id	sid	tid	token	lemma	pos
fcis-6909	1	1	frontiers	frontier	NOUN
fcis-6909	1	2	in	in	ADP
fcis-6909	1	3	computing	computing	NOUN
fcis-6909	1	4	and	and	CCONJ
fcis-6909	1	5	intelligent	intelligent	ADJ
fcis-6909	1	6	systems	system	NOUN
fcis-6909	1	7	issn	issn	VERB
fcis-6909	1	8	:	:	PUNCT
fcis-6909	1	9	2832	2832	NUM
fcis-6909	1	10	-	-	SYM
fcis-6909	1	11	6024	6024	NUM
fcis-6909	1	12	|	|	NOUN
fcis-6909	1	13	vol	vol	NOUN
fcis-6909	1	14	.	.	PROPN
fcis-6909	2	1	3	3	NUM
fcis-6909	2	2	,	,	PUNCT
fcis-6909	2	3	no	no	INTJ
fcis-6909	2	4	.	.	NOUN
fcis-6909	2	5	2	2	NUM
fcis-6909	2	6	,	,	PUNCT
fcis-6909	2	7	2023	2023	NUM
fcis-6909	2	8	10	10	NUM
fcis-6909	2	9	bias	bias	NOUN
fcis-6909	2	10	-	-	PUNCT
fcis-6909	2	11	based	base	VERB
fcis-6909	2	12	denoising	denoising	NOUN
fcis-6909	2	13	causal	causal	ADJ
fcis-6909	2	14	recommendation	recommendation	NOUN
fcis-6909	2	15	algorithm	algorithm	PROPN
fcis-6909	3	1	xu	xu	PROPN
fcis-6909	3	2	yang	yang	PROPN
fcis-6909	3	3	school	school	PROPN
fcis-6909	3	4	of	of	ADP
fcis-6909	3	5	computer	computer	NOUN
fcis-6909	3	6	science	science	NOUN
fcis-6909	3	7	and	and	CCONJ
fcis-6909	3	8	technology	technology	NOUN
fcis-6909	3	9	,	,	PUNCT
fcis-6909	3	10	henan	henan	PROPN
fcis-6909	3	11	polytechnic	polytechnic	PROPN
fcis-6909	3	12	university	university	PROPN
fcis-6909	3	13	,	,	PUNCT
fcis-6909	3	14	jiaozuo	jiaozuo	PROPN
fcis-6909	3	15	454003	454003	NUM
fcis-6909	3	16	,	,	PUNCT
fcis-6909	3	17	china	china	PROPN
fcis-6909	3	18	.	.	PUNCT
fcis-6909	4	1	abstract	abstract	ADJ
fcis-6909	4	2	:	:	PUNCT
fcis-6909	5	1	traditional	traditional	ADJ
fcis-6909	5	2	recommendation	recommendation	NOUN
fcis-6909	5	3	algorithms	algorithm	NOUN
fcis-6909	5	4	,	,	PUNCT
fcis-6909	5	5	such	such	ADJ
fcis-6909	5	6	as	as	ADP
fcis-6909	5	7	collaborative	collaborative	ADJ
fcis-6909	5	8	filtering	filtering	NOUN
fcis-6909	5	9	,	,	PUNCT
fcis-6909	5	10	make	make	VERB
fcis-6909	5	11	recommendations	recommendation	NOUN
fcis-6909	5	12	by	by	ADP
fcis-6909	5	13	learning	learn	VERB
fcis-6909	5	14	the	the	DET
fcis-6909	5	15	relevant	relevant	ADJ
fcis-6909	5	16	relationships	relationship	NOUN
fcis-6909	5	17	between	between	ADP
fcis-6909	5	18	users	user	NOUN
fcis-6909	5	19	and	and	CCONJ
fcis-6909	5	20	items	item	NOUN
fcis-6909	5	21	.	.	PUNCT
fcis-6909	6	1	however	however	ADV
fcis-6909	6	2	,	,	PUNCT
fcis-6909	6	3	considering	consider	VERB
fcis-6909	6	4	only	only	ADV
fcis-6909	6	5	the	the	DET
fcis-6909	6	6	relationships	relationship	NOUN
fcis-6909	6	7	without	without	ADP
fcis-6909	6	8	considering	consider	VERB
fcis-6909	6	9	the	the	DET
fcis-6909	6	10	underlying	underlying	ADJ
fcis-6909	6	11	causal	causal	ADJ
fcis-6909	6	12	mechanisms	mechanism	NOUN
fcis-6909	6	13	would	would	AUX
fcis-6909	6	14	be	be	AUX
fcis-6909	6	15	unfair	unfair	ADJ
fcis-6909	6	16	,	,	PUNCT
fcis-6909	6	17	uninterpretable	uninterpretable	ADJ
fcis-6909	6	18	,	,	PUNCT
fcis-6909	6	19	and	and	CCONJ
fcis-6909	6	20	would	would	AUX
fcis-6909	6	21	lead	lead	VERB
fcis-6909	6	22	to	to	PART
fcis-6909	6	23	bias	bias	VERB
fcis-6909	6	24	.	.	PUNCT
fcis-6909	7	1	in	in	ADP
fcis-6909	7	2	this	this	DET
fcis-6909	7	3	paper	paper	NOUN
fcis-6909	7	4	,	,	PUNCT
fcis-6909	7	5	we	we	PRON
fcis-6909	7	6	propose	propose	VERB
fcis-6909	7	7	bias	bias	NOUN
fcis-6909	7	8	-	-	PUNCT
fcis-6909	7	9	based	base	VERB
fcis-6909	7	10	denoising	denoising	NOUN
fcis-6909	7	11	causal	causal	ADJ
fcis-6909	7	12	recommendation	recommendation	NOUN
fcis-6909	7	13	algorithm	algorithm	NOUN
fcis-6909	7	14	(	(	PUNCT
fcis-6909	7	15	bdcr	bdcr	NOUN
fcis-6909	7	16	)	)	PUNCT
fcis-6909	7	17	.	.	PUNCT
fcis-6909	8	1	first	first	ADV
fcis-6909	8	2	,	,	PUNCT
fcis-6909	8	3	the	the	DET
fcis-6909	8	4	method	method	NOUN
fcis-6909	8	5	dynamically	dynamically	ADV
fcis-6909	8	6	transforms	transform	VERB
fcis-6909	8	7	the	the	DET
fcis-6909	8	8	explicit	explicit	ADJ
fcis-6909	8	9	user	user	NOUN
fcis-6909	8	10	-	-	PUNCT
fcis-6909	8	11	item	item	NOUN
fcis-6909	8	12	feedback	feedback	NOUN
fcis-6909	8	13	into	into	ADP
fcis-6909	8	14	implicit	implicit	ADJ
fcis-6909	8	15	feedback	feedback	NOUN
fcis-6909	8	16	with	with	ADP
fcis-6909	8	17	an	an	DET
fcis-6909	8	18	embedded	embed	VERB
fcis-6909	8	19	representation	representation	NOUN
fcis-6909	8	20	.	.	PUNCT
fcis-6909	9	1	then	then	ADV
fcis-6909	9	2	,	,	PUNCT
fcis-6909	9	3	a	a	DET
fcis-6909	9	4	truncation	truncation	NOUN
fcis-6909	9	5	function	function	NOUN
fcis-6909	9	6	based	base	VERB
fcis-6909	9	7	on	on	ADP
fcis-6909	9	8	causal	causal	ADJ
fcis-6909	9	9	inference	inference	NOUN
fcis-6909	9	10	is	be	AUX
fcis-6909	9	11	constructed	construct	VERB
fcis-6909	9	12	to	to	PART
fcis-6909	9	13	remove	remove	VERB
fcis-6909	9	14	false	false	ADJ
fcis-6909	9	15	positive	positive	ADJ
fcis-6909	9	16	noise	noise	NOUN
fcis-6909	9	17	.	.	PUNCT
fcis-6909	10	1	in	in	ADP
fcis-6909	10	2	addition	addition	NOUN
fcis-6909	10	3	,	,	PUNCT
fcis-6909	10	4	traditional	traditional	ADJ
fcis-6909	10	5	recommendations	recommendation	NOUN
fcis-6909	10	6	and	and	CCONJ
fcis-6909	10	7	denoised	denoise	VERB
fcis-6909	10	8	causal	causal	ADJ
fcis-6909	10	9	recommendations	recommendation	NOUN
fcis-6909	10	10	are	be	AUX
fcis-6909	10	11	aggregated	aggregate	VERB
fcis-6909	10	12	to	to	PART
fcis-6909	10	13	obtain	obtain	VERB
fcis-6909	10	14	predictive	predictive	ADJ
fcis-6909	10	15	scores	score	NOUN
fcis-6909	10	16	.	.	PUNCT
fcis-6909	11	1	finally	finally	ADV
fcis-6909	11	2	,	,	PUNCT
fcis-6909	11	3	experimental	experimental	ADJ
fcis-6909	11	4	results	result	NOUN
fcis-6909	11	5	on	on	ADP
fcis-6909	11	6	two	two	NUM
fcis-6909	11	7	real	real	ADJ
fcis-6909	11	8	datasets	dataset	NOUN
fcis-6909	11	9	show	show	VERB
fcis-6909	11	10	that	that	SCONJ
fcis-6909	11	11	the	the	DET
fcis-6909	11	12	bdcr	bdcr	NOUN
fcis-6909	11	13	algorithm	algorithm	NOUN
fcis-6909	11	14	outperforms	outperform	VERB
fcis-6909	11	15	the	the	DET
fcis-6909	11	16	classical	classical	ADJ
fcis-6909	11	17	algorithm	algorithm	NOUN
fcis-6909	11	18	in	in	ADP
fcis-6909	11	19	terms	term	NOUN
fcis-6909	11	20	of	of	ADP
fcis-6909	11	21	recall	recall	NOUN
fcis-6909	11	22	and	and	CCONJ
fcis-6909	11	23	ndcg	ndcg	NOUN
fcis-6909	11	24	metrics	metric	NOUN
fcis-6909	11	25	.	.	PUNCT
fcis-6909	12	1	keywords	keyword	NOUN
fcis-6909	12	2	:	:	PUNCT
fcis-6909	13	1	causal	causal	ADV
fcis-6909	13	2	;	;	PUNCT
fcis-6909	13	3	denoising	denoising	NOUN
fcis-6909	13	4	;	;	PUNCT
fcis-6909	13	5	bias	bias	NOUN
fcis-6909	13	6	;	;	PUNCT
fcis-6909	13	7	positive	positive	ADJ
fcis-6909	13	8	;	;	PUNCT
fcis-6909	13	9	recommendation	recommendation	NOUN
fcis-6909	13	10	.	.	PUNCT
fcis-6909	14	1	1	1	X
fcis-6909	14	2	.	.	X
fcis-6909	14	3	introduction	introduction	NOUN
fcis-6909	14	4	today	today	NOUN
fcis-6909	14	5	's	's	PART
fcis-6909	14	6	era	era	NOUN
fcis-6909	14	7	is	be	AUX
fcis-6909	14	8	an	an	DET
fcis-6909	14	9	information	information	NOUN
fcis-6909	14	10	explosion	explosion	NOUN
fcis-6909	14	11	,	,	PUNCT
fcis-6909	14	12	and	and	CCONJ
fcis-6909	14	13	recommendation	recommendation	NOUN
fcis-6909	14	14	systems	system	NOUN
fcis-6909	14	15	are	be	AUX
fcis-6909	14	16	used	use	VERB
fcis-6909	14	17	to	to	PART
fcis-6909	14	18	solve	solve	VERB
fcis-6909	14	19	the	the	DET
fcis-6909	14	20	user	user	NOUN
fcis-6909	14	21	's	's	PART
fcis-6909	14	22	exploration	exploration	NOUN
fcis-6909	14	23	problem	problem	NOUN
fcis-6909	14	24	as	as	ADP
fcis-6909	14	25	an	an	DET
fcis-6909	14	26	information	information	NOUN
fcis-6909	14	27	filtering	filter	VERB
fcis-6909	14	28	system	system	NOUN
fcis-6909	14	29	,	,	PUNCT
fcis-6909	14	30	i.e.	i.e.	X
fcis-6909	14	31	,	,	PUNCT
fcis-6909	14	32	to	to	PART
fcis-6909	14	33	find	find	VERB
fcis-6909	14	34	the	the	DET
fcis-6909	14	35	most	most	ADV
fcis-6909	14	36	suitable	suitable	ADJ
fcis-6909	14	37	one	one	NUM
fcis-6909	14	38	among	among	ADP
fcis-6909	14	39	the	the	DET
fcis-6909	14	40	huge	huge	ADJ
fcis-6909	14	41	number	number	NOUN
fcis-6909	14	42	of	of	ADP
fcis-6909	14	43	available	available	ADJ
fcis-6909	14	44	options	option	NOUN
fcis-6909	14	45	,	,	PUNCT
fcis-6909	14	46	thus	thus	ADV
fcis-6909	14	47	saving	save	VERB
fcis-6909	14	48	the	the	DET
fcis-6909	14	49	user	user	NOUN
fcis-6909	14	50	's	's	PART
fcis-6909	14	51	time	time	NOUN
fcis-6909	14	52	and	and	CCONJ
fcis-6909	14	53	effort	effort	NOUN
fcis-6909	14	54	.	.	PUNCT
fcis-6909	15	1	traditional	traditional	ADJ
fcis-6909	15	2	recommender	recommender	NOUN
fcis-6909	15	3	systems	system	NOUN
fcis-6909	15	4	include	include	VERB
fcis-6909	15	5	three	three	NUM
fcis-6909	15	6	categories	category	NOUN
fcis-6909	15	7	:	:	PUNCT
fcis-6909	15	8	collaborative	collaborative	ADJ
fcis-6909	15	9	filtering	filtering	NOUN
fcis-6909	15	10	,	,	PUNCT
fcis-6909	15	11	content	content	NOUN
fcis-6909	15	12	-	-	PUNCT
fcis-6909	15	13	based	base	VERB
fcis-6909	15	14	recommendation	recommendation	NOUN
fcis-6909	15	15	and	and	CCONJ
fcis-6909	15	16	hybrid	hybrid	NOUN
fcis-6909	15	17	recommendation	recommendation	NOUN
fcis-6909	15	18	[	[	X
fcis-6909	15	19	1	1	NUM
fcis-6909	15	20	]	]	PUNCT
fcis-6909	15	21	.	.	PUNCT
fcis-6909	16	1	they	they	PRON
fcis-6909	16	2	are	be	AUX
fcis-6909	16	3	based	base	VERB
fcis-6909	16	4	on	on	ADP
fcis-6909	16	5	mining	mining	NOUN
fcis-6909	16	6	or	or	CCONJ
fcis-6909	16	7	learning	learn	VERB
fcis-6909	16	8	relevant	relevant	ADJ
fcis-6909	16	9	patterns	pattern	NOUN
fcis-6909	16	10	in	in	ADP
fcis-6909	16	11	data	datum	NOUN
fcis-6909	16	12	.	.	PUNCT
fcis-6909	17	1	in	in	ADP
fcis-6909	17	2	recent	recent	ADJ
fcis-6909	17	3	years	year	NOUN
fcis-6909	17	4	,	,	PUNCT
fcis-6909	17	5	researchers	researcher	NOUN
fcis-6909	17	6	are	be	AUX
fcis-6909	17	7	not	not	PART
fcis-6909	17	8	satisfied	satisfied	ADJ
fcis-6909	17	9	with	with	ADP
fcis-6909	17	10	simply	simply	ADV
fcis-6909	17	11	discovering	discover	VERB
fcis-6909	17	12	correlations	correlation	NOUN
fcis-6909	17	13	between	between	ADP
fcis-6909	17	14	learning	learn	VERB
fcis-6909	17	15	users	user	NOUN
fcis-6909	17	16	,	,	PUNCT
fcis-6909	17	17	items	item	NOUN
fcis-6909	17	18	,	,	PUNCT
fcis-6909	17	19	and	and	CCONJ
fcis-6909	17	20	features	feature	NOUN
fcis-6909	17	21	,	,	PUNCT
fcis-6909	17	22	but	but	CCONJ
fcis-6909	17	23	have	have	AUX
fcis-6909	17	24	started	start	VERB
fcis-6909	17	25	to	to	PART
fcis-6909	17	26	consider	consider	VERB
fcis-6909	17	27	causal	causal	ADJ
fcis-6909	17	28	relationships	relationship	NOUN
fcis-6909	17	29	.	.	PUNCT
fcis-6909	18	1	causality	causality	NOUN
fcis-6909	18	2	refers	refer	VERB
fcis-6909	18	3	to	to	ADP
fcis-6909	18	4	the	the	DET
fcis-6909	18	5	effect	effect	NOUN
fcis-6909	18	6	of	of	ADP
fcis-6909	18	7	one	one	NUM
fcis-6909	18	8	event	event	NOUN
fcis-6909	18	9	or	or	CCONJ
fcis-6909	18	10	behavior	behavior	NOUN
fcis-6909	18	11	on	on	ADP
fcis-6909	18	12	other	other	ADJ
fcis-6909	18	13	events	event	NOUN
fcis-6909	18	14	or	or	CCONJ
fcis-6909	18	15	behaviors	behavior	NOUN
fcis-6909	18	16	.	.	PUNCT
fcis-6909	19	1	recommender	recommender	NOUN
fcis-6909	19	2	systems	system	NOUN
fcis-6909	19	3	combined	combine	VERB
fcis-6909	19	4	with	with	ADP
fcis-6909	19	5	causal	causal	ADJ
fcis-6909	19	6	learning	learning	NOUN
fcis-6909	19	7	are	be	AUX
fcis-6909	19	8	able	able	ADJ
fcis-6909	19	9	to	to	PART
fcis-6909	19	10	deal	deal	VERB
fcis-6909	19	11	with	with	ADP
fcis-6909	19	12	different	different	ADJ
fcis-6909	19	13	problems	problem	NOUN
fcis-6909	19	14	including	include	VERB
fcis-6909	19	15	interpretability	interpretability	NOUN
fcis-6909	19	16	,	,	PUNCT
fcis-6909	19	17	fairness	fairness	NOUN
fcis-6909	19	18	,	,	PUNCT
fcis-6909	19	19	robustness	robustness	NOUN
fcis-6909	19	20	,	,	PUNCT
fcis-6909	19	21	and	and	CCONJ
fcis-6909	19	22	unbiasedness	unbiasedness	NOUN
fcis-6909	19	23	[	[	X
fcis-6909	19	24	2	2	NUM
fcis-6909	19	25	]	]	PUNCT
fcis-6909	19	26	.	.	PUNCT
fcis-6909	20	1	there	there	PRON
fcis-6909	20	2	have	have	AUX
fcis-6909	20	3	been	be	AUX
fcis-6909	20	4	many	many	ADJ
fcis-6909	20	5	studies	study	NOUN
fcis-6909	20	6	on	on	ADP
fcis-6909	20	7	various	various	ADJ
fcis-6909	20	8	causal	causal	ADJ
fcis-6909	20	9	recommendations	recommendation	NOUN
fcis-6909	20	10	,	,	PUNCT
fcis-6909	20	11	which	which	PRON
fcis-6909	20	12	deal	deal	VERB
fcis-6909	20	13	with	with	ADP
fcis-6909	20	14	different	different	ADJ
fcis-6909	20	15	problems	problem	NOUN
fcis-6909	20	16	in	in	ADP
fcis-6909	20	17	recommendations	recommendation	NOUN
fcis-6909	20	18	separately	separately	ADV
fcis-6909	20	19	.	.	PUNCT
fcis-6909	21	1	in	in	ADP
fcis-6909	21	2	2020	2020	NUM
fcis-6909	21	3	,	,	PUNCT
fcis-6909	21	4	sato	sato	NOUN
fcis-6909	21	5	[	[	X
fcis-6909	21	6	3]et	3]et	NUM
fcis-6909	21	7	al	al	PROPN
fcis-6909	21	8	.	.	PROPN
fcis-6909	21	9	identifies	identify	VERB
fcis-6909	21	10	user	user	NOUN
fcis-6909	21	11	/	/	SYM
fcis-6909	21	12	item	item	NOUN
fcis-6909	21	13	features	feature	NOUN
fcis-6909	21	14	as	as	ADP
fcis-6909	21	15	confounders	confounder	NOUN
fcis-6909	21	16	when	when	SCONJ
fcis-6909	21	17	estimating	estimate	VERB
fcis-6909	21	18	causal	causal	ADJ
fcis-6909	21	19	effects	effect	NOUN
fcis-6909	21	20	of	of	ADP
fcis-6909	21	21	recommendation	recommendation	NOUN
fcis-6909	21	22	.	.	PUNCT
fcis-6909	22	1	in	in	ADP
fcis-6909	22	2	2021	2021	NUM
fcis-6909	22	3	,	,	PUNCT
fcis-6909	22	4	wei	wei	X
fcis-6909	23	1	[	[	X
fcis-6909	23	2	4	4	X
fcis-6909	23	3	]	]	PUNCT
fcis-6909	23	4	et	et	PROPN
fcis-6909	23	5	al	al	PROPN
fcis-6909	23	6	.	.	PROPN
fcis-6909	23	7	proposed	propose	VERB
fcis-6909	23	8	a	a	DET
fcis-6909	23	9	modelindependent	modelindependent	NOUN
fcis-6909	23	10	counterfactual	counterfactual	ADJ
fcis-6909	23	11	inference	inference	NOUN
fcis-6909	23	12	method	method	NOUN
fcis-6909	23	13	to	to	PART
fcis-6909	23	14	construct	construct	VERB
fcis-6909	23	15	multi	multi	ADJ
fcis-6909	23	16	-	-	NOUN
fcis-6909	23	17	task	task	ADJ
fcis-6909	23	18	learning	learning	NOUN
fcis-6909	23	19	to	to	PART
fcis-6909	23	20	learn	learn	VERB
fcis-6909	23	21	the	the	DET
fcis-6909	23	22	effect	effect	NOUN
fcis-6909	23	23	of	of	ADP
fcis-6909	23	24	item	item	NOUN
fcis-6909	23	25	attributes	attribute	VERB
fcis-6909	23	26	on	on	ADP
fcis-6909	23	27	prediction	prediction	NOUN
fcis-6909	23	28	results	result	NOUN
fcis-6909	23	29	and	and	CCONJ
fcis-6909	23	30	user	user	NOUN
fcis-6909	23	31	attributes	attribute	VERB
fcis-6909	23	32	on	on	ADP
fcis-6909	23	33	prediction	prediction	NOUN
fcis-6909	23	34	results	result	NOUN
fcis-6909	23	35	separately	separately	ADV
fcis-6909	23	36	,	,	PUNCT
fcis-6909	23	37	and	and	CCONJ
fcis-6909	23	38	thus	thus	ADV
fcis-6909	23	39	obtain	obtain	VERB
fcis-6909	23	40	counterfactual	counterfactual	ADJ
fcis-6909	23	41	ranking	ranking	ADJ
fcis-6909	23	42	scores	score	NOUN
fcis-6909	23	43	using	use	VERB
fcis-6909	23	44	only	only	ADJ
fcis-6909	23	45	item	item	NOUN
fcis-6909	23	46	attributes	attribute	NOUN
fcis-6909	23	47	,	,	PUNCT
fcis-6909	23	48	and	and	CCONJ
fcis-6909	23	49	by	by	ADP
fcis-6909	23	50	deducting	deduct	VERB
fcis-6909	23	51	the	the	DET
fcis-6909	23	52	counterfactual	counterfactual	NOUN
fcis-6909	23	53	from	from	ADP
fcis-6909	23	54	the	the	DET
fcis-6909	23	55	overall	overall	ADJ
fcis-6909	23	56	score	score	NOUN
fcis-6909	23	57	scores	score	NOUN
fcis-6909	23	58	,	,	PUNCT
fcis-6909	23	59	eliminating	eliminate	VERB
fcis-6909	23	60	the	the	DET
fcis-6909	23	61	problem	problem	NOUN
fcis-6909	23	62	of	of	ADP
fcis-6909	23	63	popularity	popularity	NOUN
fcis-6909	23	64	bias	bias	NOUN
fcis-6909	23	65	from	from	ADP
fcis-6909	23	66	a	a	DET
fcis-6909	23	67	causal	causal	ADJ
fcis-6909	23	68	perspective	perspective	NOUN
fcis-6909	23	69	.	.	PUNCT
fcis-6909	24	1	in	in	ADP
fcis-6909	24	2	2022	2022	NUM
fcis-6909	24	3	,	,	PUNCT
fcis-6909	24	4	xu	xu	PROPN
fcis-6909	25	1	[	[	X
fcis-6909	25	2	5	5	NUM
fcis-6909	25	3	]	]	PUNCT
fcis-6909	25	4	et	et	PROPN
fcis-6909	25	5	al	al	PROPN
fcis-6909	25	6	.	.	PUNCT
fcis-6909	25	7	use	use	VERB
fcis-6909	25	8	counterfactual	counterfactual	ADJ
