	evaluate	evaluate	VERB
abrj-637	249	15	them	they	PRON
abrj-637	249	16	comparatively	comparatively	ADV
abrj-637	249	17	:	:	PUNCT
abrj-637	249	18	•	•	NUM
abrj-637	249	19	random	random	ADJ
abrj-637	249	20	forest	forest	NOUN
abrj-637	249	21	(	(	PUNCT
abrj-637	249	22	rf	rf	NUM
abrj-637	249	23	):	):	PUNCT
abrj-637	249	24	this	this	PRON
abrj-637	249	25	is	be	AUX
abrj-637	249	26	a	a	DET
abrj-637	249	27	baseline	baseline	ADJ
abrj-637	249	28	ensemble	ensemble	ADJ
abrj-637	249	29	classifier	classifier	NOUN
abrj-637	249	30	that	that	PRON
abrj-637	249	31	is	be	AUX
abrj-637	249	32	employed	employ	VERB
abrj-637	249	33	to	to	PART
abrj-637	249	34	model	model	VERB
abrj-637	249	35	nonlinear	nonlinear	ADJ
abrj-637	249	36	relationships	relationship	NOUN
abrj-637	249	37	between	between	ADP
abrj-637	249	38	financial	financial	ADJ
abrj-637	249	39	attributes	attribute	NOUN
abrj-637	249	40	(	(	PUNCT
abrj-637	249	41	nguyen	nguyen	NOUN
abrj-637	249	42	,	,	PUNCT
abrj-637	249	43	2025	2025	NUM
abrj-637	249	44	;	;	PUNCT
abrj-637	249	45	acharya	acharya	NOUN
abrj-637	249	46	,	,	PUNCT
abrj-637	249	47	2025	2025	NUM
abrj-637	249	48	)	)	PUNCT
abrj-637	249	49	.	.	PUNCT
abrj-637	250	1	•	•	NUM
abrj-637	250	2	xgboost	xgboost	ADV
abrj-637	250	3	:	:	PUNCT
abrj-637	250	4	it	it	PRON
abrj-637	250	5	is	be	AUX
abrj-637	250	6	a	a	DET
abrj-637	250	7	gradient	gradient	NOUN
abrj-637	250	8	-	-	PUNCT
abrj-637	250	9	boosted	boost	VERB
abrj-637	250	10	ensemble	ensemble	ADJ
abrj-637	250	11	model	model	NOUN
abrj-637	250	12	that	that	PRON
abrj-637	250	13	is	be	AUX
abrj-637	250	14	optimized	optimize	VERB
abrj-637	250	15	on	on	ADP
abrj-637	250	16	imbalanced	imbalanced	ADJ
abrj-637	250	17	data	datum	NOUN
abrj-637	250	18	(	(	PUNCT
abrj-637	250	19	zhou	zhou	X
abrj-637	250	20	et	et	PROPN
abrj-637	250	21	al	al	PROPN
abrj-637	250	22	.	.	PROPN
abrj-637	250	23	,	,	PUNCT
abrj-637	250	24	2024	2024	NUM
abrj-637	250	25	)	)	PUNCT
abrj-637	250	26	.	.	PUNCT
abrj-637	251	1	•	•	NUM
abrj-637	251	2	autoencoder	autoencoder	NOUN
abrj-637	251	3	(	(	PUNCT
abrj-637	251	4	ae	ae	PROPN
abrj-637	251	5	):	):	PUNCT
abrj-637	251	6	this	this	PRON
abrj-637	251	7	is	be	AUX
abrj-637	251	8	an	an	DET
abrj-637	251	9	unsupervised	unsupervised	ADJ
abrj-637	251	10	deep	deep	ADJ
abrj-637	251	11	learning	learning	NOUN
abrj-637	251	12	model	model	NOUN
abrj-637	251	13	that	that	PRON
abrj-637	251	14	is	be	AUX
abrj-637	251	15	trained	train	VERB
abrj-637	251	16	to	to	PART
abrj-637	251	17	understand	understand	VERB
abrj-637	251	18	the	the	DET
abrj-637	251	19	normal	normal	ADJ
abrj-637	251	20	operation	operation	NOUN
abrj-637	251	21	of	of	ADP
abrj-637	251	22	the	the	DET
abrj-637	251	23	financial	financial	ADJ
abrj-637	251	24	aspect	aspect	NOUN
abrj-637	251	25	of	of	ADP
abrj-637	251	26	companies	company	NOUN
abrj-637	251	27	and	and	CCONJ
abrj-637	251	28	identifies	identify	VERB
abrj-637	251	29	anomalies	anomaly	NOUN
abrj-637	251	30	by	by	ADP
abrj-637	251	31	the	the	DET
abrj-637	251	32	error	error	NOUN
abrj-637	251	33	of	of	ADP
abrj-637	251	34	reconstruction	reconstruction	NOUN
abrj-637	251	35	(	(	PUNCT
abrj-637	251	36	choudhary	choudhary	PROPN
abrj-637	251	37	,	,	PUNCT
abrj-637	251	38	2025	2025	NUM
abrj-637	251	39	;	;	PUNCT
abrj-637	251	40	mehta	mehta	PROPN
abrj-637	251	41	et	et	PROPN
abrj-637	251	42	al	al	PROPN
abrj-637	251	43	.	.	PROPN
abrj-637	251	44	,	,	PUNCT
abrj-637	251	45	2022	2022	NUM
abrj-637	251	46	)	)	PUNCT
abrj-637	251	47	.	.	PUNCT
abrj-637	252	1	•	•	NUM
abrj-637	252	2	the	the	DET
abrj-637	252	3	model	model	NOUN
abrj-637	252	4	parameters	parameter	NOUN
abrj-637	252	5	were	be	AUX
abrj-637	252	6	optimized	optimize	VERB
abrj-637	252	7	to	to	PART
abrj-637	252	8	reach	reach	VERB
abrj-637	252	9	the	the	DET
abrj-637	252	10	best	good	ADJ
abrj-637	252	11	precision	precision	NOUN
abrj-637	252	12	-	-	PUNCT
abrj-637	252	13	recall	recall	NOUN
abrj-637	252	14	tradeoff	tradeoff	NOUN
abrj-637	252	15	by	by	ADP
abrj-637	252	16	the	the	DET
abrj-637	252	17	methods	method	NOUN
abrj-637	252	18	of	of	ADP
abrj-637	252	19	grid	grid	NOUN
abrj-637	252	20	search	search	NOUN
abrj-637	252	21	and	and	CCONJ
abrj-637	252	22	the	the	DET
abrj-637	252	23	5	5	NUM
abrj-637	252	24	-	-	ADJ
abrj-637	252	25	fold	fold	ADJ
abrj-637	252	26	cross	cross	NOUN
abrj-637	252	27	-	-	NOUN
abrj-637	252	28	validation	validation	NOUN
abrj-637	252	29	.	.	PUNCT
abrj-637	253	1	4.4	4.4	NUM
abrj-637	253	2	.	.	PUNCT
abrj-637	253	3	results	result	NOUN
abrj-637	253	4	and	and	CCONJ
abrj-637	253	5	performance	performance	NOUN
abrj-637	253	6	evaluation	evaluation	NOUN
abrj-637	253	7	performance	performance	NOUN
abrj-637	253	8	metrics	metric	NOUN
abrj-637	253	9	were	be	AUX
abrj-637	253	10	derived	derive	VERB
abrj-637	253	11	from	from	ADP
abrj-637	253	12	the	the	DET
abrj-637	253	13	test	test	NOUN
abrj-637	253	14	set	set	VERB
abrj-637	253	15	using	use	VERB
abrj-637	253	16	precision	precision	NOUN
abrj-637	253	17	,	,	PUNCT
abrj-637	253	18	recall	recall	NOUN
abrj-637	253	19	,	,	PUNCT
abrj-637	253	20	f1	f1	NOUN
abrj-637	253	21	-	-	PUNCT
abrj-637	253	22	score	score	NOUN
abrj-637	253	23	,	,	PUNCT
abrj-637	253	24	and	and	CCONJ
abrj-637	253	25	roc	roc	NOUN
abrj-637	253	26	-	-	PUNCT
abrj-637	253	27	auc	auc	NOUN
abrj-637	253	28	values	value	NOUN
abrj-637	253	29	(	(	PUNCT
abrj-637	253	30	martínez	martínez	NOUN
abrj-637	253	31	,	,	PUNCT
abrj-637	253	32	2025	2025	NUM
abrj-637	253	33	)	)	PUNCT
abrj-637	253	34	.	.	PUNCT
abrj-637	254	1	table	table	NOUN
abrj-637	254	2	4	4	NUM
abrj-637	254	3	.	.	PUNCT
abrj-637	254	4	performance	performance	NOUN
abrj-637	254	5	metrics	metric	NOUN
abrj-637	254	6	.	.	PUNCT
abrj-637	255	1	model	model	NOUN
abrj-637	255	2	precision	precision	PROPN
abrj-637	255	3	recall	recall	PROPN
abrj-637	255	4	f1	f1	NOUN
abrj-637	255	5	-	-	PUNCT
abrj-637	255	6	score	score	NOUN
abrj-637	255	7	auc	auc	NOUN
abrj-637	255	8	random	random	ADJ
abrj-637	255	9	forest	forest	NOUN
abrj-637	255	10	0.91	0.91	NUM
abrj-637	255	11	0.76	0.76	NUM
abrj-637	255	12	0.83	0.83	NUM
abrj-637	255	13	0.94	0.94	NUM
abrj-637	255	14	xgboost	xgboost	ADP
abrj-637	256	1	0.89	0.89	NUM
abrj-637	256	2	0.82	0.82	NUM
abrj-637	256	3	0.85	0.85	NUM
abrj-637	256	4	0.96	0.96	NUM
abrj-637	256	5	autoencoder	autoencoder	NOUN
abrj-637	256	6	0.72	0.72	NUM
abrj-637	256	7	0.88	0.88	NUM
abrj-637	256	8	0.79	0.79	NUM
abrj-637	256	9	0.89	0.89	NUM
abrj-637	256	10	xgboost	xgboost	ADV
abrj-637	256	11	was	be	AUX
abrj-637	256	12	found	find	VERB
abrj-637	256	13	to	to	PART
abrj-637	256	14	be	be	AUX
abrj-637	256	15	the	the	DET
abrj-637	256	16	best	good	ADJ
abrj-637	256	17	service	service	NOUN
abrj-637	256	18	overall	overall	ADV
abrj-637	256	19	,	,	PUNCT
abrj-637	256	20	and	and	CCONJ
abrj-637	256	21	it	it	PRON
abrj-637	256	22	comes	come	VERB
abrj-637	256	23	with	with	ADP
abrj-637	256	24	a	a	DET
abrj-637	256	25	good	good	ADJ
abrj-637	256	26	balance	balance	NOUN
abrj-637	256	27	between	between	ADP
abrj-637	256	28	precision	precision	NOUN
abrj-637	256	29	and	and	CCONJ
abrj-637	256	30	recall	recall	NOUN
abrj-637	256	31	.	.	PUNCT
abrj-637	257	1	the	the	DET
abrj-637	257	2	unsupervised	unsupervised	ADJ
abrj-637	257	3	autoencoder	autoencoder	NOUN
abrj-637	257	4	was	be	AUX
abrj-637	257	5	useful	useful	ADJ
abrj-637	257	6	in	in	ADP
abrj-637	257	7	the	the	DET
abrj-637	257	8	detection	detection	NOUN
abrj-637	257	9	of	of	ADP
abrj-637	257	10	the	the	DET
abrj-637	257	11	hidden	hide	VERB
abrj-637	257	12	anomalies	anomaly	NOUN
abrj-637	257	13	that	that	SCONJ
abrj-637	257	14	the	the	DET
abrj-637	257	15	supervised	supervised	ADJ
abrj-637	257	16	models	model	NOUN
abrj-637	257	17	were	be	AUX
abrj-637	257	18	unable	unable	ADJ
abrj-637	257	19	to	to	PART
abrj-637	257	20	detect	detect	VERB
abrj-637	257	21	at	at	ADP
abrj-637	257	22	times	time	NOUN
abrj-637	257	23	.	.	PUNCT
abrj-637	258	1	these	these	DET
abrj-637	258	2	findings	finding	NOUN
abrj-637	258	3	are	be	AUX
abrj-637	258	4	consistent	consistent	ADJ
abrj-637	258	5	with	with	ADP
abrj-637	258	6	the	the	DET
abrj-637	258	7	studies	study	NOUN
abrj-637	258	8	of	of	ADP
abrj-637	258	9	zhou	zhou	PROPN
abrj-637	258	10	et	et	PROPN
abrj-637	258	11	al	al	PROPN
abrj-637	258	12	.	.	PROPN
abrj-637	259	1	(	(	PUNCT
abrj-637	259	2	2024	2024	NUM
abrj-637	259	3	)	)	PUNCT
abrj-637	259	4	and	and	CCONJ
abrj-637	259	5	zhang	zhang	PROPN
abrj-637	259	6	et	et	PROPN
abrj-637	259	7	al	al	PROPN
abrj-637	259	8	.	.	PROPN
abrj-637	260	1	(	(	PUNCT
abrj-637	260	2	2025	2025	NUM
abrj-637	260	3	)	)	PUNCT
abrj-637	260	4	,	,	PUNCT
abrj-637	260	5	who	who	PRON
abrj-637	260	6	highlighted	highlight	VERB
abrj-637	260	7	the	the	DET
abrj-637	260	8	relevance	relevance	NOUN
abrj-637	260	9	of	of	ADP
abrj-637	260	10	ensemble	ensemble	ADJ
abrj-637	260	11	and	and	CCONJ
abrj-637	260	12	hybrid	hybrid	ADJ
abrj-637	260	13	learning	learning	NOUN
abrj-637	260	14	methods	method	NOUN
abrj-637	260	15	in	in	ADP
abrj-637	260	16	terms	term	NOUN
abrj-637	260	17	of	of	ADP
abrj-637	260	18	the	the	DET
abrj-637	260	19	highest	high	ADJ
abrj-637	260	20	fraud	fraud	NOUN
abrj-637	260	21	detection	detection	NOUN
abrj-637	260	22	rate	rate	NOUN
abrj-637	260	23	.	.	PUNCT
abrj-637	261	1	4.5	4.5	NUM
abrj-637	261	2	.	.	PUNCT
abrj-637	261	3	scoring	scoring	NOUN
abrj-637	261	4	and	and	CCONJ
abrj-637	261	5	interpretation	interpretation	NOUN
abrj-637	261	6	of	of	ADP
abrj-637	261	7	risk	risk	NOUN
abrj-637	261	8	risk	risk	NOUN
abrj-637	261	9	scoring	scoring	NOUN
abrj-637	261	10	system	system	NOUN
abrj-637	261	11	was	be	AUX
abrj-637	261	12	adopted	adopt	VERB
abrj-637	261	13	after	after	ADP
abrj-637	261	14	the	the	DET
abrj-637	261	15	analysis	analysis	NOUN
abrj-637	261	16	of	of	ADP
abrj-637	261	17	the	the	DET
abrj-637	261	18	models	model	NOUN
abrj-637	261	19	to	to	PART
abrj-637	261	20	convert	convert	VERB
abrj-637	261	21	the	the	DET
abrj-637	261	22	anticipated	anticipate	VERB
abrj-637	261	23	probabilities	probability	NOUN
abrj-637	261	24	into	into	ADP
abrj-637	261	25	understandable	understandable	ADJ
abrj-637	261	26	categories	category	NOUN
abrj-637	261	27	:	:	PUNCT
abrj-637	261	28	asian	asian	ADJ
abrj-637	261	29	business	business	NOUN
abrj-637	261	30	research	research	NOUN
abrj-637	261	31	journal	journal	NOUN
abrj-637	261	32	,	,	PUNCT
abrj-637	261	33	2025	2025	NUM
abrj-637	261	34	,	,	PUNCT
abrj-637	261	35	10(11	10(11	NUM
abrj-637	261	36	):	):	PUNCT
abrj-637	261	37	1	1	NUM
abrj-637	261	38	-	-	SYM
abrj-637	261	39	13	13	NUM
abrj-637	261	40	11	11	NUM
abrj-637	261	41	©	©	ADP
abrj-637	261	42	2025	2025	NUM
abrj-637	261	43	by	by	ADP
abrj-637	261	44	the	the	DET
abrj-637	261	45	authors	author	NOUN
abrj-637	261	46	;	;	PUNCT
abrj-637	261	47	licensee	licensee	NOUN
abrj-637	261	48	eastern	eastern	ADJ
abrj-637	261	49	centre	centre	NOUN
abrj-637	261	50	of	of	ADP
abrj-637	261	51	science	science	NOUN
abrj-637	261	52	and	and	CCONJ
abrj-637	261	53	education	education	NOUN
abrj-637	261	54	,	,	PUNCT
abrj-637	261	55	usa	usa	PROPN
abrj-637	261	56	•	•	NUM
abrj-637	262	1	low	low	ADJ
abrj-637	262	2	risk	risk	NOUN
abrj-637	262	3	(	(	PUNCT
abrj-637	262	4	0	0	NUM
abrj-637	262	5	-0.3	-0.3	NUM
abrj-637	262	6	):	):	PUNCT
abrj-637	262	7	filings	filing	NOUN
abrj-637	262	8	of	of	ADP
abrj-637	262	9	routine	routine	ADJ
abrj-637	262	10	nature	nature	NOUN
abrj-637	262	11	with	with	ADP
abrj-637	262	12	uniform	uniform	ADJ
abrj-637	262	13	trends	trend	NOUN
abrj-637	262	14	.	.	PUNCT
abrj-637	263	1	•	•	NUM
abrj-637	263	2	medium	medium	NOUN
abrj-637	263	3	risk	risk	NOUN
abrj-637	263	4	(	(	PUNCT
abrj-637	263	5	0.31	0.31	NUM
abrj-637	263	6	-0.7	-0.7	NOUN
abrj-637	263	7	):	):	PUNCT
abrj-637	263	8	minor	minor	ADJ
abrj-637	263	9	inconsistencies	inconsistency	NOUN
abrj-637	263	10	that	that	PRON
abrj-637	263	11	need	need	VERB
abrj-637	263	12	to	to	PART
abrj-637	263	13	be	be	AUX
abrj-637	263	14	reviewed	review	VERB
abrj-637	263	15	partially	partially	ADV
abrj-637	263	16	.	.	PUNCT
abrj-637	264	1	•	•	NUM
abrj-637	264	2	high	high	ADJ
abrj-637	264	3	risk	risk	NOUN
abrj-637	264	4	(	(	PUNCT
abrj-637	264	5	0.711.0	0.711.0	NUM
abrj-637	264	6	):	):	PUNCT
abrj-637	264	7	there	there	PRON
abrj-637	264	8	are	be	VERB
abrj-637	264	9	great	great	ADJ
abrj-637	264	10	signals	signal	NOUN
abrj-637	264	11	of	of	ADP
abrj-637	264	12	potential	potential	ADJ
abrj-637	264	13	evasion	evasion	NOUN
abrj-637	264	14	or	or	CCONJ
abrj-637	264	15	misstatement	misstatement	NOUN
abrj-637	264	16	.	.	PUNCT
abrj-637	265	1	each	each	PRON
abrj-637	265	2	of	of	ADP
abrj-637	265	3	the	the	DET
abrj-637	265	4	models	model	NOUN
abrj-637	265	5	predictions	prediction	NOUN
abrj-637	265	6	has	have	AUX
abrj-637	265	7	been	be	AUX
abrj-637	265	8	interpreted	interpret	VERB
abrj-637	265	9	with	with	ADP
abrj-637	265	10	the	the	DET
abrj-637	265	11	help	help	NOUN
abrj-637	265	12	of	of	ADP
abrj-637	265	13	explainable	explainable	ADJ
abrj-637	265	14	ai	ai	NOUN
abrj-637	265	15	(	(	PUNCT
abrj-637	265	16	xai	xai	PROPN
abrj-637	265	17	)	)	PUNCT
abrj-637	265	18	methods	method	NOUN
abrj-637	265	19	,	,	PUNCT
abrj-637	265	20	specifically	specifically	ADV
abrj-637	265	21	,	,	PUNCT
abrj-637	265	22	shap	shap	NOUN
abrj-637	265	23	values	value	NOUN
abrj-637	265	24	(	(	PUNCT
abrj-637	265	25	zhou	zhou	X
abrj-637	265	26	et	et	PROPN
abrj-637	265	27	al	al	PROPN
abrj-637	265	28	.	.	PROPN
abrj-637	265	29	,	,	PUNCT
abrj-637	265	30	2024	2024	NUM
abrj-637	265	31	)	)	PUNCT
abrj-637	265	32	.	.	PUNCT
abrj-637	266	1	the	the	DET
abrj-637	266	2	strongest	strong	ADJ
abrj-637	266	3	characteristics	characteristic	NOUN
abrj-637	266	4	that	that	PRON
abrj-637	266	5	have	have	AUX
abrj-637	266	6	led	lead	VERB
abrj-637	266	7	to	to	ADP
abrj-637	266	8	high	high	ADJ
abrj-637	266	9	-	-	PUNCT
abrj-637	266	10	risk	risk	NOUN
abrj-637	266	11	predictions	prediction	NOUN
abrj-637	266	12	were	be	AUX
abrj-637	266	13	:	:	PUNCT
abrj-637	266	14	•	•	NUM
abrj-637	266	15	great	great	ADJ
abrj-637	266	16	declines	decline	NOUN
abrj-637	266	17	in	in	ADP
abrj-637	266	18	reported	report	VERB
abrj-637	266	19	taxable	taxable	ADJ
abrj-637	266	20	income	income	NOUN
abrj-637	266	21	.	.	PUNCT
abrj-637	267	1	•	•	NUM
abrj-637	267	2	unusual	unusual	ADJ
abrj-637	267	3	changes	change	NOUN
abrj-637	267	4	in	in	ADP
abrj-637	267	5	expense	expense	NOUN
abrj-637	267	6	to	to	ADP
abrj-637	267	7	revenue	revenue	NOUN
abrj-637	267	8	ratios	ratio	NOUN
abrj-637	267	9	.	.	PUNCT
abrj-637	268	1	•	•	NUM
abrj-637	268	2	too	too	ADV
abrj-637	268	3	much	much	ADJ
abrj-637	268	4	use	use	NOUN
abrj-637	268	5	of	of	ADP
abrj-637	268	6	positive	positive	ADJ
abrj-637	268	7	language	language	NOUN
abrj-637	268	8	in	in	ADP
abrj-637	268	9	writing	writing	NOUN
abrj-637	268	10	reports	report	NOUN
abrj-637	268	11	.	.	PUNCT
abrj-637	269	1	these	these	DET
abrj-637	269	2	features	feature	NOUN
abrj-637	269	3	were	be	AUX
abrj-637	269	4	presented	present	VERB
abrj-637	269	5	in	in	ADP
abrj-637	269	6	visual	visual	ADJ
abrj-637	269	7	dashboards	dashboard	NOUN
abrj-637	269	8	where	where	SCONJ
abrj-637	269	9	auditors	auditor	NOUN
abrj-637	269	10	could	could	AUX
abrj-637	269	11	see	see	VERB
abrj-637	269	12	why	why	SCONJ
abrj-637	269	13	a	a	DET
abrj-637	269	14	specific	specific	ADJ
abrj-637	269	15	filing	filing	NOUN
abrj-637	269	16	was	be	AUX
abrj-637	269	17	rated	rate	VERB
abrj-637	269	18	as	as	ADP
abrj-637	269	19	high	high	ADJ
abrj-637	269	20	-	-	PUNCT
abrj-637	269	21	risk	risk	NOUN
abrj-637	269	22	,	,	PUNCT
abrj-637	269	23	and	and	CCONJ
abrj-637	269	24	could	could	AUX
abrj-637	269	25	plan	plan	VERB
abrj-637	269	26	audits	audits	NOUN
abrj-637	269	27	based	base	VERB
abrj-637	269	28	on	on	ADP
abrj-637	269	29	that	that	PRON
abrj-637	269	30	(	(	PUNCT
abrj-637	269	31	breslin	breslin	PROPN
abrj-637	269	32	,	,	PUNCT
abrj-637	269	33	2021	2021	NUM
abrj-637	269	34	;	;	PUNCT
abrj-637	269	35	wahyono	wahyono	PROPN
abrj-637	269	36	and	and	CCONJ
abrj-637	269	37	david	david	PROPN
abrj-637	269	38	,	,	PUNCT
abrj-637	269	39	2025	2025	NUM
abrj-637	269	40	)	)	PUNCT
abrj-637	269	41	.	.	PUNCT
abrj-637	270	1	4.6	4.6	NUM
abrj-637	270	2	.	.	PUNCT
abrj-637	270	3	discussion	discussion	NOUN
abrj-637	270	4	of	of	ADP
abrj-637	270	5	findings	finding	NOUN
abrj-637	270	6	the	the	DET
abrj-637	270	7	findings	finding	NOUN
abrj-637	270	8	indicate	indicate	VERB
abrj-637	270	9	the	the	DET
abrj-637	270	10	importance	importance	NOUN
abrj-637	270	11	of	of	ADP
abrj-637	270	12	combining	combine	VERB
abrj-637	270	13	ensemble	ensemble	ADJ
abrj-637	270	14	learning	learning	NOUN
abrj-637	270	15	and	and	CCONJ
abrj-637	270	16	anomaly	anomaly	NOUN
abrj-637	270	17	detection	detection	NOUN
abrj-637	270	18	to	to	PART
abrj-637	270	19	identify	identify	VERB
abrj-637	270	20	tax	tax	NOUN
abrj-637	270	21	fraud	fraud	NOUN
abrj-637	270	22	with	with	ADP
abrj-637	270	23	high	high	ADJ
abrj-637	270	24	accuracy	accuracy	NOUN
abrj-637	270	25	.	.	PUNCT
abrj-637	271	1	supervised	supervised	ADJ
abrj-637	271	2	models	model	NOUN
abrj-637	271	3	were	be	AUX
abrj-637	271	4	more	more	ADV
abrj-637	271	5	accurate	accurate	ADJ
abrj-637	271	6	but	but	CCONJ
abrj-637	271	7	unsupervised	unsupervised	ADJ
abrj-637	271	8	models	model	NOUN
abrj-637	271	9	were	be	AUX
abrj-637	271	10	necessary	necessary	ADJ
abrj-637	271	11	to	to	PART
abrj-637	271	12	reveal	reveal	VERB
abrj-637	271	13	new	new	ADJ
abrj-637	271	14	patterns	pattern	NOUN
abrj-637	271	15	of	of	ADP
abrj-637	271	16	fraudsupporting	fraudsupporte	VERB
abrj-637	271	17	the	the	DET
abrj-637	271	18	findings	finding	NOUN
abrj-637	271	19	of	of	ADP
abrj-637	271	20	shujaaddeen	shujaaddeen	PROPN
abrj-637	271	21	et	et	PROPN
abrj-637	271	22	al	al	PROPN
abrj-637	271	23	.	.	PROPN
abrj-637	272	1	(	(	PUNCT
abrj-637	272	2	2024	2024	NUM
abrj-637	272	3	)	)	PUNCT
abrj-637	272	4	and	and	CCONJ
abrj-637	272	5	mehta	mehta	PROPN
abrj-637	272	6	et	et	PROPN
abrj-637	272	7	al	al	PROPN
abrj-637	272	8	.	.	PROPN
abrj-637	273	1	(	(	PUNCT
abrj-637	273	2	2022	2022	NUM
abrj-637	273	3	)	)	PUNCT
abrj-637	273	4	.	.	PUNCT
abrj-637	274	1	additionally	additionally	ADV
abrj-637	274	2	,	,	PUNCT
abrj-637	274	3	the	the	DET
abrj-637	274	4	combination	combination	NOUN
abrj-637	274	5	of	of	ADP
abrj-637	274	6	textual	textual	ADJ
abrj-637	274	7	analytics	analytic	NOUN
abrj-637	274	8	enhanced	enhance	VERB
abrj-637	274	9	general	general	ADJ
abrj-637	274	10	recall	recall	NOUN
abrj-637	274	11	,	,	PUNCT
abrj-637	274	12	which	which	PRON
abrj-637	274	13	proves	prove	VERB
abrj-637	274	14	the	the	DET
abrj-637	274	15	argument	argument	NOUN
abrj-637	274	16	that	that	SCONJ
abrj-637	274	17	the	the	DET
abrj-637	274	18	qualitative	qualitative	ADJ
abrj-637	274	19	disclosure	disclosure	NOUN
abrj-637	274	20	has	have	AUX
abrj-637	274	21	a	a	DET
abrj-637	274	22	very	very	ADV
abrj-637	274	23	strong	strong	ADJ
abrj-637	274	24	predictive	predictive	ADJ
abrj-637	274	25	capacity	capacity	NOUN
abrj-637	274	26	(	(	PUNCT
abrj-637	274	27	zhang	zhang	X
abrj-637	274	28	et	et	PROPN
abrj-637	274	29	al	al	PROPN
abrj-637	274	30	.	.	PROPN
abrj-637	274	31	,	,	PUNCT
abrj-637	274	32	2024	2024	NUM
abrj-637	274	33	)	)	PUNCT
abrj-637	274	34	.	.	PUNCT
abrj-637	275	1	significantly	significantly	ADV
abrj-637	275	2	,	,	PUNCT
abrj-637	275	3	the	the	DET
abrj-637	275	4	explainability	explainability	NOUN
abrj-637	275	5	layer	layer	NOUN
abrj-637	275	6	allowed	allow	VERB
abrj-637	275	7	human	human	ADJ
abrj-637	275	8	auditors	auditor	NOUN
abrj-637	275	9	to	to	PART
abrj-637	275	10	test	test	VERB
abrj-637	275	11	system	system	NOUN
abrj-637	275	12	outputs	output	NOUN
abrj-637	275	13	,	,	PUNCT
abrj-637	275	14	which	which	PRON
abrj-637	275	15	strengthened	strengthen	VERB
abrj-637	275	16	the	the	DET
abrj-637	275	17	trust	trust	NOUN
abrj-637	275	18	and	and	CCONJ
abrj-637	275	19	accountability	accountability	NOUN
abrj-637	275	20	.	.	PUNCT
abrj-637	276	1	using	use	VERB
abrj-637	276	2	the	the	DET
abrj-637	276	3	combination	combination	NOUN
abrj-637	276	4	of	of	ADP
abrj-637	276	5	automation	automation	NOUN
abrj-637	276	6	and	and	CCONJ
abrj-637	276	7	interpretability	interpretability	NOUN
abrj-637	276	8	,	,	PUNCT
abrj-637	276	9	the	the	DET
abrj-637	276	10	offer	offer	NOUN
abrj-637	276	11	system	system	NOUN
abrj-637	276	12	can	can	AUX
abrj-637	276	13	offer	offer	VERB
abrj-637	276	14	a	a	DET
abrj-637	276	15	sensible	sensible	ADJ
abrj-637	276	16	balance	balance	NOUN
abrj-637	276	17	between	between	ADP
abrj-637	276	18	efficiency	efficiency	NOUN
abrj-637	276	19	and	and	CCONJ
abrj-637	276	20	ethics	ethic	NOUN
abrj-637	276	21	(	(	PUNCT
abrj-637	276	22	tagbo	tagbo	NOUN
abrj-637	276	23	and	and	CCONJ
abrj-637	276	24	adekoya	adekoya	NOUN
abrj-637	276	25	,	,	PUNCT
abrj-637	276	26	2023	2023	NUM
abrj-637	276	27	;	;	PUNCT
abrj-637	276	28	zhou	zhou	PROPN
abrj-637	276	29	et	et	PROPN
abrj-637	276	30	al	al	PROPN
abrj-637	276	31	.	.	PROPN
abrj-637	276	32	,	,	PUNCT
abrj-637	276	33	2024	2024	NUM
abrj-637	276	34	)	)	PUNCT
abrj-637	276	35	.	.	PUNCT
abrj-637	277	1	4.7	4.7	NUM
abrj-637	277	2	.	.	PUNCT
abrj-637	278	1	summary	summary	VERB
abrj-637	278	2	the	the	DET
abrj-637	278	3	case	case	NOUN
abrj-637	278	4	study	study	NOUN
abrj-637	278	5	illustrates	illustrate	VERB
abrj-637	278	6	how	how	SCONJ
abrj-637	278	7	machine	machine	NOUN
abrj-637	278	8	learning	learning	NOUN
abrj-637	278	9	could	could	AUX
abrj-637	278	10	be	be	AUX
abrj-637	278	11	applied	apply	VERB
abrj-637	278	12	to	to	PART
abrj-637	278	13	detect	detect	VERB
abrj-637	278	14	tax	tax	NOUN
abrj-637	278	15	fraud	fraud	NOUN
abrj-637	278	16	in	in	ADP
abrj-637	278	17	terms	term	NOUN
abrj-637	278	18	of	of	ADP
abrj-637	278	19	its	its	PRON
abrj-637	278	20	operational	operational	ADJ
abrj-637	278	21	viability	viability	NOUN
abrj-637	278	22	and	and	CCONJ
abrj-637	278	23	analytical	analytical	ADJ
abrj-637	278	24	depth	depth	NOUN
abrj-637	278	25	.	.	PUNCT
abrj-637	279	1	the	the	DET
abrj-637	279	2	findings	finding	NOUN
abrj-637	279	3	indicate	indicate	VERB
abrj-637	279	4	that	that	SCONJ
abrj-637	279	5	the	the	DET
abrj-637	279	6	ensemble	ensemble	ADJ
abrj-637	279	7	models	model	NOUN
abrj-637	279	8	such	such	ADJ
abrj-637	279	9	as	as	ADP
abrj-637	279	10	the	the	DET
abrj-637	279	11	xgboost	xgboost	NOUN
abrj-637	279	12	,	,	PUNCT
abrj-637	279	13	a	a	DET
abrj-637	279	14	combination	combination	NOUN
abrj-637	279	15	with	with	ADP
abrj-637	279	16	the	the	DET
abrj-637	279	17	anomaly	anomaly	NOUN
abrj-637	279	18	detectors	detector	NOUN
abrj-637	279	19	and	and	CCONJ
abrj-637	279	20	the	the	DET
abrj-637	279	21	nlp	nlp	NOUN
abrj-637	279	22	-	-	PUNCT
abrj-637	279	23	driven	drive	VERB
abrj-637	279	24	insights	insight	NOUN
abrj-637	279	25	would	would	AUX
abrj-637	279	26	greatly	greatly	ADV
abrj-637	279	27	improve	improve	VERB
abrj-637	279	28	the	the	DET
abrj-637	279	29	accuracy	accuracy	NOUN
abrj-637	279	30	and	and	CCONJ
abrj-637	279	31	explainability	explainability	NOUN
abrj-637	279	32	of	of	ADP
abrj-637	279	33	the	the	DET
abrj-637	279	34	fraud	fraud	NOUN
abrj-637	279	35	detection	detection	NOUN
abrj-637	279	36	systems	system	NOUN
abrj-637	279	37	.	.	PUNCT
abrj-637	280	1	such	such	ADJ
abrj-637	280	2	results	result	NOUN
abrj-637	280	3	confirm	confirm	VERB
abrj-637	280	4	the	the	DET
abrj-637	280	5	relevance	relevance	NOUN
abrj-637	280	6	of	of	ADP
abrj-637	280	7	the	the	DET
abrj-637	280	8	framework	framework	NOUN
abrj-637	280	9	in	in	ADP
abrj-637	280	10	the	the	DET
abrj-637	280	11	actual	actual	ADJ
abrj-637	280	12	audit	audit	NOUN
abrj-637	280	13	setting	set	VERB
abrj-637	280	14	and	and	CCONJ
abrj-637	280	15	precondition	precondition	VERB
abrj-637	280	16	the	the	DET
abrj-637	280	17	scope	scope	NOUN
abrj-637	280	18	of	of	ADP
abrj-637	280	19	its	its	PRON
abrj-637	280	20	scaling	scaling	NOUN
abrj-637	280	21	to	to	ADP
abrj-637	280	22	the	the	DET
abrj-637	280	23	tax	tax	NOUN
abrj-637	280	24	authorities	authority	NOUN
abrj-637	280	25	and	and	CCONJ
abrj-637	280	26	corporate	corporate	ADJ
abrj-637	280	27	compliance	compliance	NOUN
abrj-637	280	28	systems	system	NOUN
abrj-637	280	29	.	.	PUNCT
abrj-637	281	1	6	6	NUM
abrj-637	281	2	.	.	X
abrj-637	281	3	discussion	discussion	NOUN
abrj-637	281	4	and	and	CCONJ
abrj-637	281	5	policy	policy	NOUN
abrj-637	281	6	implications	implication	NOUN
abrj-637	281	7	the	the	DET
abrj-637	281	8	introduction	introduction	NOUN
abrj-637	281	9	of	of	ADP
abrj-637	281	10	machine	machine	NOUN
abrj-637	281	11	learning	learning	NOUN
abrj-637	281	12	(	(	PUNCT
abrj-637	281	13	ml	ml	NOUN
abrj-637	281	14	)	)	PUNCT
abrj-637	281	15	to	to	ADP
abrj-637	281	16	the	the	DET
abrj-637	281	17	tax	tax	NOUN
abrj-637	281	18	fraud	fraud	NOUN
abrj-637	281	19	detection	detection	NOUN
abrj-637	281	20	systems	system	NOUN
abrj-637	281	21	is	be	AUX
abrj-637	281	22	a	a	DET
abrj-637	281	23	paradigm	paradigm	NOUN
abrj-637	281	24	shift	shift	NOUN
abrj-637	281	25	in	in	ADP
abrj-637	281	26	how	how	SCONJ
abrj-637	281	27	the	the	DET
abrj-637	281	28	government	government	NOUN
abrj-637	281	29	and	and	CCONJ
abrj-637	281	30	regulating	regulate	VERB
abrj-637	281	31	bodies	body	NOUN
abrj-637	281	32	approach	approach	NOUN
abrj-637	281	33	compliance	compliance	NOUN
abrj-637	281	34	and	and	CCONJ
abrj-637	281	35	risk	risk	NOUN
abrj-637	281	36	evaluation	evaluation	NOUN
abrj-637	281	37	.	.	PUNCT
abrj-637	282	1	in	in	ADP
abrj-637	282	2	addition	addition	NOUN
abrj-637	282	3	to	to	ADP
abrj-637	282	4	enhancing	enhance	VERB
abrj-637	282	5	the	the	DET
abrj-637	282	6	accuracy	accuracy	NOUN
abrj-637	282	7	of	of	ADP
abrj-637	282	8	detection	detection	NOUN
abrj-637	282	9	,	,	PUNCT
abrj-637	282	10	ml	ml	NOUN
abrj-637	282	11	-	-	PUNCT
abrj-637	282	12	driven	drive	VERB
abrj-637	282	13	models	model	NOUN
abrj-637	282	14	can	can	AUX
abrj-637	282	15	turn	turn	VERB
abrj-637	282	16	tax	tax	NOUN
abrj-637	282	17	administration	administration	NOUN
abrj-637	282	18	into	into	ADP
abrj-637	282	19	an	an	DET
abrj-637	282	20	active	active	ADJ
abrj-637	282	21	process	process	NOUN
abrj-637	282	22	rather	rather	ADV
abrj-637	282	23	than	than	ADP
abrj-637	282	24	a	a	DET
abrj-637	282	25	passive	passive	ADJ
abrj-637	282	26	one	one	NUM
abrj-637	282	27	based	base	VERB
abrj-637	282	28	on	on	ADP
abrj-637	282	29	reactive	reactive	ADJ
abrj-637	282	30	auditing	auditing	NOUN
abrj-637	282	31	and	and	CCONJ
abrj-637	282	32	instead	instead	ADV
abrj-637	282	33	on	on	ADP