fcis-6909	25	9	explainable	explainable	ADJ
fcis-6909	25	10	recommendation	recommendation	NOUN
fcis-6909	25	11	(	(	PUNCT
fcis-6909	25	12	counter	counter	NOUN
fcis-6909	25	13	)	)	PUNCT
fcis-6909	25	14	,	,	PUNCT
fcis-6909	25	15	which	which	PRON
fcis-6909	25	16	introduces	introduce	VERB
fcis-6909	25	17	counterfactual	counterfactual	ADJ
fcis-6909	25	18	reasoning	reasoning	NOUN
fcis-6909	25	19	from	from	ADP
fcis-6909	25	20	causal	causal	ADJ
fcis-6909	25	21	inference	inference	NOUN
fcis-6909	25	22	to	to	ADP
fcis-6909	25	23	explainable	explainable	ADJ
fcis-6909	25	24	recommendations	recommendation	NOUN
fcis-6909	25	25	.	.	PUNCT
fcis-6909	26	1	although	although	SCONJ
fcis-6909	26	2	the	the	DET
fcis-6909	26	3	above	above	ADJ
fcis-6909	26	4	methods	method	NOUN
fcis-6909	26	5	are	be	AUX
fcis-6909	26	6	able	able	ADJ
fcis-6909	26	7	to	to	PART
fcis-6909	26	8	improve	improve	VERB
fcis-6909	26	9	the	the	DET
fcis-6909	26	10	performance	performance	NOUN
fcis-6909	26	11	of	of	ADP
fcis-6909	26	12	recommendation	recommendation	NOUN
fcis-6909	26	13	algorithms	algorithm	NOUN
fcis-6909	26	14	through	through	ADP
fcis-6909	26	15	causal	causal	ADJ
fcis-6909	26	16	inference	inference	NOUN
fcis-6909	26	17	,	,	PUNCT
fcis-6909	26	18	they	they	PRON
fcis-6909	26	19	fail	fail	VERB
fcis-6909	26	20	to	to	PART
fcis-6909	26	21	remove	remove	VERB
fcis-6909	26	22	noise	noise	NOUN
fcis-6909	26	23	,	,	PUNCT
fcis-6909	26	24	such	such	ADJ
fcis-6909	26	25	as	as	ADP
fcis-6909	26	26	the	the	DET
fcis-6909	26	27	effect	effect	NOUN
fcis-6909	26	28	of	of	ADP
fcis-6909	26	29	false	false	ADJ
fcis-6909	26	30	-	-	PUNCT
fcis-6909	26	31	positive	positive	ADJ
fcis-6909	26	32	interactions	interaction	NOUN
fcis-6909	26	33	,	,	PUNCT
fcis-6909	26	34	in	in	ADP
fcis-6909	26	35	the	the	DET
fcis-6909	26	36	original	original	ADJ
fcis-6909	26	37	dataset	dataset	NOUN
fcis-6909	26	38	in	in	ADP
fcis-6909	26	39	a	a	DET
fcis-6909	26	40	timely	timely	ADJ
fcis-6909	26	41	manner	manner	NOUN
fcis-6909	26	42	.	.	PUNCT
fcis-6909	27	1	2	2	X
fcis-6909	27	2	.	.	X
fcis-6909	27	3	related	relate	VERB
fcis-6909	27	4	work	work	NOUN
fcis-6909	27	5	causal	causal	NOUN
fcis-6909	27	6	reasoning	reasoning	NOUN
fcis-6909	27	7	has	have	AUX
fcis-6909	27	8	been	be	AUX
fcis-6909	27	9	developed	develop	VERB
fcis-6909	27	10	from	from	ADP
fcis-6909	27	11	correlation	correlation	NOUN
fcis-6909	27	12	studies	study	NOUN
fcis-6909	27	13	and	and	CCONJ
fcis-6909	27	14	contains	contain	VERB
fcis-6909	27	15	two	two	NUM
fcis-6909	27	16	main	main	ADJ
fcis-6909	27	17	model	model	NOUN
fcis-6909	27	18	frameworks	framework	NOUN
fcis-6909	27	19	,	,	PUNCT
fcis-6909	27	20	the	the	DET
fcis-6909	27	21	first	first	ADJ
fcis-6909	27	22	is	be	AUX
fcis-6909	27	23	the	the	DET
fcis-6909	27	24	potential	potential	ADJ
fcis-6909	27	25	outcomes	outcome	NOUN
fcis-6909	27	26	framework	framework	NOUN
fcis-6909	27	27	(	(	PUNCT
fcis-6909	27	28	pof	pof	PROPN
fcis-6909	27	29	)	)	PUNCT
fcis-6909	27	30	,	,	PUNCT
fcis-6909	27	31	also	also	ADV
fcis-6909	27	32	known	know	VERB
fcis-6909	27	33	as	as	ADP
fcis-6909	27	34	rubin	rubin	PROPN
fcis-6909	27	35	causal	causal	PROPN
fcis-6909	27	36	model	model	PROPN
fcis-6909	27	37	(	(	PUNCT
fcis-6909	27	38	rcm	rcm	PROPN
fcis-6909	27	39	)	)	PUNCT
fcis-6909	27	40	,	,	PUNCT
fcis-6909	27	41	proposed	propose	VERB
fcis-6909	27	42	by	by	ADP
fcis-6909	27	43	rubin	rubin	PROPN
fcis-6909	27	44	et	et	PROPN
fcis-6909	27	45	al	al	PROPN
fcis-6909	27	46	.	.	PROPN
fcis-6909	28	1	in	in	ADP
fcis-6909	28	2	1974	1974	NUM
fcis-6909	28	3	[	[	X
fcis-6909	28	4	6	6	NUM
fcis-6909	28	5	]	]	PUNCT
fcis-6909	28	6	,	,	PUNCT
fcis-6909	28	7	and	and	CCONJ
fcis-6909	28	8	the	the	DET
fcis-6909	28	9	other	other	ADJ
fcis-6909	28	10	is	be	AUX
fcis-6909	28	11	the	the	DET
fcis-6909	28	12	structural	structural	ADJ
fcis-6909	28	13	causal	causal	ADJ
fcis-6909	28	14	model	model	NOUN
fcis-6909	28	15	(	(	PUNCT
fcis-6909	28	16	scm	scm	PROPN
fcis-6909	28	17	)	)	PUNCT
fcis-6909	29	1	[	[	X
fcis-6909	29	2	7	7	NUM
fcis-6909	29	3	]	]	PUNCT
fcis-6909	29	4	.	.	PUNCT
fcis-6909	30	1	structural	structural	ADJ
fcis-6909	30	2	causal	causal	ADJ
fcis-6909	30	3	framework	framework	NOUN
fcis-6909	30	4	contains	contain	VERB
fcis-6909	30	5	two	two	NUM
fcis-6909	30	6	basic	basic	ADJ
fcis-6909	30	7	concepts	concept	NOUN
fcis-6909	30	8	:	:	PUNCT
fcis-6909	30	9	intervention	intervention	NOUN
fcis-6909	30	10	and	and	CCONJ
fcis-6909	30	11	counterfactual	counterfactual	ADJ
fcis-6909	30	12	,	,	PUNCT
fcis-6909	30	13	intervention	intervention	NOUN
fcis-6909	30	14	methods	method	NOUN
fcis-6909	30	15	including	include	VERB
fcis-6909	30	16	front	front	ADJ
fcis-6909	30	17	door	door	NOUN
fcis-6909	30	18	adjustment	adjustment	NOUN
fcis-6909	30	19	,	,	PUNCT
fcis-6909	30	20	back	back	ADJ
fcis-6909	30	21	door	door	NOUN
fcis-6909	30	22	adjustment	adjustment	NOUN
fcis-6909	30	23	,	,	PUNCT
fcis-6909	30	24	and	and	CCONJ
fcis-6909	30	25	potential	potential	ADJ
fcis-6909	30	26	outcome	outcome	NOUN
fcis-6909	30	27	framework	framework	NOUN
fcis-6909	30	28	to	to	PART
fcis-6909	30	29	estimate	estimate	VERB
fcis-6909	30	30	potential	potential	ADJ
fcis-6909	30	31	outcomes	outcome	NOUN
fcis-6909	30	32	and	and	CCONJ
fcis-6909	30	33	treatment	treatment	NOUN
fcis-6909	30	34	effects	effect	NOUN
fcis-6909	30	35	,	,	PUNCT
fcis-6909	30	36	including	include	VERB
fcis-6909	30	37	methods	method	NOUN
fcis-6909	30	38	such	such	ADJ
fcis-6909	30	39	as	as	ADP
fcis-6909	30	40	ipw	ipw	PROPN
fcis-6909	30	41	(	(	PUNCT
fcis-6909	30	42	inverse	inverse	NOUN
fcis-6909	30	43	probability	probability	NOUN
fcis-6909	30	44	weighting	weighting	NOUN
fcis-6909	30	45	)	)	PUNCT
fcis-6909	30	46	,	,	PUNCT
fcis-6909	30	47	and	and	CCONJ
fcis-6909	30	48	double	double	ADJ
fcis-6909	30	49	robustness	robustness	NOUN
fcis-6909	30	50	.	.	PUNCT
fcis-6909	31	1	as	as	SCONJ
fcis-6909	31	2	pearl	pearl	NOUN
fcis-6909	31	3	stated	state	VERB
fcis-6909	31	4	,	,	PUNCT
fcis-6909	31	5	these	these	DET
fcis-6909	31	6	two	two	NUM
fcis-6909	31	7	frameworks	framework	NOUN
fcis-6909	31	8	are	be	AUX
fcis-6909	31	9	logically	logically	ADV
fcis-6909	31	10	equivalent	equivalent	ADJ
fcis-6909	31	11	[	[	X
fcis-6909	31	12	8	8	NUM
fcis-6909	31	13	]	]	PUNCT
fcis-6909	31	14	.	.	PUNCT
fcis-6909	32	1	theorems	theorem	NOUN
fcis-6909	32	2	and	and	CCONJ
fcis-6909	32	3	assumptions	assumption	NOUN
fcis-6909	32	4	in	in	ADP
fcis-6909	32	5	one	one	NUM
fcis-6909	32	6	framework	framework	NOUN
fcis-6909	32	7	can	can	AUX
fcis-6909	32	8	be	be	AUX
fcis-6909	32	9	equally	equally	ADV
fcis-6909	32	10	translated	translate	VERB
fcis-6909	32	11	into	into	ADP
fcis-6909	32	12	the	the	DET
fcis-6909	32	13	language	language	NOUN
fcis-6909	32	14	of	of	ADP
fcis-6909	32	15	the	the	DET
fcis-6909	32	16	other	other	ADJ
fcis-6909	32	17	framework	framework	NOUN
fcis-6909	32	18	.	.	PUNCT
fcis-6909	33	1	in	in	ADP
fcis-6909	33	2	1998	1998	NUM
fcis-6909	33	3	,	,	PUNCT
fcis-6909	33	4	breese	breese	PROPN
fcis-6909	33	5	et	et	PROPN
fcis-6909	33	6	al	al	PROPN
fcis-6909	34	1	[	[	X
fcis-6909	34	2	9	9	NUM
fcis-6909	34	3	]	]	PUNCT
fcis-6909	34	4	applied	apply	VERB
fcis-6909	34	5	bayesian	bayesian	NOUN
fcis-6909	34	6	networks	network	NOUN
fcis-6909	34	7	in	in	ADP
fcis-6909	34	8	causal	causal	ADJ
fcis-6909	34	9	inference	inference	NOUN
fcis-6909	34	10	to	to	ADP
fcis-6909	34	11	personalized	personalized	ADJ
fcis-6909	34	12	recommendation	recommendation	NOUN
fcis-6909	34	13	and	and	CCONJ
fcis-6909	34	14	demonstrated	demonstrate	VERB
fcis-6909	34	15	the	the	DET
fcis-6909	34	16	superiority	superiority	NOUN
fcis-6909	34	17	of	of	ADP
fcis-6909	34	18	bayesian	bayesian	NOUN
fcis-6909	34	19	networks	network	NOUN
fcis-6909	34	20	under	under	ADP
fcis-6909	34	21	extensive	extensive	ADJ
fcis-6909	34	22	comparative	comparative	ADJ
fcis-6909	34	23	experiments	experiment	NOUN
fcis-6909	34	24	.	.	PUNCT
fcis-6909	35	1	in	in	ADP
fcis-6909	35	2	2020	2020	NUM
fcis-6909	35	3	,	,	PUNCT
fcis-6909	35	4	zhang	zhang	PROPN
fcis-6909	35	5	et	et	PROPN
fcis-6909	35	6	al	al	PROPN
fcis-6909	35	7	.	.	PROPN
fcis-6909	35	8	integrated	integrate	VERB
fcis-6909	35	9	the	the	DET
fcis-6909	35	10	learning	learning	NOUN
fcis-6909	35	11	of	of	ADP
fcis-6909	35	12	propensity	propensity	NOUN
fcis-6909	35	13	and	and	CCONJ
fcis-6909	35	14	recommendation	recommendation	NOUN
fcis-6909	35	15	models	model	NOUN
fcis-6909	35	16	into	into	ADP
fcis-6909	35	17	a	a	DET
fcis-6909	35	18	multi	multi	ADJ
fcis-6909	35	19	-	-	ADJ
fcis-6909	35	20	task	task	ADJ
fcis-6909	35	21	learning	learn	VERB
fcis-6909	35	22	framework	framework	NOUN
fcis-6909	35	23	[	[	X
fcis-6909	35	24	10	10	NUM
fcis-6909	35	25	]	]	PUNCT
fcis-6909	35	26	,	,	PUNCT
fcis-6909	35	27	which	which	PRON
fcis-6909	35	28	is	be	AUX
fcis-6909	35	29	more	more	ADV
fcis-6909	35	30	advantageous	advantageous	ADJ
fcis-6909	35	31	than	than	ADP
fcis-6909	35	32	learning	learn	VERB
fcis-6909	35	33	alone	alone	ADV
fcis-6909	35	34	.	.	PUNCT
fcis-6909	36	1	in	in	ADP
fcis-6909	36	2	2022	2022	NUM
fcis-6909	36	3	,	,	PUNCT
fcis-6909	36	4	wang	wang	PROPN
fcis-6909	36	5	et	et	PROPN
fcis-6909	36	6	al	al	PROPN
fcis-6909	37	1	[	[	X
fcis-6909	37	2	11	11	NUM
fcis-6909	37	3	]	]	PUNCT
fcis-6909	37	4	took	take	VERB
fcis-6909	37	5	a	a	DET
fcis-6909	37	6	pioneering	pioneering	ADJ
fcis-6909	37	7	step	step	NOUN
fcis-6909	37	8	in	in	ADP
fcis-6909	37	9	considering	consider	VERB
fcis-6909	37	10	exposure	exposure	NOUN
fcis-6909	37	11	bias	bias	NOUN
fcis-6909	37	12	in	in	ADP
fcis-6909	37	13	sequence	sequence	NOUN
fcis-6909	37	14	recommendation	recommendation	NOUN
fcis-6909	37	15	by	by	ADP
fcis-6909	37	16	proposing	propose	VERB
fcis-6909	37	17	an	an	DET
fcis-6909	37	18	ipw	ipw	PROPN
fcis-6909	37	19	-	-	PUNCT
fcis-6909	37	20	based	base	VERB
fcis-6909	37	21	usr	usr	NOUN
fcis-6909	37	22	approach	approach	NOUN
fcis-6909	37	23	(	(	PUNCT
fcis-6909	37	24	unbiased	unbiased	ADJ
fcis-6909	37	25	sequential	sequential	ADJ
fcis-6909	37	26	recommender	recommender	NOUN
fcis-6909	37	27	,	,	PUNCT
fcis-6909	37	28	unbiased	unbiased	ADJ
fcis-6909	37	29	sequential	sequential	ADJ
fcis-6909	37	30	recommendation	recommendation	NOUN
fcis-6909	37	31	model	model	NOUN
fcis-6909	37	32	)	)	PUNCT
fcis-6909	37	33	to	to	PART
fcis-6909	37	34	mitigate	mitigate	VERB
fcis-6909	37	35	the	the	DET
fcis-6909	37	36	confounding	confound	VERB
fcis-6909	37	37	factors	factor	NOUN
fcis-6909	37	38	in	in	ADP
fcis-6909	37	39	sequence	sequence	NOUN
fcis-6909	37	40	behavior	behavior	NOUN
fcis-6909	37	41	.	.	PUNCT
fcis-6909	38	1	despite	despite	SCONJ
fcis-6909	38	2	their	their	PRON
fcis-6909	38	3	good	good	ADJ
fcis-6909	38	4	performance	performance	NOUN
fcis-6909	38	5	in	in	ADP
fcis-6909	38	6	causal	causal	ADJ
fcis-6909	38	7	modeling	modeling	NOUN
fcis-6909	38	8	,	,	PUNCT
fcis-6909	38	9	they	they	PRON
fcis-6909	38	10	still	still	ADV
fcis-6909	38	11	do	do	AUX
fcis-6909	38	12	not	not	PART
fcis-6909	38	13	model	model	VERB
fcis-6909	38	14	noise	noise	NOUN
fcis-6909	38	15	in	in	ADP
fcis-6909	38	16	complex	complex	ADJ
fcis-6909	38	17	raw	raw	ADJ
fcis-6909	38	18	data	datum	NOUN
fcis-6909	38	19	well	well	ADV
fcis-6909	38	20	,	,	PUNCT
fcis-6909	38	21	while	while	SCONJ
fcis-6909	38	22	our	our	PRON
fcis-6909	38	23	approach	approach	NOUN
fcis-6909	38	24	uses	use	VERB
fcis-6909	38	25	a	a	DET
fcis-6909	38	26	truncation	truncation	NOUN
fcis-6909	38	27	function	function	NOUN
fcis-6909	38	28	to	to	PART
fcis-6909	38	29	remove	remove	VERB
fcis-6909	38	30	noise	noise	NOUN
fcis-6909	38	31	.	.	PUNCT
fcis-6909	39	1	many	many	ADJ
fcis-6909	39	2	previous	previous	ADJ
fcis-6909	39	3	works	work	NOUN
fcis-6909	39	4	have	have	AUX
fcis-6909	39	5	experimentally	experimentally	ADV
fcis-6909	39	6	demonstrated	demonstrate	VERB
fcis-6909	39	7	the	the	DET
fcis-6909	39	8	severity	severity	NOUN
fcis-6909	39	9	of	of	ADP
fcis-6909	39	10	data	datum	NOUN
fcis-6909	39	11	noise	noise	NOUN
fcis-6909	39	12	and	and	CCONJ
fcis-6909	39	13	its	its	PRON
fcis-6909	39	14	negative	negative	ADJ
fcis-6909	39	15	impact	impact	NOUN
fcis-6909	39	16	on	on	ADP
fcis-6909	39	17	recommender	recommender	NOUN
fcis-6909	39	18	systems	system	NOUN
fcis-6909	39	19	.	.	PUNCT
fcis-6909	40	1	cosley	cosley	PROPN
fcis-6909	40	2	et	et	PROPN
fcis-6909	40	3	al	al	PROPN
fcis-6909	41	1	[	[	X
fcis-6909	41	2	12	12	NUM
fcis-6909	41	3	]	]	PUNCT
fcis-6909	41	4	showed	show	VERB
fcis-6909	41	5	that	that	SCONJ
fcis-6909	41	6	only	only	ADV
fcis-6909	41	7	60	60	NUM
fcis-6909	41	8	%	%	NOUN
fcis-6909	41	9	of	of	ADP
fcis-6909	41	10	users	user	NOUN
fcis-6909	41	11	maintain	maintain	VERB
fcis-6909	41	12	their	their	PRON
fcis-6909	41	13	ratings	rating	NOUN
fcis-6909	41	14	when	when	SCONJ
fcis-6909	41	15	asked	ask	VERB
fcis-6909	41	16	to	to	PART
fcis-6909	41	17	re	re	VERB
fcis-6909	41	18	-	-	VERB
fcis-6909	41	19	rate	rate	VERB
fcis-6909	41	20	the	the	DET
fcis-6909	41	21	same	same	ADJ
fcis-6909	41	22	movie	movie	NOUN
fcis-6909	41	23	.	.	PUNCT
fcis-6909	42	1	amatriain	amatriain	PROPN
fcis-6909	42	2	et	et	PROPN
fcis-6909	42	3	al	al	PROPN
fcis-6909	43	1	[	[	X
fcis-6909	43	2	13	13	NUM
fcis-6909	43	3	]	]	PUNCT
fcis-6909	43	4	showed	show	VERB
fcis-6909	43	5	that	that	SCONJ
fcis-6909	43	6	compared	compare	VERB
fcis-6909	43	7	to	to	ADP
fcis-6909	43	8	noiseless	noiseless	ADJ
fcis-6909	43	9	data	datum	NOUN
fcis-6909	43	10	,	,	PUNCT
fcis-6909	43	11	recommendation	recommendation	NOUN
fcis-6909	43	12	performance	performance	NOUN
fcis-6909	43	13	with	with	ADP
fcis-6909	43	14	noisy	noisy	ADJ
fcis-6909	43	15	data	datum	NOUN
fcis-6909	43	16	is	be	AUX
fcis-6909	43	17	is	be	AUX
fcis-6909	43	18	significantly	significantly	ADV
fcis-6909	43	19	affected	affect	VERB
fcis-6909	43	20	,	,	PUNCT
fcis-6909	43	21	with	with	ADP
fcis-6909	43	22	a	a	DET
fcis-6909	43	23	difference	difference	NOUN
fcis-6909	43	24	of	of	ADP
fcis-6909	43	25	about	about	ADV
fcis-6909	43	26	40	40	NUM
fcis-6909	43	27	%	%	NOUN
fcis-6909	43	28	in	in	ADP
fcis-6909	43	29	rmse	rmse	NOUN
fcis-6909	43	30	.	.	PUNCT
fcis-6909	44	1	some	some	DET
fcis-6909	44	2	work	work	NOUN
fcis-6909	44	3	,	,	PUNCT
fcis-6909	44	4	such	such	ADJ
fcis-6909	44	5	as	as	ADP
fcis-6909	44	6	negative	negative	ADJ
fcis-6909	44	7	experience	experience	NOUN
fcis-6909	44	8	identification	identification	NOUN
fcis-6909	44	9	notes	note	VERB
fcis-6909	44	10	the	the	DET
fcis-6909	44	11	identification	identification	NOUN
fcis-6909	44	12	of	of	ADP
fcis-6909	44	13	negative	negative	ADJ
fcis-6909	44	14	experiences	experience	NOUN
fcis-6909	44	15	in	in	ADP
fcis-6909	44	16	implicit	implicit	ADJ
fcis-6909	44	17	signals	signal	NOUN
fcis-6909	44	18	,	,	PUNCT
fcis-6909	44	19	and	and	CCONJ
fcis-6909	44	20	previous	previous	ADJ
fcis-6909	44	21	work	work	NOUN
fcis-6909	44	22	typically	typically	ADV
fcis-6909	44	23	collects	collect	VERB
fcis-6909	44	24	feedback	feedback	NOUN
fcis-6909	44	25	from	from	ADP
fcis-6909	44	26	different	different	ADJ
fcis-6909	44	27	users	user	NOUN
fcis-6909	44	28	(	(	PUNCT
fcis-6909	44	29	e.g.	e.g.	ADV
fcis-6909	44	30	,	,	PUNCT
fcis-6909	44	31	dwell	dwell	NOUN
fcis-6909	44	32	time	time	NOUN
fcis-6909	44	33	,	,	PUNCT
fcis-6909	44	34	gaze	gaze	NOUN
fcis-6909	44	35	patterns	pattern	NOUN
fcis-6909	44	36	,	,	PUNCT
fcis-6909	44	37	and	and	CCONJ
fcis-6909	44	38	more	more	ADV
fcis-6909	44	39	detailed	detailed	ADJ
fcis-6909	44	40	item	item	NOUN
fcis-6909	44	41	characteristics	characteristic	NOUN
fcis-6909	44	42	)	)	PUNCT
fcis-6909	44	43	to	to	PART
fcis-6909	44	44	predict	predict	VERB
fcis-6909	44	45	user	user	NOUN
fcis-6909	44	46	satisfaction	satisfaction	NOUN
fcis-6909	44	47	[	[	X
fcis-6909	44	48	14	14	NUM
fcis-6909	44	49	]	]	PUNCT
fcis-6909	44	50	.	.	PUNCT
fcis-6909	45	1	however	however	ADV
fcis-6909	45	2	,	,	PUNCT
fcis-6909	45	3	these	these	DET
fcis-6909	45	4	approaches	approach	NOUN
fcis-6909	45	5	require	require	VERB
fcis-6909	45	6	11	11	NUM
fcis-6909	45	7	additional	additional	ADJ
fcis-6909	45	8	feedback	feedback	NOUN
fcis-6909	45	9	and	and	CCONJ