abrj-637	282	34	data	data	NOUN
abrj-637	282	35	-	-	PUNCT
abrj-637	282	36	driven	drive	VERB
abrj-637	282	37	decision	decision	NOUN
abrj-637	282	38	-	-	PUNCT
abrj-637	282	39	making	making	NOUN
abrj-637	282	40	(	(	PUNCT
abrj-637	282	41	acharya	acharya	NOUN
abrj-637	282	42	,	,	PUNCT
abrj-637	282	43	2025	2025	NUM
abrj-637	282	44	;	;	PUNCT
abrj-637	282	45	breslin	breslin	PROPN
abrj-637	282	46	,	,	PUNCT
abrj-637	282	47	2021	2021	NUM
abrj-637	282	48	)	)	PUNCT
abrj-637	282	49	.	.	PUNCT
abrj-637	283	1	this	this	DET
abrj-637	283	2	part	part	NOUN
abrj-637	283	3	will	will	AUX
abrj-637	283	4	look	look	VERB
abrj-637	283	5	at	at	ADP
abrj-637	283	6	the	the	DET
abrj-637	283	7	implication	implication	NOUN
abrj-637	283	8	of	of	ADP
abrj-637	283	9	such	such	ADJ
abrj-637	283	10	technologies	technology	NOUN
abrj-637	283	11	in	in	ADP
abrj-637	283	12	terms	term	NOUN
abrj-637	283	13	of	of	ADP
abrj-637	283	14	operational	operational	ADJ
abrj-637	283	15	,	,	PUNCT
abrj-637	283	16	regulatory	regulatory	ADJ
abrj-637	283	17	,	,	PUNCT
abrj-637	283	18	and	and	CCONJ
abrj-637	283	19	ethical	ethical	ADJ
abrj-637	283	20	aspects	aspect	NOUN
abrj-637	283	21	and	and	CCONJ
abrj-637	283	22	present	present	VERB
abrj-637	283	23	the	the	DET
abrj-637	283	24	policy	policy	NOUN
abrj-637	283	25	implications	implication	NOUN
abrj-637	283	26	that	that	PRON
abrj-637	283	27	should	should	AUX
abrj-637	283	28	be	be	AUX
abrj-637	283	29	put	put	VERB
abrj-637	283	30	in	in	ADP
abrj-637	283	31	place	place	NOUN
abrj-637	283	32	in	in	ADP
abrj-637	283	33	order	order	NOUN
abrj-637	283	34	to	to	PART
abrj-637	283	35	adopt	adopt	VERB
abrj-637	283	36	such	such	ADJ
abrj-637	283	37	technologies	technology	NOUN
abrj-637	283	38	responsibly	responsibly	ADV
abrj-637	283	39	.	.	PUNCT
abrj-637	284	1	6.1	6.1	NUM
abrj-637	284	2	.	.	PUNCT
abrj-637	284	3	increasing	increase	VERB
abrj-637	284	4	audit	audit	NOUN
abrj-637	284	5	support	support	NOUN
abrj-637	284	6	and	and	CCONJ
abrj-637	284	7	efficiency	efficiency	NOUN
abrj-637	284	8	machine	machine	NOUN
abrj-637	284	9	learning	learn	VERB
abrj-637	284	10	solutions	solution	NOUN
abrj-637	284	11	help	help	VERB
abrj-637	284	12	to	to	PART
abrj-637	284	13	improve	improve	VERB
abrj-637	284	14	audit	audit	NOUN
abrj-637	284	15	efficiency	efficiency	NOUN
abrj-637	284	16	greatly	greatly	ADV
abrj-637	284	17	through	through	ADP
abrj-637	284	18	the	the	DET
abrj-637	284	19	automation	automation	NOUN
abrj-637	284	20	of	of	ADP
abrj-637	284	21	detecting	detect	VERB
abrj-637	284	22	red	red	ADJ
abrj-637	284	23	flags	flag	NOUN
abrj-637	284	24	within	within	ADP
abrj-637	284	25	corporate	corporate	ADJ
abrj-637	284	26	filings	filing	NOUN
abrj-637	284	27	(	(	PUNCT
abrj-637	284	28	wu	wu	PROPN
abrj-637	284	29	et	et	PROPN
abrj-637	284	30	al	al	PROPN
abrj-637	284	31	.	.	PROPN
abrj-637	284	32	,	,	PUNCT
abrj-637	284	33	2012	2012	NUM
abrj-637	284	34	;	;	PUNCT
abrj-637	284	35	ippolito	ippolito	PROPN
abrj-637	284	36	and	and	CCONJ
abrj-637	284	37	lozano	lozano	PROPN
abrj-637	284	38	,	,	PUNCT
abrj-637	284	39	2020	2020	NUM
abrj-637	284	40	)	)	PUNCT
abrj-637	284	41	.	.	PUNCT
abrj-637	285	1	in	in	ADP
abrj-637	285	2	comparison	comparison	NOUN
abrj-637	285	3	to	to	ADP
abrj-637	285	4	traditional	traditional	ADJ
abrj-637	285	5	systems	system	NOUN
abrj-637	285	6	,	,	PUNCT
abrj-637	285	7	where	where	SCONJ
abrj-637	285	8	reviews	review	NOUN
abrj-637	285	9	and	and	CCONJ
abrj-637	285	10	strict	strict	ADJ
abrj-637	285	11	rules	rule	NOUN
abrj-637	285	12	are	be	AUX
abrj-637	285	13	needed	need	VERB
abrj-637	285	14	,	,	PUNCT
abrj-637	285	15	ml	ml	ADP
abrj-637	285	16	models	model	NOUN
abrj-637	285	17	are	be	AUX
abrj-637	285	18	dynamically	dynamically	ADV
abrj-637	285	19	adjusted	adjust	VERB
abrj-637	285	20	to	to	ADP
abrj-637	285	21	new	new	ADJ
abrj-637	285	22	data	datum	NOUN
abrj-637	285	23	,	,	PUNCT
abrj-637	285	24	which	which	PRON
abrj-637	285	25	allows	allow	VERB
abrj-637	285	26	tax	tax	NOUN
abrj-637	285	27	authorities	authority	NOUN
abrj-637	285	28	to	to	PART
abrj-637	285	29	concentrate	concentrate	VERB
abrj-637	285	30	on	on	ADP
abrj-637	285	31	the	the	DET
abrj-637	285	32	risky	risky	ADJ
abrj-637	285	33	cases	case	NOUN
abrj-637	285	34	.	.	PUNCT
abrj-637	286	1	it	it	PRON
abrj-637	286	2	has	have	AUX
abrj-637	286	3	been	be	AUX
abrj-637	286	4	shown	show	VERB
abrj-637	286	5	that	that	SCONJ
abrj-637	286	6	supervised	supervise	VERB
abrj-637	286	7	and	and	CCONJ
abrj-637	286	8	ensemble	ensemble	ADJ
abrj-637	286	9	algorithms	algorithm	NOUN
abrj-637	286	10	(	(	PUNCT
abrj-637	286	11	random	random	ADJ
abrj-637	286	12	forest	forest	NOUN
abrj-637	286	13	and	and	CCONJ
abrj-637	286	14	xgboost	xgboost	NUM
abrj-637	286	15	)	)	PUNCT
abrj-637	286	16	can	can	AUX
abrj-637	286	17	minimize	minimize	VERB
abrj-637	286	18	false	false	ADJ
abrj-637	286	19	positives	positive	NOUN
abrj-637	286	20	and	and	CCONJ
abrj-637	286	21	maximize	maximize	VERB
abrj-637	286	22	the	the	DET
abrj-637	286	23	ranking	ranking	NOUN
abrj-637	286	24	of	of	ADP
abrj-637	286	25	audit	audit	NOUN
abrj-637	286	26	targets	target	NOUN
abrj-637	286	27	(	(	PUNCT
abrj-637	286	28	zhou	zhou	X
abrj-637	286	29	et	et	PROPN
abrj-637	286	30	al	al	PROPN
abrj-637	286	31	.	.	PROPN
abrj-637	286	32	,	,	PUNCT
abrj-637	286	33	2024	2024	NUM
abrj-637	286	34	;	;	PUNCT
abrj-637	286	35	zhang	zhang	PROPN
abrj-637	286	36	et	et	PROPN
abrj-637	286	37	al	al	PROPN
abrj-637	286	38	.	.	PROPN
abrj-637	286	39	,	,	PUNCT
abrj-637	286	40	2025	2025	NUM
abrj-637	286	41	)	)	PUNCT
abrj-637	286	42	.	.	PUNCT
abrj-637	287	1	practically	practically	ADV
abrj-637	287	2	,	,	PUNCT
abrj-637	287	3	it	it	PRON
abrj-637	287	4	enables	enable	VERB
abrj-637	287	5	the	the	DET
abrj-637	287	6	auditors	auditor	NOUN
abrj-637	287	7	to	to	PART
abrj-637	287	8	move	move	VERB
abrj-637	287	9	away	away	ADV
abrj-637	287	10	to	to	ADP
abrj-637	287	11	the	the	DET
abrj-637	287	12	exhaustive	exhaustive	ADJ
abrj-637	287	13	verification	verification	NOUN
abrj-637	287	14	approach	approach	NOUN
abrj-637	287	15	to	to	ADP
abrj-637	287	16	risk	risk	NOUN
abrj-637	287	17	based	base	VERB
abrj-637	287	18	auditing	auditing	NOUN
abrj-637	287	19	such	such	ADJ
abrj-637	287	20	that	that	PRON
abrj-637	287	21	enhance	enhance	VERB
abrj-637	287	22	the	the	DET
abrj-637	287	23	cost	cost	NOUN
abrj-637	287	24	efficiency	efficiency	NOUN
abrj-637	287	25	and	and	CCONJ
abrj-637	287	26	compliance	compliance	NOUN
abrj-637	287	27	coverage	coverage	NOUN
abrj-637	287	28	.	.	PUNCT
abrj-637	288	1	moreover	moreover	ADV
abrj-637	288	2	,	,	PUNCT
abrj-637	288	3	risk	risk	NOUN
abrj-637	288	4	scoring	scoring	NOUN
abrj-637	288	5	models	model	NOUN
abrj-637	288	6	do	do	AUX
abrj-637	288	7	not	not	PART
abrj-637	288	8	only	only	ADV
abrj-637	288	9	convert	convert	VERB
abrj-637	288	10	the	the	DET
abrj-637	288	11	intricate	intricate	ADJ
abrj-637	288	12	outputs	output	NOUN
abrj-637	288	13	of	of	ADP
abrj-637	288	14	algorithms	algorithm	NOUN
abrj-637	288	15	into	into	ADP
abrj-637	288	16	interpretable	interpretable	ADJ
abrj-637	288	17	indicators	indicator	NOUN
abrj-637	288	18	but	but	CCONJ
abrj-637	288	19	also	also	ADV
abrj-637	288	20	enable	enable	VERB
abrj-637	288	21	the	the	DET
abrj-637	288	22	non	non	ADJ
abrj-637	288	23	-	-	ADJ
abrj-637	288	24	technical	technical	ADJ
abrj-637	288	25	staff	staff	NOUN
abrj-637	288	26	to	to	PART
abrj-637	288	27	base	base	VERB
abrj-637	288	28	their	their	PRON
abrj-637	288	29	decisions	decision	NOUN
abrj-637	288	30	on	on	ADP
abrj-637	288	31	the	the	DET
abrj-637	288	32	data	datum	NOUN
abrj-637	288	33	.	.	PUNCT
abrj-637	289	1	the	the	DET
abrj-637	289	2	interaction	interaction	NOUN
abrj-637	289	3	of	of	ADP
abrj-637	289	4	the	the	DET
abrj-637	289	5	ai	ai	PROPN
abrj-637	289	6	systems	system	NOUN
abrj-637	289	7	with	with	ADP
abrj-637	289	8	human	human	ADJ
abrj-637	289	9	auditors	auditor	NOUN
abrj-637	289	10	,	,	PUNCT
abrj-637	289	11	therefore	therefore	ADV
abrj-637	289	12	,	,	PUNCT
abrj-637	289	13	forms	form	VERB
abrj-637	289	14	a	a	DET
abrj-637	289	15	hybrid	hybrid	NOUN
abrj-637	289	16	setting	setting	NOUN
abrj-637	289	17	in	in	ADP
abrj-637	289	18	which	which	PRON
abrj-637	289	19	the	the	DET
abrj-637	289	20	technology	technology	NOUN
abrj-637	289	21	supports	support	VERB
abrj-637	289	22	,	,	PUNCT
abrj-637	289	23	not	not	PART
abrj-637	289	24	omits	omit	VERB
abrj-637	289	25	human	human	ADJ
abrj-637	289	26	experience	experience	NOUN
abrj-637	289	27	(	(	PUNCT
abrj-637	289	28	breslin	breslin	PROPN
abrj-637	289	29	,	,	PUNCT
abrj-637	289	30	2021	2021	NUM
abrj-637	289	31	;	;	PUNCT
abrj-637	289	32	tagbo	tagbo	NOUN
abrj-637	289	33	and	and	CCONJ
abrj-637	289	34	adekoya	adekoya	NOUN
abrj-637	289	35	,	,	PUNCT
abrj-637	289	36	2023	2023	NUM
abrj-637	289	37	)	)	PUNCT
abrj-637	289	38	.	.	PUNCT
abrj-637	290	1	6.2	6.2	NUM
abrj-637	290	2	.	.	PUNCT
abrj-637	291	1	regulatory	regulatory	ADJ
abrj-637	291	2	and	and	CCONJ
abrj-637	291	3	governance	governance	NOUN
abrj-637	291	4	nature	nature	NOUN
abrj-637	291	5	.	.	PUNCT
abrj-637	292	1	the	the	DET
abrj-637	292	2	implementation	implementation	NOUN
abrj-637	292	3	of	of	ADP
abrj-637	292	4	ml	ml	NOUN
abrj-637	292	5	systems	system	NOUN
abrj-637	292	6	in	in	ADP
abrj-637	292	7	taxation	taxation	NOUN
abrj-637	292	8	needs	need	VERB
abrj-637	292	9	strong	strong	ADJ
abrj-637	292	10	governance	governance	NOUN
abrj-637	292	11	systems	system	NOUN
abrj-637	292	12	in	in	ADP
abrj-637	292	13	place	place	NOUN
abrj-637	292	14	to	to	PART
abrj-637	292	15	facilitate	facilitate	VERB
abrj-637	292	16	fairness	fairness	NOUN
abrj-637	292	17	,	,	PUNCT
abrj-637	292	18	transparency	transparency	NOUN
abrj-637	292	19	,	,	PUNCT
abrj-637	292	20	and	and	CCONJ
abrj-637	292	21	accountability	accountability	NOUN
abrj-637	292	22	.	.	PUNCT
abrj-637	293	1	tax	tax	NOUN
abrj-637	293	2	information	information	NOUN
abrj-637	293	3	is	be	AUX
abrj-637	293	4	very	very	ADV
abrj-637	293	5	confidential	confidential	ADJ
abrj-637	293	6	,	,	PUNCT
abrj-637	293	7	and	and	CCONJ
abrj-637	293	8	as	as	SCONJ
abrj-637	293	9	it	it	PRON
abrj-637	293	10	is	be	AUX
abrj-637	293	11	utilized	utilize	VERB
abrj-637	293	12	in	in	ADP
abrj-637	293	13	the	the	DET
abrj-637	293	14	automated	automate	VERB
abrj-637	293	15	system	system	NOUN
abrj-637	293	16	,	,	PUNCT
abrj-637	293	17	it	it	PRON
abrj-637	293	18	brings	bring	VERB
abrj-637	293	19	about	about	ADP
abrj-637	293	20	a	a	DET
abrj-637	293	21	possibility	possibility	NOUN
abrj-637	293	22	of	of	ADP
abrj-637	293	23	risk	risk	NOUN
abrj-637	293	24	,	,	PUNCT
abrj-637	293	25	such	such	ADJ
abrj-637	293	26	as	as	ADP
abrj-637	293	27	bias	bias	NOUN
abrj-637	293	28	,	,	PUNCT
abrj-637	293	29	misuse	misuse	NOUN
abrj-637	293	30	and	and	CCONJ
abrj-637	293	31	over	over	ADP
abrj-637	293	32	-	-	PUNCT
abrj-637	293	33	reliance	reliance	NOUN
abrj-637	293	34	on	on	ADP
abrj-637	293	35	opaque	opaque	ADJ
abrj-637	293	36	algorithms	algorithm	NOUN
abrj-637	293	37	(	(	PUNCT
abrj-637	293	38	wahyono	wahyono	PROPN
abrj-637	293	39	&	&	CCONJ
abrj-637	293	40	david	david	PROPN
abrj-637	293	41	,	,	PUNCT
abrj-637	293	42	2025	2025	NUM
abrj-637	293	43	)	)	PUNCT
abrj-637	293	44	.	.	PUNCT
abrj-637	294	1	the	the	DET
abrj-637	294	2	regulatory	regulatory	ADJ
abrj-637	294	3	agencies	agency	NOUN
abrj-637	294	4	should	should	AUX
abrj-637	294	5	develop	develop	VERB
abrj-637	294	6	effective	effective	ADJ
abrj-637	294	7	guidelines	guideline	NOUN
abrj-637	294	8	on	on	ADP
abrj-637	294	9	:	:	PUNCT
abrj-637	294	10	•	•	NUM
abrj-637	294	11	data	data	NOUN
abrj-637	294	12	governance	governance	NOUN
abrj-637	294	13	:	:	PUNCT
abrj-637	294	14	the	the	DET
abrj-637	294	15	control	control	NOUN
abrj-637	294	16	of	of	ADP
abrj-637	294	17	the	the	DET
abrj-637	294	18	safety	safety	NOUN
abrj-637	294	19	of	of	ADP
abrj-637	294	20	tax	tax	NOUN
abrj-637	294	21	and	and	CCONJ
abrj-637	294	22	financial	financial	ADJ
abrj-637	294	23	data	datum	NOUN
abrj-637	294	24	and	and	CCONJ
abrj-637	294	25	the	the	DET
abrj-637	294	26	limitations	limitation	NOUN
abrj-637	294	27	of	of	ADP
abrj-637	294	28	their	their	PRON
abrj-637	294	29	utilization	utilization	NOUN
abrj-637	294	30	to	to	ADP
abrj-637	294	31	the	the	DET
abrj-637	294	32	justifiable	justifiable	ADJ
abrj-637	294	33	reasons	reason	NOUN
abrj-637	294	34	of	of	ADP
abrj-637	294	35	the	the	DET
abrj-637	294	36	regulation	regulation	NOUN
abrj-637	294	37	(	(	PUNCT
abrj-637	294	38	breslin	breslin	NOUN
abrj-637	294	39	,	,	PUNCT
abrj-637	294	40	2021	2021	NUM
abrj-637	294	41	)	)	PUNCT
abrj-637	294	42	.	.	PUNCT
abrj-637	295	1	•	•	NUM
abrj-637	295	2	algorithmic	algorithmic	ADJ
abrj-637	295	3	accountability	accountability	NOUN
abrj-637	295	4	:	:	PUNCT
abrj-637	295	5	introducing	introduce	VERB
abrj-637	295	6	model	model	NOUN
abrj-637	295	7	decisions	decision	NOUN
abrj-637	295	8	based	base	VERB
abrj-637	295	9	on	on	ADP
abrj-637	295	10	audit	audit	NOUN
abrj-637	295	11	trails	trail	NOUN
abrj-637	295	12	,	,	PUNCT
abrj-637	295	13	with	with	ADP
abrj-637	295	14	explainable	explainable	ADJ
abrj-637	295	15	ai	ai	NOUN
abrj-637	295	16	tools	tool	NOUN
abrj-637	295	17	to	to	PART
abrj-637	295	18	justify	justify	VERB
abrj-637	295	19	the	the	DET
abrj-637	295	20	results	result	NOUN
abrj-637	295	21	(	(	PUNCT
abrj-637	295	22	zhou	zhou	X
abrj-637	295	23	et	et	PROPN
abrj-637	295	24	al	al	PROPN
abrj-637	295	25	.	.	PROPN
abrj-637	295	26	,	,	PUNCT
abrj-637	295	27	2024	2024	NUM
abrj-637	295	28	)	)	PUNCT
abrj-637	295	29	.	.	PUNCT
abrj-637	296	1	•	•	NUM
abrj-637	296	2	bias	bias	NOUN
abrj-637	296	3	detection	detection	NOUN
abrj-637	296	4	and	and	CCONJ
abrj-637	296	5	mitigation	mitigation	NOUN
abrj-637	296	6	:	:	PUNCT
abrj-637	296	7	organizing	organize	VERB
abrj-637	296	8	frequent	frequent	ADJ
abrj-637	296	9	fairness	fairness	NOUN
abrj-637	296	10	audits	audits	NOUN
abrj-637	296	11	to	to	PART
abrj-637	296	12	avoid	avoid	VERB
abrj-637	296	13	the	the	DET
abrj-637	296	14	discriminatory	discriminatory	ADJ
abrj-637	296	15	treatment	treatment	NOUN
abrj-637	296	16	of	of	ADP
abrj-637	296	17	certain	certain	ADJ
abrj-637	296	18	sectors	sector	NOUN
abrj-637	296	19	or	or	CCONJ
abrj-637	296	20	groups	group	NOUN
abrj-637	296	21	of	of	ADP
abrj-637	296	22	taxpayers	taxpayer	NOUN
abrj-637	296	23	(	(	PUNCT
abrj-637	296	24	tagbo	tagbo	NOUN
abrj-637	296	25	&	&	CCONJ
abrj-637	296	26	adekoya	adekoya	PROPN
abrj-637	296	27	,	,	PUNCT
abrj-637	296	28	2023	2023	NUM
abrj-637	296	29	)	)	PUNCT
abrj-637	296	30	.	.	PUNCT
abrj-637	297	1	•	•	NOUN
abrj-637	297	2	interoperability	interoperability	NOUN
abrj-637	297	3	standards	standard	NOUN
abrj-637	297	4	:	:	PUNCT
abrj-637	297	5	creation	creation	NOUN
abrj-637	297	6	of	of	ADP
abrj-637	297	7	standardized	standardized	ADJ
abrj-637	297	8	data	data	NOUN
abrj-637	297	9	forms	form	NOUN
abrj-637	297	10	and	and	CCONJ
abrj-637	297	11	apis	apis	ADJ
abrj-637	297	12	to	to	PART
abrj-637	297	13	enable	enable	VERB
abrj-637	297	14	data	datum	NOUN
abrj-637	297	15	sharing	sharing	NOUN
abrj-637	297	16	between	between	ADP
abrj-637	297	17	ml	ml	NOUN
abrj-637	297	18	systems	system	NOUN
abrj-637	297	19	and	and	CCONJ
abrj-637	297	20	the	the	DET
abrj-637	297	21	current	current	ADJ
abrj-637	297	22	audit	audit	NOUN
abrj-637	297	23	infrastructure	infrastructure	NOUN
abrj-637	297	24	(	(	PUNCT
abrj-637	297	25	nguyen	nguyen	NOUN
abrj-637	297	26	,	,	PUNCT
abrj-637	297	27	2025	2025	NUM
abrj-637	297	28	)	)	PUNCT
abrj-637	297	29	.	.	PUNCT
abrj-637	298	1	these	these	DET
abrj-637	298	2	regulatory	regulatory	ADJ
abrj-637	298	3	frameworks	framework	NOUN
abrj-637	298	4	are	be	AUX
abrj-637	298	5	related	relate	VERB
abrj-637	298	6	to	to	ADP
abrj-637	298	7	the	the	DET
abrj-637	298	8	overall	overall	ADJ
abrj-637	298	9	trend	trend	NOUN
abrj-637	298	10	of	of	ADP
abrj-637	298	11	algorithmic	algorithmic	ADJ
abrj-637	298	12	governance	governance	NOUN
abrj-637	298	13	,	,	PUNCT
abrj-637	298	14	in	in	ADP
abrj-637	298	15	which	which	PRON
abrj-637	298	16	transparency	transparency	NOUN
abrj-637	298	17	and	and	CCONJ
abrj-637	298	18	interpretability	interpretability	NOUN
abrj-637	298	19	are	be	AUX
abrj-637	298	20	valued	value	VERB
abrj-637	298	21	above	above	ADP
abrj-637	298	22	accuracy	accuracy	NOUN
abrj-637	298	23	.	.	PUNCT
abrj-637	299	1	the	the	DET
abrj-637	299	2	explainable	explainable	ADJ
abrj-637	299	3	ai	ai	NOUN
abrj-637	299	4	(	(	PUNCT
abrj-637	299	5	xai	xai	PROPN
abrj-637	299	6	)	)	PUNCT
abrj-637	299	7	aspects	aspect	NOUN
abrj-637	299	8	suggested	suggest	VERB
abrj-637	299	9	in	in	ADP
abrj-637	299	10	the	the	DET
abrj-637	299	11	present	present	ADJ
abrj-637	299	12	research	research	NOUN
abrj-637	299	13	are	be	AUX
abrj-637	299	14	at	at	ADP
abrj-637	299	15	the	the	DET
abrj-637	299	16	heart	heart	NOUN
abrj-637	299	17	of	of	ADP
abrj-637	299	18	ensuring	ensure	VERB
abrj-637	299	19	these	these	DET
abrj-637	299	20	guidelines	guideline	NOUN
abrj-637	299	21	as	as	SCONJ
abrj-637	299	22	they	they	PRON
abrj-637	299	23	offer	offer	VERB
abrj-637	299	24	human	human	ADJ
abrj-637	299	25	-	-	PUNCT
abrj-637	299	26	understandable	understandable	ADJ
abrj-637	299	27	explanations	explanation	NOUN
abrj-637	299	28	of	of	ADP
abrj-637	299	29	model	model	NOUN
abrj-637	299	30	predictions	prediction	NOUN
abrj-637	299	31	(	(	PUNCT
abrj-637	299	32	zhou	zhou	X
abrj-637	299	33	et	et	PROPN
abrj-637	299	34	al	al	PROPN
abrj-637	299	35	.	.	PROPN
abrj-637	299	36	,	,	PUNCT
abrj-637	299	37	2024	2024	NUM
abrj-637	299	38	;	;	PUNCT
abrj-637	299	39	wahyono	wahyono	PROPN
abrj-637	299	40	and	and	CCONJ
abrj-637	299	41	david	david	PROPN
abrj-637	299	42	,	,	PUNCT
abrj-637	299	43	2025	2025	NUM
abrj-637	299	44	)	)	PUNCT
abrj-637	299	45	.	.	PUNCT
abrj-637	300	1	asian	asian	ADJ
abrj-637	300	2	business	business	PROPN
abrj-637	300	3	research	research	NOUN
abrj-637	300	4	journal	journal	NOUN
abrj-637	300	5	,	,	PUNCT
abrj-637	300	6	2025	2025	NUM
abrj-637	300	7	,	,	PUNCT
abrj-637	300	8	10(11	10(11	NUM
abrj-637	300	9	):	):	PUNCT
abrj-637	300	10	1	1	NUM
abrj-637	300	11	-	-	SYM
abrj-637	300	12	13	13	NUM
abrj-637	300	13	12	12	NUM
abrj-637	300	14	©	©	PROPN
abrj-637	300	15	2025	2025	NUM
abrj-637	300	16	by	by	ADP
abrj-637	300	17	the	the	DET
abrj-637	300	18	authors	author	NOUN
abrj-637	300	19	;	;	PUNCT
abrj-637	300	20	licensee	licensee	NOUN
abrj-637	300	21	eastern	eastern	ADJ
abrj-637	300	22	centre	centre	NOUN
abrj-637	300	23	of	of	ADP
abrj-637	300	24	science	science	NOUN
abrj-637	300	25	and	and	CCONJ
abrj-637	300	26	education	education	NOUN
abrj-637	300	27	,	,	PUNCT
abrj-637	300	28	usa	usa	PROPN
abrj-637	300	29	6.3	6.3	NUM
abrj-637	300	30	.	.	PUNCT
abrj-637	301	1	ethical	ethical	ADJ
abrj-637	301	2	and	and	CCONJ
abrj-637	301	3	social	social	ADJ
abrj-637	301	4	implications	implication	NOUN
abrj-637	301	5	taxation	taxation	NOUN
abrj-637	301	6	with	with	ADP
abrj-637	301	7	ai	ai	NOUN
abrj-637	301	8	and	and	CCONJ
abrj-637	301	9	ml	ml	AUX
abrj-637	301	10	involves	involve	VERB
abrj-637	301	11	the	the	DET
abrj-637	301	12	development	development	NOUN
abrj-637	301	13	of	of	ADP
abrj-637	301	14	important	important	ADJ
abrj-637	301	15	ethical	ethical	ADJ
abrj-637	301	16	issues	issue	NOUN
abrj-637	301	17	.	.	PUNCT
abrj-637	302	1	this	this	PRON
abrj-637	302	2	is	be	AUX
abrj-637	302	3	because	because	SCONJ
abrj-637	302	4	automation	automation	NOUN
abrj-637	302	5	systems	system	NOUN
abrj-637	302	6	should	should	AUX
abrj-637	302	7	run	run	VERB
abrj-637	302	8	according	accord	VERB
abrj-637	302	9	to	to	ADP
abrj-637	302	10	high	high	ADJ
abrj-637	302	11	principles	principle	NOUN
abrj-637	302	12	of	of	ADP
abrj-637	302	13	fairness	fairness	NOUN
abrj-637	302	14	,	,	PUNCT
abrj-637	302	15	privacy	privacy	NOUN
abrj-637	302	16	and	and	CCONJ
abrj-637	302	17	proportionality	proportionality	NOUN
abrj-637	302	18	to	to	PART
abrj-637	302	19	ensure	ensure	VERB
abrj-637	302	20	the	the	DET
abrj-637	302	21	confidence	confidence	NOUN
abrj-637	302	22	of	of	ADP
abrj-637	302	23	the	the	DET
abrj-637	302	24	people	people	NOUN
abrj-637	302	25	.	.	PUNCT
abrj-637	303	1	as	as	ADP
abrj-637	303	2	an	an	DET
abrj-637	303	3	example	example	NOUN
abrj-637	303	4	,	,	PUNCT
abrj-637	303	5	the	the	DET
abrj-637	303	6	models	model	NOUN
abrj-637	303	7	are	be	AUX
abrj-637	303	8	not	not	PART
abrj-637	303	9	supposed	suppose	VERB
abrj-637	303	10	to	to	PART
abrj-637	303	11	punish	punish	VERB
abrj-637	303	12	taxpayers	taxpayer	NOUN
abrj-637	303	13	using	use	VERB
abrj-637	303	14	demographic	demographic	ADJ
abrj-637	303	15	or	or	CCONJ
abrj-637	303	16	geographic	geographic	ADJ
abrj-637	303	17	proxies	proxy	NOUN
abrj-637	303	18	accidentally	accidentally	ADV
abrj-637	303	19	included	include	VERB
abrj-637	303	20	in	in	ADP
abrj-637	303	21	training	training	NOUN
abrj-637	303	22	data	datum	NOUN
abrj-637	303	23	(	(	PUNCT
abrj-637	303	24	tagbo	tagbo	NOUN
abrj-637	303	25	&	&	CCONJ
abrj-637	303	26	adekoya	adekoya	PROPN
abrj-637	303	27	,	,	PUNCT
abrj-637	303	28	2023	2023	NUM
abrj-637	303	29	)	)	PUNCT
abrj-637	303	30	.	.	PUNCT
abrj-637	304	1	furthermore	furthermore	ADV
abrj-637	304	2	,	,	PUNCT
abrj-637	304	3	even	even	ADV
abrj-637	304	4	though	though	SCONJ
abrj-637	304	5	automation	automation	NOUN
abrj-637	304	6	increases	increase	VERB
abrj-637	304	7	efficiency	efficiency	NOUN
abrj-637	304	8	,	,	PUNCT
abrj-637	304	9	it	it	PRON
abrj-637	304	10	is	be	AUX
abrj-637	304	11	a	a	DET
abrj-637	304	12	source	source	NOUN
abrj-637	304	13	of	of	ADP
abrj-637	304	14	over	over	ADP
abrj-637	304	15	-	-	PUNCT
abrj-637	304	16	reliance	reliance	NOUN
abrj-637	304	17	on	on	ADP
abrj-637	304	18	algorithmic	algorithmic	ADJ
abrj-637	304	19	choices	choice	NOUN
abrj-637	304	20	.	.	PUNCT
abrj-637	305	1	to	to	PART
abrj-637	305	2	avoid	avoid	VERB
abrj-637	305	3	this	this	PRON
abrj-637	305	4	,	,	PUNCT
abrj-637	305	5	people	people	NOUN
abrj-637	305	6	control	control	NOUN
abrj-637	305	7	must	must	AUX
abrj-637	305	8	also	also	ADV
abrj-637	305	9	be	be	AUX
abrj-637	305	10	a	a	DET
abrj-637	305	11	core	core	NOUN
abrj-637	305	12	element	element	NOUN
abrj-637	305	13	in	in	ADP
abrj-637	305	14	audit	audit	NOUN
abrj-637	305	15	activities	activity	NOUN
abrj-637	305	16	,	,	PUNCT
abrj-637	305	17	whereby	whereby	SCONJ
abrj-637	305	18	models	model	NOUN
abrj-637	305	19	identify	identify	VERB
abrj-637	305	20	highrisk	highrisk	NOUN
abrj-637	305	21	cases	case	NOUN
abrj-637	305	22	that	that	PRON
abrj-637	305	23	must	must	AUX
abrj-637	305	24	be	be	AUX
abrj-637	305	25	reviewed	review	VERB
abrj-637	305	26	by	by	ADP
abrj-637	305	27	auditors	auditor	NOUN
abrj-637	305	28	and	and	CCONJ
abrj-637	305	29	then	then	ADV
abrj-637	305	30	enforcement	enforcement	NOUN
abrj-637	305	31	actions	action	NOUN
abrj-637	305	32	followed	follow	VERB
abrj-637	305	33	(	(	PUNCT
abrj-637	305	34	breslin	breslin	PROPN
abrj-637	305	35	,	,	PUNCT
abrj-637	305	36	2021	2021	NUM
abrj-637	305	37	;	;	PUNCT
abrj-637	305	38	wahyono	wahyono	PROPN
abrj-637	305	39	and	and	CCONJ
abrj-637	305	40	david	david	PROPN
abrj-637	305	41	,	,	PUNCT
abrj-637	305	42	2025	2025	NUM
abrj-637	305	43	)	)	PUNCT
abrj-637	305	44	.	.	PUNCT
abrj-637	306	1	this	this	DET
abrj-637	306	2	hybrid	hybrid	ADJ
abrj-637	306	3	oversight	oversight	NOUN
abrj-637	306	4	system	system	NOUN
abrj-637	306	5	is	be	AUX
abrj-637	306	6	efficient	efficient	ADJ
abrj-637	306	7	but	but	CCONJ
abrj-637	306	8	does	do	AUX
abrj-637	306	9	not	not	PART
abrj-637	306	10	infringe	infringe	VERB
abrj-637	306	11	on	on	ADP
abrj-637	306	12	due	due	ADJ
abrj-637	306	13	process	process	NOUN
abrj-637	306	14	since	since	SCONJ
abrj-637	306	15	the	the	DET
abrj-637	306	16	technology	technology	NOUN
abrj-637	306	17	does	do	AUX
abrj-637	306	18	not	not	PART
abrj-637	306	19	substitute	substitute	VERB
abrj-637	306	20	ethical	ethical	ADJ
abrj-637	306	21	decision	decision	NOUN
abrj-637	306	22	-	-	PUNCT
abrj-637	306	23	making	making	NOUN
abrj-637	306	24	but	but	CCONJ
abrj-637	306	25	enhances	enhance	VERB
abrj-637	306	26	it	it	PRON
abrj-637	306	27	.	.	PUNCT
abrj-637	307	1	the	the	DET
abrj-637	307	2	other	other	ADJ
abrj-637	307	3	important	important	ADJ
abrj-637	307	4	dimension	dimension	NOUN
abrj-637	307	5	is	be	AUX
abrj-637	307	6	data	datum	NOUN
abrj-637	307	7	privacy	privacy	NOUN
abrj-637	307	8	.	.	PUNCT
abrj-637	308	1	this	this	DET
abrj-637	308	2	data	datum	NOUN
abrj-637	308	3	of	of	ADP
abrj-637	308	4	taxpayers	taxpayer	NOUN
abrj-637	308	5	should	should	AUX
abrj-637	308	6	be	be	AUX
abrj-637	308	7	anonymized	anonymize	VERB
abrj-637	308	8	,	,	PUNCT
abrj-637	308	9	encrypted	encrypt	VERB
abrj-637	308	10	and	and	CCONJ
abrj-637	308	11	processed	process	VERB
abrj-637	308	12	under	under	ADP
abrj-637	308	13	the	the	DET
abrj-637	308	14	international	international	ADJ
abrj-637	308	15	data	datum	NOUN
abrj-637	308	16	protection	protection	NOUN
abrj-637	308	17	laws	law	NOUN
abrj-637	308	18	like	like	ADP
abrj-637	308	19	gdpr	gdpr	NOUN
abrj-637	308	20	.	.	PUNCT
abrj-637	309	1	the	the	DET
abrj-637	309	2	introduction	introduction	NOUN
abrj-637	309	3	of	of	ADP
abrj-637	309	4	federated	federated	ADJ
abrj-637	309	5	learning	learning	NOUN
abrj-637	309	6	systems	system	NOUN
abrj-637	309	7	can	can	AUX
abrj-637	309	8	aid	aid	VERB
abrj-637	309	9	in	in	ADP
abrj-637	309	10	safeguarding	safeguard	VERB
abrj-637	309	11	sensitive	sensitive	ADJ
abrj-637	309	12	data	datum	NOUN
abrj-637	309	13	as	as	SCONJ
abrj-637	309	14	it	it	PRON
abrj-637	309	15	prevents	prevent	VERB
abrj-637	309	16	the	the	DET
abrj-637	309	17	transfer	transfer	NOUN
abrj-637	309	18	of	of	ADP
abrj-637	309	19	raw	raw	ADJ
abrj-637	309	20	information	information	NOUN
abrj-637	309	21	,	,	PUNCT
abrj-637	309	22	as	as	SCONJ
abrj-637	309	23	it	it	PRON
abrj-637	309	24	is	be	AUX
abrj-637	309	25	possible	possible	ADJ
abrj-637	309	26	to	to	PART
abrj-637	309	27	train	train	VERB
abrj-637	309	28	the	the	DET
abrj-637	309	29	models	model	NOUN
abrj-637	309	30	on	on	ADP
abrj-637	309	31	decentralized	decentralized	ADJ
abrj-637	309	32	databases	database	NOUN
abrj-637	309	33	(	(	PUNCT
abrj-637	309	34	acharya	acharya	NOUN
abrj-637	309	35	,	,	PUNCT
abrj-637	309	36	2025	2025	NUM
abrj-637	309	37	;	;	PUNCT
abrj-637	309	38	mehta	mehta	PROPN
abrj-637	309	39	et	et	PROPN
abrj-637	309	40	al	al	PROPN
abrj-637	309	41	.	.	PROPN
abrj-637	309	42	,	,	PUNCT
abrj-637	309	43	2022	2022	NUM
abrj-637	309	44	)	)	PUNCT
abrj-637	309	45	.	.	PUNCT