fcis-6909	45	10	extensive	extensive	ADJ
fcis-6909	45	11	manual	manual	ADJ
fcis-6909	45	12	marking	mark	VERB
fcis-6909	45	13	efforts	effort	NOUN
fcis-6909	45	14	,	,	PUNCT
fcis-6909	45	15	e.g.	e.g.	ADV
fcis-6909	45	16	,	,	PUNCT
fcis-6909	45	17	users	user	NOUN
fcis-6909	45	18	must	must	AUX
fcis-6909	45	19	tell	tell	VERB
fcis-6909	45	20	whether	whether	SCONJ
fcis-6909	45	21	they	they	PRON
fcis-6909	45	22	are	be	AUX
fcis-6909	45	23	satisfied	satisfied	ADJ
fcis-6909	45	24	with	with	ADP
fcis-6909	45	25	each	each	DET
fcis-6909	45	26	interaction	interaction	NOUN
fcis-6909	45	27	,	,	PUNCT
fcis-6909	45	28	and	and	CCONJ
fcis-6909	45	29	features	feature	VERB
fcis-6909	45	30	rely	rely	VERB
fcis-6909	45	31	on	on	ADP
fcis-6909	45	32	manual	manual	ADJ
fcis-6909	45	33	design	design	NOUN
fcis-6909	45	34	.	.	PUNCT
fcis-6909	46	1	wen	wen	PROPN
fcis-6909	46	2	et	et	PROPN
fcis-6909	46	3	al	al	PROPN
fcis-6909	47	1	[	[	X
fcis-6909	47	2	15	15	NUM
fcis-6909	47	3	]	]	PUNCT
fcis-6909	47	4	suggest	suggest	AUX
fcis-6909	47	5	using	use	VERB
fcis-6909	47	6	"	"	PUNCT
fcis-6909	47	7	click	click	VERB
fcis-6909	47	8	-	-	PUNCT
fcis-6909	47	9	to	to	AUX
fcis-6909	47	10	-	-	PUNCT
fcis-6909	47	11	complete	complete	ADJ
fcis-6909	47	12	"	"	PUNCT
fcis-6909	47	13	,	,	PUNCT
fcis-6909	47	14	"	"	PUNCT
fcis-6909	47	15	click	click	VERB
fcis-6909	47	16	-	-	PUNCT
fcis-6909	47	17	to	to	AUX
fcis-6909	47	18	-	-	PUNCT
fcis-6909	47	19	skip	skip	VERB
fcis-6909	47	20	"	"	PUNCT
fcis-6909	47	21	and	and	CCONJ
fcis-6909	47	22	"	"	PUNCT
fcis-6909	47	23	nonclick	nonclick	ADJ
fcis-6909	47	24	"	"	PUNCT
fcis-6909	47	25	items	item	NOUN
fcis-6909	47	26	to	to	PART
fcis-6909	47	27	train	train	VERB
fcis-6909	47	28	recommenders	recommender	NOUN
fcis-6909	47	29	.	.	PUNCT
fcis-6909	48	1	the	the	DET
fcis-6909	48	2	latter	latter	ADJ
fcis-6909	48	3	two	two	NUM
fcis-6909	48	4	items	item	NOUN
fcis-6909	48	5	are	be	AUX
fcis-6909	48	6	considered	consider	VERB
fcis-6909	48	7	as	as	ADP
fcis-6909	48	8	negative	negative	ADJ
fcis-6909	48	9	,	,	PUNCT
fcis-6909	48	10	but	but	CCONJ
fcis-6909	48	11	with	with	ADP
fcis-6909	48	12	different	different	ADJ
fcis-6909	48	13	weights	weight	NOUN
fcis-6909	48	14	.	.	PUNCT
fcis-6909	49	1	by	by	ADP
fcis-6909	49	2	jointly	jointly	ADV
fcis-6909	49	3	considering	consider	VERB
fcis-6909	49	4	different	different	ADJ
fcis-6909	49	5	feedbacks	feedback	NOUN
fcis-6909	49	6	,	,	PUNCT
fcis-6909	49	7	noise	noise	NOUN
fcis-6909	49	8	is	be	AUX
fcis-6909	49	9	also	also	ADV
fcis-6909	49	10	removed	remove	VERB
fcis-6909	49	11	to	to	ADP
fcis-6909	49	12	some	some	DET
fcis-6909	49	13	extent	extent	NOUN
fcis-6909	49	14	,	,	PUNCT
fcis-6909	49	15	but	but	CCONJ
fcis-6909	49	16	sometimes	sometimes	ADV
fcis-6909	49	17	this	this	DET
fcis-6909	49	18	feedback	feedback	NOUN
fcis-6909	49	19	is	be	AUX
fcis-6909	49	20	difficult	difficult	ADJ
fcis-6909	49	21	to	to	PART
fcis-6909	49	22	obtain	obtain	VERB
fcis-6909	49	23	.	.	PUNCT
fcis-6909	50	1	the	the	DET
fcis-6909	50	2	literature	literature	NOUN
fcis-6909	50	3	[	[	X
fcis-6909	50	4	16	16	NUM
fcis-6909	50	5	]	]	PUNCT
fcis-6909	50	6	explores	explore	NOUN
fcis-6909	50	7	denoising	denoise	VERB
fcis-6909	50	8	without	without	ADP
fcis-6909	50	9	adding	add	VERB
fcis-6909	50	10	information	information	NOUN
fcis-6909	50	11	during	during	ADP
fcis-6909	50	12	implicit	implicit	ADJ
fcis-6909	50	13	feedback	feedback	NOUN
fcis-6909	50	14	training	training	NOUN
fcis-6909	50	15	,	,	PUNCT
fcis-6909	50	16	pointing	point	VERB
fcis-6909	50	17	to	to	ADP
fcis-6909	50	18	a	a	DET
fcis-6909	50	19	new	new	ADJ
fcis-6909	50	20	research	research	NOUN
fcis-6909	50	21	direction	direction	NOUN
fcis-6909	50	22	for	for	ADP
fcis-6909	50	23	noise	noise	NOUN
fcis-6909	50	24	removal	removal	NOUN
fcis-6909	50	25	without	without	ADP
fcis-6909	50	26	using	use	VERB
fcis-6909	50	27	additional	additional	ADJ
fcis-6909	50	28	feedback	feedback	NOUN
fcis-6909	50	29	.	.	PUNCT
fcis-6909	51	1	although	although	SCONJ
fcis-6909	51	2	these	these	DET
fcis-6909	51	3	methods	method	NOUN
fcis-6909	51	4	are	be	AUX
fcis-6909	51	5	good	good	ADJ
fcis-6909	51	6	at	at	ADP
fcis-6909	51	7	removing	remove	VERB
fcis-6909	51	8	noise	noise	NOUN
fcis-6909	51	9	,	,	PUNCT
fcis-6909	51	10	they	they	PRON
fcis-6909	51	11	either	either	CCONJ
fcis-6909	51	12	require	require	VERB
fcis-6909	51	13	additional	additional	ADJ
fcis-6909	51	14	information	information	NOUN
fcis-6909	51	15	or	or	CCONJ
fcis-6909	51	16	do	do	AUX
fcis-6909	51	17	not	not	PART
fcis-6909	51	18	consider	consider	VERB
fcis-6909	51	19	causality	causality	NOUN
fcis-6909	51	20	,	,	PUNCT
fcis-6909	51	21	and	and	CCONJ
fcis-6909	51	22	this	this	DET
fcis-6909	51	23	paper	paper	NOUN
fcis-6909	51	24	designs	design	VERB
fcis-6909	51	25	a	a	DET
fcis-6909	51	26	denoising	denoising	NOUN
fcis-6909	51	27	method	method	NOUN
fcis-6909	51	28	that	that	PRON
fcis-6909	51	29	incorporates	incorporate	VERB
fcis-6909	51	30	causal	causal	ADJ
fcis-6909	51	31	inference	inference	NOUN
fcis-6909	51	32	.	.	PUNCT
fcis-6909	52	1	3	3	X
fcis-6909	52	2	.	.	X
fcis-6909	52	3	bdcr	bdcr	NOUN
fcis-6909	52	4	algorithm	algorithm	PROPN
fcis-6909	52	5	3.1	3.1	NUM
fcis-6909	52	6	.	.	PUNCT
fcis-6909	53	1	bdcr	bdcr	NOUN
fcis-6909	53	2	algorithm	algorithm	NOUN
fcis-6909	53	3	framework	framework	NOUN
fcis-6909	53	4	the	the	DET
fcis-6909	53	5	bdcr	bdcr	NOUN
fcis-6909	53	6	algorithm	algorithm	NOUN
fcis-6909	53	7	includes	include	VERB
fcis-6909	53	8	embedding	embed	VERB
fcis-6909	53	9	layer	layer	NOUN
fcis-6909	53	10	,	,	PUNCT
fcis-6909	53	11	bdcr	bdcr	NOUN
fcis-6909	53	12	layer	layer	NOUN
fcis-6909	53	13	,	,	PUNCT
fcis-6909	53	14	and	and	CCONJ
fcis-6909	53	15	prediction	prediction	NOUN
fcis-6909	53	16	layer	layer	NOUN
fcis-6909	53	17	.	.	PUNCT
fcis-6909	54	1	the	the	DET
fcis-6909	54	2	algorithm	algorithm	NOUN
fcis-6909	54	3	framework	framework	NOUN
fcis-6909	54	4	is	be	AUX
fcis-6909	54	5	shown	show	VERB
fcis-6909	54	6	in	in	ADP
fcis-6909	54	7	figure	figure	NOUN
fcis-6909	54	8	1	1	NUM
fcis-6909	54	9	.	.	PUNCT
fcis-6909	55	1	in	in	ADP
fcis-6909	55	2	embedding	embed	VERB
fcis-6909	55	3	layer	layer	NOUN
fcis-6909	55	4	,	,	PUNCT
fcis-6909	55	5	the	the	DET
fcis-6909	55	6	original	original	ADJ
fcis-6909	55	7	dataset	dataset	NOUN
fcis-6909	55	8	is	be	AUX
fcis-6909	55	9	processed	process	VERB
fcis-6909	55	10	as	as	ADP
fcis-6909	55	11	implicit	implicit	ADJ
fcis-6909	55	12	feedback	feedback	NOUN
fcis-6909	55	13	input	input	NOUN
fcis-6909	55	14	,	,	PUNCT
fcis-6909	55	15	and	and	CCONJ
fcis-6909	55	16	the	the	DET
fcis-6909	55	17	user	user	NOUN
fcis-6909	55	18	and	and	CCONJ
fcis-6909	55	19	item	item	NOUN
fcis-6909	55	20	features	feature	NOUN
fcis-6909	55	21	and	and	CCONJ
fcis-6909	55	22	eigenvalues	eigenvalue	NOUN
fcis-6909	55	23	are	be	AUX
fcis-6909	55	24	input	input	ADJ
fcis-6909	55	25	to	to	PART
fcis-6909	55	26	obtain	obtain	VERB
fcis-6909	55	27	the	the	DET
fcis-6909	55	28	embedding	embedding	NOUN
fcis-6909	55	29	;	;	PUNCT
fcis-6909	55	30	in	in	ADP
fcis-6909	55	31	the	the	DET
fcis-6909	55	32	bdcr	bdcr	NOUN
fcis-6909	55	33	layer	layer	NOUN
fcis-6909	55	34	,	,	PUNCT
fcis-6909	55	35	the	the	DET
fcis-6909	55	36	traditional	traditional	ADJ
fcis-6909	55	37	rating	rating	NOUN
fcis-6909	55	38	is	be	AUX
fcis-6909	55	39	first	first	ADV
fcis-6909	55	40	entered	enter	VERB
fcis-6909	55	41	into	into	ADP
fcis-6909	55	42	the	the	DET
fcis-6909	55	43	traditional	traditional	ADJ
fcis-6909	55	44	model	model	NOUN
fcis-6909	55	45	to	to	PART
fcis-6909	55	46	calculate	calculate	VERB
fcis-6909	55	47	the	the	DET
fcis-6909	55	48	traditional	traditional	ADJ
fcis-6909	55	49	score	score	NOUN
fcis-6909	55	50	,	,	PUNCT
fcis-6909	55	51	and	and	CCONJ
fcis-6909	55	52	the	the	DET
fcis-6909	55	53	false	false	ADJ
fcis-6909	55	54	positive	positive	ADJ
fcis-6909	55	55	noise	noise	NOUN
fcis-6909	55	56	data	datum	NOUN
fcis-6909	55	57	is	be	AUX
fcis-6909	55	58	removed	remove	VERB
fcis-6909	55	59	according	accord	VERB
fcis-6909	55	60	to	to	ADP
fcis-6909	55	61	the	the	DET
fcis-6909	55	62	rating	rating	NOUN
fcis-6909	55	63	,	,	PUNCT
fcis-6909	55	64	and	and	CCONJ
fcis-6909	55	65	the	the	DET
fcis-6909	55	66	causal	causal	ADJ
fcis-6909	55	67	debiasing	debiase	VERB
fcis-6909	55	68	model	model	NOUN
fcis-6909	55	69	is	be	AUX
fcis-6909	55	70	input	input	NOUN
fcis-6909	55	71	to	to	PART
fcis-6909	55	72	aggregate	aggregate	VERB
fcis-6909	55	73	the	the	DET
fcis-6909	55	74	calculated	calculated	ADJ
fcis-6909	55	75	m	m	NOUN
fcis-6909	55	76	-	-	NOUN
fcis-6909	55	77	values	value	NOUN
fcis-6909	55	78	by	by	ADP
fcis-6909	55	79	adding	add	VERB
fcis-6909	55	80	them	they	PRON
fcis-6909	55	81	to	to	ADP
fcis-6909	55	82	the	the	DET
fcis-6909	55	83	backbone	backbone	NOUN
fcis-6909	55	84	model	model	NOUN
fcis-6909	55	85	;	;	PUNCT
fcis-6909	55	86	in	in	ADP
fcis-6909	55	87	the	the	DET
fcis-6909	55	88	prediction	prediction	NOUN
fcis-6909	55	89	layer	layer	NOUN
fcis-6909	55	90	,	,	PUNCT
fcis-6909	55	91	the	the	DET
fcis-6909	55	92	inner	inner	ADJ
fcis-6909	55	93	product	product	NOUN
fcis-6909	55	94	calculates	calculate	VERB
fcis-6909	55	95	the	the	DET
fcis-6909	55	96	denoised	denoised	ADJ
fcis-6909	55	97	causal	causal	ADJ
fcis-6909	55	98	score	score	NOUN
fcis-6909	55	99	,	,	PUNCT
fcis-6909	55	100	interest	interest	NOUN
fcis-6909	55	101	drift	drift	NOUN
fcis-6909	55	102	degree	degree	NOUN
fcis-6909	55	103	,	,	PUNCT
fcis-6909	55	104	and	and	CCONJ
fcis-6909	55	105	then	then	ADV
fcis-6909	55	106	the	the	DET
fcis-6909	55	107	fusion	fusion	NOUN
fcis-6909	55	108	of	of	ADP
fcis-6909	55	109	the	the	DET
fcis-6909	55	110	traditional	traditional	ADJ
fcis-6909	55	111	algorithm	algorithm	NOUN
fcis-6909	55	112	with	with	ADP
fcis-6909	55	113	the	the	DET
fcis-6909	55	114	bdcr	bdcr	NOUN
fcis-6909	55	115	algorithm	algorithm	NOUN
fcis-6909	55	116	to	to	PART
fcis-6909	55	117	derive	derive	VERB
fcis-6909	55	118	the	the	DET
fcis-6909	55	119	final	final	ADJ
fcis-6909	55	120	prediction	prediction	NOUN
fcis-6909	55	121	.	.	PUNCT
fcis-6909	56	1	3.2	3.2	NUM
fcis-6909	56	2	.	.	PUNCT
fcis-6909	56	3	embedding	embed	VERB
fcis-6909	56	4	layer	layer	NOUN
fcis-6909	56	5	the	the	DET
fcis-6909	56	6	publicly	publicly	ADV
fcis-6909	56	7	available	available	ADJ
fcis-6909	56	8	datasets	dataset	NOUN
fcis-6909	56	9	ml-1	ml-1	NOUN
fcis-6909	56	10	m	m	PROPN
fcis-6909	56	11	and	and	CCONJ
fcis-6909	56	12	amazon	amazon	NOUN
fcis-6909	56	13	-	-	PUNCT
fcis-6909	56	14	book	book	NOUN
fcis-6909	56	15	are	be	AUX
fcis-6909	56	16	explicit	explicit	ADJ
fcis-6909	56	17	rating	rating	NOUN
fcis-6909	56	18	data	datum	NOUN
fcis-6909	56	19	,	,	PUNCT
fcis-6909	56	20	and	and	CCONJ
fcis-6909	56	21	the	the	DET
fcis-6909	56	22	rating	rating	NOUN
fcis-6909	56	23	are	be	AUX
fcis-6909	56	24	both	both	PRON
fcis-6909	56	25	on	on	ADP
fcis-6909	56	26	a	a	DET
fcis-6909	56	27	scale	scale	NOUN
fcis-6909	56	28	of	of	ADP
fcis-6909	56	29	15	15	NUM
fcis-6909	56	30	.	.	PUNCT
fcis-6909	57	1	since	since	SCONJ
fcis-6909	57	2	most	most	ADJ
fcis-6909	57	3	of	of	ADP
fcis-6909	57	4	the	the	DET
fcis-6909	57	5	existing	exist	VERB
fcis-6909	57	6	recommendation	recommendation	NOUN
fcis-6909	57	7	systems	system	NOUN
fcis-6909	57	8	use	use	VERB
fcis-6909	57	9	implicit	implicit	ADJ
fcis-6909	57	10	feedback	feedback	NOUN
fcis-6909	57	11	recommendations	recommendation	NOUN
fcis-6909	57	12	,	,	PUNCT
fcis-6909	57	13	they	they	PRON
fcis-6909	57	14	need	need	VERB
fcis-6909	57	15	to	to	PART
fcis-6909	57	16	be	be	AUX
fcis-6909	57	17	converted	convert	VERB
fcis-6909	57	18	,	,	PUNCT
fcis-6909	57	19	and	and	CCONJ
fcis-6909	57	20	the	the	DET
fcis-6909	57	21	conversion	conversion	NOUN
fcis-6909	57	22	chosen	choose	VERB
fcis-6909	57	23	for	for	ADP
fcis-6909	57	24	the	the	DET
fcis-6909	57	25	algorithm	algorithm	NOUN
fcis-6909	57	26	in	in	ADP
fcis-6909	57	27	this	this	DET
fcis-6909	57	28	chapter	chapter	NOUN
fcis-6909	57	29	is	be	AUX
fcis-6909	57	30	shown	show	VERB
fcis-6909	57	31	in	in	ADP
fcis-6909	57	32	equation	equation	NOUN
fcis-6909	57	33	1	1	NUM
fcis-6909	57	34	.	.	NOUN
fcis-6909	57	35	1	1	NUM
fcis-6909	57	36	,	,	PUNCT
fcis-6909	57	37	5	5	NUM
fcis-6909	57	38	0	0	NUM
fcis-6909	57	39	,	,	PUNCT
fcis-6909	57	40	ui	ui	NOUN
fcis-6909	57	41	r	r	NOUN
fcis-6909	57	42	y	y	PROPN
fcis-6909	57	43	else	else	ADV
fcis-6909	57	44	=	=	NOUN
fcis-6909	57	45			NOUN
fcis-6909	57	46	=	=	PUNCT
fcis-6909	58	1			NUM
fcis-6909	58	2			PROPN
fcis-6909	58	3	(	(	PUNCT
fcis-6909	58	4	1	1	NUM
fcis-6909	58	5	)	)	PUNCT
fcis-6909	58	6	where	where	SCONJ
fcis-6909	58	7	r	r	NOUN
fcis-6909	58	8	denotes	denote	VERB
fcis-6909	58	9	the	the	DET
fcis-6909	58	10	original	original	ADJ
fcis-6909	58	11	rating	rating	NOUN
fcis-6909	58	12	,	,	PUNCT
fcis-6909	58	13	yui	yui	PROPN
fcis-6909	58	14	denotes	denote	VERB
fcis-6909	58	15	the	the	DET
fcis-6909	58	16	modified	modify	VERB
fcis-6909	58	17	rating	rating	NOUN
fcis-6909	58	18	used	use	VERB
fcis-6909	58	19	in	in	ADP
fcis-6909	58	20	the	the	DET
fcis-6909	58	21	training	training	NOUN
fcis-6909	58	22	process	process	NOUN
fcis-6909	58	23	,	,	PUNCT
fcis-6909	58	24	i.e.	i.e.	X
fcis-6909	58	25	,	,	PUNCT
fcis-6909	58	26	a	a	DET
fcis-6909	58	27	full	full	ADJ
fcis-6909	58	28	rating	rating	NOUN
fcis-6909	58	29	is	be	AUX
fcis-6909	58	30	regarded	regard	VERB
fcis-6909	58	31	as	as	ADP
fcis-6909	58	32	a	a	DET
fcis-6909	58	33	positive	positive	ADJ
fcis-6909	58	34	example	example	NOUN
fcis-6909	58	35	,	,	PUNCT
fcis-6909	58	36	otherwise	otherwise	ADV
fcis-6909	58	37	it	it	PRON
fcis-6909	58	38	is	be	AUX
fcis-6909	58	39	regarded	regard	VERB
fcis-6909	58	40	as	as	ADP
fcis-6909	58	41	a	a	DET
fcis-6909	58	42	negative	negative	ADJ
fcis-6909	58	43	example	example	NOUN
fcis-6909	58	44	,	,	PUNCT
fcis-6909	58	45	and	and	CCONJ
fcis-6909	58	46	any	any	DET
fcis-6909	58	47	item	item	NOUN
fcis-6909	58	48	that	that	PRON
fcis-6909	58	49	the	the	DET
fcis-6909	58	50	user	user	NOUN
fcis-6909	58	51	is	be	AUX
fcis-6909	58	52	not	not	PART
fcis-6909	58	53	satisfied	satisfied	ADJ
fcis-6909	58	54	with	with	ADP
fcis-6909	58	55	is	be	AUX
fcis-6909	58	56	not	not	PART
fcis-6909	58	57	given	give	VERB
fcis-6909	58	58	a	a	DET
fcis-6909	58	59	full	full	ADJ
fcis-6909	58	60	rating	rating	NOUN
fcis-6909	58	61	.	.	PUNCT
fcis-6909	59	1	fig	fig	NOUN
fcis-6909	59	2	.	.	PUNCT
fcis-6909	60	1	1	1	NUM
fcis-6909	60	2	bdcr	bdcr	NOUN
fcis-6909	60	3	algorithm	algorithm	NOUN
fcis-6909	60	4	framework	framework	NOUN
fcis-6909	60	5	subsequently	subsequently	ADV
fcis-6909	60	6	,	,	PUNCT
fcis-6909	60	7	the	the	DET
fcis-6909	60	8	user	user	NOUN
fcis-6909	60	9	features	feature	NOUN
fcis-6909	60	10	are	be	AUX
fcis-6909	60	11	jointly	jointly	ADV
fcis-6909	60	12	coded	code	VERB
fcis-6909	60	13	with	with	ADP
fcis-6909	60	14	the	the	DET
fcis-6909	60	15	item	item	NOUN
fcis-6909	60	16	features	feature	VERB
fcis-6909	60	17	,	,	PUNCT
fcis-6909	60	18	with	with	ADP
fcis-6909	60	19	all	all	DET
fcis-6909	60	20	user	user	NOUN
fcis-6909	60	21	feature	feature	NOUN
fcis-6909	60	22	feature	feature	NOUN
fcis-6909	60	23	values	value	NOUN
fcis-6909	60	24	being	be	AUX
fcis-6909	60	25	1	1	NUM
fcis-6909	60	26	and	and	CCONJ
fcis-6909	60	27	the	the	DET
fcis-6909	60	28	item	item	NOUN
fcis-6909	60	29	feature	feature	NOUN
fcis-6909	60	30	values	value	NOUN
fcis-6909	60	31	being	be	AUX
fcis-6909	60	32	determined	determine	VERB
fcis-6909	60	33	according	accord	VERB
fcis-6909	60	34	to	to	ADP
fcis-6909	60	35	the	the	DET
fcis-6909	60	36	category	category	NOUN
fcis-6909	60	37	to	to	PART
fcis-6909	60	38	which	which	PRON
fcis-6909	60	39	the	the	DET
fcis-6909	60	40	item	item	NOUN
fcis-6909	60	41	belongs	belong	VERB
fcis-6909	60	42	.	.	PUNCT
fcis-6909	61	1	specifically	specifically	ADV
fcis-6909	61	2	,	,	PUNCT
fcis-6909	61	3	an	an	DET
fcis-6909	61	4	item	item	NOUN
fcis-6909	61	5	may	may	AUX
fcis-6909	61	6	belong	belong	VERB
fcis-6909	61	7	to	to	ADP
fcis-6909	61	8	many	many	ADJ
fcis-6909	61	9	categories	category	NOUN
fcis-6909	61	10	,	,	PUNCT