abrj-637	310	1	6.4	6.4	NUM
abrj-637	310	2	.	.	PUNCT
abrj-637	311	1	capacity	capacity	NOUN
abrj-637	311	2	building	building	NOUN
abrj-637	311	3	and	and	CCONJ
abrj-637	311	4	institutional	institutional	ADJ
abrj-637	311	5	readiness	readiness	NOUN
abrj-637	311	6	the	the	DET
abrj-637	311	7	willingness	willingness	NOUN
abrj-637	311	8	of	of	ADP
abrj-637	311	9	institutions	institution	NOUN
abrj-637	311	10	and	and	CCONJ
abrj-637	311	11	staffs	staff	NOUN
abrj-637	311	12	is	be	AUX
abrj-637	311	13	also	also	ADV
abrj-637	311	14	a	a	DET
abrj-637	311	15	requirement	requirement	NOUN
abrj-637	311	16	to	to	ADP
abrj-637	311	17	a	a	DET
abrj-637	311	18	successful	successful	ADJ
abrj-637	311	19	implementation	implementation	NOUN
abrj-637	311	20	of	of	ADP
abrj-637	311	21	ml	ml	PROPN
abrj-637	311	22	.	.	PUNCT
abrj-637	312	1	a	a	DET
abrj-637	312	2	significant	significant	ADJ
abrj-637	312	3	number	number	NOUN
abrj-637	312	4	of	of	ADP
abrj-637	312	5	tax	tax	NOUN
abrj-637	312	6	authorities	authority	NOUN
abrj-637	312	7	have	have	VERB
abrj-637	312	8	skill	skill	NOUN
abrj-637	312	9	deficiencies	deficiency	NOUN
abrj-637	312	10	in	in	ADP
abrj-637	312	11	data	datum	NOUN
abrj-637	312	12	science	science	NOUN
abrj-637	312	13	,	,	PUNCT
abrj-637	312	14	model	model	NOUN
abrj-637	312	15	governance	governance	NOUN
abrj-637	312	16	and	and	CCONJ
abrj-637	312	17	ai	ai	VERB
abrj-637	312	18	ethics	ethic	NOUN
abrj-637	312	19	(	(	PUNCT
abrj-637	312	20	tagbo	tagbo	PROPN
abrj-637	312	21	&	&	CCONJ
abrj-637	312	22	adekoya	adekoya	PROPN
abrj-637	312	23	,	,	PUNCT
abrj-637	312	24	2023	2023	NUM
abrj-637	312	25	)	)	PUNCT
abrj-637	312	26	.	.	PUNCT
abrj-637	313	1	the	the	DET
abrj-637	313	2	governments	government	NOUN
abrj-637	313	3	therefore	therefore	ADV
abrj-637	313	4	need	need	VERB
abrj-637	313	5	to	to	PART
abrj-637	313	6	invest	invest	VERB
abrj-637	313	7	in	in	ADP
abrj-637	313	8	capacity	capacity	NOUN
abrj-637	313	9	building	building	NOUN
abrj-637	313	10	programs	program	NOUN
abrj-637	313	11	-	-	PUNCT
abrj-637	313	12	training	training	NOUN
abrj-637	313	13	of	of	ADP
abrj-637	313	14	auditors	auditor	NOUN
abrj-637	313	15	,	,	PUNCT
abrj-637	313	16	analysts	analyst	NOUN
abrj-637	313	17	and	and	CCONJ
abrj-637	313	18	policymakers	policymaker	NOUN
abrj-637	313	19	on	on	ADP
abrj-637	313	20	the	the	DET
abrj-637	313	21	knowledge	knowledge	NOUN
abrj-637	313	22	of	of	ADP
abrj-637	313	23	how	how	SCONJ
abrj-637	313	24	to	to	PART
abrj-637	313	25	evaluate	evaluate	VERB
abrj-637	313	26	and	and	CCONJ
abrj-637	313	27	supervise	supervise	VERB
abrj-637	313	28	the	the	DET
abrj-637	313	29	ml	ml	NOUN
abrj-637	313	30	systems	system	NOUN
abrj-637	313	31	.	.	PUNCT
abrj-637	314	1	moreover	moreover	ADV
abrj-637	314	2	,	,	PUNCT
abrj-637	314	3	public	public	ADJ
abrj-637	314	4	-	-	PUNCT
abrj-637	314	5	private	private	ADJ
abrj-637	314	6	partnership	partnership	NOUN
abrj-637	314	7	may	may	AUX
abrj-637	314	8	promote	promote	VERB
abrj-637	314	9	the	the	DET
abrj-637	314	10	transfer	transfer	NOUN
abrj-637	314	11	of	of	ADP
abrj-637	314	12	technology	technology	NOUN
abrj-637	314	13	and	and	CCONJ
abrj-637	314	14	the	the	DET
abrj-637	314	15	implementation	implementation	NOUN
abrj-637	314	16	of	of	ADP
abrj-637	314	17	the	the	DET
abrj-637	314	18	best	good	ADJ
abrj-637	314	19	practices	practice	NOUN
abrj-637	314	20	in	in	ADP
abrj-637	314	21	academic	academic	ADJ
abrj-637	314	22	and	and	CCONJ
abrj-637	314	23	corporate	corporate	ADJ
abrj-637	314	24	research	research	NOUN
abrj-637	314	25	(	(	PUNCT
abrj-637	314	26	zhang	zhang	X
abrj-637	314	27	et	et	PROPN
abrj-637	314	28	al	al	PROPN
abrj-637	314	29	.	.	PROPN
abrj-637	314	30	,	,	PUNCT
abrj-637	314	31	2025	2025	NUM
abrj-637	314	32	)	)	PUNCT
abrj-637	314	33	.	.	PUNCT
abrj-637	315	1	in	in	ADP
abrj-637	315	2	-	-	PUNCT
abrj-637	315	3	house	house	NOUN
abrj-637	315	4	construction	construction	NOUN
abrj-637	315	5	not	not	PART
abrj-637	315	6	only	only	ADV
abrj-637	315	7	leads	lead	VERB
abrj-637	315	8	to	to	ADP
abrj-637	315	9	less	less	ADJ
abrj-637	315	10	reliance	reliance	NOUN
abrj-637	315	11	on	on	ADP
abrj-637	315	12	third	third	ADJ
abrj-637	315	13	party	party	NOUN
abrj-637	315	14	vendors	vendor	NOUN
abrj-637	315	15	,	,	PUNCT
abrj-637	315	16	but	but	CCONJ
abrj-637	315	17	also	also	ADV
abrj-637	315	18	encourages	encourage	VERB
abrj-637	315	19	change	change	NOUN
abrj-637	315	20	based	base	VERB
abrj-637	315	21	on	on	ADP
abrj-637	315	22	local	local	ADJ
abrj-637	315	23	tax	tax	NOUN
abrj-637	315	24	conditions	condition	NOUN
abrj-637	315	25	.	.	PUNCT
abrj-637	316	1	6.5	6.5	NUM
abrj-637	316	2	.	.	PUNCT
abrj-637	317	1	future	future	NOUN
abrj-637	317	2	of	of	ADP
abrj-637	317	3	tax	tax	NOUN
abrj-637	317	4	compliance	compliance	NOUN
abrj-637	317	5	and	and	CCONJ
abrj-637	317	6	ai	ai	VERB
abrj-637	317	7	introductions	introduction	NOUN
abrj-637	317	8	the	the	DET
abrj-637	317	9	role	role	NOUN
abrj-637	317	10	of	of	ADP
abrj-637	317	11	ml	ml	NOUN
abrj-637	317	12	in	in	ADP
abrj-637	317	13	tax	tax	NOUN
abrj-637	317	14	administration	administration	NOUN
abrj-637	317	15	will	will	AUX
abrj-637	317	16	keep	keep	VERB
abrj-637	317	17	increasing	increase	VERB
abrj-637	317	18	as	as	ADP
abrj-637	317	19	the	the	DET
abrj-637	317	20	corporate	corporate	ADJ
abrj-637	317	21	data	data	NOUN
abrj-637	317	22	ecosystem	ecosystem	NOUN
abrj-637	317	23	changes	change	NOUN
abrj-637	317	24	.	.	PUNCT
abrj-637	318	1	deep	deep	ADJ
abrj-637	318	2	learning	learning	NOUN
abrj-637	318	3	with	with	ADP
abrj-637	318	4	natural	natural	ADJ
abrj-637	318	5	language	language	NOUN
abrj-637	318	6	processing	processing	NOUN
abrj-637	318	7	(	(	PUNCT
abrj-637	318	8	nlp	nlp	NOUN
abrj-637	318	9	)	)	PUNCT
abrj-637	318	10	and	and	CCONJ
abrj-637	318	11	anomaly	anomaly	NOUN
abrj-637	318	12	detection	detection	NOUN
abrj-637	318	13	will	will	AUX
abrj-637	318	14	result	result	VERB
abrj-637	318	15	in	in	ADP
abrj-637	318	16	systems	system	NOUN
abrj-637	318	17	that	that	PRON
abrj-637	318	18	are	be	AUX
abrj-637	318	19	able	able	ADJ
abrj-637	318	20	to	to	PART
abrj-637	318	21	monitor	monitor	VERB
abrj-637	318	22	corporate	corporate	ADJ
abrj-637	318	23	compliance	compliance	NOUN
abrj-637	318	24	in	in	ADP
abrj-637	318	25	near	near	ADP
abrj-637	318	26	real	real	ADJ
abrj-637	318	27	-	-	PUNCT
abrj-637	318	28	time	time	NOUN
abrj-637	318	29	(	(	PUNCT
abrj-637	318	30	zhang	zhang	X
abrj-637	318	31	et	et	PROPN
abrj-637	318	32	al	al	PROPN
abrj-637	318	33	.	.	PROPN
abrj-637	318	34	,	,	PUNCT
abrj-637	318	35	2024	2024	NUM
abrj-637	318	36	;	;	PUNCT
abrj-637	318	37	choudhary	choudhary	PROPN
abrj-637	318	38	,	,	PUNCT
abrj-637	318	39	2025	2025	NUM
abrj-637	318	40	)	)	PUNCT
abrj-637	318	41	.	.	PUNCT
abrj-637	319	1	the	the	DET
abrj-637	319	2	development	development	NOUN
abrj-637	319	3	of	of	ADP
abrj-637	319	4	federated	federated	ADJ
abrj-637	319	5	ai	ai	NOUN
abrj-637	319	6	and	and	CCONJ
abrj-637	319	7	blockchain	blockchain	PROPN
abrj-637	319	8	integration	integration	NOUN
abrj-637	319	9	in	in	ADP
abrj-637	319	10	the	the	DET
abrj-637	319	11	future	future	NOUN
abrj-637	319	12	can	can	AUX
abrj-637	319	13	also	also	ADV
abrj-637	319	14	improve	improve	VERB
abrj-637	319	15	the	the	DET
abrj-637	319	16	integrity	integrity	NOUN
abrj-637	319	17	of	of	ADP
abrj-637	319	18	data	datum	NOUN
abrj-637	319	19	and	and	CCONJ
abrj-637	319	20	traceability	traceability	NOUN
abrj-637	319	21	of	of	ADP
abrj-637	319	22	audits	audits	NOUN
abrj-637	319	23	.	.	PUNCT
abrj-637	320	1	nonetheless	nonetheless	ADV
abrj-637	320	2	,	,	PUNCT
abrj-637	320	3	the	the	PRON
abrj-637	320	4	closer	close	ADJ
abrj-637	320	5	ai	ai	VERB
abrj-637	320	6	is	be	AUX
abrj-637	320	7	integrated	integrate	VERB
abrj-637	320	8	into	into	ADP
abrj-637	320	9	governance	governance	NOUN
abrj-637	320	10	,	,	PUNCT
abrj-637	320	11	the	the	PRON
abrj-637	320	12	more	more	ADJ
abrj-637	320	13	policymakers	policymaker	NOUN
abrj-637	320	14	need	need	VERB
abrj-637	320	15	to	to	PART
abrj-637	320	16	make	make	VERB
abrj-637	320	17	sure	sure	ADJ
abrj-637	320	18	that	that	SCONJ
abrj-637	320	19	ethical	ethical	ADJ
abrj-637	320	20	standards	standard	NOUN
abrj-637	320	21	,	,	PUNCT
abrj-637	320	22	transparency	transparency	NOUN
abrj-637	320	23	,	,	PUNCT
abrj-637	320	24	and	and	CCONJ
abrj-637	320	25	human	human	ADJ
abrj-637	320	26	analysis	analysis	NOUN
abrj-637	320	27	are	be	AUX
abrj-637	320	28	in	in	ADP
abrj-637	320	29	the	the	DET
abrj-637	320	30	frontline	frontline	NOUN
abrj-637	320	31	.	.	PUNCT
abrj-637	321	1	the	the	DET
abrj-637	321	2	regulatory	regulatory	ADJ
abrj-637	321	3	adaptation	adaptation	NOUN
abrj-637	321	4	,	,	PUNCT
abrj-637	321	5	consultation	consultation	NOUN
abrj-637	321	6	with	with	ADP
abrj-637	321	7	stakeholders	stakeholder	NOUN
abrj-637	321	8	,	,	PUNCT
abrj-637	321	9	and	and	CCONJ
abrj-637	321	10	international	international	ADJ
abrj-637	321	11	collaboration	collaboration	NOUN
abrj-637	321	12	will	will	AUX
abrj-637	321	13	be	be	AUX
abrj-637	321	14	required	require	VERB
abrj-637	321	15	on	on	ADP
abrj-637	321	16	a	a	DET
abrj-637	321	17	continuous	continuous	ADJ
abrj-637	321	18	basis	basis	NOUN
abrj-637	321	19	to	to	PART
abrj-637	321	20	ensure	ensure	VERB
abrj-637	321	21	the	the	DET
abrj-637	321	22	fair	fair	ADJ
abrj-637	321	23	and	and	CCONJ
abrj-637	321	24	responsible	responsible	ADJ
abrj-637	321	25	systems	system	NOUN
abrj-637	321	26	of	of	ADP
abrj-637	321	27	tax	tax	NOUN
abrj-637	321	28	enforcement	enforcement	NOUN
abrj-637	321	29	(	(	PUNCT
abrj-637	321	30	wahyono	wahyono	PROPN
abrj-637	321	31	&	&	CCONJ
abrj-637	321	32	david	david	PROPN
abrj-637	321	33	,	,	PUNCT
abrj-637	321	34	2025	2025	NUM
abrj-637	321	35	;	;	PUNCT
abrj-637	321	36	zhou	zhou	PROPN
abrj-637	321	37	et	et	PROPN
abrj-637	321	38	al	al	PROPN
abrj-637	321	39	.	.	PROPN
abrj-637	321	40	,	,	PUNCT
abrj-637	321	41	2024	2024	NUM
abrj-637	321	42	)	)	PUNCT
abrj-637	321	43	.	.	PUNCT
abrj-637	322	1	6.6	6.6	NUM
abrj-637	322	2	.	.	PUNCT
abrj-637	323	1	summary	summary	VERB
abrj-637	323	2	the	the	DET
abrj-637	323	3	machine	machine	NOUN
abrj-637	323	4	learning	learning	NOUN
abrj-637	323	5	can	can	AUX
abrj-637	323	6	transform	transform	VERB
abrj-637	323	7	the	the	DET
abrj-637	323	8	way	way	NOUN
abrj-637	323	9	tax	tax	NOUN
abrj-637	323	10	frauds	fraud	NOUN
abrj-637	323	11	are	be	AUX
abrj-637	323	12	detected	detect	VERB
abrj-637	323	13	and	and	CCONJ
abrj-637	323	14	compliance	compliance	NOUN
abrj-637	323	15	is	be	AUX
abrj-637	323	16	monitored	monitor	VERB
abrj-637	323	17	by	by	ADP
abrj-637	323	18	enhancing	enhance	VERB
abrj-637	323	19	accuracy	accuracy	NOUN
abrj-637	323	20	,	,	PUNCT
abrj-637	323	21	scalability	scalability	NOUN
abrj-637	323	22	and	and	CCONJ
abrj-637	323	23	transparency	transparency	NOUN
abrj-637	323	24	.	.	PUNCT
abrj-637	324	1	however	however	ADV
abrj-637	324	2	,	,	PUNCT
abrj-637	324	3	it	it	PRON
abrj-637	324	4	requires	require	VERB
abrj-637	324	5	strong	strong	ADJ
abrj-637	324	6	governance	governance	NOUN
abrj-637	324	7	,	,	PUNCT
abrj-637	324	8	ethical	ethical	ADJ
abrj-637	324	9	protection	protection	NOUN
abrj-637	324	10	,	,	PUNCT
abrj-637	324	11	and	and	CCONJ
abrj-637	324	12	preparedness	preparedness	NOUN
abrj-637	324	13	of	of	ADP
abrj-637	324	14	institutions	institution	NOUN
abrj-637	324	15	to	to	PART
abrj-637	324	16	be	be	AUX
abrj-637	324	17	successful	successful	ADJ
abrj-637	324	18	in	in	ADP
abrj-637	324	19	this	this	DET
abrj-637	324	20	adoption	adoption	NOUN
abrj-637	324	21	.	.	PUNCT
abrj-637	325	1	through	through	ADP
abrj-637	325	2	technological	technological	ADJ
abrj-637	325	3	innovation	innovation	NOUN
abrj-637	325	4	and	and	CCONJ
abrj-637	325	5	regulatory	regulatory	ADJ
abrj-637	325	6	integrity	integrity	NOUN
abrj-637	325	7	,	,	PUNCT
abrj-637	325	8	the	the	DET
abrj-637	325	9	tax	tax	NOUN
abrj-637	325	10	authorities	authority	NOUN
abrj-637	325	11	would	would	AUX
abrj-637	325	12	be	be	AUX
abrj-637	325	13	able	able	ADJ
abrj-637	325	14	to	to	PART
abrj-637	325	15	establish	establish	VERB
abrj-637	325	16	a	a	DET
abrj-637	325	17	new	new	ADJ
abrj-637	325	18	taxation	taxation	NOUN
abrj-637	325	19	paradigm	paradigm	NOUN
abrj-637	325	20	grounded	ground	VERB
abrj-637	325	21	in	in	ADP
abrj-637	325	22	data	data	NOUN
abrj-637	325	23	-	-	PUNCT
abrj-637	325	24	driven	drive	VERB
abrj-637	325	25	,	,	PUNCT
abrj-637	325	26	equitable	equitable	ADJ
abrj-637	325	27	and	and	CCONJ
abrj-637	325	28	responsible	responsible	ADJ
abrj-637	325	29	taxation	taxation	NOUN
abrj-637	325	30	.	.	PUNCT
abrj-637	326	1	the	the	DET
abrj-637	326	2	facts	fact	NOUN
abrj-637	326	3	provided	provide	VERB
abrj-637	326	4	in	in	ADP
abrj-637	326	5	this	this	DET
abrj-637	326	6	study	study	NOUN
abrj-637	326	7	support	support	VERB
abrj-637	326	8	the	the	DET
abrj-637	326	9	idea	idea	NOUN
abrj-637	326	10	that	that	PRON
abrj-637	326	11	ml	ml	VERB
abrj-637	326	12	can	can	AUX
abrj-637	326	13	not	not	PART
abrj-637	326	14	be	be	AUX
abrj-637	326	15	used	use	VERB
abrj-637	326	16	as	as	ADP
abrj-637	326	17	an	an	DET
abrj-637	326	18	alternative	alternative	NOUN
abrj-637	326	19	to	to	ADP
abrj-637	326	20	human	human	ADJ