fcis-6909	61	11	and	and	CCONJ
fcis-6909	61	12	the	the	DET
fcis-6909	61	13	setting	setting	NOUN
fcis-6909	61	14	of	of	ADP
fcis-6909	61	15	each	each	DET
fcis-6909	61	16	item	item	NOUN
fcis-6909	61	17	feature	feature	NOUN
fcis-6909	61	18	value	value	NOUN
fcis-6909	61	19	affects	affect	VERB
fcis-6909	61	20	the	the	DET
fcis-6909	61	21	prediction	prediction	NOUN
fcis-6909	61	22	score	score	NOUN
fcis-6909	61	23	.	.	PUNCT
fcis-6909	62	1	however	however	ADV
fcis-6909	62	2	,	,	PUNCT
fcis-6909	62	3	for	for	ADP
fcis-6909	62	4	information	information	NOUN
fcis-6909	62	5	statistics	statistic	NOUN
fcis-6909	62	6	,	,	PUNCT
fcis-6909	62	7	the	the	DET
fcis-6909	62	8	more	more	ADV
fcis-6909	62	9	important	important	ADJ
fcis-6909	62	10	categories	category	NOUN
fcis-6909	62	11	are	be	AUX
fcis-6909	62	12	given	give	VERB
fcis-6909	62	13	priority	priority	NOUN
fcis-6909	62	14	in	in	ADP
fcis-6909	62	15	the	the	DET
fcis-6909	62	16	statistics	statistic	NOUN
fcis-6909	62	17	.	.	PUNCT
fcis-6909	63	1	take	take	VERB
fcis-6909	63	2	toy	toy	NOUN
fcis-6909	63	3	story	story	NOUN
fcis-6909	63	4	in	in	ADP
fcis-6909	63	5	movielens	movielen	NOUN
fcis-6909	63	6	movie	movie	NOUN
fcis-6909	63	7	dataset	dataset	VERB
fcis-6909	63	8	as	as	ADP
fcis-6909	63	9	an	an	DET
fcis-6909	63	10	example	example	NOUN
fcis-6909	63	11	,	,	PUNCT
fcis-6909	63	12	it	it	PRON
fcis-6909	63	13	is	be	AUX
fcis-6909	63	14	firstly	firstly	ADV
fcis-6909	63	15	an	an	DET
fcis-6909	63	16	animated	animate	VERB
fcis-6909	63	17	movie	movie	NOUN
fcis-6909	63	18	,	,	PUNCT
fcis-6909	63	19	then	then	ADV
fcis-6909	63	20	a	a	DET
fcis-6909	63	21	children	child	NOUN
fcis-6909	63	22	's	's	PART
fcis-6909	63	23	movie	movie	NOUN
fcis-6909	63	24	,	,	PUNCT
fcis-6909	63	25	and	and	CCONJ
fcis-6909	63	26	finally	finally	ADV
fcis-6909	63	27	a	a	DET
fcis-6909	63	28	comedy	comedy	NOUN
fcis-6909	63	29	movie	movie	NOUN
fcis-6909	63	30	.	.	PUNCT
fcis-6909	64	1	after	after	ADP
fcis-6909	64	2	comprehensive	comprehensive	ADJ
fcis-6909	64	3	consideration	consideration	NOUN
fcis-6909	64	4	,	,	PUNCT
fcis-6909	64	5	the	the	DET
fcis-6909	64	6	probability	probability	NOUN
fcis-6909	64	7	feature	feature	NOUN
fcis-6909	64	8	that	that	SCONJ
fcis-6909	64	9	toy	toy	NOUN
fcis-6909	64	10	story	story	NOUN
fcis-6909	64	11	belongs	belong	VERB
fcis-6909	64	12	to	to	ADP
fcis-6909	64	13	the	the	DET
fcis-6909	64	14	animation	animation	NOUN
fcis-6909	64	15	category	category	NOUN
fcis-6909	64	16	is	be	AUX
fcis-6909	64	17	set	set	VERB
fcis-6909	64	18	to	to	ADP
fcis-6909	64	19	the	the	DET
fcis-6909	64	20	maximum	maximum	ADJ
fcis-6909	64	21	,	,	PUNCT
fcis-6909	64	22	while	while	SCONJ
fcis-6909	64	23	the	the	DET
fcis-6909	64	24	probability	probability	NOUN
fcis-6909	64	25	of	of	ADP
fcis-6909	64	26	the	the	DET
fcis-6909	64	27	children	child	NOUN
fcis-6909	64	28	's	's	PART
fcis-6909	64	29	category	category	NOUN
fcis-6909	64	30	and	and	CCONJ
fcis-6909	64	31	the	the	DET
fcis-6909	64	32	probability	probability	NOUN
fcis-6909	64	33	of	of	ADP
fcis-6909	64	34	the	the	DET
fcis-6909	64	35	comedy	comedy	NOUN
fcis-6909	64	36	category	category	NOUN
fcis-6909	64	37	are	be	AUX
fcis-6909	64	38	set	set	VERB
fcis-6909	64	39	to	to	PART
fcis-6909	64	40	decrease	decrease	VERB
fcis-6909	64	41	in	in	ADP
fcis-6909	64	42	that	that	DET
fcis-6909	64	43	order	order	NOUN
fcis-6909	64	44	.	.	PUNCT
fcis-6909	65	1	3.3	3.3	NUM
fcis-6909	65	2	.	.	PUNCT
fcis-6909	66	1	bdcr	bdcr	NOUN
fcis-6909	66	2	layer	layer	NOUN
fcis-6909	66	3	3.3.1	3.3.1	X
fcis-6909	66	4	.	.	PUNCT
fcis-6909	67	1	removal	removal	NOUN
fcis-6909	67	2	of	of	ADP
fcis-6909	67	3	false	false	ADJ
fcis-6909	67	4	positives	positive	NOUN
fcis-6909	67	5	following	follow	VERB
fcis-6909	67	6	the	the	DET
fcis-6909	67	7	literature	literature	NOUN
fcis-6909	67	8	[	[	X
fcis-6909	67	9	16	16	NUM
fcis-6909	67	10	]	]	PUNCT
fcis-6909	67	11	,	,	PUNCT
fcis-6909	67	12	the	the	DET
fcis-6909	67	13	training	training	NOUN
fcis-6909	67	14	loss	loss	NOUN
fcis-6909	67	15	values	value	NOUN
fcis-6909	67	16	for	for	ADP
fcis-6909	67	17	false	false	ADJ
fcis-6909	67	18	positives	positive	NOUN
fcis-6909	67	19	are	be	AUX
fcis-6909	67	20	larger	large	ADJ
fcis-6909	67	21	,	,	PUNCT
fcis-6909	67	22	and	and	CCONJ
fcis-6909	67	23	the	the	DET
fcis-6909	67	24	loss	loss	NOUN
fcis-6909	67	25	values	value	NOUN
fcis-6909	67	26	are	be	AUX
fcis-6909	67	27	used	use	VERB
fcis-6909	67	28	to	to	PART
fcis-6909	67	29	distinguish	distinguish	VERB
fcis-6909	67	30	between	between	ADP
fcis-6909	67	31	true	true	ADJ
fcis-6909	67	32	and	and	CCONJ
fcis-6909	67	33	false	false	ADJ
fcis-6909	67	34	positive	positive	ADJ
fcis-6909	67	35	interactions	interaction	NOUN
fcis-6909	67	36	in	in	ADP
fcis-6909	67	37	the	the	DET
fcis-6909	67	38	data	datum	NOUN
fcis-6909	67	39	by	by	ADP
fcis-6909	67	40	setting	set	VERB
fcis-6909	67	41	the	the	DET
fcis-6909	67	42	loss	loss	NOUN
fcis-6909	67	43	values	value	NOUN
fcis-6909	67	44	as	as	SCONJ
fcis-6909	67	45	follows	follow	VERB
fcis-6909	67	46	.	.	PUNCT
fcis-6909	68	1	(	(	PUNCT
fcis-6909	68	2	)	)	PUNCT
fcis-6909	68	3	(	(	PUNCT
fcis-6909	68	4	)	)	PUNCT
fcis-6909	68	5	(	(	PUNCT
fcis-6909	68	6	)	)	PUNCT
fcis-6909	68	7	0	0	NUM
fcis-6909	68	8	,	,	PUNCT
fcis-6909	68	9	,	,	PUNCT
fcis-6909	68	10	1	1	NUM
fcis-6909	68	11	,	,	PUNCT
fcis-6909	68	12	,	,	PUNCT
fcis-6909	68	13	,	,	PUNCT
fcis-6909	68	14	otherwise	otherwise	ADV
fcis-6909	68	15	ui	ui	PROPN
fcis-6909	68	16	de	de	PROPN
fcis-6909	68	17	l	l	X
fcis-6909	68	18	u	u	NOUN
fcis-6909	69	1	i	i	NOUN
fcis-6909	69	2	y	y	VERB
fcis-6909	69	3	l	l	NOUN
fcis-6909	69	4	u	u	X
fcis-6909	70	1	i	i	PRON
fcis-6909	70	2	lce	lce	VERB
fcis-6909	70	3	u	u	NOUN
fcis-6909	70	4	i	i	PRON
fcis-6909	70	5			VERB
fcis-6909	70	6			PROPN
fcis-6909	71	1	=	=	NUM
fcis-6909	71	2			NOUN
fcis-6909	71	3	=	=	PUNCT
fcis-6909	71	4			NUM
fcis-6909	71	5			NOUN
fcis-6909	71	6	(	(	PUNCT
fcis-6909	71	7	2	2	NUM
fcis-6909	71	8	)	)	PUNCT
fcis-6909	71	9	where	where	SCONJ
fcis-6909	71	10	τ	τ	PROPN
fcis-6909	71	11	is	be	AUX
fcis-6909	71	12	the	the	DET
fcis-6909	71	13	threshold	threshold	NOUN
fcis-6909	71	14	value	value	NOUN
fcis-6909	71	15	,	,	PUNCT
fcis-6909	71	16	and	and	CCONJ
fcis-6909	71	17	since	since	SCONJ
fcis-6909	71	18	the	the	DET
fcis-6909	71	19	loss	loss	NOUN
fcis-6909	71	20	value	value	NOUN
fcis-6909	71	21	must	must	AUX
fcis-6909	71	22	decrease	decrease	VERB
fcis-6909	71	23	with	with	ADP
fcis-6909	71	24	iteration	iteration	NOUN
fcis-6909	71	25	,	,	PUNCT
fcis-6909	71	26	it	it	PRON
fcis-6909	71	27	is	be	AUX
fcis-6909	71	28	set	set	VERB
fcis-6909	71	29	to	to	ADP
fcis-6909	71	30	a	a	DET
fcis-6909	71	31	dynamic	dynamic	ADJ
fcis-6909	71	32	,	,	PUNCT
fcis-6909	71	33	varying	vary	VERB
fcis-6909	71	34	threshold	threshold	NOUN
fcis-6909	71	35	value	value	NOUN
fcis-6909	71	36	with	with	ADP
fcis-6909	71	37	the	the	DET
fcis-6909	71	38	number	number	NOUN
fcis-6909	71	39	of	of	ADP
fcis-6909	71	40	iterations	iteration	NOUN
fcis-6909	71	41	(	(	PUNCT
fcis-6909	71	42	)	)	PUNCT
fcis-6909	71	43	t	t	NOUN
fcis-6909	71	44	.	.	PUNCT
fcis-6909	72	1	since	since	SCONJ
fcis-6909	72	2	the	the	DET
fcis-6909	72	3	loss	loss	NOUN
fcis-6909	72	4	value	value	NOUN
fcis-6909	72	5	varies	vary	VERB
fcis-6909	72	6	across	across	ADP
fcis-6909	72	7	data	datum	NOUN
fcis-6909	72	8	sets	set	NOUN
fcis-6909	72	9	,	,	PUNCT
fcis-6909	72	10	it	it	PRON
fcis-6909	72	11	is	be	AUX
fcis-6909	72	12	also	also	ADV
fcis-6909	72	13	necessary	necessary	ADJ
fcis-6909	72	14	to	to	PART
fcis-6909	72	15	set	set	VERB
fcis-6909	72	16	(	(	PUNCT
fcis-6909	72	17	)	)	PUNCT
fcis-6909	72	18	t	t	NOUN
fcis-6909	72	19	as	as	ADP
fcis-6909	72	20	a	a	DET
fcis-6909	72	21	function	function	NOUN
fcis-6909	72	22	of	of	ADP
fcis-6909	72	23	the	the	DET
fcis-6909	72	24	discard	discard	ADJ
fcis-6909	72	25	rate	rate	NOUN
fcis-6909	72	26			PROPN
fcis-6909	72	27	:	:	PUNCT
fcis-6909	72	28	(	(	PUNCT
fcis-6909	72	29	)	)	PUNCT
fcis-6909	72	30	(	(	PUNCT
fcis-6909	72	31	)	)	PUNCT
fcis-6909	72	32	maxmin	maxmin	NOUN
fcis-6909	72	33	,	,	PUNCT
fcis-6909	72	34	t	t	PROPN
fcis-6909	72	35	t	t	NOUN
fcis-6909	72	36	=	=	PROPN
fcis-6909	72	37	(	(	PUNCT
fcis-6909	72	38	3	3	NUM
fcis-6909	72	39	)	)	PUNCT
fcis-6909	72	40	where	where	SCONJ
fcis-6909	72	41	𝜀𝑚𝑎𝑥	𝜀𝑚𝑎𝑥	NOUN
fcis-6909	72	42	is	be	AUX
fcis-6909	72	43	the	the	DET
fcis-6909	72	44	upper	upper	ADJ
fcis-6909	72	45	discard	discard	NOUN
fcis-6909	72	46	rate	rate	NOUN
fcis-6909	72	47	limit	limit	NOUN
fcis-6909	72	48	,	,	PUNCT
fcis-6909	72	49	and	and	CCONJ
fcis-6909	72	50			NOUN
fcis-6909	72	51	is	be	AUX
fcis-6909	72	52	a	a	DET
fcis-6909	72	53	hyperparameter	hyperparameter	NOUN
fcis-6909	72	54	to	to	PART
fcis-6909	72	55	adjust	adjust	VERB
fcis-6909	72	56	the	the	DET
fcis-6909	72	57	speed	speed	NOUN
fcis-6909	72	58	to	to	PART
fcis-6909	72	59	reach	reach	VERB
fcis-6909	72	60	the	the	DET
fcis-6909	72	61	maximum	maximum	ADJ
fcis-6909	72	62	discard	discard	ADJ
fcis-6909	72	63	rate	rate	NOUN
fcis-6909	72	64	.	.	PUNCT
fcis-6909	73	1	3.3.2	3.3.2	X
fcis-6909	73	2	.	.	X
fcis-6909	73	3	causal	causal	ADJ
fcis-6909	73	4	inference	inference	NOUN
fcis-6909	73	5	with	with	ADP
fcis-6909	73	6	distribution	distribution	NOUN
fcis-6909	73	7	bias	bias	NOUN
fcis-6909	73	8	considering	consider	VERB
fcis-6909	73	9	the	the	DET
fcis-6909	73	10	causal	causal	ADJ
fcis-6909	73	11	graph	graph	NOUN
fcis-6909	73	12	shown	show	VERB
fcis-6909	73	13	in	in	ADP
fcis-6909	73	14	figure	figure	NOUN
fcis-6909	73	15	1	1	NUM
fcis-6909	73	16	,	,	PUNCT
fcis-6909	73	17	there	there	PRON
fcis-6909	73	18	are	be	VERB
fcis-6909	73	19	two	two	NUM
fcis-6909	73	20	paths	path	NOUN
fcis-6909	73	21	for	for	ADP
fcis-6909	73	22	user	user	NOUN
fcis-6909	73	23	u	u	NOUN
fcis-6909	73	24	to	to	PART
fcis-6909	73	25	influence	influence	VERB
fcis-6909	73	26	the	the	DET
fcis-6909	73	27	matching	matching	NOUN
fcis-6909	73	28	score	score	NOUN
fcis-6909	73	29	:	:	PUNCT
fcis-6909	73	30	the	the	DET
fcis-6909	73	31	first	first	ADJ
fcis-6909	73	32	path	path	NOUN
fcis-6909	73	33	is	be	AUX
fcis-6909	73	34	u→y	u→y	PROPN
fcis-6909	73	35	,	,	PUNCT
fcis-6909	73	36	which	which	PRON
fcis-6909	73	37	simply	simply	ADV
fcis-6909	73	38	indicates	indicate	VERB
fcis-6909	73	39	u	u	NOUN
fcis-6909	73	40	's	's	PART
fcis-6909	73	41	true	true	ADJ
fcis-6909	73	42	preference	preference	NOUN
fcis-6909	73	43	for	for	ADP
fcis-6909	73	44	the	the	DET
fcis-6909	73	45	item	item	NOUN
fcis-6909	73	46	;	;	PUNCT
fcis-6909	73	47	the	the	DET
fcis-6909	73	48	second	second	ADJ
fcis-6909	73	49	path	path	NOUN
fcis-6909	73	50	is	be	AUX
fcis-6909	73	51	u→m→y	u→m→y	NOUN
fcis-6909	73	52	,	,	PUNCT
fcis-6909	73	53	which	which	PRON
fcis-6909	73	54	indicates	indicate	VERB
fcis-6909	73	55	that	that	SCONJ
fcis-6909	73	56	the	the	DET
fcis-6909	73	57	user	user	NOUN
fcis-6909	73	58	's	's	PART
fcis-6909	73	59	preference	preference	NOUN
fcis-6909	73	60	for	for	ADP
fcis-6909	73	61	the	the	DET
fcis-6909	73	62	category	category	NOUN
fcis-6909	73	63	influences	influence	VERB
fcis-6909	73	64	the	the	DET
fcis-6909	73	65	predicted	predict	VERB
fcis-6909	73	66	score	score	NOUN
fcis-6909	73	67	y.	y.	PROPN
fcis-6909	73	68	since	since	SCONJ
fcis-6909	73	69	there	there	PRON
fcis-6909	73	70	is	be	VERB
fcis-6909	73	71	a	a	DET
fcis-6909	73	72	backdoor	backdoor	NOUN
fcis-6909	73	73	path	path	NOUN
fcis-6909	73	74	u←d→m→y	u←d→m→y	INTJ
fcis-6909	73	75	in	in	ADP
fcis-6909	73	76	the	the	DET
fcis-6909	73	77	graph	graph	NOUN
fcis-6909	73	78	,	,	PUNCT
fcis-6909	73	79	it	it	PRON
fcis-6909	73	80	can	can	AUX
fcis-6909	73	81	be	be	AUX
fcis-6909	73	82	learned	learn	VERB
fcis-6909	73	83	that	that	SCONJ
fcis-6909	73	84	both	both	CCONJ
fcis-6909	73	85	d	d	NOUN
fcis-6909	73	86	and	and	CCONJ
fcis-6909	73	87	m	m	PROPN
fcis-6909	73	88	are	be	AUX
fcis-6909	73	89	confounders	confounder	NOUN
fcis-6909	73	90	,	,	PUNCT
fcis-6909	73	91	leading	lead	VERB
fcis-6909	73	92	to	to	ADP
fcis-6909	73	93	amplification	amplification	NOUN
fcis-6909	73	94	of	of	ADP
fcis-6909	73	95	the	the	DET
fcis-6909	73	96	spurious	spurious	ADJ
fcis-6909	73	97	bias	bias	NOUN
fcis-6909	73	98	and	and	CCONJ
fcis-6909	73	99	requiring	require	VERB
fcis-6909	73	100	intervention	intervention	NOUN
fcis-6909	73	101	to	to	PART
fcis-6909	73	102	remove	remove	VERB
fcis-6909	73	103	the	the	DET
fcis-6909	73	104	confounding	confounding	NOUN
fcis-6909	73	105	.	.	PUNCT
fcis-6909	74	1	according	accord	VERB
fcis-6909	74	2	to	to	ADP
fcis-6909	74	3	the	the	DET
fcis-6909	74	4	literature	literature	NOUN
fcis-6909	74	5	[	[	X
fcis-6909	74	6	17	17	NUM
fcis-6909	74	7	]	]	PUNCT
fcis-6909	74	8	,	,	PUNCT
fcis-6909	74	9	the	the	DET
fcis-6909	74	10	prediction	prediction	NOUN
fcis-6909	74	11	probability	probability	NOUN
fcis-6909	74	12	p(y|u	p(y|u	NOUN
fcis-6909	74	13	,	,	PUNCT
fcis-6909	74	14	i	i	PRON
fcis-6909	74	15	)	)	PUNCT
fcis-6909	74	16	is	be	AUX
fcis-6909	74	17	as	as	SCONJ
fcis-6909	74	18	follows	follow	VERB
fcis-6909	74	19	.	.	PUNCT
fcis-6909	75	1	(	(	PUNCT
fcis-6909	75	2	)	)	PUNCT
fcis-6909	75	3	(	(	PUNCT
fcis-6909	75	4	,	,	PUNCT
fcis-6909	75	5	,	,	PUNCT
fcis-6909	75	6	(	(	PUNCT
fcis-6909	75	7	,	,	PUNCT
fcis-6909	75	8	)	)	PUNCT
fcis-6909	75	9	)	)	PUNCT
fcis-6909	75	10	(	(	PUNCT
fcis-6909	75	11	,	,	PUNCT
fcis-6909	75	12	,	,	PUNCT
fcis-6909	75	13	(	(	PUNCT
fcis-6909	75	14	(	(	PUNCT
fcis-6909	75	15	)	)	PUNCT
fcis-6909	75	16	,	,	PUNCT
fcis-6909	75	17	)	)	PUNCT
fcis-6909	75	18	)	)	PUNCT
fcis-6909	76	1	(	(	PUNCT
fcis-6909	76	2	|	|	ADV
fcis-6909	76	3	(	(	PUNCT
fcis-6909	76	4	)	)	PUNCT
fcis-6909	76	5	,	,	PUNCT
fcis-6909	76	6	)	)	PUNCT
fcis-6909	77	1	d	d	PUNCT
fcis-6909	77	2	d	d	PUNCT
fcis-6909	77	3	d	d	X
fcis-6909	77	4	d	d	X
fcis-6909	77	5	p	p	X
fcis-6909	77	6	y	y	PROPN
fcis-6909	77	7	do	do	AUX
fcis-6909	77	8	p	p	PROPN
fcis-6909	78	1	d	d	X
fcis-6909	78	2	f	f	X
fcis-6909	78	3	u	u	NOUN
fcis-6909	78	4	i	i	NOUN
fcis-6909	78	5	m	m	VERB
fcis-6909	78	6	d	d	NOUN
fcis-6909	78	7	u	u	NOUN
fcis-6909	78	8	f	f	PROPN
fcis-6909	78	9	u	u	NOUN
fcis-6909	78	10	i	i	NOUN
fcis-6909	78	11	m	m	VERB
fcis-6909	78	12	u	u	NOUN
fcis-6909	78	13	u	u	NOUN
fcis-6909	78	14	i	i	PRON
fcis-6909	78	15	i	i	PRON
fcis-6909	78	16	p	p	X
fcis-6909	78	17	d	d	X
fcis-6909	78	18	d	d	X
fcis-6909	78	19	u	u	PROPN
fcis-6909	78	20			NOUN
fcis-6909	78	21			NOUN
fcis-6909	78	22			X
fcis-6909	78	23	=	=	SYM
fcis-6909	78	24	=	=	SYM
fcis-6909	78	25	=	=	SYM
fcis-6909	78	26			X
fcis-6909	78	27			X
fcis-6909	78	28	(	(	PUNCT
fcis-6909	78	29	4	4	NUM
fcis-6909	78	30	)	)	PUNCT
fcis-6909	78	31	where	where	SCONJ
fcis-6909	78	32	,	,	PUNCT
fcis-6909	78	33	(	(	PUNCT
fcis-6909	78	34	.)f	.)f	PROPN
fcis-6909	78	35	is	be	AUX
fcis-6909	78	36	an	an	DET
fcis-6909	78	37	arbitrary	arbitrary	ADJ
fcis-6909	78	38	recommended	recommend	VERB
fcis-6909	78	39	model	model	NOUN
fcis-6909	78	40	,	,	PUNCT
fcis-6909	78	41	m	m	VERB
fcis-6909	78	42	is	be	AUX
fcis-6909	78	43	used	use	VERB
fcis-6909	78	44	as	as	ADP
fcis-6909	78	45	an	an	DET
fcis-6909	78	46	additional	additional	ADJ
fcis-6909	78	47	input	input	NOUN
fcis-6909	78	48	to	to	ADP
fcis-6909	78	49	the	the	DET
fcis-6909	78	50	model	model	NOUN
fcis-6909	78	51	,	,	PUNCT
fcis-6909	78	52	and	and	CCONJ
fcis-6909	78	53	m	m	PROPN
fcis-6909	78	54	is	be	AUX
fcis-6909	78	55	calculated	calculate	VERB
fcis-6909	78	56	as	as	SCONJ
fcis-6909	78	57	follows	follow	VERB
fcis-6909	78	58	.	.	PUNCT
fcis-6909	79	1	(	(	PUNCT
fcis-6909	79	2	)	)	PUNCT
fcis-6909	79	3	,	,	PUNCT
fcis-6909	79	4	1	1	NUM
fcis-6909	79	5	1	1	NUM
fcis-6909	79	6	(	(	PUNCT
fcis-6909	79	7	,	,	PUNCT
fcis-6909	79	8	)	)	PUNCT
fcis-6909	79	9	n	n	CCONJ
fcis-6909	80	1	k	k	PROPN
fcis-6909	80	2	a	a	DET
fcis-6909	80	3	a	a	DET
fcis-6909	80	4	u	u	NOUN
fcis-6909	80	5	b	b	PROPN
fcis-6909	80	6	b	b	PROPN
fcis-6909	80	7	a	a	DET
fcis-6909	80	8	b	b	NOUN
fcis-6909	80	9	m	m	NOUN
fcis-6909	80	10	p	p	NOUN
fcis-6909	80	11	g	g	ADJ
fcis-6909	80	12	v	v	NOUN
fcis-6909	80	13	x	x	PUNCT
fcis-6909	80	14	=	=	PUNCT
fcis-6909	80	15	=	=	PUNCT
fcis-6909	80	16	=	=	X
fcis-6909	80	17	d	d	SYM
fcis-6909	80	18	u	u	NOUN
fcis-6909	80	19	u	u	NOUN
fcis-6909	80	20	(	(	PUNCT
fcis-6909	80	21	5	5	NUM
fcis-6909	80	22	)	)	PUNCT
fcis-6909	80	23	the	the	DET
fcis-6909	80	24	probability	probability	NOUN
fcis-6909	80	25	values	value	NOUN
fcis-6909	80	26	of	of	ADP
fcis-6909	80	27	the	the	DET
fcis-6909	80	28	n	n	PRON
fcis-6909	80	29	categories	category	NOUN
fcis-6909	80	30	are	be	AUX
fcis-6909	80	31	multiplied	multiply	VERB
fcis-6909	80	32	by	by	ADP
fcis-6909	80	33	the	the	DET
fcis-6909	80	34	category	category	NOUN
fcis-6909	80	35	representations	representation	NOUN
fcis-6909	80	36	and	and	CCONJ
fcis-6909	80	37	then	then	ADV
fcis-6909	80	38	multiplied	multiply	VERB
fcis-6909	80	39	with	with	ADP
fcis-6909	80	40	the	the	DET
fcis-6909	80	41	user	user	NOUN
fcis-6909	80	42	's	's	PART
fcis-6909	80	43	feature	feature	NOUN
fcis-6909	80	44	values	value	NOUN
fcis-6909	80	45	and	and	CCONJ
fcis-6909	80	46	feature	feature	NOUN
fcis-6909	80	47	representations	representation	NOUN
fcis-6909	80	48	as	as	ADP
fcis-6909	80	49	dot	dot	NOUN
fcis-6909	80	50	12	12	NUM
fcis-6909	80	51	products	product	NOUN
fcis-6909	80	52	.	.	PUNCT
fcis-6909	81	1	3.3.3	3.3.3	NUM
fcis-6909	81	2	.	.	PUNCT
fcis-6909	81	3	causal	causal	ADJ
fcis-6909	81	4	inference	inference	NOUN
fcis-6909	81	5	fusion	fusion	NOUN
fcis-6909	81	6	denoising	denoising	NOUN
fcis-6909	81	7	using	use	VERB
fcis-6909	81	8	an	an	DET
fcis-6909	81	9	implicit	implicit	ADJ
fcis-6909	81	10	feedback	feedback	NOUN
fcis-6909	81	11	denoising	denoising	NOUN
fcis-6909	81	12	method	method	NOUN
fcis-6909	81	13	to	to	PART
fcis-6909	81	14	obtain	obtain	VERB
fcis-6909	81	15	noisy	noisy	ADJ
fcis-6909	81	16	data	datum	NOUN
fcis-6909	81	17	in	in	ADP
fcis-6909	81	18	the	the	DET
fcis-6909	81	19	dataset	dataset	NOUN
fcis-6909	81	20	,	,	PUNCT
fcis-6909	81	21	interactions	interaction	NOUN
fcis-6909	81	22	with	with	ADP
fcis-6909	81	23	large	large	ADJ
fcis-6909	81	24	loss	loss	NOUN
fcis-6909	81	25	values	value	NOUN
fcis-6909	81	26	are	be	AUX
fcis-6909	81	27	dynamically	dynamically	ADV
fcis-6909	81	28	discarded	discard	VERB
fcis-6909	81	29	,	,	PUNCT
fcis-6909	81	30	allowing	allow	VERB
fcis-6909	81	31	the	the	DET
fcis-6909	81	32	dataset	dataset	NOUN
fcis-6909	81	33	to	to	PART
fcis-6909	81	34	eliminate	eliminate	VERB
fcis-6909	81	35	false	false	ADJ
fcis-6909	81	36	positive	positive	ADJ
fcis-6909	81	37	samples	sample	NOUN
fcis-6909	81	38	.	.	PUNCT
fcis-6909	82	1	the	the	DET
fcis-6909	82	2	data	datum	NOUN
fcis-6909	82	3	are	be	AUX
fcis-6909	82	4	then	then	ADV
fcis-6909	82	5	fed	feed	VERB
fcis-6909	82	6	into	into	ADP
fcis-6909	82	7	the	the	DET
fcis-6909	82	8	causal	causal	ADJ
fcis-6909	82	9	debiasing	debiase	VERB
fcis-6909	82	10	model	model	NOUN
fcis-6909	82	11	so	so	SCONJ
fcis-6909	82	12	that	that	SCONJ
fcis-6909	82	13	the	the	DET
fcis-6909	82	14	data	datum	NOUN
fcis-6909	82	15	are	be	AUX
fcis-6909	82	16	no	no	ADV
fcis-6909	82	17	longer	long	ADV
fcis-6909	82	18	affected	affect	VERB
fcis-6909	82	19	by	by	ADP
fcis-6909	82	20	item	item	NOUN
fcis-6909	82	21	category	category	NOUN
fcis-6909	82	22	bias	bias	NOUN
fcis-6909	82	23	.	.	PUNCT
fcis-6909	83	1	first	first	ADV
fcis-6909	83	2	,	,	PUNCT
fcis-6909	83	3	the	the	DET
fcis-6909	83	4	discard	discard	ADJ
fcis-6909	83	5	rate	rate	NOUN
fcis-6909	83	6	that	that	PRON
fcis-6909	83	7	increases	increase	VERB
fcis-6909	83	8	linearly	linearly	ADV
fcis-6909	83	9	with	with	ADP
fcis-6909	83	10	the	the	DET
fcis-6909	83	11	number	number	NOUN
fcis-6909	83	12	of	of	ADP
fcis-6909	83	13	iterations	iteration	NOUN
fcis-6909	83	14	is	be	AUX
fcis-6909	83	15	set	set	VERB
fcis-6909	83	16	.	.	PUNCT
fcis-6909	84	1	after	after	SCONJ
fcis-6909	84	2	the	the	DET
fcis-6909	84	3	causal	causal	ADJ
fcis-6909	84	4	model	model	NOUN
fcis-6909	84	5	obtains	obtain	VERB
fcis-6909	84	6	the	the	DET
fcis-6909	84	7	prediction	prediction	NOUN
fcis-6909	84	8	scores	score	NOUN
fcis-6909	84	9	based	base	VERB
fcis-6909	84	10	on	on	ADP
fcis-6909	84	11	the	the	DET
fcis-6909	84	12	features	feature	NOUN
fcis-6909	84	13	and	and	CCONJ
fcis-6909	84	14	eigenvalues	eigenvalue	NOUN
fcis-6909	84	15	,	,	PUNCT
fcis-6909	84	16	the	the	DET
fcis-6909	84	17	prediction	prediction	NOUN
fcis-6909	84	18	scores	score	NOUN
fcis-6909	84	19	and	and	CCONJ
fcis-6909	84	20	the	the	DET
fcis-6909	84	21	true	true	ADJ
fcis-6909	84	22	labels	label	NOUN
fcis-6909	84	23	are	be	AUX
fcis-6909	84	24	entered	enter	VERB
fcis-6909	84	25	into	into	ADP
fcis-6909	84	26	the	the	DET
fcis-6909	84	27	following	follow	VERB
fcis-6909	84	28	equation	equation	NOUN
fcis-6909	84	29	.	.	PUNCT
fcis-6909	85	1	1	1	NUM
fcis-6909	85	2	ˆlog	ˆlog	NOUN
fcis-6909	85	3	n	n	X
fcis-6909	85	4	ce	ce	NOUN
fcis-6909	86	1	i	i	PRON
fcis-6909	86	2	i	i	INTJ
fcis-6909	87	1	i	i	VERB
fcis-6909	87	2	l	l	VERB
fcis-6909	88	1	y	y	NOUN
fcis-6909	88	2	y	y	PROPN
fcis-6909	88	3	=	=	PUNCT
fcis-6909	88	4	=	=	X
fcis-6909	88	5	−	−	NOUN
fcis-6909	88	6	(	(	PUNCT
fcis-6909	88	7	6	6	NUM
fcis-6909	88	8	)	)	PUNCT
fcis-6909	88	9	next	next	ADV
fcis-6909	88	10	,	,	PUNCT
fcis-6909	88	11	the	the	DET
fcis-6909	88	12	lce	lce	PROPN
fcis-6909	88	13	is	be	AUX
fcis-6909	88	14	sorted	sort	VERB
fcis-6909	88	15	by	by	ADP
fcis-6909	88	16	the	the	DET
fcis-6909	88	17	principle	principle	NOUN
fcis-6909	88	18	of	of	ADP
fcis-6909	88	19	smallest	small	ADJ
fcis-6909	88	20	to	to	PART
fcis-6909	88	21	largest	large	ADJ
fcis-6909	88	22	to	to	PART
fcis-6909	88	23	get	get	VERB
fcis-6909	88	24	the	the	DET
fcis-6909	88	25	corresponding	corresponding	ADJ
fcis-6909	88	26	index	index	NOUN
fcis-6909	88	27	.	.	PUNCT
fcis-6909	89	1	the	the	DET
fcis-6909	89	2	retention	retention	NOUN
fcis-6909	89	3	rate	rate	NOUN
fcis-6909	89	4	is	be	AUX
fcis-6909	89	5	obtained	obtain	VERB
fcis-6909	89	6	from	from	ADP
fcis-6909	89	7	the	the	DET
fcis-6909	89	8	discard	discard	ADJ
fcis-6909	89	9	rate	rate	NOUN
fcis-6909	89	10	,	,	PUNCT
fcis-6909	89	11	and	and	CCONJ
fcis-6909	89	12	the	the	DET
fcis-6909	89	13	index	index	NOUN
fcis-6909	89	14	corresponding	correspond	VERB
fcis-6909	89	15	to	to	ADP
fcis-6909	89	16	the	the	DET
fcis-6909	89	17	first	first	ADJ
fcis-6909	89	18	n	n	PRON
fcis-6909	89	19	loss	loss	NOUN
fcis-6909	89	20	values	value	NOUN
fcis-6909	89	21	is	be	AUX
fcis-6909	89	22	retained	retain	VERB
fcis-6909	89	23	.	.	PUNCT
fcis-6909	90	1	finally	finally	ADV
fcis-6909	90	2	,	,	PUNCT
fcis-6909	90	3	the	the	DET
fcis-6909	90	4	predicted	predict	VERB
fcis-6909	90	5	values	value	NOUN
fcis-6909	90	6	and	and	CCONJ
fcis-6909	90	7	labels	label	NOUN
fcis-6909	90	8	corresponding	correspond	VERB
fcis-6909	90	9	to	to	ADP
fcis-6909	90	10	the	the	DET
fcis-6909	90	11	retained	retain	VERB
fcis-6909	90	12	indexes	index	NOUN
fcis-6909	90	13	are	be	AUX
fcis-6909	90	14	taken	take	VERB
fcis-6909	90	15	out	out	ADP
fcis-6909	90	16	,	,	PUNCT
fcis-6909	90	17	and	and	CCONJ
fcis-6909	90	18	the	the	DET
fcis-6909	90	19	false	false	ADJ
fcis-6909	90	20	positive	positive	ADJ
fcis-6909	90	21	interactions	interaction	NOUN
fcis-6909	90	22	are	be	AUX
fcis-6909	90	23	considered	consider	VERB
fcis-6909	90	24	to	to	PART
fcis-6909	90	25	have	have	AUX
fcis-6909	90	26	been	be	AUX
fcis-6909	90	27	removed	remove	VERB
fcis-6909	90	28	,	,	PUNCT
fcis-6909	90	29	and	and	CCONJ
fcis-6909	90	30	the	the	DET
fcis-6909	90	31	cross	cross	NOUN
fcis-6909	90	32	first	first	ADJ
fcis-6909	90	33	loss	loss	NOUN
fcis-6909	90	34	is	be	AUX
fcis-6909	90	35	recalculated	recalculate	VERB
fcis-6909	90	36	by	by	ADP
fcis-6909	90	37	them	they	PRON
fcis-6909	90	38	,	,	PUNCT
fcis-6909	90	39	and	and	CCONJ
fcis-6909	90	40	the	the	DET
fcis-6909	90	41	recalculated	recalculated	PROPN
fcis-6909	90	42	cross	cross	PROPN
fcis-6909	90	43	first	first	ADJ
fcis-6909	90	44	loss	loss	NOUN
fcis-6909	90	45	is	be	AUX
fcis-6909	90	46	returned	return	VERB
fcis-6909	90	47	to	to	ADP
fcis-6909	90	48	the	the	DET
fcis-6909	90	49	causal	causal	ADJ
fcis-6909	90	50	model	model	NOUN
fcis-6909	90	51	for	for	ADP
fcis-6909	90	52	back	back	ADJ
fcis-6909	90	53	propagation	propagation	NOUN
fcis-6909	90	54	.	.	PUNCT
fcis-6909	91	1	all	all	PRON
fcis-6909	91	2	in	in	ADP
fcis-6909	91	3	all	all	PRON
fcis-6909	91	4	,	,	PUNCT
fcis-6909	91	5	the	the	DET
fcis-6909	91	6	false	false	ADJ
fcis-6909	91	7	positives	positive	NOUN
fcis-6909	91	8	are	be	AUX
fcis-6909	91	9	calculated	calculate	VERB
fcis-6909	91	10	as	as	SCONJ
fcis-6909	91	11	follows	follow	VERB
fcis-6909	91	12	.	.	PUNCT
fcis-6909	92	1	(	(	PUNCT
fcis-6909	92	2	)	)	PUNCT
fcis-6909	92	3	1	1	NUM
fcis-6909	92	4	,	,	PUNCT
fcis-6909	92	5	|	|	ADV
fcis-6909	92	6	|	|	ADV
fcis-6909	92	7	(	(	PUNCT
fcis-6909	92	8	)	)	PUNCT
fcis-6909	92	9	*	*	PUNCT
fcis-6909	92	10	(	(	PUNCT
fcis-6909	92	11	,	,	PUNCT
fcis-6909	92	12	)	)	PUNCT
fcis-6909	92	13	arg	arg	NOUN
fcis-6909	92	14	ma	ma	PROPN
fcis-6909	92	15	,	,	PUNCT
fcis-6909	92	16	x	x	PROPN
fcis-6909	92	17	pos	pos	NOUN
fcis-6909	92	18	t	t	PROPN
fcis-6909	92	19	ce	ce	PROPN
fcis-6909	93	1	t	t	PROPN
fcis-6909	93	2	t	t	PROPN
fcis-6909	93	3	u	u	X
fcis-6909	93	4	i	i	PRON
fcis-6909	93	5	u	u	X
fcis-6909	93	6	il	il	PROPN
fcis-6909	93	7	−	−	PROPN
fcis-6909	93	8			PROPN
fcis-6909	93	9	=	=	PROPN
fcis-6909	93	10			NOUN
fcis-6909	93	11			NOUN
fcis-6909	93	12			NOUN
fcis-6909	93	13			PROPN
fcis-6909	93	14	=	=	PUNCT
fcis-6909	94	1			PROPN
fcis-6909	94	2			PROPN
fcis-6909	94	3	(	(	PUNCT
fcis-6909	94	4	7	7	NUM
fcis-6909	94	5	)	)	PUNCT
fcis-6909	94	6	where	where	SCONJ
fcis-6909	94	7	d	d	NOUN
fcis-6909	94	8	is	be	AUX
fcis-6909	94	9	the	the	DET
fcis-6909	94	10	false	false	ADJ
fcis-6909	94	11	positive	positive	ADJ
fcis-6909	94	12	in	in	ADP
fcis-6909	94	13	the	the	DET
fcis-6909	94	14	positive	positive	ADJ
fcis-6909	94	15	dpos	dpos	NOUN
fcis-6909	94	16	and	and	CCONJ
fcis-6909	94	17	dt	dt	PROPN
fcis-6909	94	18	is	be	AUX
fcis-6909	94	19	the	the	DET
fcis-6909	94	20	training	training	NOUN
fcis-6909	94	21	data	datum	NOUN
fcis-6909	94	22	that	that	PRON
fcis-6909	94	23	contains	contain	VERB
fcis-6909	94	24	not	not	PART
fcis-6909	94	25	only	only	ADV
fcis-6909	94	26	dpos	dpos	NOUN
fcis-6909	94	27	but	but	CCONJ
fcis-6909	94	28	also	also	ADV
fcis-6909	94	29	the	the	DET
fcis-6909	94	30	negative	negative	ADJ
fcis-6909	94	31	sampling	sample	VERB
fcis-6909	94	32	data	datum	NOUN
fcis-6909	94	33	.	.	PUNCT
fcis-6909	95	1	after	after	SCONJ
fcis-6909	95	2	the	the	DET
fcis-6909	95	3	false	false	ADJ
fcis-6909	95	4	positives	positive	NOUN
fcis-6909	95	5	are	be	AUX
fcis-6909	95	6	removed	remove	VERB
fcis-6909	95	7	from	from	ADP
fcis-6909	95	8	the	the	DET
fcis-6909	95	9	training	training	NOUN
fcis-6909	95	10	data	datum	NOUN
fcis-6909	95	11	,	,	PUNCT
fcis-6909	95	12	denoted	denote	VERB
fcis-6909	95	13	as	as	ADP
fcis-6909	95	14	d	d	PROPN
fcis-6909	95	15	*	*	PROPN
fcis-6909	95	16	,	,	PUNCT
fcis-6909	95	17	the	the	DET
fcis-6909	95	18	parameters	parameter	NOUN
fcis-6909	95	19	are	be	AUX
fcis-6909	95	20	updated	update	VERB
fcis-6909	95	21	by	by	ADP
fcis-6909	95	22	the	the	DET
fcis-6909	95	23	following	follow	VERB
fcis-6909	95	24	equation	equation	NOUN
fcis-6909	95	25	.	.	PUNCT
fcis-6909	96	1	(	(	PUNCT
fcis-6909	96	2	)	)	PUNCT
fcis-6909	96	3	(	(	PUNCT
fcis-6909	96	4	)	)	PUNCT
fcis-6909	96	5	*	*	PUNCT
fcis-6909	96	6	,	,	PUNCT
fcis-6909	96	7	,	,	PUNCT
fcis-6909	96	8	,	,	PUNCT
fcis-6909	96	9	,	,	PUNCT
fcis-6909	96	10	arg	arg	NOUN
fcis-6909	96	11	min	min	PROPN
fcis-6909	96	12	(	(	PUNCT
fcis-6909	96	13	,	,	PUNCT
fcis-6909	96	14	,	,	PUNCT
fcis-6909	96	15	(	(	PUNCT
fcis-6909	96	16	,	,	PUNCT
fcis-6909	96	17	)	)	PUNCT
fcis-6909	96	18	)	)	PUNCT
fcis-6909	96	19	,	,	PUNCT
fcis-6909	96	20	u	u	NOUN
fcis-6909	96	21	i	i	NOUN
fcis-6909	96	22	u	u	VERB
fcis-6909	96	23	i	i	PRON
fcis-6909	96	24	u	u	VERB
fcis-6909	96	25	i	i	VERB
fcis-6909	96	26	y	y	PROPN
fcis-6909	96	27	l	l	NOUN
fcis-6909	96	28	f	f	X
fcis-6909	97	1	u	u	INTJ
fcis-6909	98	1	i	i	PRON
fcis-6909	98	2	m	m	VERB
fcis-6909	98	3	d	d	VERB
fcis-6909	98	4	u	u	X
fcis-6909	98	5	y	y	PROPN
fcis-6909	98	6			PROPN
fcis-6909	98	7			NOUN
fcis-6909	98	8			X
fcis-6909	98	9	=	=	SYM
fcis-6909	98	10			X
fcis-6909	98	11	(	(	PUNCT
fcis-6909	98	12	8)	8)	NUM
fcis-6909	98	13	3.4	3.4	NUM
fcis-6909	98	14	.	.	PUNCT
fcis-6909	98	15	prediction	prediction	NOUN
fcis-6909	98	16	layer	layer	NOUN
fcis-6909	98	17	sometimes	sometimes	ADV
fcis-6909	98	18	bias	bias	VERB
fcis-6909	98	19	amplification	amplification	NOUN
fcis-6909	98	20	is	be	AUX
fcis-6909	98	21	beneficial	beneficial	ADJ
fcis-6909	98	22	to	to	ADP
fcis-6909	98	23	users	user	NOUN
fcis-6909	98	24	,	,	PUNCT
fcis-6909	98	25	so	so	ADV
fcis-6909	98	26	to	to	PART
fcis-6909	98	27	dynamically	dynamically	ADV
fcis-6909	98	28	adjust	adjust	VERB
fcis-6909	98	29	the	the	DET
fcis-6909	98	30	impact	impact	NOUN
fcis-6909	98	31	of	of	ADP
fcis-6909	98	32	causal	causal	ADJ
fcis-6909	98	33	inference	inference	NOUN
fcis-6909	98	34	,	,	PUNCT
fcis-6909	98	35	scatter	scatter	NOUN
fcis-6909	98	36	is	be	AUX
fcis-6909	98	37	used	use	VERB
fcis-6909	98	38	to	to	PART
fcis-6909	98	39	measure	measure	VERB
fcis-6909	98	40	interest	interest	NOUN
fcis-6909	98	41	drift	drift	NOUN
fcis-6909	98	42	in	in	ADP
fcis-6909	98	43	the	the	DET
fcis-6909	98	44	sequence	sequence	NOUN
fcis-6909	98	45	of	of	ADP
fcis-6909	98	46	user	user	NOUN
fcis-6909	98	47	interactions	interaction	NOUN
fcis-6909	98	48	,	,	PUNCT
fcis-6909	98	49	and	and	CCONJ
fcis-6909	98	50	thus	thus	ADV
fcis-6909	98	51	to	to	PART
fcis-6909	98	52	adjust	adjust	VERB
fcis-6909	98	53	the	the	DET
fcis-6909	98	54	blend	blend	NOUN
fcis-6909	98	55	of	of	ADP
fcis-6909	98	56	traditional	traditional	ADJ
fcis-6909	98	57	and	and	CCONJ
fcis-6909	98	58	causal	causal	ADJ
fcis-6909	98	59	recommendations	recommendation	NOUN
fcis-6909	98	60	.	.	PUNCT
fcis-6909	99	1	kl	kl	PROPN
fcis-6909	99	2	scatter	scatter	NOUN
fcis-6909	99	3	is	be	AUX
fcis-6909	99	4	used	use	VERB
fcis-6909	99	5	to	to	PART
fcis-6909	99	6	measure	measure	VERB
fcis-6909	99	7	the	the	DET
fcis-6909	99	8	difference	difference	NOUN
fcis-6909	99	9	between	between	ADP
fcis-6909	99	10	two	two	NUM
fcis-6909	99	11	distributions	distribution	NOUN
fcis-6909	99	12	and	and	CCONJ
fcis-6909	99	13	is	be	AUX
fcis-6909	99	14	equal	equal	ADJ
fcis-6909	99	15	to	to	ADP
fcis-6909	99	16	a	a	DET
fcis-6909	99	17	cross	cross	NOUN
fcis-6909	99	18	-	-	NOUN
fcis-6909	99	19	entropy	entropy	NOUN
fcis-6909	99	20	minus	minus	ADP
fcis-6909	99	21	an	an	DET
fcis-6909	99	22	information	information	NOUN
fcis-6909	99	23	entropy	entropy	NOUN
fcis-6909	99	24	,	,	PUNCT
fcis-6909	99	25	but	but	CCONJ
fcis-6909	99	26	it	it	PRON
fcis-6909	99	27	can	can	AUX
fcis-6909	99	28	not	not	PART
fcis-6909	99	29	be	be	AUX
fcis-6909	99	30	used	use	VERB
fcis-6909	99	31	properly	properly	ADV
fcis-6909	99	32	in	in	ADP
fcis-6909	99	33	recommender	recommender	NOUN
fcis-6909	99	34	systems	system	NOUN
fcis-6909	99	35	due	due	ADP
fcis-6909	99	36	to	to	ADP
fcis-6909	99	37	its	its	PRON
fcis-6909	99	38	asymmetry	asymmetry	NOUN
fcis-6909	99	39	,	,	PUNCT
fcis-6909	99	40	i.e.	i.e.	X
fcis-6909	99	41	,	,	PUNCT
fcis-6909	99	42	kl(p||q	kl(p||q	ADJ
fcis-6909	99	43	)	)	PUNCT
fcis-6909	100	1	≠	≠	PROPN
fcis-6909	100	2	kl(q||p	kl(q||p	NOUN
fcis-6909	100	3	)	)	PUNCT
fcis-6909	100	4	.	.	PUNCT
fcis-6909	101	1	jsd	jsd	PROPN
fcis-6909	101	2	(	(	PUNCT
fcis-6909	101	3	jensen	jensen	PROPN
fcis-6909	101	4	-	-	PUNCT
fcis-6909	101	5	shannon	shannon	PROPN
fcis-6909	101	6	divergence	divergence	PROPN
fcis-6909	101	7	,	,	PUNCT
fcis-6909	101	8	js	js	ADJ
fcis-6909	101	9	scatter	scatter	NOUN
fcis-6909	101	10	)	)	PUNCT
fcis-6909	101	11	is	be	AUX
fcis-6909	101	12	also	also	ADV
fcis-6909	101	13	a	a	DET
fcis-6909	101	14	measure	measure	NOUN
fcis-6909	101	15	for	for	ADP
fcis-6909	101	16	similarity	similarity	NOUN
fcis-6909	101	17	or	or	CCONJ
fcis-6909	101	18	difference	difference	NOUN
fcis-6909	101	19	between	between	ADP
fcis-6909	101	20	two	two	NUM
fcis-6909	101	21	probability	probability	NOUN
fcis-6909	101	22	distributions	distribution	NOUN