abrj-637	326	21	auditors	auditor	NOUN
abrj-637	326	22	but	but	CCONJ
abrj-637	326	23	rather	rather	ADV
abrj-637	326	24	as	as	ADP
abrj-637	326	25	an	an	DET
abrj-637	326	26	effective	effective	ADJ
abrj-637	326	27	,	,	PUNCT
abrj-637	326	28	smart	smart	ADJ
abrj-637	326	29	partner	partner	NOUN
abrj-637	326	30	to	to	PART
abrj-637	326	31	foster	foster	VERB
abrj-637	326	32	financial	financial	ADJ
abrj-637	326	33	integrity	integrity	NOUN
abrj-637	326	34	and	and	CCONJ
abrj-637	326	35	trust	trust	NOUN
abrj-637	326	36	in	in	ADP
abrj-637	326	37	the	the	DET
abrj-637	326	38	populations	population	NOUN
abrj-637	326	39	.	.	PUNCT
abrj-637	327	1	7	7	X
abrj-637	327	2	.	.	X
abrj-637	327	3	conclusion	conclusion	NOUN
abrj-637	327	4	and	and	CCONJ
abrj-637	327	5	future	future	ADJ
abrj-637	327	6	work	work	NOUN
abrj-637	327	7	the	the	DET
abrj-637	327	8	growing	grow	VERB
abrj-637	327	9	sophistication	sophistication	NOUN
abrj-637	327	10	of	of	ADP
abrj-637	327	11	corporate	corporate	ADJ
abrj-637	327	12	tax	tax	NOUN
abrj-637	327	13	and	and	CCONJ
abrj-637	327	14	financial	financial	ADJ
abrj-637	327	15	reporting	reporting	NOUN
abrj-637	327	16	has	have	AUX
abrj-637	327	17	rendered	render	VERB
abrj-637	327	18	the	the	DET
abrj-637	327	19	old	old	ADJ
abrj-637	327	20	methods	method	NOUN
abrj-637	327	21	of	of	ADP
abrj-637	327	22	detecting	detect	VERB
abrj-637	327	23	fraud	fraud	NOUN
abrj-637	327	24	insufficient	insufficient	ADJ
abrj-637	327	25	in	in	ADP
abrj-637	327	26	the	the	DET
abrj-637	327	27	presence	presence	NOUN
abrj-637	327	28	of	of	ADP
abrj-637	327	29	huge	huge	ADJ
abrj-637	327	30	,	,	PUNCT
abrj-637	327	31	many	many	ADJ
abrj-637	327	32	-	-	PUNCT
abrj-637	327	33	dimensional	dimensional	ADJ
abrj-637	327	34	information	information	NOUN
abrj-637	327	35	.	.	PUNCT
abrj-637	328	1	this	this	DET
abrj-637	328	2	paper	paper	NOUN
abrj-637	328	3	has	have	AUX
abrj-637	328	4	shown	show	VERB
abrj-637	328	5	that	that	PRON
abrj-637	328	6	machine	machine	NOUN
abrj-637	328	7	learning	learning	NOUN
abrj-637	328	8	(	(	PUNCT
abrj-637	328	9	ml	ml	NOUN
abrj-637	328	10	)	)	PUNCT
abrj-637	328	11	may	may	AUX
abrj-637	328	12	transform	transform	VERB
abrj-637	328	13	the	the	DET
abrj-637	328	14	process	process	NOUN
abrj-637	328	15	of	of	ADP
abrj-637	328	16	fraud	fraud	NOUN
abrj-637	328	17	detection	detection	NOUN
abrj-637	328	18	and	and	CCONJ
abrj-637	328	19	scoring	scoring	NOUN
abrj-637	328	20	of	of	ADP
abrj-637	328	21	risks	risk	NOUN
abrj-637	328	22	in	in	ADP
abrj-637	328	23	taxation	taxation	NOUN
abrj-637	328	24	by	by	ADP
abrj-637	328	25	automating	automate	VERB
abrj-637	328	26	pattern	pattern	NOUN
abrj-637	328	27	recognition	recognition	NOUN
abrj-637	328	28	,	,	PUNCT
abrj-637	328	29	anomaly	anomaly	NOUN
abrj-637	328	30	detection	detection	NOUN
abrj-637	328	31	,	,	PUNCT
abrj-637	328	32	and	and	CCONJ
abrj-637	328	33	predictive	predictive	ADJ
abrj-637	328	34	analytics	analytic	NOUN
abrj-637	328	35	in	in	ADP
abrj-637	328	36	corporate	corporate	ADJ
abrj-637	328	37	filings	filing	NOUN
abrj-637	328	38	.	.	PUNCT
abrj-637	329	1	with	with	ADP
abrj-637	329	2	the	the	DET
abrj-637	329	3	help	help	NOUN
abrj-637	329	4	of	of	ADP
abrj-637	329	5	both	both	DET
abrj-637	329	6	structured	structured	ADJ
abrj-637	329	7	financial	financial	ADJ
abrj-637	329	8	and	and	CCONJ
abrj-637	329	9	unstructured	unstructured	ADJ
abrj-637	329	10	text	text	NOUN
abrj-637	329	11	information	information	NOUN
abrj-637	329	12	,	,	PUNCT
abrj-637	329	13	the	the	DET
abrj-637	329	14	ml	ml	NOUN
abrj-637	329	15	-	-	PUNCT
abrj-637	329	16	based	base	VERB
abrj-637	329	17	systems	system	NOUN
abrj-637	329	18	can	can	AUX
abrj-637	329	19	help	help	VERB
abrj-637	329	20	the	the	DET
abrj-637	329	21	auditors	auditor	NOUN
abrj-637	329	22	and	and	CCONJ
abrj-637	329	23	tax	tax	NOUN
abrj-637	329	24	authorities	authority	NOUN
abrj-637	329	25	to	to	PART
abrj-637	329	26	go	go	VERB
abrj-637	329	27	beyond	beyond	ADP
abrj-637	329	28	reactive	reactive	NOUN
abrj-637	329	29	,	,	PUNCT
abrj-637	329	30	manual	manual	ADJ
abrj-637	329	31	processes	process	NOUN
abrj-637	329	32	to	to	ADP
abrj-637	329	33	proactive	proactive	ADJ
abrj-637	329	34	,	,	PUNCT
abrj-637	329	35	data	data	NOUN
abrj-637	329	36	-	-	PUNCT
abrj-637	329	37	driven	drive	VERB
abrj-637	329	38	decision	decision	NOUN
abrj-637	329	39	-	-	PUNCT
abrj-637	329	40	making	making	NOUN
abrj-637	329	41	(	(	PUNCT
abrj-637	329	42	acharya	acharya	NOUN
abrj-637	329	43	,	,	PUNCT
abrj-637	329	44	2025	2025	NUM
abrj-637	329	45	;	;	PUNCT
abrj-637	329	46	galla	galla	PROPN
abrj-637	329	47	,	,	PUNCT
abrj-637	329	48	2023	2023	NUM
abrj-637	329	49	;	;	PUNCT
abrj-637	329	50	zhou	zhou	PROPN
abrj-637	329	51	et	et	PROPN
abrj-637	329	52	al	al	PROPN
abrj-637	329	53	.	.	PROPN
abrj-637	329	54	,	,	PUNCT
abrj-637	329	55	2024	2024	NUM
abrj-637	329	56	)	)	PUNCT
abrj-637	329	57	.	.	PUNCT
abrj-637	330	1	the	the	DET
abrj-637	330	2	framework	framework	NOUN
abrj-637	330	3	proposed	propose	VERB
abrj-637	330	4	is	be	AUX
abrj-637	330	5	a	a	DET
abrj-637	330	6	combination	combination	NOUN
abrj-637	330	7	of	of	ADP
abrj-637	330	8	supervised	supervised	ADJ
abrj-637	330	9	learning	learning	NOUN
abrj-637	330	10	,	,	PUNCT
abrj-637	330	11	unsupervised	unsupervised	ADJ
abrj-637	330	12	learning	learning	NOUN
abrj-637	330	13	,	,	PUNCT
abrj-637	330	14	and	and	CCONJ
abrj-637	330	15	ensemble	ensemble	ADJ
abrj-637	330	16	learning	learning	NOUN
abrj-637	330	17	to	to	PART
abrj-637	330	18	provide	provide	VERB
abrj-637	330	19	high	high	ADJ
abrj-637	330	20	-	-	PUNCT
abrj-637	330	21	precision	precision	NOUN
abrj-637	330	22	and	and	CCONJ
abrj-637	330	23	flexibility	flexibility	NOUN
abrj-637	330	24	in	in	ADP
abrj-637	330	25	fraud	fraud	NOUN
abrj-637	330	26	detection	detection	NOUN
abrj-637	330	27	(	(	PUNCT
abrj-637	330	28	random	random	ADJ
abrj-637	330	29	forest	forest	NOUN
abrj-637	330	30	,	,	PUNCT
abrj-637	330	31	xgboost	xgboost	ADV
abrj-637	330	32	,	,	PUNCT
abrj-637	330	33	and	and	CCONJ
abrj-637	330	34	autoencoders	autoencoder	NOUN
abrj-637	330	35	)	)	PUNCT
abrj-637	330	36	.	.	PUNCT
abrj-637	331	1	explainable	explainable	ADJ
abrj-637	331	2	artificial	artificial	ADJ
abrj-637	331	3	intelligence	intelligence	NOUN
abrj-637	331	4	(	(	PUNCT
abrj-637	331	5	xai	xai	PROPN
abrj-637	331	6	)	)	PUNCT
abrj-637	331	7	is	be	AUX
abrj-637	331	8	also	also	ADV
abrj-637	331	9	used	use	VERB
abrj-637	331	10	to	to	PART
abrj-637	331	11	promote	promote	VERB
abrj-637	331	12	transparency	transparency	NOUN
abrj-637	331	13	and	and	CCONJ
abrj-637	331	14	accountability	accountability	NOUN
abrj-637	331	15	to	to	PART
abrj-637	331	16	ensure	ensure	VERB
abrj-637	331	17	that	that	SCONJ
abrj-637	331	18	human	human	ADJ
abrj-637	331	19	experts	expert	NOUN
abrj-637	331	20	are	be	AUX
abrj-637	331	21	able	able	ADJ
abrj-637	331	22	to	to	PART
abrj-637	331	23	interpret	interpret	VERB
abrj-637	331	24	and	and	CCONJ
abrj-637	331	25	audit	audit	NOUN
abrj-637	331	26	model	model	NOUN
abrj-637	331	27	outputs	output	NOUN
abrj-637	331	28	(	(	PUNCT
abrj-637	331	29	wahyono	wahyono	PROPN
abrj-637	331	30	and	and	CCONJ
abrj-637	331	31	david	david	PROPN
abrj-637	331	32	,	,	PUNCT
abrj-637	331	33	2025	2025	NUM
abrj-637	331	34	)	)	PUNCT
abrj-637	331	35	.	.	PUNCT
abrj-637	332	1	the	the	DET
abrj-637	332	2	system	system	NOUN
abrj-637	332	3	enables	enable	VERB
abrj-637	332	4	the	the	DET
abrj-637	332	5	detection	detection	NOUN
abrj-637	332	6	of	of	ADP
abrj-637	332	7	fraud	fraud	NOUN
abrj-637	332	8	to	to	PART
abrj-637	332	9	go	go	VERB
abrj-637	332	10	beyond	beyond	ADP
abrj-637	332	11	numerical	numerical	ADJ
abrj-637	332	12	anomalies	anomaly	NOUN
abrj-637	332	13	and	and	CCONJ
abrj-637	332	14	detects	detect	NOUN
abrj-637	332	15	deceptive	deceptive	ADJ
abrj-637	332	16	linguistic	linguistic	ADJ
abrj-637	332	17	indicators	indicator	NOUN
abrj-637	332	18	in	in	ADP
abrj-637	332	19	financial	financial	ADJ
abrj-637	332	20	accounts	account	NOUN
abrj-637	332	21	by	by	ADP
abrj-637	332	22	including	include	VERB
abrj-637	332	23	nlp	nlp	NOUN
abrj-637	332	24	-	-	PUNCT
abrj-637	332	25	based	base	VERB
abrj-637	332	26	text	text	NOUN
abrj-637	332	27	analytics	analytic	NOUN
abrj-637	332	28	(	(	PUNCT
abrj-637	332	29	zhang	zhang	X
abrj-637	332	30	et	et	PROPN
abrj-637	332	31	al	al	PROPN
abrj-637	332	32	.	.	PROPN
abrj-637	332	33	,	,	PUNCT
abrj-637	332	34	2024	2024	NUM
abrj-637	332	35	;	;	PUNCT
abrj-637	332	36	ji	ji	PROPN
abrj-637	332	37	et	et	PROPN
abrj-637	332	38	al	al	PROPN
abrj-637	332	39	.	.	PROPN
abrj-637	332	40	,	,	PUNCT
abrj-637	332	41	2024	2024	NUM
abrj-637	332	42	)	)	PUNCT
abrj-637	332	43	.	.	PUNCT
abrj-637	333	1	the	the	DET
abrj-637	333	2	policy	policy	NOUN
abrj-637	333	3	and	and	CCONJ
abrj-637	333	4	governance	governance	NOUN
abrj-637	333	5	implications	implication	NOUN
abrj-637	333	6	of	of	ADP
abrj-637	333	7	the	the	DET
abrj-637	333	8	findings	finding	NOUN
abrj-637	333	9	are	be	AUX
abrj-637	333	10	that	that	DET
abrj-637	333	11	regulatory	regulatory	ADJ
abrj-637	333	12	harmonisation	harmonisation	NOUN
abrj-637	333	13	,	,	PUNCT
abrj-637	333	14	ethical	ethical	ADJ
abrj-637	333	15	protection	protection	NOUN
abrj-637	333	16	,	,	PUNCT
abrj-637	333	17	and	and	CCONJ
abrj-637	333	18	capacity	capacity	NOUN
abrj-637	333	19	building	building	NOUN
abrj-637	333	20	is	be	AUX
abrj-637	333	21	required	require	VERB
abrj-637	333	22	to	to	PART
abrj-637	333	23	make	make	VERB
abrj-637	333	24	ai	ai	VERB
abrj-637	333	25	use	use	NOUN
abrj-637	333	26	in	in	ADP
abrj-637	333	27	taxation	taxation	NOUN
abrj-637	333	28	responsible	responsible	ADJ
abrj-637	333	29	.	.	PUNCT
abrj-637	334	1	institutions	institution	NOUN
abrj-637	334	2	and	and	CCONJ
abrj-637	334	3	governments	government	NOUN
abrj-637	334	4	have	have	VERB
abrj-637	334	5	to	to	PART
abrj-637	334	6	weigh	weigh	VERB
abrj-637	334	7	the	the	DET
abrj-637	334	8	benefits	benefit	NOUN
abrj-637	334	9	of	of	ADP
abrj-637	334	10	efficiency	efficiency	NOUN
abrj-637	334	11	against	against	ADP
abrj-637	334	12	fairness	fairness	NOUN
abrj-637	334	13	,	,	PUNCT
abrj-637	334	14	privacy	privacy	NOUN
abrj-637	334	15	,	,	PUNCT
abrj-637	334	16	and	and	CCONJ
abrj-637	334	17	due	due	ADJ
abrj-637	334	18	process	process	NOUN
abrj-637	334	19	.	.	PUNCT
abrj-637	335	1	data	datum	NOUN
abrj-637	335	2	scientists	scientist	NOUN
abrj-637	335	3	,	,	PUNCT
abrj-637	335	4	auditors	auditor	NOUN
abrj-637	335	5	,	,	PUNCT
abrj-637	335	6	and	and	CCONJ
abrj-637	335	7	policymakers	policymaker	NOUN
abrj-637	335	8	will	will	AUX
abrj-637	335	9	have	have	VERB
abrj-637	335	10	to	to	PART
abrj-637	335	11	collaborate	collaborate	VERB
abrj-637	335	12	to	to	PART
abrj-637	335	13	keep	keep	VERB
abrj-637	335	14	accountability	accountability	NOUN
abrj-637	335	15	and	and	CCONJ
abrj-637	335	16	foster	foster	ADJ
abrj-637	335	17	trust	trust	NOUN
abrj-637	335	18	among	among	ADP
abrj-637	335	19	the	the	DET
abrj-637	335	20	population	population	NOUN
abrj-637	335	21	as	as	SCONJ
abrj-637	335	22	more	more	ADJ
abrj-637	335	23	and	and	CCONJ
abrj-637	335	24	more	more	ADJ
abrj-637	335	25	fiscal	fiscal	ADJ
abrj-637	335	26	systems	system	NOUN
abrj-637	335	27	become	become	VERB
abrj-637	335	28	automated	automated	ADJ
abrj-637	335	29	(	(	PUNCT
abrj-637	335	30	tagbo	tagbo	NOUN
abrj-637	335	31	&	&	CCONJ
abrj-637	335	32	adekoya	adekoya	PROPN
abrj-637	335	33	,	,	PUNCT
abrj-637	335	34	2023	2023	NUM
abrj-637	335	35	;	;	PUNCT
abrj-637	335	36	breslin	breslin	PROPN
abrj-637	335	37	,	,	PUNCT
abrj-637	335	38	2021	2021	NUM
abrj-637	335	39	)	)	PUNCT
abrj-637	335	40	.	.	PUNCT
abrj-637	336	1	references	reference	NOUN
abrj-637	336	2	acharya	acharya	PROPN
abrj-637	336	3	,	,	PUNCT
abrj-637	336	4	p.	p.	NOUN
abrj-637	336	5	(	(	PUNCT
abrj-637	336	6	2025	2025	NUM
abrj-637	336	7	)	)	PUNCT
abrj-637	336	8	.	.	PUNCT
abrj-637	337	1	using	use	VERB
abrj-637	337	2	machine	machine	NOUN
abrj-637	337	3	learning	learn	VERB
abrj-637	337	4	to	to	PART
abrj-637	337	5	detect	detect	VERB
abrj-637	337	6	financial	financial	ADJ
abrj-637	337	7	fraud	fraud	NOUN
abrj-637	337	8	in	in	ADP
abrj-637	337	9	corporate	corporate	ADJ
abrj-637	337	10	filings	filing	NOUN
abrj-637	337	11	.	.	PUNCT
abrj-637	338	1	ssrn	ssrn	PROPN
abrj-637	338	2	.	.	PUNCT
abrj-637	339	1	https://doi.org/10.2139/ssrn.5207888	https://doi.org/10.2139/ssrn.5207888	PROPN
abrj-637	339	2	breslin	breslin	PROPN
abrj-637	339	3	,	,	PUNCT
abrj-637	339	4	j.	j.	PROPN
abrj-637	339	5	(	(	PUNCT
abrj-637	339	6	2021	2021	NUM
abrj-637	339	7	)	)	PUNCT
abrj-637	339	8	.	.	PUNCT
abrj-637	340	1	improving	improve	VERB
abrj-637	340	2	tax	tax	NOUN
abrj-637	340	3	audit	audit	NOUN
abrj-637	340	4	efficiency	efficiency	NOUN
abrj-637	340	5	using	use	VERB
abrj-637	340	6	machine	machine	NOUN
abrj-637	340	7	learning	learning	NOUN
abrj-637	340	8	.	.	PUNCT
abrj-637	341	1	journal	journal	PROPN
abrj-637	341	2	of	of	ADP
abrj-637	341	3	forensic	forensic	ADJ
abrj-637	341	4	accounting	accounting	NOUN
abrj-637	341	5	research	research	NOUN
abrj-637	341	6	,	,	PUNCT
abrj-637	341	7	2(1	2(1	NUM
abrj-637	341	8	)	)	PUNCT
abrj-637	341	9	,	,	PUNCT
abrj-637	341	10	34–52	34–52	NUM
abrj-637	341	11	.	.	PUNCT