fcis-6909	101	23	as	as	ADP
fcis-6909	101	24	a	a	DET
fcis-6909	101	25	metric	metric	NOUN
fcis-6909	101	26	.	.	PUNCT
fcis-6909	102	1	it	it	PRON
fcis-6909	102	2	can	can	AUX
fcis-6909	102	3	use	use	VERB
fcis-6909	102	4	kl	kl	NOUN
fcis-6909	102	5	(	(	PUNCT
fcis-6909	102	6	kullbackleibler	kullbackleibler	NOUN
fcis-6909	102	7	)	)	PUNCT
fcis-6909	102	8	scatter	scatter	NOUN
fcis-6909	102	9	to	to	PART
fcis-6909	102	10	measure	measure	VERB
fcis-6909	102	11	its	its	PRON
fcis-6909	102	12	similarity	similarity	NOUN
fcis-6909	102	13	and	and	CCONJ
fcis-6909	102	14	it	it	PRON
fcis-6909	102	15	can	can	AUX
fcis-6909	102	16	be	be	AUX
fcis-6909	102	17	weakened	weaken	VERB
fcis-6909	102	18	again	again	ADV
fcis-6909	102	19	in	in	ADP
fcis-6909	102	20	a	a	DET
fcis-6909	102	21	different	different	ADJ
fcis-6909	102	22	dimension	dimension	NOUN
fcis-6909	102	23	.	.	PUNCT
fcis-6909	103	1	since	since	SCONJ
fcis-6909	103	2	the	the	DET
fcis-6909	103	3	kl	kl	PROPN
fcis-6909	103	4	scatter	scatter	NOUN
fcis-6909	103	5	is	be	AUX
fcis-6909	103	6	asymmetric	asymmetric	ADJ
fcis-6909	103	7	,	,	PUNCT
fcis-6909	103	8	a	a	DET
fcis-6909	103	9	slight	slight	ADJ
fcis-6909	103	10	modification	modification	NOUN
fcis-6909	103	11	of	of	ADP
fcis-6909	103	12	it	it	PRON
fcis-6909	103	13	can	can	AUX
fcis-6909	103	14	be	be	AUX
fcis-6909	103	15	transformed	transform	VERB
fcis-6909	103	16	into	into	ADP
fcis-6909	103	17	a	a	DET
fcis-6909	103	18	symmetric	symmetric	ADJ
fcis-6909	103	19	js	js	NOUN
fcis-6909	103	20	.	.	PUNCT
fcis-6909	104	1	first	first	ADV
fcis-6909	104	2	,	,	PUNCT
fcis-6909	104	3	the	the	DET
fcis-6909	104	4	js	js	ADJ
fcis-6909	104	5	scatter	scatter	NOUN
fcis-6909	104	6	is	be	AUX
fcis-6909	104	7	defined	define	VERB
fcis-6909	104	8	as	as	ADP
fcis-6909	104	9	follows	follow	VERB
fcis-6909	104	10	.	.	PUNCT
fcis-6909	105	1	(	(	PUNCT
fcis-6909	105	2	)	)	PUNCT
fcis-6909	105	3	(	(	PUNCT
fcis-6909	105	4	)	)	PUNCT
fcis-6909	105	5	(	(	PUNCT
fcis-6909	105	6	)	)	PUNCT
fcis-6909	105	7	1	1	NUM
fcis-6909	105	8	1	1	NUM
fcis-6909	105	9	2	2	NUM
fcis-6909	105	10	2	2	NUM
fcis-6909	105	11	jsd	jsd	NOUN
fcis-6909	105	12	p	p	X
fcis-6909	105	13	q	q	PROPN
fcis-6909	105	14	kl	kl	NOUN
fcis-6909	105	15	p	p	NOUN
fcis-6909	105	16	m	m	PROPN
fcis-6909	105	17	kl	kl	NOUN
fcis-6909	105	18	q	q	X
fcis-6909	105	19	m=	m=	X
fcis-6909	105	20	+	+	CCONJ
fcis-6909	105	21	(	(	PUNCT
fcis-6909	105	22	9	9	NUM
fcis-6909	105	23	)	)	PUNCT
fcis-6909	105	24	where	where	SCONJ
fcis-6909	105	25	m=1/2(p+q	m=1/2(p+q	NOUN
fcis-6909	105	26	)	)	PUNCT
fcis-6909	105	27	and	and	CCONJ
fcis-6909	105	28	p	p	NOUN
fcis-6909	105	29	and	and	CCONJ
fcis-6909	105	30	q	q	NOUN
fcis-6909	105	31	denote	denote	NOUN
fcis-6909	105	32	1	1	NUM
fcis-6909	105	33	ud	ud	ADV
fcis-6909	105	34	and	and	CCONJ
fcis-6909	105	35	2	2	NUM
fcis-6909	105	36	ud	ud	INTJ
fcis-6909	105	37	,	,	PUNCT
fcis-6909	105	38	respectively	respectively	ADV
fcis-6909	105	39	,	,	PUNCT
fcis-6909	105	40	the	the	DET
fcis-6909	105	41	class	class	NOUN
fcis-6909	105	42	distribution	distribution	NOUN
fcis-6909	105	43	of	of	ADP
fcis-6909	105	44	user	user	NOUN
fcis-6909	105	45	u	u	NOUN
fcis-6909	105	46	on	on	ADP
fcis-6909	105	47	the	the	DET
fcis-6909	105	48	two	two	NUM
fcis-6909	105	49	segments	segment	NOUN
fcis-6909	105	50	before	before	ADP
fcis-6909	105	51	and	and	CCONJ
fcis-6909	105	52	after	after	ADP
fcis-6909	105	53	the	the	DET
fcis-6909	105	54	interaction	interaction	NOUN
fcis-6909	105	55	sequence	sequence	NOUN
fcis-6909	105	56	.	.	PUNCT
fcis-6909	106	1	in	in	ADP
fcis-6909	106	2	general	general	ADJ
fcis-6909	106	3	,	,	PUNCT
fcis-6909	106	4	the	the	DET
fcis-6909	106	5	js	js	ADJ
fcis-6909	106	6	scatter	scatter	NOUN
fcis-6909	106	7	is	be	AUX
fcis-6909	106	8	symmetric	symmetric	ADJ
fcis-6909	106	9	and	and	CCONJ
fcis-6909	106	10	takes	take	VERB
fcis-6909	106	11	values	value	NOUN
fcis-6909	106	12	between	between	ADP
fcis-6909	106	13	0	0	NUM
fcis-6909	106	14	and	and	CCONJ
fcis-6909	106	15	1	1	NUM
fcis-6909	106	16	.	.	PUNCT
fcis-6909	107	1	the	the	DET
fcis-6909	107	2	js	js	ADJ
fcis-6909	107	3	scatter	scatter	NOUN
fcis-6909	107	4	and	and	CCONJ
fcis-6909	107	5	the	the	DET
fcis-6909	107	6	kl	kl	PROPN
fcis-6909	107	7	scatter	scatter	NOUN
fcis-6909	107	8	have	have	VERB
fcis-6909	107	9	some	some	DET
fcis-6909	107	10	similarities	similarity	NOUN
fcis-6909	107	11	but	but	CCONJ
fcis-6909	107	12	also	also	ADV
fcis-6909	107	13	some	some	DET
fcis-6909	107	14	significant	significant	ADJ
fcis-6909	107	15	differences	difference	NOUN
fcis-6909	107	16	.	.	PUNCT
fcis-6909	108	1	the	the	DET
fcis-6909	108	2	js	js	PROPN
fcis-6909	108	3	scatter	scatter	NOUN
fcis-6909	108	4	can	can	AUX
fcis-6909	108	5	consider	consider	VERB
fcis-6909	108	6	the	the	DET
fcis-6909	108	7	overlap	overlap	NOUN
fcis-6909	108	8	between	between	ADP
fcis-6909	108	9	the	the	DET
fcis-6909	108	10	two	two	NUM
fcis-6909	108	11	probability	probability	NOUN
fcis-6909	108	12	distributions	distribution	NOUN
fcis-6909	108	13	at	at	ADP
fcis-6909	108	14	the	the	DET
fcis-6909	108	15	same	same	ADJ
fcis-6909	108	16	time	time	NOUN
fcis-6909	108	17	and	and	CCONJ
fcis-6909	108	18	thus	thus	ADV
fcis-6909	108	19	calculate	calculate	VERB
fcis-6909	108	20	their	their	PRON
fcis-6909	108	21	similarity	similarity	NOUN
fcis-6909	108	22	more	more	ADV
fcis-6909	108	23	accurately	accurately	ADV
fcis-6909	108	24	,	,	PUNCT
fcis-6909	108	25	whereas	whereas	SCONJ
fcis-6909	108	26	the	the	DET
fcis-6909	108	27	kl	kl	PROPN
fcis-6909	108	28	scatter	scatter	NOUN
fcis-6909	108	29	treats	treat	VERB
fcis-6909	108	30	them	they	PRON
fcis-6909	108	31	as	as	ADP
fcis-6909	108	32	independent	independent	ADJ
fcis-6909	108	33	probability	probability	NOUN
fcis-6909	108	34	distributions	distribution	NOUN
fcis-6909	108	35	and	and	CCONJ
fcis-6909	108	36	thus	thus	ADV
fcis-6909	108	37	can	can	AUX
fcis-6909	108	38	only	only	ADV
fcis-6909	108	39	be	be	AUX
fcis-6909	108	40	used	use	VERB
fcis-6909	108	41	to	to	PART
fcis-6909	108	42	calculate	calculate	VERB
fcis-6909	108	43	the	the	DET
fcis-6909	108	44	variability	variability	NOUN
fcis-6909	108	45	between	between	ADP
fcis-6909	108	46	different	different	ADJ
fcis-6909	108	47	probability	probability	NOUN
fcis-6909	108	48	distributions	distribution	NOUN
fcis-6909	108	49	.	.	PUNCT
fcis-6909	109	1	after	after	ADP
fcis-6909	109	2	normalizing	normalize	VERB
fcis-6909	109	3	the	the	DET
fcis-6909	109	4	jsd	jsd	PROPN
fcis-6909	109	5	:	:	PUNCT
fcis-6909	109	6	min	min	PROPN
fcis-6909	109	7	max	max	PROPN
fcis-6909	109	8	min	min	PROPN
fcis-6909	109	9	ˆ	ˆ	ADP
fcis-6909	109	10	u	u	NOUN
fcis-6909	109	11	u	u	PROPN
fcis-6909	109	12	jsd	jsd	PROPN
fcis-6909	109	13	j	j	PROPN
fcis-6909	109	14	jsd	jsd	PROPN
fcis-6909	109	15	jsd	jsd	PROPN
fcis-6909	109	16	jsd	jsd	PROPN
fcis-6909	109	17			NOUN
fcis-6909	109	18			NOUN
fcis-6909	109	19	−	−	NOUN
fcis-6909	109	20	=	=	SYM
fcis-6909	109	21			PROPN
fcis-6909	110	1			INTJ
fcis-6909	110	2	−	−	NUM
fcis-6909	110	3			NOUN
fcis-6909	110	4	(	(	PUNCT
fcis-6909	110	5	10	10	NUM
fcis-6909	110	6	)	)	PUNCT
fcis-6909	110	7	where	where	SCONJ
fcis-6909	110	8	ˆ	ˆ	NOUN
fcis-6909	110	9	uj	uj	PROPN
fcis-6909	110	10	is	be	AUX
fcis-6909	110	11	the	the	DET
fcis-6909	110	12	normalized	normalize	VERB
fcis-6909	110	13	value	value	NOUN
fcis-6909	110	14	of	of	ADP
fcis-6909	110	15	jsd	jsd	PROPN
fcis-6909	110	16	,	,	PUNCT
fcis-6909	110	17	minjsd	minjsd	PROPN
fcis-6909	110	18	and	and	CCONJ
fcis-6909	110	19	maxjsd	maxjsd	PROPN
fcis-6909	110	20	are	be	AUX
fcis-6909	110	21	the	the	DET
fcis-6909	110	22	maximum	maximum	ADJ
fcis-6909	110	23	and	and	CCONJ
fcis-6909	110	24	minimum	minimum	ADJ
fcis-6909	110	25	values	value	NOUN
fcis-6909	110	26	of	of	ADP
fcis-6909	110	27	jsd	jsd	PROPN
fcis-6909	110	28	for	for	ADP
fcis-6909	110	29	all	all	DET
fcis-6909	110	30	users	user	NOUN
fcis-6909	110	31	,	,	PUNCT
fcis-6909	110	32	respectively	respectively	ADV
fcis-6909	110	33	.	.	PUNCT
fcis-6909	111	1	it	it	PRON
fcis-6909	111	2	can	can	AUX
fcis-6909	111	3	be	be	AUX
fcis-6909	111	4	used	use	VERB
fcis-6909	111	5	as	as	ADP
fcis-6909	111	6	a	a	DET
fcis-6909	111	7	moderating	moderate	VERB
fcis-6909	111	8	factor	factor	NOUN
fcis-6909	111	9	to	to	PART
fcis-6909	111	10	reconcile	reconcile	VERB
fcis-6909	111	11	the	the	DET
fcis-6909	111	12	traditional	traditional	ADJ
fcis-6909	111	13	model	model	NOUN
fcis-6909	111	14	with	with	ADP
fcis-6909	111	15	the	the	DET
fcis-6909	111	16	bdcr	bdcr	NOUN
fcis-6909	111	17	model	model	NOUN
fcis-6909	111	18	as	as	SCONJ
fcis-6909	111	19	shown	show	VERB
fcis-6909	111	20	in	in	ADP
fcis-6909	111	21	equation	equation	NOUN
fcis-6909	111	22	11	11	NUM
fcis-6909	111	23	.	.	PUNCT
fcis-6909	112	1	(	(	PUNCT
fcis-6909	112	2	)	)	PUNCT
fcis-6909	112	3	,	,	PUNCT
fcis-6909	112	4	,	,	PUNCT
fcis-6909	112	5	,	,	PUNCT
fcis-6909	112	6	ˆ	ˆ	ADV
fcis-6909	112	7	ˆ1	ˆ1	NOUN
fcis-6909	112	8	*	*	PUNCT
fcis-6909	112	9	*	*	PUNCT
fcis-6909	112	10	rs	rs	PROPN
fcis-6909	112	11	bdcr	bdcr	NOUN
fcis-6909	112	12	u	u	NOUN
fcis-6909	112	13	i	i	PRON
fcis-6909	112	14	u	u	VERB
fcis-6909	112	15	u	u	VERB
fcis-6909	112	16	i	i	PRON
fcis-6909	112	17	u	u	VERB
fcis-6909	112	18	u	u	NOUN
fcis-6909	112	19	iy	iy	PROPN
fcis-6909	112	20	j	j	PROPN
fcis-6909	112	21	y	y	PROPN
fcis-6909	112	22	j	j	PROPN
fcis-6909	113	1	y=	y=	NUM
fcis-6909	113	2	−	−	PROPN
fcis-6909	114	1	+	+	CCONJ
fcis-6909	114	2	(	(	PUNCT
fcis-6909	114	3	11	11	NUM
fcis-6909	114	4	)	)	PUNCT
fcis-6909	114	5	,	,	PUNCT
fcis-6909	114	6	rs	rs	INTJ
fcis-6909	114	7	u	u	PRON
fcis-6909	114	8	iy	iy	PROPN
fcis-6909	114	9	is	be	AUX
fcis-6909	114	10	the	the	DET
fcis-6909	114	11	score	score	NOUN
fcis-6909	114	12	predicted	predict	VERB
fcis-6909	114	13	by	by	ADP
fcis-6909	114	14	the	the	DET
fcis-6909	114	15	traditional	traditional	ADJ
fcis-6909	114	16	model	model	NOUN
fcis-6909	114	17	,	,	PUNCT
fcis-6909	114	18	and	and	CCONJ
fcis-6909	114	19	,	,	PUNCT
fcis-6909	114	20	bdcr	bdcr	VERB
fcis-6909	114	21	u	u	PROPN
fcis-6909	114	22	iy	iy	PROPN
fcis-6909	114	23	is	be	AUX
fcis-6909	114	24	the	the	DET
fcis-6909	114	25	score	score	NOUN
fcis-6909	114	26	predicted	predict	VERB
fcis-6909	114	27	by	by	ADP
fcis-6909	114	28	the	the	DET
fcis-6909	114	29	bdcr	bdcr	NOUN
fcis-6909	114	30	.	.	PUNCT
fcis-6909	115	1	4	4	X
fcis-6909	115	2	.	.	X
fcis-6909	115	3	experimental	experimental	ADJ
fcis-6909	115	4	analysis	analysis	NOUN
fcis-6909	115	5	the	the	DET
fcis-6909	115	6	experiment	experiment	NOUN
fcis-6909	115	7	focuses	focus	VERB
fcis-6909	115	8	on	on	ADP
fcis-6909	115	9	the	the	DET
fcis-6909	115	10	following	follow	VERB
fcis-6909	115	11	questions	question	NOUN
fcis-6909	115	12	.	.	PUNCT
fcis-6909	116	1	rq1	rq1	NOUN
fcis-6909	116	2	:	:	PUNCT
fcis-6909	116	3	how	how	SCONJ
fcis-6909	116	4	does	do	AUX
fcis-6909	116	5	our	our	PRON
fcis-6909	116	6	method	method	NOUN
fcis-6909	116	7	perform	perform	NOUN
fcis-6909	116	8	compared	compare	VERB
fcis-6909	116	9	with	with	ADP
fcis-6909	116	10	baseline	baseline	ADJ
fcis-6909	116	11	methods	method	NOUN
fcis-6909	116	12	?	?	PUNCT
fcis-6909	117	1	rq2	rq2	NOUN
fcis-6909	117	2	:	:	PUNCT
fcis-6909	117	3	how	how	SCONJ
fcis-6909	117	4	do	do	AUX
fcis-6909	117	5	different	different	ADJ
fcis-6909	117	6	hyperparameter	hyperparameter	NOUN
fcis-6909	117	7	affect	affect	VERB
fcis-6909	117	8	recommendation	recommendation	NOUN
fcis-6909	117	9	performance	performance	NOUN
fcis-6909	117	10	?	?	PUNCT
fcis-6909	118	1	rq3	rq3	NOUN
fcis-6909	118	2	:	:	PUNCT
fcis-6909	118	3	how	how	SCONJ
fcis-6909	118	4	does	do	AUX
fcis-6909	118	5	different	different	ADJ
fcis-6909	118	6	sub	sub	NOUN
fcis-6909	118	7	-	-	ADJ
fcis-6909	118	8	model	model	ADJ
fcis-6909	118	9	affect	affect	VERB
fcis-6909	118	10	model	model	NOUN
fcis-6909	118	11	performance	performance	NOUN
fcis-6909	118	12	?	?	PUNCT
fcis-6909	119	1	4.1	4.1	NUM
fcis-6909	119	2	.	.	PUNCT
fcis-6909	119	3	data	datum	NOUN
fcis-6909	119	4	set	set	VERB
fcis-6909	119	5	and	and	CCONJ
fcis-6909	119	6	evaluation	evaluation	NOUN
fcis-6909	119	7	using	use	VERB
fcis-6909	119	8	the	the	DET
fcis-6909	119	9	generic	generic	ADJ
fcis-6909	119	10	datasets	dataset	NOUN
fcis-6909	119	11	ml-1	ml-1	NOUN
fcis-6909	119	12	m	m	PROPN
fcis-6909	119	13	and	and	CCONJ
fcis-6909	119	14	amazon	amazon	NOUN
fcis-6909	119	15	-	-	PUNCT
fcis-6909	119	16	book	book	NOUN
fcis-6909	119	17	,	,	PUNCT
fcis-6909	119	18	ml-1	ml-1	PRON
fcis-6909	119	19	m	m	VERB
fcis-6909	119	20	is	be	AUX
fcis-6909	119	21	a	a	DET
fcis-6909	119	22	movie	movie	NOUN
fcis-6909	119	23	recommendation	recommendation	NOUN
fcis-6909	119	24	dataset	dataset	VERB
fcis-6909	119	25	with	with	ADP
fcis-6909	119	26	rich	rich	ADJ
fcis-6909	119	27	features	feature	NOUN
fcis-6909	119	28	,	,	PUNCT
fcis-6909	119	29	such	such	ADJ
fcis-6909	119	30	as	as	ADP
fcis-6909	119	31	user	user	NOUN
fcis-6909	119	32	gender	gender	NOUN
fcis-6909	119	33	and	and	CCONJ
fcis-6909	119	34	movie	movie	NOUN
fcis-6909	119	35	genre	genre	NOUN
fcis-6909	119	36	,	,	PUNCT
fcis-6909	119	37	grouped	group	VERB
fcis-6909	119	38	by	by	ADP
fcis-6909	119	39	movie	movie	NOUN
fcis-6909	119	40	genre	genre	NOUN
fcis-6909	119	41	,	,	PUNCT
fcis-6909	119	42	and	and	CCONJ
fcis-6909	119	43	amazon	amazon	NOUN
fcis-6909	119	44	-	-	PUNCT
fcis-6909	119	45	book	book	NOUN
fcis-6909	119	46	is	be	AUX
fcis-6909	119	47	an	an	DET
fcis-6909	119	48	amazon	amazon	NOUN
fcis-6909	119	49	book	book	NOUN
fcis-6909	119	50	product	product	NOUN
fcis-6909	119	51	dataset	dataset	NOUN
fcis-6909	119	52	grouped	group	VERB
fcis-6909	119	53	by	by	ADP
fcis-6909	119	54	book	book	NOUN
fcis-6909	119	55	category	category	NOUN
fcis-6909	119	56	,	,	PUNCT
fcis-6909	119	57	with	with	ADP
fcis-6909	119	58	the	the	DET
fcis-6909	119	59	dataset	dataset	NOUN
fcis-6909	119	60	description	description	NOUN
fcis-6909	119	61	shown	show	VERB
fcis-6909	119	62	in	in	ADP
fcis-6909	119	63	table	table	NOUN
fcis-6909	119	64	1	1	NUM
fcis-6909	119	65	.	.	PUNCT
fcis-6909	120	1	for	for	ADP
fcis-6909	120	2	each	each	DET
fcis-6909	120	3	interaction	interaction	NOUN
fcis-6909	120	4	in	in	ADP
fcis-6909	120	5	the	the	DET
fcis-6909	120	6	dataset	dataset	NOUN
fcis-6909	120	7	with	with	ADP
fcis-6909	120	8	rating>=5	rating>=5	PROPN
fcis-6909	120	9	,	,	PUNCT
fcis-6909	120	10	it	it	PRON
fcis-6909	120	11	is	be	AUX
fcis-6909	120	12	considered	consider	VERB
fcis-6909	120	13	as	as	ADP
fcis-6909	120	14	a	a	DET
fcis-6909	120	15	positive	positive	ADJ
fcis-6909	120	16	example	example	NOUN
fcis-6909	120	17	,	,	PUNCT
fcis-6909	120	18	and	and	CCONJ
fcis-6909	120	19	users	user	NOUN
fcis-6909	120	20	and	and	CCONJ
fcis-6909	120	21	items	item	NOUN
fcis-6909	120	22	with	with	ADP
fcis-6909	120	23	less	less	ADJ
fcis-6909	120	24	than	than	ADP
fcis-6909	120	25	20	20	NUM
fcis-6909	120	26	interactions	interaction	NOUN
fcis-6909	120	27	are	be	AUX
fcis-6909	120	28	discarded	discard	VERB
fcis-6909	120	29	.	.	PUNCT
fcis-6909	121	1	the	the	DET
fcis-6909	121	2	ratio	ratio	NOUN
fcis-6909	121	3	of	of	ADP
fcis-6909	121	4	training	training	NOUN
fcis-6909	121	5	set	set	NOUN
fcis-6909	121	6	,	,	PUNCT
fcis-6909	121	7	validation	validation	NOUN
fcis-6909	121	8	set	set	NOUN
fcis-6909	121	9	and	and	CCONJ
fcis-6909	121	10	test	test	NOUN
fcis-6909	121	11	set	set	VERB
fcis-6909	121	12	was	be	AUX
fcis-6909	121	13	8:1:1	8:1:1	NUM
fcis-6909	121	14	after	after	ADP
fcis-6909	121	15	sorting	sort	VERB
fcis-6909	121	16	according	accord	VERB
fcis-6909	121	17	to	to	ADP
fcis-6909	121	18	the	the	DET
fcis-6909	121	19	interaction	interaction	NOUN
fcis-6909	121	20	timestamp	timestamp	NOUN
fcis-6909	121	21	of	of	ADP
fcis-6909	121	22	each	each	DET
fcis-6909	121	23	user	user	NOUN
fcis-6909	121	24	.	.	PUNCT
fcis-6909	122	1	random	random	ADJ
fcis-6909	122	2	negative	negative	ADJ
fcis-6909	122	3	sampling	sampling	NOUN
fcis-6909	122	4	was	be	AUX
fcis-6909	122	5	used	use	VERB
fcis-6909	122	6	as	as	ADP
fcis-6909	122	7	negative	negative	ADJ
fcis-6909	122	8	samples	sample	NOUN