abrj-637	342	1	https://doi.org/10.1080/08839514.2021.2012002	https://doi.org/10.1080/08839514.2021.2012002	PROPN
abrj-637	343	1	choudhary	choudhary	PROPN
abrj-637	343	2	,	,	PUNCT
abrj-637	343	3	h.	h.	PROPN
abrj-637	343	4	(	(	PUNCT
abrj-637	343	5	2025	2025	NUM
abrj-637	343	6	)	)	PUNCT
abrj-637	343	7	.	.	PUNCT
abrj-637	344	1	hybrid	hybrid	ADJ
abrj-637	344	2	approach	approach	NOUN
abrj-637	344	3	to	to	ADP
abrj-637	344	4	tax	tax	NOUN
abrj-637	344	5	fraud	fraud	NOUN
abrj-637	344	6	detection	detection	NOUN
abrj-637	344	7	using	use	VERB
abrj-637	344	8	machine	machine	NOUN
abrj-637	344	9	learning	learning	NOUN
abrj-637	344	10	.	.	PUNCT
abrj-637	345	1	ssrn	ssrn	PROPN
abrj-637	345	2	.	.	PUNCT
abrj-637	346	1	https://doi.org/10.2139/ssrn.5366577	https://doi.org/10.2139/ssrn.5366577	PROPN
abrj-637	346	2	craja	craja	PROPN
abrj-637	346	3	,	,	PUNCT
abrj-637	346	4	p.	p.	PROPN
abrj-637	346	5	,	,	PUNCT
abrj-637	346	6	kim	kim	PROPN
abrj-637	346	7	,	,	PUNCT
abrj-637	346	8	r.	r.	PROPN
abrj-637	346	9	,	,	PUNCT
abrj-637	346	10	lessmann	lessmann	PROPN
abrj-637	346	11	,	,	PUNCT
abrj-637	346	12	s.	s.	PROPN
abrj-637	346	13	,	,	PUNCT
abrj-637	346	14	&	&	CCONJ
abrj-637	346	15	vetter	vetter	PROPN
abrj-637	346	16	,	,	PUNCT
abrj-637	346	17	t.	t.	PROPN
abrj-637	346	18	(	(	PUNCT
abrj-637	346	19	2020	2020	NUM
abrj-637	346	20	)	)	PUNCT
abrj-637	346	21	.	.	PUNCT
abrj-637	347	1	deep	deep	ADJ
abrj-637	347	2	learning	learning	NOUN
abrj-637	347	3	for	for	ADP
abrj-637	347	4	detecting	detect	VERB
abrj-637	347	5	financial	financial	ADJ
abrj-637	347	6	statement	statement	NOUN
abrj-637	347	7	fraud	fraud	NOUN
abrj-637	347	8	.	.	PUNCT
abrj-637	348	1	journal	journal	PROPN
abrj-637	348	2	of	of	ADP
abrj-637	348	3	empirical	empirical	ADJ
abrj-637	348	4	finance	finance	NOUN
abrj-637	348	5	,	,	PUNCT
abrj-637	348	6	57	57	NUM
abrj-637	348	7	,	,	PUNCT
abrj-637	348	8	292–308	292–308	NUM
abrj-637	348	9	.	.	PUNCT
abrj-637	349	1	https://doi.org/10.1016/j.jempfin.2020.01.008	https://doi.org/10.1016/j.jempfin.2020.01.008	PROPN
abrj-637	349	2	galla	galla	PROPN
abrj-637	349	3	,	,	PUNCT
abrj-637	349	4	e.	e.	PROPN
abrj-637	349	5	p.	p.	PROPN
abrj-637	349	6	(	(	PUNCT
abrj-637	349	7	2023	2023	NUM
abrj-637	349	8	)	)	PUNCT
abrj-637	349	9	.	.	PUNCT
abrj-637	350	1	enhancing	enhance	VERB
abrj-637	350	2	performance	performance	NOUN
abrj-637	350	3	of	of	ADP
abrj-637	350	4	financial	financial	ADJ
abrj-637	350	5	fraud	fraud	NOUN
abrj-637	350	6	detection	detection	NOUN
abrj-637	350	7	using	use	VERB
abrj-637	350	8	machine	machine	NOUN
abrj-637	350	9	learning	learning	NOUN
abrj-637	350	10	.	.	PUNCT
abrj-637	351	1	ssrn	ssrn	PROPN
abrj-637	351	2	.	.	PUNCT
abrj-637	352	1	https://doi.org/10.2139/ssrn.4993827	https://doi.org/10.2139/ssrn.4993827	PROPN
abrj-637	352	2	https://doi.org/10.2139/ssrn.5207888	https://doi.org/10.2139/ssrn.5207888	X
abrj-637	353	1	https://doi.org/10.1080/08839514.2021.2012002	https://doi.org/10.1080/08839514.2021.2012002	X
abrj-637	353	2	https://doi.org/10.2139/ssrn.5366577	https://doi.org/10.2139/ssrn.5366577	PROPN
abrj-637	353	3	https://doi.org/10.1016/j.jempfin.2020.01.008	https://doi.org/10.1016/j.jempfin.2020.01.008	PROPN
abrj-637	353	4	https://doi.org/10.2139/ssrn.4993827	https://doi.org/10.2139/ssrn.4993827	ADP
abrj-637	353	5	asian	asian	PROPN
abrj-637	353	6	business	business	PROPN
abrj-637	353	7	research	research	NOUN
abrj-637	353	8	journal	journal	NOUN
abrj-637	353	9	,	,	PUNCT
abrj-637	353	10	2025	2025	NUM
abrj-637	353	11	,	,	PUNCT
abrj-637	353	12	10(11	10(11	NUM
abrj-637	353	13	):	):	PUNCT
abrj-637	353	14	1	1	NUM
abrj-637	353	15	-	-	SYM
abrj-637	353	16	13	13	NUM
abrj-637	353	17	13	13	NUM
abrj-637	353	18	©	©	PROPN
abrj-637	353	19	2025	2025	NUM
abrj-637	353	20	by	by	ADP
abrj-637	353	21	the	the	DET
abrj-637	353	22	authors	author	NOUN
abrj-637	353	23	;	;	PUNCT
abrj-637	353	24	licensee	licensee	NOUN
abrj-637	353	25	eastern	eastern	ADJ
abrj-637	353	26	centre	centre	NOUN
abrj-637	353	27	of	of	ADP
abrj-637	353	28	science	science	NOUN
abrj-637	353	29	and	and	CCONJ
abrj-637	353	30	education	education	NOUN
abrj-637	353	31	,	,	PUNCT
abrj-637	353	32	usa	usa	PROPN
abrj-637	353	33	ippolito	ippolito	PROPN
abrj-637	353	34	,	,	PUNCT
abrj-637	353	35	a.	a.	PROPN
abrj-637	353	36	,	,	PUNCT
abrj-637	353	37	&	&	CCONJ
abrj-637	353	38	lozano	lozano	PROPN
abrj-637	353	39	,	,	PUNCT
abrj-637	354	1	a.	a.	PROPN
abrj-637	354	2	c.	c.	PROPN
abrj-637	354	3	g.	g.	PROPN
abrj-637	354	4	(	(	PUNCT
abrj-637	354	5	2020	2020	NUM
abrj-637	354	6	)	)	PUNCT
abrj-637	354	7	.	.	PUNCT
abrj-637	355	1	tax	tax	NOUN
abrj-637	355	2	crime	crime	NOUN
abrj-637	355	3	prediction	prediction	NOUN
abrj-637	355	4	with	with	ADP
abrj-637	355	5	machine	machine	NOUN
abrj-637	355	6	learning	learning	NOUN
abrj-637	355	7	:	:	PUNCT
abrj-637	355	8	a	a	DET
abrj-637	355	9	case	case	NOUN
abrj-637	355	10	study	study	NOUN
abrj-637	355	11	in	in	ADP
abrj-637	355	12	the	the	DET
abrj-637	355	13	municipality	municipality	NOUN
abrj-637	355	14	of	of	ADP
abrj-637	355	15	são	são	PROPN
abrj-637	355	16	paulo	paulo	PROPN
abrj-637	355	17	.	.	PUNCT
abrj-637	356	1	in	in	ADP
abrj-637	356	2	proceedings	proceeding	NOUN
abrj-637	356	3	of	of	ADP
abrj-637	356	4	the	the	DET
abrj-637	356	5	12th	12th	ADJ
abrj-637	356	6	international	international	ADJ
abrj-637	356	7	conference	conference	NOUN
abrj-637	356	8	on	on	ADP
abrj-637	356	9	agents	agent	NOUN
abrj-637	356	10	and	and	CCONJ
abrj-637	356	11	artificial	artificial	ADJ
abrj-637	356	12	intelligence	intelligence	NOUN
abrj-637	356	13	(	(	PUNCT
abrj-637	356	14	icaart	icaart	NOUN
abrj-637	356	15	2020	2020	NUM
abrj-637	356	16	)	)	PUNCT
abrj-637	356	17	(	(	PUNCT
abrj-637	356	18	pp	pp	ADP
abrj-637	356	19	.	.	PUNCT
abrj-637	357	1	156–163	156–163	NUM
abrj-637	357	2	)	)	PUNCT
abrj-637	357	3	.	.	PUNCT
abrj-637	358	1	scitepress	scitepress	NOUN
abrj-637	358	2	.	.	PUNCT
abrj-637	358	3	https://doi.org/10.5220/0009564701560163	https://doi.org/10.5220/0009564701560163	PUNCT
abrj-637	359	1	ji	ji	PROPN
abrj-637	359	2	,	,	PUNCT
abrj-637	359	3	j.	j.	PROPN
abrj-637	359	4	,	,	PUNCT
abrj-637	359	5	li	li	PROPN
abrj-637	359	6	,	,	PUNCT
abrj-637	359	7	x.	x.	PROPN
abrj-637	359	8	,	,	PUNCT
abrj-637	359	9	wang	wang	PROPN
abrj-637	359	10	,	,	PUNCT
abrj-637	359	11	m.	m.	NOUN
abrj-637	359	12	,	,	PUNCT
abrj-637	359	13	&	&	CCONJ
abrj-637	359	14	zhou	zhou	PROPN
abrj-637	359	15	,	,	PUNCT
abrj-637	359	16	m.	m.	NOUN
abrj-637	359	17	(	(	PUNCT
abrj-637	359	18	2024	2024	NUM
abrj-637	359	19	)	)	PUNCT
abrj-637	359	20	.	.	PUNCT
abrj-637	360	1	a	a	DET
abrj-637	360	2	machine	machine	NOUN
abrj-637	360	3	learning	learn	VERB
abrj-637	360	4	approach	approach	NOUN
abrj-637	360	5	to	to	ADP
abrj-637	360	6	detecting	detect	VERB
abrj-637	360	7	financial	financial	ADJ
abrj-637	360	8	anomalies	anomaly	NOUN
abrj-637	360	9	.	.	PUNCT
abrj-637	361	1	in	in	ADP
abrj-637	361	2	proceedings	proceeding	NOUN
abrj-637	361	3	of	of	ADP
abrj-637	361	4	the	the	DET
abrj-637	361	5	acm	acm	NOUN
abrj-637	361	6	symposium	symposium	NOUN
abrj-637	361	7	on	on	ADP
abrj-637	361	8	applied	apply	VERB
abrj-637	361	9	computing	computing	NOUN
abrj-637	361	10	(	(	PUNCT
abrj-637	361	11	pp	pp	ADJ
abrj-637	361	12	.	.	PUNCT
abrj-637	361	13	3718751–3718820	3718751–3718820	NUM
abrj-637	361	14	)	)	PUNCT
abrj-637	361	15	.	.	PUNCT
abrj-637	362	1	acm	acm	PROPN
abrj-637	362	2	.	.	PUNCT
abrj-637	362	3	https://doi.org/10.1145/3718751.3718820	https://doi.org/10.1145/3718751.3718820	NOUN
abrj-637	362	4	martínez	martínez	PROPN
abrj-637	362	5	,	,	PUNCT
abrj-637	362	6	p.	p.	NOUN
abrj-637	362	7	m.	m.	NOUN
abrj-637	363	1	p.	p.	NOUN
abrj-637	363	2	(	(	PUNCT
abrj-637	363	3	2025	2025	NUM
abrj-637	363	4	)	)	PUNCT
abrj-637	363	5	.	.	PUNCT
abrj-637	364	1	comparative	comparative	ADJ
abrj-637	364	2	analysis	analysis	NOUN
abrj-637	364	3	of	of	ADP
abrj-637	364	4	machine	machine	NOUN
abrj-637	364	5	learning	learning	NOUN
abrj-637	364	6	models	model	NOUN
abrj-637	364	7	for	for	ADP
abrj-637	364	8	financial	financial	ADJ
abrj-637	364	9	fraud	fraud	NOUN
abrj-637	364	10	detection	detection	NOUN
abrj-637	364	11	.	.	PUNCT
abrj-637	365	1	international	international	ADJ
abrj-637	365	2	journal	journal	PROPN
abrj-637	365	3	of	of	ADP
abrj-637	365	4	economics	economic	NOUN
abrj-637	365	5	and	and	CCONJ
abrj-637	365	6	finance	finance	NOUN
abrj-637	365	7	,	,	PUNCT
abrj-637	365	8	7(7	7(7	NUM
abrj-637	365	9	)	)	PUNCT
abrj-637	365	10	,	,	PUNCT
abrj-637	365	11	178–193	178–193	NUM
abrj-637	365	12	.	.	PUNCT
abrj-637	366	1	https://doi.org/10.5539/ijef.v7n7p178	https://doi.org/10.5539/ijef.v7n7p178	PROPN
abrj-637	366	2	mehta	mehta	PROPN
abrj-637	366	3	,	,	PUNCT
abrj-637	366	4	p.	p.	PROPN
abrj-637	366	5	,	,	PUNCT
abrj-637	366	6	kumar	kumar	PROPN
abrj-637	366	7	,	,	PUNCT
abrj-637	366	8	s.	s.	PROPN
abrj-637	366	9	,	,	PUNCT
abrj-637	366	10	kumar	kumar	PROPN
abrj-637	366	11	,	,	PUNCT
abrj-637	366	12	r.	r.	PROPN
abrj-637	366	13	,	,	PUNCT
abrj-637	366	14	&	&	CCONJ
abrj-637	366	15	babu	babu	PROPN
abrj-637	366	16	,	,	PUNCT
abrj-637	366	17	c.	c.	PROPN
abrj-637	366	18	s.	s.	PROPN
abrj-637	366	19	(	(	PUNCT
abrj-637	366	20	2022	2022	NUM
abrj-637	366	21	)	)	PUNCT
abrj-637	366	22	.	.	PUNCT
abrj-637	367	1	enhancement	enhancement	NOUN
abrj-637	367	2	to	to	ADP
abrj-637	367	3	training	training	NOUN
abrj-637	367	4	of	of	ADP
abrj-637	367	5	bidirectional	bidirectional	ADJ
abrj-637	367	6	gan	gan	NOUN
abrj-637	367	7	:	:	PUNCT
abrj-637	367	8	an	an	DET
abrj-637	367	9	approach	approach	NOUN
abrj-637	367	10	to	to	PART
abrj-637	367	11	demystify	demystify	VERB
abrj-637	367	12	tax	tax	NOUN
abrj-637	367	13	fraud	fraud	NOUN
abrj-637	367	14	.	.	PUNCT
abrj-637	368	1	arxiv	arxiv	PROPN
abrj-637	368	2	preprint	preprint	PROPN
abrj-637	368	3	,	,	PUNCT
abrj-637	368	4	arxiv:2208.07675	arxiv:2208.07675	PROPN
abrj-637	368	5	.	.	PUNCT
abrj-637	369	1	https://arxiv.org/abs/2208.07675	https://arxiv.org/abs/2208.07675	PROPN
abrj-637	369	2	nguyen	nguyen	PROPN
abrj-637	369	3	,	,	PUNCT
abrj-637	369	4	c.	c.	PROPN
abrj-637	369	5	t.	t.	PROPN
abrj-637	369	6	(	(	PUNCT
abrj-637	369	7	2025	2025	NUM
abrj-637	369	8	)	)	PUNCT
abrj-637	369	9	.	.	PUNCT
abrj-637	370	1	predicting	predict	VERB
abrj-637	370	2	financial	financial	ADJ
abrj-637	370	3	reports	report	NOUN
abrj-637	370	4	fraud	fraud	NOUN
abrj-637	370	5	by	by	ADP
abrj-637	370	6	machine	machine	NOUN
abrj-637	370	7	learning	learning	NOUN
abrj-637	370	8	.	.	PUNCT
abrj-637	371	1	cogent	cogent	NOUN
abrj-637	371	2	business	business	PROPN
abrj-637	371	3	&	&	CCONJ
abrj-637	371	4	management	management	PROPN
abrj-637	371	5	,	,	PUNCT
abrj-637	371	6	12(1	12(1	NUM
abrj-637	371	7	)	)	PUNCT
abrj-637	371	8	,	,	PUNCT
abrj-637	371	9	2510556	2510556	NUM
abrj-637	371	10	.	.	PUNCT
abrj-637	372	1	https://doi.org/10.1080/23311975.2025.2510556	https://doi.org/10.1080/23311975.2025.2510556	PROPN
abrj-637	372	2	olushola	olushola	PROPN
abrj-637	372	3	,	,	PUNCT
abrj-637	372	4	a.	a.	NOUN
abrj-637	372	5	(	(	PUNCT
abrj-637	372	6	2024	2024	NUM
abrj-637	372	7	)	)	PUNCT
abrj-637	372	8	.	.	PUNCT
abrj-637	373	1	fraud	fraud	NOUN
abrj-637	373	2	detection	detection	NOUN
abrj-637	373	3	using	use	VERB
abrj-637	373	4	machine	machine	NOUN
abrj-637	373	5	learning	learning	NOUN
abrj-637	373	6	.	.	PUNCT
abrj-637	374	1	scienceopen	scienceopen	ADJ
abrj-637	374	2	preprints	preprint	NOUN
abrj-637	374	3	.	.	PUNCT
abrj-637	375	1	https://doi.org/10.14293/pr2199.000647.v1	https://doi.org/10.14293/pr2199.000647.v1	PROPN
abrj-637	375	2	shujaaddeen	shujaaddeen	PROPN
abrj-637	375	3	,	,	PUNCT
abrj-637	375	4	a.	a.	NOUN
abrj-637	375	5	,	,	PUNCT
abrj-637	375	6	ba	ba	NOUN
abrj-637	375	7	-	-	PUNCT
abrj-637	375	8	alwi	alwi	PROPN
abrj-637	375	9	,	,	PUNCT
abrj-637	375	10	f.	f.	PROPN
abrj-637	375	11	m.	m.	PROPN
abrj-637	375	12	,	,	PUNCT
abrj-637	375	13	&	&	CCONJ
abrj-637	375	14	al	al	PROPN
abrj-637	375	15	-	-	PUNCT
abrj-637	375	16	gaphari	gaphari	PROPN
abrj-637	375	17	,	,	PUNCT
abrj-637	375	18	g.	g.	PROPN
abrj-637	375	19	(	(	PUNCT
abrj-637	375	20	2024	2024	NUM
abrj-637	375	21	)	)	PUNCT
abrj-637	375	22	.	.	PUNCT
abrj-637	376	1	a	a	DET
abrj-637	376	2	new	new	ADJ
abrj-637	376	3	machine	machine	NOUN
abrj-637	376	4	learning	learning	NOUN
abrj-637	376	5	model	model	NOUN
abrj-637	376	6	for	for	ADP
abrj-637	376	7	detecting	detect	VERB
abrj-637	376	8	levels	level	NOUN
abrj-637	376	9	of	of	ADP
abrj-637	376	10	tax	tax	NOUN
abrj-637	376	11	evasion	evasion	NOUN
abrj-637	376	12	based	base	VERB
abrj-637	376	13	on	on	ADP
abrj-637	376	14	hybrid	hybrid	ADJ
abrj-637	376	15	neural	neural	ADJ
abrj-637	376	16	network	network	NOUN
abrj-637	376	17	.	.	PUNCT
abrj-637	377	1	international	international	ADJ
abrj-637	377	2	journal	journal	NOUN
abrj-637	377	3	of	of	ADP
abrj-637	377	4	intelligent	intelligent	ADJ
abrj-637	377	5	systems	system	NOUN
abrj-637	377	6	and	and	CCONJ
abrj-637	377	7	applications	application	NOUN
abrj-637	377	8	in	in	ADP
abrj-637	377	9	engineering	engineering	NOUN
abrj-637	377	10	,	,	PUNCT
abrj-637	377	11	12(11	12(11	NUM
abrj-637	377	12	suppl	suppl	PROPN
abrj-637	377	13	.	.	PUNCT
abrj-637	377	14	)	)	PUNCT
abrj-637	377	15	,	,	PUNCT
abrj-637	377	16	450–468	450–468	NUM