fcis-6909	122	9	during	during	ADP
fcis-6909	122	10	the	the	DET
fcis-6909	122	11	training	training	NOUN
fcis-6909	122	12	process	process	NOUN
fcis-6909	122	13	.	.	PUNCT
fcis-6909	123	1	evaluation	evaluation	NOUN
fcis-6909	123	2	metrics	metric	NOUN
fcis-6909	123	3	:	:	PUNCT
fcis-6909	123	4	since	since	SCONJ
fcis-6909	123	5	the	the	DET
fcis-6909	123	6	top	top	ADJ
fcis-6909	123	7	-	-	PUNCT
fcis-6909	123	8	k	k	NOUN
fcis-6909	123	9	ranking	ranking	NOUN
fcis-6909	123	10	is	be	AUX
fcis-6909	123	11	used	use	VERB
fcis-6909	123	12	,	,	PUNCT
fcis-6909	123	13	recall	recall	NOUN
fcis-6909	123	14	,	,	PUNCT
fcis-6909	123	15	ndcg	ndcg	PROPN
fcis-6909	123	16	is	be	AUX
fcis-6909	123	17	selected	select	VERB
fcis-6909	123	18	as	as	ADP
fcis-6909	123	19	the	the	DET
fcis-6909	123	20	evaluation	evaluation	NOUN
fcis-6909	123	21	metric	metric	NOUN
fcis-6909	123	22	.	.	PUNCT
fcis-6909	124	1	an	an	DET
fcis-6909	124	2	early	early	ADJ
fcis-6909	124	3	stop	stop	NOUN
fcis-6909	124	4	strategy	strategy	NOUN
fcis-6909	124	5	is	be	AUX
fcis-6909	124	6	adopted	adopt	VERB
fcis-6909	124	7	,	,	PUNCT
fcis-6909	124	8	and	and	CCONJ
fcis-6909	124	9	recall@10	recall@10	NOUN
fcis-6909	124	10	on	on	ADP
fcis-6909	124	11	the	the	DET
fcis-6909	124	12	validation	validation	NOUN
fcis-6909	124	13	set	set	NOUN
fcis-6909	124	14	is	be	AUX
fcis-6909	124	15	used	use	VERB
fcis-6909	124	16	as	as	ADP
fcis-6909	124	17	the	the	DET
fcis-6909	124	18	criterion	criterion	NOUN
fcis-6909	124	19	for	for	ADP
fcis-6909	124	20	selecting	select	VERB
fcis-6909	124	21	the	the	DET
fcis-6909	124	22	best	good	ADJ
fcis-6909	124	23	model	model	NOUN
fcis-6909	124	24	.	.	PUNCT
fcis-6909	125	1	table	table	NOUN
fcis-6909	125	2	1	1	NUM
fcis-6909	125	3	.	.	PUNCT
fcis-6909	125	4	description	description	NOUN
fcis-6909	125	5	of	of	ADP
fcis-6909	125	6	the	the	DET
fcis-6909	125	7	data	datum	NOUN
fcis-6909	125	8	set	set	VERB
fcis-6909	125	9	dataset	dataset	NOUN
fcis-6909	125	10	users	user	NOUN
fcis-6909	125	11	items	item	NOUN
fcis-6909	125	12	interactions	interaction	NOUN
fcis-6909	125	13	features	feature	VERB
fcis-6909	125	14	group	group	NOUN
fcis-6909	125	15	ml-1	ml-1	PROPN
fcis-6909	125	16	m	m	VERB
fcis-6909	125	17	3,883	3,883	NUM
fcis-6909	125	18	6,040	6,040	NUM
fcis-6909	125	19	575,276	575,276	NUM
fcis-6909	125	20	13,408	13,408	NUM
fcis-6909	125	21	18	18	NUM
fcis-6909	125	22	amazonbook	amazonbook	NOUN
fcis-6909	125	23	29115	29115	NUM
fcis-6909	125	24	16,845	16,845	NUM
fcis-6909	125	25	1,712,409	1,712,409	NUM
fcis-6909	125	26	46,213	46,213	NUM
fcis-6909	125	27	253	253	NUM
fcis-6909	125	28	4.2	4.2	NUM
fcis-6909	125	29	.	.	PUNCT
fcis-6909	126	1	baseline	baseline	VERB
fcis-6909	126	2	the	the	DET
fcis-6909	126	3	proposed	propose	VERB
fcis-6909	126	4	method	method	NOUN
fcis-6909	126	5	is	be	AUX
fcis-6909	126	6	generalized	generalize	VERB
fcis-6909	126	7	and	and	CCONJ
fcis-6909	126	8	instantiated	instantiate	VERB
fcis-6909	126	9	in	in	ADP
fcis-6909	126	10	two	two	NUM
fcis-6909	126	11	representative	representative	ADJ
fcis-6909	126	12	recommendation	recommendation	NOUN
fcis-6909	126	13	models	model	NOUN
fcis-6909	126	14	,	,	PUNCT
fcis-6909	126	15	fm	fm	PROPN
fcis-6909	126	16	and	and	CCONJ
fcis-6909	126	17	nfm	nfm	PROPN
fcis-6909	126	18	,	,	PUNCT
fcis-6909	126	19	to	to	PART
fcis-6909	126	20	mitigate	mitigate	VERB
fcis-6909	126	21	noise	noise	NOUN
fcis-6909	126	22	and	and	CCONJ
fcis-6909	126	23	bias	bias	NOUN
fcis-6909	126	24	amplification	amplification	NOUN
fcis-6909	126	25	.	.	PUNCT
fcis-6909	127	1	the	the	DET
fcis-6909	127	2	benchmark	benchmark	ADJ
fcis-6909	127	3	algorithms	algorithms	NOUN
fcis-6909	127	4	for	for	ADP
fcis-6909	127	5	the	the	DET
fcis-6909	127	6	comparison	comparison	NOUN
fcis-6909	127	7	are	be	AUX
fcis-6909	127	8	as	as	SCONJ
fcis-6909	127	9	follows	follow	VERB
fcis-6909	127	10	.	.	PUNCT
fcis-6909	128	1	ips	ip	NOUN
fcis-6909	129	1	[	[	X
fcis-6909	129	2	18	18	NUM
fcis-6909	129	3	]	]	PUNCT
fcis-6909	129	4	is	be	AUX
fcis-6909	129	5	the	the	DET
fcis-6909	129	6	classical	classical	ADJ
fcis-6909	129	7	approach	approach	NOUN
fcis-6909	129	8	in	in	ADP
fcis-6909	129	9	causal	causal	ADJ
fcis-6909	129	10	13	13	NUM
fcis-6909	129	11	recommendation	recommendation	NOUN
fcis-6909	129	12	.	.	PUNCT
fcis-6909	130	1	in	in	SCONJ
fcis-6909	130	2	order	order	NOUN
fcis-6909	130	3	to	to	PART
fcis-6909	130	4	reduce	reduce	VERB
fcis-6909	130	5	the	the	DET
fcis-6909	130	6	weight	weight	NOUN
fcis-6909	130	7	of	of	ADP
fcis-6909	130	8	items	item	NOUN
fcis-6909	130	9	in	in	ADP
fcis-6909	130	10	the	the	DET
fcis-6909	130	11	multi	multi	NOUN
fcis-6909	130	12	-	-	NOUN
fcis-6909	130	13	array	array	NOUN
fcis-6909	130	14	during	during	ADP
fcis-6909	130	15	the	the	DET
fcis-6909	130	16	debiased	debiase	VERB
fcis-6909	130	17	training	training	NOUN
fcis-6909	130	18	,	,	PUNCT
fcis-6909	130	19	𝑃(𝒅𝑢	𝑃(𝒅𝑢	PROPN
fcis-6909	130	20	)	)	PUNCT
fcis-6909	130	21	is	be	AUX
fcis-6909	130	22	used	use	VERB
fcis-6909	130	23	here	here	ADV
fcis-6909	130	24	as	as	ADP
fcis-6909	130	25	the	the	DET
fcis-6909	130	26	user	user	NOUN
fcis-6909	130	27	's	's	PART
fcis-6909	130	28	propensity	propensity	NOUN
fcis-6909	130	29	𝑢	𝑢	NOUN
fcis-6909	130	30	,	,	PUNCT
fcis-6909	130	31	and	and	CCONJ
fcis-6909	130	32	the	the	DET
fcis-6909	130	33	propensity	propensity	NOUN
fcis-6909	130	34	clipping	clip	VERB
fcis-6909	130	35	technique	technique	NOUN
fcis-6909	130	36	[	[	X
fcis-6909	130	37	33	33	NUM
fcis-6909	130	38	]	]	PUNCT
fcis-6909	130	39	is	be	AUX
fcis-6909	130	40	used	use	VERB
fcis-6909	130	41	to	to	PART
fcis-6909	130	42	reduce	reduce	VERB
fcis-6909	130	43	the	the	DET
fcis-6909	130	44	propensity	propensity	NOUN
fcis-6909	130	45	variance	variance	NOUN
fcis-6909	130	46	,	,	PUNCT
fcis-6909	130	47	where	where	SCONJ
fcis-6909	130	48	the	the	DET
fcis-6909	130	49	clipping	clip	VERB
fcis-6909	130	50	threshold	threshold	NOUN
fcis-6909	130	51	in	in	ADP
fcis-6909	130	52	{	{	PUNCT
fcis-6909	130	53	2	2	NUM
fcis-6909	130	54	,	,	PUNCT
fcis-6909	130	55	3	3	NUM
fcis-6909	130	56	,	,	PUNCT
fcis-6909	130	57	...	...	PUNCT
fcis-6909	131	1	10	10	NUM
fcis-6909	131	2	}	}	PUNCT
fcis-6909	131	3	in	in	ADP
fcis-6909	131	4	the	the	DET
fcis-6909	131	5	search	search	NOUN
fcis-6909	131	6	adjustment	adjustment	NOUN
fcis-6909	131	7	.	.	PUNCT
fcis-6909	132	1	dice	dice	NOUN
fcis-6909	133	1	[	[	X
fcis-6909	133	2	19	19	NUM
fcis-6909	133	3	]	]	PUNCT
fcis-6909	133	4	utilizes	utilizes	ADJ
fcis-6909	133	5	causal	causal	ADJ
fcis-6909	133	6	separation	separation	NOUN
fcis-6909	133	7	of	of	ADP
fcis-6909	133	8	interest	interest	NOUN
fcis-6909	133	9	and	and	CCONJ
fcis-6909	133	10	consistent	consistent	ADJ
fcis-6909	133	11	representation	representation	NOUN
fcis-6909	133	12	embeddings	embedding	NOUN
fcis-6909	133	13	.	.	PUNCT
fcis-6909	134	1	these	these	DET
fcis-6909	134	2	embeddings	embedding	NOUN
fcis-6909	134	3	are	be	AUX
fcis-6909	134	4	obtained	obtain	VERB
fcis-6909	134	5	by	by	ADP
fcis-6909	134	6	training	training	NOUN
fcis-6909	134	7	on	on	ADP
fcis-6909	134	8	cause	cause	NOUN
fcis-6909	134	9	-	-	PUNCT
fcis-6909	134	10	specific	specific	ADJ
fcis-6909	134	11	data	datum	NOUN
fcis-6909	134	12	obtained	obtain	VERB
fcis-6909	134	13	from	from	ADP
fcis-6909	134	14	causal	causal	ADJ
fcis-6909	134	15	inference	inference	NOUN
fcis-6909	134	16	,	,	PUNCT
fcis-6909	134	17	and	and	CCONJ
fcis-6909	134	18	each	each	DET
fcis-6909	134	19	embedding	embed	VERB
fcis-6909	134	20	will	will	AUX
fcis-6909	134	21	capture	capture	VERB
fcis-6909	134	22	only	only	ADV
fcis-6909	134	23	one	one	NUM
fcis-6909	134	24	cause	cause	NOUN
fcis-6909	134	25	.	.	PUNCT
fcis-6909	135	1	choose	choose	VERB
fcis-6909	135	2	0.1	0.1	PUNCT
fcis-6909	136	1	=	=	SYM
fcis-6909	136	2	,	,	PUNCT
fcis-6909	136	3	0.01	0.01	NOUN
fcis-6909	136	4	=	=	PUNCT
fcis-6909	136	5	,	,	PUNCT
fcis-6909	136	6	and	and	CCONJ
fcis-6909	136	7	the	the	DET
fcis-6909	136	8	loss	loss	NOUN
fcis-6909	136	9	function	function	NOUN
fcis-6909	136	10	as	as	ADP
fcis-6909	136	11	bpr	bpr	PROPN
fcis-6909	136	12	.	.	PUNCT
fcis-6909	137	1	pda	pda	NOUN
fcis-6909	138	1	[	[	X
fcis-6909	138	2	20	20	NUM
fcis-6909	138	3	]	]	PUNCT
fcis-6909	138	4	considers	consider	VERB
fcis-6909	138	5	that	that	SCONJ
fcis-6909	138	6	the	the	DET
fcis-6909	138	7	popularity	popularity	NOUN
fcis-6909	138	8	bias	bias	NOUN
fcis-6909	138	9	affects	affect	VERB
fcis-6909	138	10	items	item	NOUN
fcis-6909	138	11	and	and	CCONJ
fcis-6909	138	12	ratings	rating	NOUN
fcis-6909	138	13	through	through	ADP
fcis-6909	138	14	exposure	exposure	NOUN
fcis-6909	138	15	and	and	CCONJ
fcis-6909	138	16	herding	herding	NOUN
fcis-6909	138	17	respectively	respectively	ADV
fcis-6909	138	18	,	,	PUNCT
fcis-6909	138	19	and	and	CCONJ
fcis-6909	138	20	then	then	ADV
fcis-6909	138	21	uses	use	VERB
fcis-6909	138	22	backdoor	backdoor	NOUN
fcis-6909	138	23	adjustment	adjustment	NOUN
fcis-6909	138	24	to	to	PART
fcis-6909	138	25	eliminate	eliminate	VERB
fcis-6909	138	26	the	the	DET
fcis-6909	138	27	effect	effect	NOUN
fcis-6909	138	28	on	on	ADP
fcis-6909	138	29	item	item	NOUN
fcis-6909	138	30	representation	representation	NOUN
fcis-6909	138	31	,	,	PUNCT
fcis-6909	138	32	and	and	CCONJ
fcis-6909	138	33	finally	finally	ADV
fcis-6909	138	34	adds	add	VERB
fcis-6909	138	35	a	a	DET
fcis-6909	138	36	reasonable	reasonable	ADJ
fcis-6909	138	37	popularity	popularity	NOUN
fcis-6909	138	38	to	to	ADP
fcis-6909	138	39	the	the	DET
fcis-6909	138	40	prediction	prediction	NOUN
fcis-6909	138	41	to	to	PART
fcis-6909	138	42	achieve	achieve	VERB
fcis-6909	138	43	full	full	ADJ
fcis-6909	138	44	utilization	utilization	NOUN
fcis-6909	138	45	of	of	ADP
fcis-6909	138	46	popularity	popularity	NOUN
fcis-6909	138	47	.	.	PUNCT
fcis-6909	139	1	the	the	DET
fcis-6909	139	2	popularity	popularity	NOUN
fcis-6909	139	3	smoothing	smooth	VERB
fcis-6909	139	4	parameter	parameter	NOUN
fcis-6909	139	5	𝛾	𝛾	PROPN
fcis-6909	139	6	is	be	AUX
fcis-6909	139	7	chosen	choose	VERB
fcis-6909	139	8	in	in	ADP
fcis-6909	139	9	steps	step	NOUN
fcis-6909	139	10	of	of	ADP
fcis-6909	139	11	0.02	0.02	NUM
fcis-6909	139	12	in	in	ADP
fcis-6909	139	13	{	{	PUNCT
fcis-6909	139	14	0.02	0.02	NUM
fcis-6909	139	15	,	,	PUNCT
fcis-6909	139	16	...	...	PUNCT
fcis-6909	139	17	,	,	PUNCT
fcis-6909	139	18	0.25	0.25	NUM
fcis-6909	139	19	}	}	PUNCT
fcis-6909	139	20	.	.	PUNCT
fcis-6909	140	1	decrs	decrs	PROPN
fcis-6909	141	1	[	[	X
fcis-6909	141	2	17	17	NUM
fcis-6909	141	3	]	]	PUNCT
fcis-6909	141	4	models	model	NOUN
fcis-6909	141	5	the	the	DET
fcis-6909	141	6	causal	causal	ADJ
fcis-6909	141	7	impact	impact	NOUN
fcis-6909	141	8	of	of	ADP
fcis-6909	141	9	user	user	NOUN
fcis-6909	141	10	representation	representation	NOUN
fcis-6909	141	11	on	on	ADP
fcis-6909	141	12	prediction	prediction	NOUN
fcis-6909	141	13	scores	score	NOUN
fcis-6909	141	14	.	.	PUNCT
fcis-6909	142	1	an	an	DET
fcis-6909	142	2	approximation	approximation	NOUN
fcis-6909	142	3	operator	operator	NOUN
fcis-6909	142	4	for	for	ADP
fcis-6909	142	5	backdoor	backdoor	NOUN
fcis-6909	142	6	adjustment	adjustment	NOUN
fcis-6909	142	7	is	be	AUX
fcis-6909	142	8	contributed	contribute	VERB
fcis-6909	142	9	,	,	PUNCT
fcis-6909	142	10	which	which	PRON
fcis-6909	142	11	can	can	AUX
fcis-6909	142	12	be	be	AUX
fcis-6909	142	13	easily	easily	ADV
fcis-6909	142	14	inserted	insert	VERB
fcis-6909	142	15	into	into	ADP
fcis-6909	142	16	most	most	ADJ
fcis-6909	142	17	recommendation	recommendation	NOUN
fcis-6909	142	18	models	model	NOUN
fcis-6909	142	19	.	.	PUNCT
fcis-6909	143	1	finally	finally	ADV
fcis-6909	143	2	,	,	PUNCT
fcis-6909	143	3	an	an	DET
fcis-6909	143	4	inference	inference	NOUN
fcis-6909	143	5	strategy	strategy	NOUN
fcis-6909	143	6	is	be	AUX
fcis-6909	143	7	designed	design	VERB
fcis-6909	143	8	to	to	PART
fcis-6909	143	9	dynamically	dynamically	ADV
fcis-6909	143	10	adjust	adjust	VERB
fcis-6909	143	11	the	the	DET
fcis-6909	143	12	backdoor	backdoor	NOUN
fcis-6909	143	13	adjustment	adjustment	NOUN
fcis-6909	143	14	according	accord	VERB
fcis-6909	143	15	to	to	ADP
fcis-6909	143	16	the	the	DET
fcis-6909	143	17	user	user	NOUN
fcis-6909	143	18	's	's	PART
fcis-6909	143	19	state	state	NOUN
fcis-6909	143	20	.	.	PUNCT
fcis-6909	144	1	4.3	4.3	NUM
fcis-6909	144	2	.	.	PUNCT
fcis-6909	145	1	overall	overall	ADJ
fcis-6909	145	2	performance	performance	NOUN
fcis-6909	145	3	comparison	comparison	NOUN
fcis-6909	145	4	(	(	PUNCT
fcis-6909	145	5	rq1	rq1	NOUN
fcis-6909	145	6	)	)	PUNCT
fcis-6909	145	7	table	table	NOUN
fcis-6909	145	8	2	2	NUM
fcis-6909	145	9	gives	give	VERB
fcis-6909	145	10	the	the	DET
fcis-6909	145	11	recommended	recommend	VERB
fcis-6909	145	12	performance	performance	NOUN
fcis-6909	145	13	of	of	ADP
fcis-6909	145	14	the	the	DET
fcis-6909	145	15	compared	compare	VERB
fcis-6909	145	16	methods	method	NOUN
fcis-6909	145	17	for	for	ADP
fcis-6909	145	18	hr@10	hr@10	NOUN
fcis-6909	145	19	,	,	PUNCT
fcis-6909	145	20	recall@20	recall@20	PROPN
fcis-6909	145	21	,	,	PUNCT
fcis-6909	145	22	ndcg@10	ndcg@10	PROPN
fcis-6909	145	23	,	,	PUNCT
fcis-6909	145	24	and	and	CCONJ
fcis-6909	145	25	ndcg@20	ndcg@20	PROPN
fcis-6909	145	26	.	.	PUNCT
fcis-6909	146	1	bold	bold	PROPN
fcis-6909	146	2	indicates	indicate	VERB
fcis-6909	146	3	the	the	DET
fcis-6909	146	4	best	good	ADJ
fcis-6909	146	5	results	result	NOUN
fcis-6909	146	6	,	,	PUNCT
fcis-6909	146	7	horizontal	horizontal	ADJ
fcis-6909	146	8	lines	line	NOUN
fcis-6909	146	9	are	be	AUX
fcis-6909	146	10	added	add	VERB
fcis-6909	146	11	under	under	ADP
fcis-6909	146	12	the	the	DET
fcis-6909	146	13	second	second	ADV
fcis-6909	146	14	-	-	PUNCT
fcis-6909	146	15	best	good	ADJ
fcis-6909	146	16	results	result	NOUN
fcis-6909	146	17	,	,	PUNCT
fcis-6909	146	18	and	and	CCONJ
fcis-6909	146	19	percentages	percentage	NOUN
fcis-6909	146	20	indicate	indicate	VERB
fcis-6909	146	21	how	how	SCONJ
fcis-6909	146	22	much	much	ADJ
fcis-6909	146	23	the	the	DET
fcis-6909	146	24	best	good	ADJ
fcis-6909	146	25	results	result	NOUN
fcis-6909	146	26	improve	improve	VERB
fcis-6909	146	27	performance	performance	NOUN
fcis-6909	146	28	compared	compare	VERB
fcis-6909	146	29	to	to	ADP
fcis-6909	146	30	the	the	DET
fcis-6909	146	31	second	second	ADV
fcis-6909	146	32	-	-	PUNCT
fcis-6909	146	33	best	good	ADJ
fcis-6909	146	34	results	result	NOUN
fcis-6909	146	35	.	.	PUNCT
fcis-6909	147	1	overall	overall	ADV
fcis-6909	147	2	,	,	PUNCT
fcis-6909	147	3	bdcr	bdcr	NOUN
fcis-6909	147	4	consistently	consistently	ADV
fcis-6909	147	5	outperforms	outperform	VERB
fcis-6909	147	6	the	the	DET
fcis-6909	147	7	other	other	ADJ
fcis-6909	147	8	methods	method	NOUN
fcis-6909	147	9	across	across	ADP
fcis-6909	147	10	all	all	DET
fcis-6909	147	11	metrics	metric	NOUN
fcis-6909	147	12	and	and	CCONJ
fcis-6909	147	13	all	all	DET
fcis-6909	147	14	data	datum	NOUN
fcis-6909	147	15	sets	set	NOUN
fcis-6909	147	16	.	.	PUNCT
fcis-6909	148	1	the	the	DET
fcis-6909	148	2	performance	performance	NOUN
fcis-6909	148	3	of	of	ADP
fcis-6909	148	4	fm	fm	PROPN
fcis-6909	148	5	and	and	CCONJ
fcis-6909	148	6	nfm	nfm	PROPN
fcis-6909	148	7	was	be	AUX
fcis-6909	148	8	improved	improve	VERB
fcis-6909	148	9	under	under	ADP
fcis-6909	148	10	decrs	decrs	PROPN
fcis-6909	148	11	,	,	PUNCT
fcis-6909	148	12	dice	dice	NOUN
fcis-6909	148	13	and	and	CCONJ
fcis-6909	148	14	pda	pda	NOUN
fcis-6909	148	15	methods	method	NOUN
fcis-6909	148	16	,	,	PUNCT
fcis-6909	148	17	which	which	PRON
fcis-6909	148	18	proves	prove	VERB
fcis-6909	148	19	the	the	DET
fcis-6909	148	20	importance	importance	NOUN
fcis-6909	148	21	of	of	ADP
fcis-6909	148	22	considering	consider	VERB
fcis-6909	148	23	bias	bias	NOUN
fcis-6909	148	24	on	on	ADP
fcis-6909	148	25	the	the	DET
fcis-6909	148	26	traditional	traditional	ADJ
fcis-6909	148	27	fm	fm	NOUN
fcis-6909	148	28	algorithm	algorithm	PROPN
fcis-6909	148	29	and	and	CCONJ
fcis-6909	148	30	nfm	nfm	PROPN
fcis-6909	148	31	algorithm	algorithm	NOUN
fcis-6909	148	32	.	.	PUNCT
fcis-6909	149	1	moreover	moreover	ADV
fcis-6909	149	2	,	,	PUNCT