abrj-637	377	17	.	.	PUNCT
abrj-637	378	1	https://ijisae.org/index.php/ijisae/article/view/4467	https://ijisae.org/index.php/ijisae/article/view/4467	PROPN
abrj-637	378	2	tagbo	tagbo	PROPN
abrj-637	378	3	,	,	PUNCT
abrj-637	378	4	s.	s.	PROPN
abrj-637	378	5	k.	k.	PROPN
abrj-637	378	6	,	,	PUNCT
abrj-637	378	7	&	&	CCONJ
abrj-637	378	8	adekoya	adekoya	PROPN
abrj-637	378	9	,	,	PUNCT
abrj-637	378	10	a.	a.	PROPN
abrj-637	378	11	f.	f.	PROPN
abrj-637	378	12	(	(	PUNCT
abrj-637	378	13	2023	2023	NUM
abrj-637	378	14	)	)	PUNCT
abrj-637	378	15	.	.	PUNCT
abrj-637	379	1	a	a	DET
abrj-637	379	2	systematic	systematic	ADJ
abrj-637	379	3	literature	literature	NOUN
abrj-637	379	4	review	review	NOUN
abrj-637	379	5	of	of	ADP
abrj-637	379	6	machine	machine	NOUN
abrj-637	379	7	learning	learn	VERB
abrj-637	379	8	techniques	technique	NOUN
abrj-637	379	9	in	in	ADP
abrj-637	379	10	financial	financial	ADJ
abrj-637	379	11	fraud	fraud	NOUN
abrj-637	379	12	prevention	prevention	NOUN
abrj-637	379	13	and	and	CCONJ
abrj-637	379	14	detection	detection	NOUN
abrj-637	379	15	.	.	PUNCT
abrj-637	380	1	international	international	ADJ
abrj-637	380	2	journal	journal	PROPN
abrj-637	380	3	of	of	ADP
abrj-637	380	4	society	society	PROPN
abrj-637	380	5	systems	system	NOUN
abrj-637	380	6	science	science	NOUN
abrj-637	380	7	,	,	PUNCT
abrj-637	380	8	14(4	14(4	NUM
abrj-637	380	9	)	)	PUNCT
abrj-637	380	10	,	,	PUNCT
abrj-637	380	11	303–348	303–348	NUM
abrj-637	380	12	.	.	PUNCT
abrj-637	381	1	https://doi.org/10.1504/ijsss.2023.135468	https://doi.org/10.1504/ijsss.2023.135468	PROPN
abrj-637	381	2	tax	tax	NOUN
abrj-637	381	3	fraud	fraud	NOUN
abrj-637	381	4	detection	detection	NOUN
abrj-637	381	5	for	for	ADP
abrj-637	381	6	under	under	NOUN
abrj-637	381	7	-	-	PUNCT
abrj-637	381	8	reporting	reporting	NOUN
abrj-637	381	9	declarations	declaration	NOUN
abrj-637	381	10	using	use	VERB
abrj-637	381	11	an	an	DET
abrj-637	381	12	unsupervised	unsupervised	ADJ
abrj-637	381	13	learning	learning	NOUN
abrj-637	381	14	approach	approach	NOUN
abrj-637	381	15	.	.	PUNCT
abrj-637	382	1	(	(	PUNCT
abrj-637	382	2	2018	2018	NUM
abrj-637	382	3	)	)	PUNCT
abrj-637	382	4	.	.	PUNCT
abrj-637	383	1	in	in	ADP
abrj-637	383	2	proceedings	proceeding	NOUN
abrj-637	383	3	of	of	ADP
abrj-637	383	4	the	the	DET
abrj-637	383	5	24th	24th	PROPN
abrj-637	383	6	acm	acm	PROPN
abrj-637	383	7	sigkdd	sigkdd	NOUN
abrj-637	383	8	international	international	ADJ
abrj-637	383	9	conference	conference	NOUN
abrj-637	383	10	on	on	ADP
abrj-637	383	11	knowledge	knowledge	PROPN
abrj-637	383	12	discovery	discovery	PROPN
abrj-637	383	13	&	&	CCONJ
abrj-637	383	14	data	data	PROPN
abrj-637	383	15	mining	mining	PROPN
abrj-637	383	16	(	(	PUNCT
abrj-637	383	17	kdd	kdd	PROPN
abrj-637	383	18	’	’	PROPN
abrj-637	383	19	18	18	NUM
abrj-637	383	20	)	)	PUNCT
abrj-637	383	21	(	(	PUNCT
abrj-637	383	22	pp	pp	ADV
abrj-637	383	23	.	.	PUNCT
abrj-637	384	1	215–223	215–223	NUM
abrj-637	384	2	)	)	PUNCT
abrj-637	384	3	.	.	PUNCT
abrj-637	385	1	acm	acm	PROPN
abrj-637	385	2	.	.	PUNCT
abrj-637	385	3	https://doi.org/10.1145/3219819.3219878	https://doi.org/10.1145/3219819.3219878	PROPN
abrj-637	385	4	wahyono	wahyono	PROPN
abrj-637	385	5	,	,	PUNCT
abrj-637	385	6	t.	t.	PROPN
abrj-637	385	7	,	,	PUNCT
abrj-637	385	8	&	&	CCONJ
abrj-637	385	9	david	david	PROPN
abrj-637	385	10	,	,	PUNCT
abrj-637	385	11	f.	f.	PROPN
abrj-637	385	12	(	(	PUNCT
abrj-637	385	13	2025	2025	NUM
abrj-637	385	14	)	)	PUNCT
abrj-637	385	15	.	.	PUNCT
abrj-637	386	1	a	a	DET
abrj-637	386	2	systematic	systematic	ADJ
abrj-637	386	3	review	review	NOUN
abrj-637	386	4	of	of	ADP
abrj-637	386	5	machine	machine	NOUN
abrj-637	386	6	-	-	PUNCT
abrj-637	386	7	learning	learning	NOUN
abrj-637	386	8	-	-	PUNCT
abrj-637	386	9	based	base	VERB
abrj-637	386	10	approaches	approach	NOUN
abrj-637	386	11	for	for	ADP
abrj-637	386	12	financial	financial	ADJ
abrj-637	386	13	fraud	fraud	NOUN
abrj-637	386	14	detection	detection	NOUN
abrj-637	386	15	.	.	PUNCT
abrj-637	387	1	journal	journal	NOUN
abrj-637	387	2	of	of	ADP
abrj-637	387	3	system	system	NOUN
abrj-637	387	4	and	and	CCONJ
abrj-637	387	5	management	management	NOUN
abrj-637	387	6	sciences	science	NOUN
abrj-637	387	7	,	,	PUNCT
abrj-637	387	8	15(1	15(1	NUM
abrj-637	387	9	)	)	PUNCT
abrj-637	387	10	,	,	PUNCT
abrj-637	387	11	69–84	69–84	PROPN
abrj-637	387	12	.	.	PUNCT
abrj-637	388	1	https://doi.org/10.33168/jsms.2025.0105	https://doi.org/10.33168/jsms.2025.0105	PROPN
abrj-637	388	2	wu	wu	PROPN
abrj-637	388	3	,	,	PUNCT
abrj-637	388	4	r.	r.	PROPN
abrj-637	388	5	s.	s.	PROPN
abrj-637	388	6	,	,	PUNCT
abrj-637	388	7	ou	ou	PROPN
abrj-637	388	8	,	,	PUNCT
abrj-637	388	9	c.	c.	PROPN
abrj-637	388	10	s.	s.	PROPN
abrj-637	388	11	,	,	PUNCT
abrj-637	388	12	lin	lin	PROPN
abrj-637	388	13	,	,	PUNCT
abrj-637	388	14	h.	h.	PROPN
abrj-637	388	15	,	,	PUNCT
abrj-637	388	16	chang	chang	PROPN
abrj-637	388	17	,	,	PUNCT
abrj-637	388	18	s.	s.	PROPN
abrj-637	388	19	i.	i.	PROPN
abrj-637	388	20	,	,	PUNCT
abrj-637	388	21	&	&	CCONJ
abrj-637	388	22	yen	yen	PROPN
abrj-637	388	23	,	,	PUNCT
abrj-637	388	24	d.	d.	PROPN
abrj-637	388	25	c.	c.	PROPN
abrj-637	388	26	(	(	PUNCT
abrj-637	388	27	2012	2012	NUM
abrj-637	388	28	)	)	PUNCT
abrj-637	388	29	.	.	PUNCT
abrj-637	389	1	using	use	VERB
abrj-637	389	2	data	data	NOUN
abrj-637	389	3	mining	mining	NOUN
abrj-637	389	4	technique	technique	NOUN
abrj-637	389	5	to	to	PART
abrj-637	389	6	enhance	enhance	VERB
abrj-637	389	7	tax	tax	NOUN
abrj-637	389	8	evasion	evasion	NOUN
abrj-637	389	9	detection	detection	NOUN
abrj-637	389	10	performance	performance	NOUN
abrj-637	389	11	.	.	PUNCT
abrj-637	390	1	expert	expert	NOUN
abrj-637	390	2	systems	system	NOUN
abrj-637	390	3	with	with	ADP
abrj-637	390	4	applications	application	NOUN
abrj-637	390	5	,	,	PUNCT
abrj-637	390	6	39(10	39(10	NUM
abrj-637	390	7	)	)	PUNCT
abrj-637	390	8	,	,	PUNCT
abrj-637	390	9	8769–8777	8769–8777	NUM
abrj-637	390	10	.	.	PUNCT
abrj-637	391	1	https://doi.org/10.1016/j.eswa.2012.01.210	https://doi.org/10.1016/j.eswa.2012.01.210	PROPN
abrj-637	391	2	zhang	zhang	PROPN
abrj-637	391	3	,	,	PUNCT
abrj-637	391	4	l.	l.	PROPN
abrj-637	391	5	,	,	PUNCT
abrj-637	391	6	nan	nan	PROPN
abrj-637	391	7	,	,	PUNCT
abrj-637	391	8	x.	x.	PROPN
abrj-637	391	9	,	,	PUNCT
abrj-637	391	10	huang	huang	PROPN
abrj-637	391	11	,	,	PUNCT
abrj-637	391	12	e.	e.	PROPN
abrj-637	391	13	,	,	PUNCT
abrj-637	391	14	&	&	CCONJ
abrj-637	391	15	liu	liu	PROPN
abrj-637	391	16	,	,	PUNCT
abrj-637	391	17	s.	s.	PROPN
abrj-637	391	18	(	(	PUNCT
abrj-637	391	19	2020	2020	NUM
abrj-637	391	20	)	)	PUNCT
abrj-637	391	21	.	.	PUNCT
abrj-637	392	1	detecting	detect	VERB
abrj-637	392	2	transaction	transaction	NOUN
abrj-637	392	3	-	-	PUNCT
abrj-637	392	4	based	base	VERB
abrj-637	392	5	tax	tax	NOUN
abrj-637	392	6	evasion	evasion	NOUN
abrj-637	392	7	activities	activity	NOUN
abrj-637	392	8	on	on	ADP
abrj-637	392	9	social	social	ADJ
abrj-637	392	10	media	medium	NOUN
abrj-637	392	11	platforms	platform	NOUN
abrj-637	392	12	using	use	VERB
abrj-637	392	13	multimodal	multimodal	ADJ
abrj-637	392	14	deep	deep	ADJ
abrj-637	392	15	neural	neural	ADJ
abrj-637	392	16	networks	network	NOUN
abrj-637	392	17	.	.	PUNCT
abrj-637	393	1	arxiv	arxiv	PROPN
abrj-637	393	2	preprint	preprint	PROPN
abrj-637	393	3	,	,	PUNCT
abrj-637	393	4	arxiv:2007.13525	arxiv:2007.13525	NOUN
abrj-637	393	5	.	.	PUNCT
abrj-637	394	1	https://arxiv.org/abs/2007.13525	https://arxiv.org/abs/2007.13525	PROPN
abrj-637	394	2	zhang	zhang	PROPN
abrj-637	394	3	,	,	PUNCT
abrj-637	394	4	y.	y.	PROPN
abrj-637	394	5	,	,	PUNCT
abrj-637	394	6	wang	wang	PROPN
abrj-637	394	7	,	,	PUNCT
abrj-637	394	8	m.	m.	NOUN
abrj-637	394	9	,	,	PUNCT
abrj-637	394	10	zhou	zhou	PROPN
abrj-637	394	11	,	,	PUNCT
abrj-637	394	12	m.	m.	NOUN
abrj-637	394	13	,	,	PUNCT
abrj-637	394	14	&	&	CCONJ
abrj-637	394	15	yang	yang	PROPN
abrj-637	394	16	,	,	PUNCT
abrj-637	394	17	y.	y.	PROPN
abrj-637	394	18	(	(	PUNCT
abrj-637	394	19	2024	2024	NUM
abrj-637	394	20	)	)	PUNCT
abrj-637	394	21	.	.	PUNCT
abrj-637	395	1	corporate	corporate	ADJ
abrj-637	395	2	fraud	fraud	NOUN
abrj-637	395	3	detection	detection	NOUN
abrj-637	395	4	based	base	VERB
abrj-637	395	5	on	on	ADP
abrj-637	395	6	linguistic	linguistic	ADJ
abrj-637	395	7	readability	readability	NOUN
abrj-637	395	8	:	:	PUNCT
abrj-637	395	9	combining	combine	VERB
abrj-637	395	10	semantic	semantic	ADJ
abrj-637	395	11	features	feature	NOUN
abrj-637	395	12	in	in	ADP
abrj-637	395	13	nlp	nlp	NOUN
abrj-637	395	14	.	.	PROPN
abrj-637	395	15	journal	journal	PROPN
abrj-637	395	16	of	of	ADP
abrj-637	395	17	corporate	corporate	ADJ
abrj-637	395	18	finance	finance	NOUN
abrj-637	395	19	,	,	PUNCT
abrj-637	395	20	85	85	NUM
abrj-637	395	21	,	,	PUNCT
abrj-637	395	22	102356	102356	NUM
abrj-637	395	23	.	.	PUNCT
abrj-637	396	1	https://doi.org/10.1016/j.jcorpfin.2024.102356	https://doi.org/10.1016/j.jcorpfin.2024.102356	PROPN
abrj-637	396	2	zhang	zhang	PROPN
abrj-637	396	3	,	,	PUNCT
abrj-637	396	4	y.	y.	PROPN
abrj-637	396	5	,	,	PUNCT
abrj-637	396	6	chen	chen	PROPN
abrj-637	396	7	,	,	PUNCT
abrj-637	396	8	x.	x.	PROPN
abrj-637	396	9	,	,	PUNCT
abrj-637	396	10	li	li	PROPN
abrj-637	396	11	,	,	PUNCT
abrj-637	396	12	h.	h.	PROPN
abrj-637	396	13	,	,	PUNCT
abrj-637	396	14	&	&	CCONJ
abrj-637	396	15	zhou	zhou	PROPN
abrj-637	396	16	,	,	PUNCT
abrj-637	396	17	m.	m.	NOUN
abrj-637	396	18	(	(	PUNCT
abrj-637	396	19	2025	2025	NUM
abrj-637	396	20	)	)	PUNCT
abrj-637	396	21	.	.	PUNCT
abrj-637	397	1	the	the	DET
abrj-637	397	2	analysis	analysis	NOUN
abrj-637	397	3	of	of	ADP
abrj-637	397	4	fraud	fraud	NOUN
abrj-637	397	5	detection	detection	NOUN
abrj-637	397	6	in	in	ADP
abrj-637	397	7	financial	financial	ADJ
abrj-637	397	8	markets	market	NOUN
abrj-637	397	9	under	under	ADP
abrj-637	397	10	stacking	stack	VERB
abrj-637	397	11	ensemble	ensemble	ADJ
abrj-637	397	12	learning	learning	NOUN
abrj-637	397	13	.	.	PUNCT
abrj-637	398	1	scientific	scientific	ADJ
abrj-637	398	2	reports	report	NOUN
abrj-637	398	3	,	,	PUNCT
abrj-637	398	4	15	15	NUM
abrj-637	398	5	,	,	PUNCT
abrj-637	398	6	15783	15783	NUM
abrj-637	398	7	.	.	PUNCT
abrj-637	399	1	https://doi.org/10.1038/s41598-025-15783-2	https://doi.org/10.1038/s41598-025-15783-2	NUM
abrj-637	400	1	zhou	zhou	PROPN
abrj-637	400	2	,	,	PUNCT
abrj-637	400	3	g.	g.	PROPN
abrj-637	400	4	,	,	PUNCT
abrj-637	400	5	wang	wang	PROPN
abrj-637	400	6	,	,	PUNCT
abrj-637	400	7	x.	x.	PROPN
abrj-637	400	8	,	,	PUNCT
abrj-637	400	9	li	li	PROPN
abrj-637	400	10	,	,	PUNCT
abrj-637	400	11	j.	j.	PROPN
abrj-637	400	12	,	,	PUNCT
abrj-637	400	13	&	&	CCONJ
abrj-637	400	14	chen	chen	PROPN
abrj-637	400	15	,	,	PUNCT
abrj-637	400	16	m.	m.	NOUN
abrj-637	400	17	(	(	PUNCT
abrj-637	400	18	2024	2024	NUM
abrj-637	400	19	)	)	PUNCT
abrj-637	400	20	.	.	PUNCT
abrj-637	401	1	advanced	advanced	ADJ
abrj-637	401	2	tax	tax	NOUN
abrj-637	401	3	fraud	fraud	NOUN
abrj-637	401	4	detection	detection	NOUN
abrj-637	401	5	:	:	PUNCT
abrj-637	401	6	a	a	DET
abrj-637	401	7	soft	soft	ADJ
abrj-637	401	8	-	-	PUNCT
abrj-637	401	9	voting	voting	NOUN
abrj-637	401	10	ensemble	ensemble	ADJ
abrj-637	401	11	based	base	VERB
abrj-637	401	12	on	on	ADP
abrj-637	401	13	encoder	encoder	NOUN
abrj-637	401	14	extraction	extraction	NOUN
abrj-637	401	15	.	.	PUNCT
abrj-637	402	1	mathematics	mathematic	NOUN
abrj-637	402	2	,	,	PUNCT
abrj-637	402	3	13(4	13(4	NUM
abrj-637	402	4	)	)	PUNCT
abrj-637	402	5	,	,	PUNCT
abrj-637	402	6	642	642	NUM
abrj-637	402	7	.	.	PUNCT
abrj-637	403	1	https://doi.org/10.3390/math13040642	https://doi.org/10.3390/math13040642	PROPN
abrj-637	403	2	https://doi.org/10.5220/0009564701560163	https://doi.org/10.5220/0009564701560163	PROPN
abrj-637	403	3	https://doi.org/10.1145/3718751.3718820	https://doi.org/10.1145/3718751.3718820	NOUN
abrj-637	403	4	https://doi.org/10.5539/ijef.v7n7p178	https://doi.org/10.5539/ijef.v7n7p178	PUNCT
abrj-637	403	5	https://arxiv.org/abs/2208.07675	https://arxiv.org/abs/2208.07675	NOUN
abrj-637	403	6	https://doi.org/10.1080/23311975.2025.2510556	https://doi.org/10.1080/23311975.2025.2510556	PROPN
abrj-637	403	7	https://doi.org/10.14293/pr2199.000647.v1	https://doi.org/10.14293/pr2199.000647.v1	PROPN
abrj-637	403	8	https://ijisae.org/index.php/ijisae/article/view/4467	https://ijisae.org/index.php/ijisae/article/view/4467	NOUN
abrj-637	403	9	https://doi.org/10.1504/ijsss.2023.135468	https://doi.org/10.1504/ijsss.2023.135468	PROPN
abrj-637	403	10	https://doi.org/10.1145/3219819.3219878	https://doi.org/10.1145/3219819.3219878	ADJ
abrj-637	403	11	https://doi.org/10.33168/jsms.2025.0105	https://doi.org/10.33168/jsms.2025.0105	PROPN
abrj-637	403	12	https://doi.org/10.1016/j.eswa.2012.01.210	https://doi.org/10.1016/j.eswa.2012.01.210	PROPN
abrj-637	403	13	https://arxiv.org/abs/2007.13525	https://arxiv.org/abs/2007.13525	NOUN
abrj-637	403	14	https://doi.org/10.1016/j.jcorpfin.2024.102356	https://doi.org/10.1016/j.jcorpfin.2024.102356	PROPN
abrj-637	403	15	https://doi.org/10.1038/s41598-025-15783-2	https://doi.org/10.1038/s41598-025-15783-2	NUM
abrj-637	403	16	https://doi.org/10.3390/math13040642	https://doi.org/10.3390/math13040642	PROPN