fcis-6909	149	3	bdcr	bdcr	NOUN
fcis-6909	149	4	can	can	AUX
fcis-6909	149	5	be	be	AUX
fcis-6909	149	6	further	far	ADV
fcis-6909	149	7	improved	improve	VERB
fcis-6909	149	8	by	by	ADP
fcis-6909	149	9	adding	add	VERB
fcis-6909	149	10	more	more	ADJ
fcis-6909	149	11	complex	complex	ADJ
fcis-6909	149	12	models	model	NOUN
fcis-6909	149	13	.	.	PUNCT
fcis-6909	150	1	however	however	ADV
fcis-6909	150	2	,	,	PUNCT
fcis-6909	150	3	the	the	DET
fcis-6909	150	4	ips	ip	NOUN
fcis-6909	150	5	has	have	VERB
fcis-6909	150	6	worse	bad	ADJ
fcis-6909	150	7	performance	performance	NOUN
fcis-6909	150	8	relative	relative	ADJ
fcis-6909	150	9	to	to	ADP
fcis-6909	150	10	the	the	DET
fcis-6909	150	11	ordinary	ordinary	ADJ
fcis-6909	150	12	fm	fm	PROPN
fcis-6909	150	13	and	and	CCONJ
fcis-6909	150	14	nfm	nfm	PROPN
fcis-6909	150	15	.	.	PUNCT
fcis-6909	151	1	the	the	DET
fcis-6909	151	2	possible	possible	ADJ
fcis-6909	151	3	reason	reason	NOUN
fcis-6909	151	4	is	be	AUX
fcis-6909	151	5	that	that	SCONJ
fcis-6909	151	6	the	the	DET
fcis-6909	151	7	ips	ips	PROPN
fcis-6909	151	8	method	method	NOUN
fcis-6909	151	9	propensity	propensity	NOUN
fcis-6909	151	10	scores	score	NOUN
fcis-6909	151	11	are	be	AUX
fcis-6909	151	12	not	not	PART
fcis-6909	151	13	accurately	accurately	ADV
fcis-6909	151	14	obtained	obtain	VERB
fcis-6909	151	15	and	and	CCONJ
fcis-6909	151	16	face	face	VERB
fcis-6909	151	17	the	the	DET
fcis-6909	151	18	effect	effect	NOUN
fcis-6909	151	19	of	of	ADP
fcis-6909	151	20	high	high	ADJ
fcis-6909	151	21	variance	variance	NOUN
fcis-6909	151	22	,	,	PUNCT
fcis-6909	151	23	and	and	CCONJ
fcis-6909	151	24	such	such	ADJ
fcis-6909	151	25	propensity	propensity	NOUN
fcis-6909	151	26	scores	score	NOUN
fcis-6909	151	27	do	do	AUX
fcis-6909	151	28	not	not	PART
fcis-6909	151	29	accurately	accurately	ADV
fcis-6909	151	30	estimate	estimate	VERB
fcis-6909	151	31	the	the	DET
fcis-6909	151	32	effect	effect	NOUN
fcis-6909	151	33	of	of	ADP
fcis-6909	151	34	d	d	PROPN
fcis-6909	151	35	on	on	ADP
fcis-6909	151	36	u.	u.	NOUN
fcis-6909	151	37	table	table	NOUN
fcis-6909	152	1	2	2	NUM
fcis-6909	152	2	.	.	PUNCT
fcis-6909	152	3	overall	overall	ADJ
fcis-6909	152	4	performance	performance	NOUN
fcis-6909	152	5	comparison	comparison	NOUN
fcis-6909	152	6	backbone	backbone	NOUN
fcis-6909	152	7	model	model	NOUN
fcis-6909	152	8	algorithm	algorithm	PROPN
fcis-6909	152	9	ml-1	ml-1	PROPN
fcis-6909	152	10	m	m	VERB
fcis-6909	152	11	amazon	amazon	NOUN
fcis-6909	152	12	-	-	PUNCT
fcis-6909	152	13	book	book	NOUN
fcis-6909	152	14	r@10	r@10	NOUN
fcis-6909	152	15	r@20	r@20	NUM
fcis-6909	152	16	n@10	n@10	NUM
fcis-6909	152	17	n@20	n@20	X
fcis-6909	152	18	r@10	r@10	NOUN
fcis-6909	152	19	r@20	r@20	NUM
fcis-6909	152	20	n@10	n@10	NUM
fcis-6909	152	21	n@20	n@20	AUX
fcis-6909	152	22	fm	fm	PROPN
fcis-6909	152	23	mf	mf	PROPN
fcis-6909	152	24	/	/	SYM
fcis-6909	152	25	nfm	nfm	PROPN
fcis-6909	152	26	0.0676	0.0676	NUM
fcis-6909	152	27	0.1162	0.1162	NUM
fcis-6909	153	1	0.0566	0.0566	NUM
fcis-6909	153	2	0.0716	0.0716	NUM
fcis-6909	153	3	0.0231	0.0231	NUM
fcis-6909	153	4	0.0409	0.0409	NUM
fcis-6909	153	5	0.0148	0.0148	NUM
fcis-6909	153	6	0.0206	0.0206	NUM
fcis-6909	153	7	ips	ip	NOUN
fcis-6909	153	8	0.0663	0.0663	NUM
fcis-6909	153	9	0.1188	0.1188	NUM
fcis-6909	153	10	0.0556	0.0556	NUM
fcis-6909	153	11	0.0718	0.0718	NUM
fcis-6909	153	12	0.0213	0.0213	NUM
fcis-6909	153	13	0.0369	0.0369	NUM
fcis-6909	153	14	0.0135	0.0135	NUM
fcis-6909	153	15	0.0187	0.0187	NUM
fcis-6909	153	16	dice	dice	NOUN
fcis-6909	153	17	0.0779	0.0779	NUM
fcis-6909	153	18	0.1209	0.1209	NUM
fcis-6909	153	19	0.0575	0.0575	NUM
fcis-6909	153	20	0.0730	0.0730	NUM
fcis-6909	153	21	0.0216	0.0216	NUM
fcis-6909	153	22	0.0377	0.0377	NUM
fcis-6909	153	23	0.0138	0.0138	NUM
fcis-6909	153	24	0.0191	0.0191	NUM
fcis-6909	153	25	pda	pda	NOUN
fcis-6909	153	26	0.0870	0.0870	NUM
fcis-6909	153	27	0.1259	0.1259	NUM
fcis-6909	153	28	0.0589	0.0589	NUM
fcis-6909	153	29	0.0781	0.0781	NUM
fcis-6909	153	30	0.0251	0.0251	NUM
fcis-6909	153	31	0.0455	0.0455	NUM
fcis-6909	153	32	0.0168	0.0168	NUM
fcis-6909	153	33	0.0238	0.0238	NUM
fcis-6909	153	34	decrs	decrs	NOUN
fcis-6909	153	35	0.0888	0.0888	NUM
fcis-6909	153	36	0.1487	0.1487	NUM
fcis-6909	153	37	0.0621	0.0621	NUM
fcis-6909	153	38	0.0799	0.0799	NUM
fcis-6909	153	39	0.0264	0.0264	NUM
fcis-6909	153	40	0.0476	0.0476	NUM
fcis-6909	153	41	0.0170	0.0170	NUM
fcis-6909	153	42	0.0240	0.0240	NUM
fcis-6909	153	43	bdcr	bdcr	NOUN
fcis-6909	153	44	0.0917	0.0917	NUM
fcis-6909	153	45	0.1489	0.1489	NUM
fcis-6909	153	46	0.0656	0.0656	NUM
fcis-6909	153	47	0.0841	0.0841	NUM
fcis-6909	154	1	0.0273	0.0273	NUM
fcis-6909	154	2	0.0477	0.0477	NUM
fcis-6909	154	3	0.0174	0.0174	NUM
fcis-6909	154	4	0.0242	0.0242	NUM
fcis-6909	154	5	%	%	NOUN
fcis-6909	154	6	improve	improve	VERB
fcis-6909	154	7	3.2	3.2	NUM
fcis-6909	154	8	0.1	0.1	NUM
fcis-6909	154	9	5.6	5.6	NUM
fcis-6909	154	10	5.2	5.2	NUM
fcis-6909	154	11	3.4	3.4	NUM
fcis-6909	154	12	0.2	0.2	NUM
fcis-6909	154	13	2.3	2.3	NUM
fcis-6909	154	14	0.8	0.8	NUM
fcis-6909	154	15	nfm	nfm	PROPN
fcis-6909	154	16	mf	mf	PROPN
fcis-6909	154	17	/	/	SYM
fcis-6909	154	18	nfm	nfm	PROPN
fcis-6909	154	19	0.067	0.067	NUM
fcis-6909	154	20	0.1176	0.1176	NUM
fcis-6909	154	21	0.056	0.056	NUM
fcis-6909	154	22	0.0717	0.0717	NUM
fcis-6909	154	23	0.0228	0.0228	NUM
fcis-6909	154	24	0.0407	0.0407	NUM
fcis-6909	154	25	0.0148	0.0148	NUM
fcis-6909	154	26	0.0207	0.0207	NUM
fcis-6909	154	27	ips	ip	NOUN
fcis-6909	154	28	0.0648	0.0648	NUM
fcis-6909	154	29	0.1135	0.1135	NUM
fcis-6909	154	30	0.0544	0.0544	NUM
fcis-6909	154	31	0.0692	0.0692	NUM
fcis-6909	154	32	0.0213	0.0213	NUM
fcis-6909	154	33	0.0370	0.0370	NUM
fcis-6909	154	34	0.0137	0.0137	NUM
fcis-6909	154	35	0.0189	0.0189	NUM
fcis-6909	154	36	dice	dice	NOUN
fcis-6909	154	37	0.0848	0.0848	NUM
fcis-6909	154	38	0.1143	0.1143	NOUN
fcis-6909	154	39	0.0556	0.0556	NUM
fcis-6909	154	40	0.0708	0.0708	NUM
fcis-6909	154	41	0.0206	0.0206	NUM
fcis-6909	154	42	0.0381	0.0381	NUM
fcis-6909	154	43	0.0133	0.0133	NUM
fcis-6909	154	44	0.0190	0.0190	NUM
fcis-6909	154	45	pda	pda	NOUN
fcis-6909	154	46	0.0934	0.0934	NUM
fcis-6909	154	47	0.1333	0.1333	NUM
fcis-6909	154	48	0.0640	0.0640	NUM
fcis-6909	154	49	0.0807	0.0807	NUM
fcis-6909	154	50	0.0215	0.0215	NUM
fcis-6909	154	51	0.0434	0.0434	NUM
fcis-6909	154	52	0.0140	0.0140	NUM
fcis-6909	154	53	0.0197	0.0197	NUM
fcis-6909	154	54	decrs	decrs	PROPN
fcis-6909	154	55	0.0936	0.0936	NUM
fcis-6909	154	56	0.1478	0.1478	NUM
fcis-6909	154	57	0.0654	0.0654	NUM
fcis-6909	154	58	0.0855	0.0855	NUM
fcis-6909	154	59	0.0244	0.0244	NUM
fcis-6909	154	60	0.0401	0.0401	NUM
fcis-6909	154	61	0.0158	0.0158	NUM
fcis-6909	154	62	0.0214	0.0214	NUM
fcis-6909	154	63	bdcr	bdcr	NOUN
fcis-6909	154	64	0.094	0.094	NUM
fcis-6909	154	65	0.1505	0.1505	NUM
fcis-6909	154	66	0.0674	0.0674	NUM
fcis-6909	154	67	0.0857	0.0857	NUM
fcis-6909	154	68	0.0245	0.0245	NUM
fcis-6909	154	69	0.0435	0.0435	NUM
fcis-6909	154	70	0.0161	0.0161	NUM
fcis-6909	154	71	0.0224	0.0224	NUM
fcis-6909	154	72	%	%	NOUN
fcis-6909	154	73	improve	improve	VERB
fcis-6909	154	74	0.4	0.4	NUM
fcis-6909	154	75	1.8	1.8	NUM
fcis-6909	154	76	3.1	3.1	NUM
fcis-6909	154	77	0.2	0.2	NUM
fcis-6909	154	78	4.9	4.9	NUM
fcis-6909	154	79	0.2	0.2	NUM
fcis-6909	154	80	1.9	1.9	NUM
fcis-6909	154	81	4.6	4.6	NUM
fcis-6909	154	82	4.4	4.4	NUM
fcis-6909	154	83	.	.	PUNCT
fcis-6909	155	1	sensitivity	sensitivity	NOUN
fcis-6909	155	2	of	of	ADP
fcis-6909	155	3	hyper	hyper	NOUN
fcis-6909	155	4	-	-	NOUN
fcis-6909	155	5	parameters	parameter	NOUN
fcis-6909	155	6	(	(	PUNCT
fcis-6909	155	7	rq2	rq2	NOUN
fcis-6909	155	8	)	)	PUNCT
fcis-6909	155	9	the	the	DET
fcis-6909	155	10	data	datum	NOUN
fcis-6909	155	11	embedding	embed	VERB
fcis-6909	155	12	representation	representation	NOUN
fcis-6909	155	13	requires	require	VERB
fcis-6909	155	14	a	a	DET
fcis-6909	155	15	reasonable	reasonable	ADJ
fcis-6909	155	16	setting	setting	NOUN
fcis-6909	155	17	of	of	ADP
fcis-6909	155	18	the	the	DET
fcis-6909	155	19	item	item	NOUN
fcis-6909	155	20	feature	feature	NOUN
fcis-6909	155	21	values	value	NOUN
fcis-6909	155	22	,	,	PUNCT
fcis-6909	155	23	and	and	CCONJ
fcis-6909	155	24	in	in	ADP
fcis-6909	155	25	order	order	NOUN
fcis-6909	155	26	to	to	PART
fcis-6909	155	27	verify	verify	VERB
fcis-6909	155	28	the	the	DET
fcis-6909	155	29	impact	impact	NOUN
fcis-6909	155	30	of	of	ADP
fcis-6909	155	31	different	different	ADJ
fcis-6909	155	32	item	item	NOUN
fcis-6909	155	33	feature	feature	NOUN
fcis-6909	155	34	values	value	NOUN
fcis-6909	155	35	on	on	ADP
fcis-6909	155	36	the	the	DET
fcis-6909	155	37	model	model	NOUN
fcis-6909	155	38	,	,	PUNCT
fcis-6909	155	39	the	the	DET
fcis-6909	155	40	ml1	ml1	PROPN
fcis-6909	155	41	m	m	NOUN
fcis-6909	155	42	dataset	dataset	NOUN
fcis-6909	155	43	is	be	AUX
fcis-6909	155	44	used	use	VERB
fcis-6909	155	45	as	as	ADP
fcis-6909	155	46	an	an	DET
fcis-6909	155	47	example	example	NOUN
fcis-6909	155	48	,	,	PUNCT
fcis-6909	155	49	with	with	SCONJ
fcis-6909	155	50	other	other	ADJ
fcis-6909	155	51	parameters	parameter	NOUN
fcis-6909	155	52	kept	keep	VERB
fcis-6909	155	53	constant	constant	ADJ
fcis-6909	155	54	,	,	PUNCT
fcis-6909	155	55	and	and	CCONJ
fcis-6909	155	56	the	the	DET
fcis-6909	155	57	probabilities	probability	NOUN
fcis-6909	155	58	are	be	AUX
fcis-6909	155	59	divided	divide	VERB
fcis-6909	155	60	sequentially	sequentially	ADV
fcis-6909	155	61	based	base	VERB
fcis-6909	155	62	on	on	ADP
fcis-6909	155	63	the	the	DET
fcis-6909	155	64	total	total	ADJ
fcis-6909	155	65	number	number	NOUN
fcis-6909	155	66	of	of	ADP
fcis-6909	155	67	categories	category	NOUN
fcis-6909	155	68	to	to	PART
fcis-6909	155	69	which	which	PRON
fcis-6909	155	70	the	the	DET
fcis-6909	155	71	items	item	NOUN
fcis-6909	155	72	belong	belong	VERB
fcis-6909	155	73	,	,	PUNCT
fcis-6909	155	74	and	and	CCONJ
fcis-6909	155	75	a	a	DET
fcis-6909	155	76	total	total	NOUN
fcis-6909	155	77	of	of	ADP
fcis-6909	155	78	five	five	NUM
fcis-6909	155	79	division	division	NOUN
fcis-6909	155	80	schemes	scheme	NOUN
fcis-6909	155	81	are	be	AUX
fcis-6909	155	82	verified	verify	VERB
fcis-6909	155	83	as	as	SCONJ
fcis-6909	155	84	shown	show	VERB
fcis-6909	155	85	in	in	ADP
fcis-6909	155	86	table	table	NOUN
fcis-6909	155	87	3	3	NUM
fcis-6909	155	88	,	,	PUNCT
fcis-6909	155	89	with	with	ADP
fcis-6909	155	90	a	a	DET
fcis-6909	155	91	total	total	ADJ
fcis-6909	155	92	probability	probability	NOUN
fcis-6909	155	93	value	value	NOUN
fcis-6909	155	94	of	of	ADP
fcis-6909	155	95	1	1	NUM
fcis-6909	155	96	for	for	ADP
fcis-6909	155	97	each	each	DET
fcis-6909	155	98	item	item	NOUN
fcis-6909	155	99	,	,	PUNCT
fcis-6909	155	100	and	and	CCONJ
fcis-6909	155	101	table	table	NOUN
fcis-6909	155	102	4	4	NUM
fcis-6909	155	103	shows	show	VERB
fcis-6909	155	104	the	the	DET
fcis-6909	155	105	recall@10	recall@10	NOUN
fcis-6909	155	106	,	,	PUNCT
fcis-6909	155	107	recall@20	recall@20	PROPN
fcis-6909	155	108	,	,	PUNCT
fcis-6909	155	109	ndcg@10	ndcg@10	PROPN
fcis-6909	155	110	and	and	CCONJ
fcis-6909	155	111	ndcg@20	ndcg@20	PROPN
fcis-6909	155	112	on	on	ADP
fcis-6909	155	113	four	four	NUM
fcis-6909	155	114	indicators	indicator	NOUN
fcis-6909	155	115	.	.	PUNCT
fcis-6909	156	1	table	table	NOUN
fcis-6909	156	2	3	3	NUM
fcis-6909	156	3	.	.	PUNCT
fcis-6909	157	1	characteristic	characteristic	ADJ
fcis-6909	157	2	values	value	NOUN
fcis-6909	157	3	based	base	VERB
fcis-6909	157	4	on	on	ADP
fcis-6909	157	5	the	the	DET
fcis-6909	157	6	general	general	ADJ
fcis-6909	157	7	category	category	NOUN
fcis-6909	157	8	of	of	ADP
fcis-6909	157	9	the	the	DET
fcis-6909	157	10	project	project	NOUN
fcis-6909	157	11	a	a	DET
fcis-6909	157	12	class	class	NOUN
fcis-6909	157	13	two	two	NUM
fcis-6909	157	14	classes	class	NOUN
fcis-6909	157	15	three	three	NUM
fcis-6909	157	16	categories	category	NOUN
fcis-6909	157	17	four	four	NUM
fcis-6909	157	18	categories	category	NOUN
fcis-6909	157	19	five	five	NUM
fcis-6909	157	20	categories	category	NOUN
fcis-6909	157	21	six	six	NUM
fcis-6909	157	22	categories	category	NOUN
fcis-6909	157	23	option	option	NOUN
fcis-6909	157	24	1	1	NUM
fcis-6909	157	25	1	1	NUM
fcis-6909	157	26	7:3	7:3	NUM
fcis-6909	157	27	5:3:2	5:3:2	NUM
fcis-6909	157	28	5:3:1.5:0.5	5:3:1.5:0.5	NUM
fcis-6909	157	29	5:3:1:0.5:0.5	5:3:1:0.5:0.5	NUM
fcis-6909	157	30	4:3:1	4:3:1	NUM
fcis-6909	157	31	:	:	PUNCT
fcis-6909	157	32	1:0.5:0.5	1:0.5:0.5	NUM
fcis-6909	157	33	option	option	NOUN
fcis-6909	157	34	2	2	NUM
fcis-6909	157	35	1	1	NUM
fcis-6909	157	36	8:2	8:2	NUM
fcis-6909	157	37	8:2:0	8:2:0	NUM
fcis-6909	157	38	8:2:0:0	8:2:0:0	NUM
fcis-6909	157	39	8:2:0	8:2:0	NUM
fcis-6909	157	40	:	:	PUNCT
fcis-6909	157	41	0:0	0:0	NUM
fcis-6909	157	42	8:2:0	8:2:0	NUM
fcis-6909	157	43	:	:	PUNCT
fcis-6909	157	44	0:0:0	0:0:0	PRON
fcis-6909	157	45	option	option	NOUN
fcis-6909	157	46	3	3	NUM
fcis-6909	157	47	1	1	NUM
fcis-6909	157	48	7	7	NUM
fcis-6909	157	49	:	:	SYM
fcis-6909	157	50	3	3	NUM
fcis-6909	157	51	7:2	7:2	NUM
fcis-6909	157	52	:	:	PUNCT
fcis-6909	157	53	1	1	NUM
fcis-6909	157	54	7:2:1	7:2:1	NUM
fcis-6909	157	55	:	:	PUNCT
fcis-6909	157	56	0	0	NUM
fcis-6909	157	57	7:2	7:2	NUM
fcis-6909	157	58	:	:	PUNCT
fcis-6909	157	59	1	1	NUM
fcis-6909	157	60	:	:	SYM
fcis-6909	157	61	0:0	0:0	NUM
fcis-6909	158	1	7:2:1:0	7:2:1:0	NUM
fcis-6909	158	2	:	:	PUNCT
fcis-6909	158	3	0:0	0:0	NUM
fcis-6909	158	4	option	option	NOUN
fcis-6909	158	5	4	4	NUM
fcis-6909	158	6	1	1	NUM
fcis-6909	158	7	2:8	2:8	NUM
fcis-6909	158	8	2:8:0	2:8:0	NUM
fcis-6909	158	9	2:8:0:0	2:8:0:0	NUM
fcis-6909	158	10	2:8:0	2:8:0	NOUN
fcis-6909	158	11	:	:	PUNCT
fcis-6909	158	12	0:0	0:0	NUM
fcis-6909	158	13	2:8:0	2:8:0	NUM
fcis-6909	158	14	:	:	PUNCT
fcis-6909	158	15	0:0:0	0:0:0	PRON
fcis-6909	158	16	option	option	NOUN
fcis-6909	158	17	5	5	NUM
fcis-6909	158	18	1	1	NUM
fcis-6909	158	19	1:0	1:0	NUM
fcis-6909	158	20	1:0:0	1:0:0	NUM
fcis-6909	158	21	1:0:0	1:0:0	NUM
fcis-6909	158	22	:	:	PUNCT
fcis-6909	158	23	0:0	0:0	NUM
fcis-6909	158	24	1:0:0	1:0:0	NUM
fcis-6909	158	25	:	:	PUNCT
fcis-6909	159	1	0:0:0	0:0:0	NUM
fcis-6909	159	2	1:0:0	1:0:0	NUM
fcis-6909	159	3	:	:	PUNCT
fcis-6909	159	4	0:0:0	0:0:0	NUM
fcis-6909	159	5	table	table	NOUN
fcis-6909	159	6	4	4	NUM
fcis-6909	159	7	.	.	PUNCT
fcis-6909	159	8	indicators	indicator	NOUN
fcis-6909	159	9	for	for	ADP
fcis-6909	159	10	each	each	DET
fcis-6909	159	11	program	program	NOUN
fcis-6909	159	12	ml-1	ml-1	NOUN
fcis-6909	159	13	m	m	VERB
fcis-6909	159	14	r@10	r@10	NOUN
fcis-6909	159	15	r@20	r@20	PUNCT
fcis-6909	159	16	n@10	n@10	NUM
fcis-6909	159	17	n@20	n@20	PUNCT
fcis-6909	159	18	option	option	NOUN
fcis-6909	159	19	1	1	NUM
fcis-6909	159	20	0.0871	0.0871	NUM
fcis-6909	159	21	0.1426	0.1426	NUM
fcis-6909	159	22	0.0629	0.0629	NUM
fcis-6909	159	23	0.0805	0.0805	NUM
fcis-6909	159	24	option	option	NOUN
fcis-6909	159	25	2	2	NUM
fcis-6909	159	26	0.0911	0.0911	NUM
fcis-6909	160	1	0.1458	0.1458	NUM
fcis-6909	160	2	0.0656	0.0656	NUM
fcis-6909	160	3	0.0828	0.0828	NUM
fcis-6909	160	4	option	option	NOUN
fcis-6909	160	5	3	3	NUM
fcis-6909	160	6	0.0917	0.0917	NUM
fcis-6909	160	7	0.1489	0.1489	NUM
fcis-6909	160	8	0.0656	0.0656	NUM
fcis-6909	160	9	0.0841	0.0841	NUM
fcis-6909	160	10	option	option	NOUN
fcis-6909	160	11	4	4	NUM
fcis-6909	160	12	0.0935	0.0935	NUM
fcis-6909	160	13	0.1458	0.1458	NUM
fcis-6909	160	14	0.0659	0.0659	NUM
fcis-6909	160	15	0.0827	0.0827	NUM
fcis-6909	160	16	option	option	NOUN
fcis-6909	160	17	5	5	NUM
fcis-6909	160	18	0.0948	0.0948	NUM
fcis-6909	160	19	0.1498	0.1498	NUM
fcis-6909	160	20	0.0671	0.0671	NUM
fcis-6909	160	21	0.085	0.085	NUM
fcis-6909	160	22	from	from	ADP
fcis-6909	160	23	the	the	DET
fcis-6909	160	24	table	table	NOUN
fcis-6909	160	25	4	4	NUM
fcis-6909	160	26	,	,	PUNCT
fcis-6909	160	27	we	we	PRON
fcis-6909	160	28	can	can	AUX
fcis-6909	160	29	see	see	VERB
fcis-6909	160	30	that	that	DET
fcis-6909	160	31	option	option	NOUN
fcis-6909	160	32	5	5	NUM
fcis-6909	160	33	works	work	NOUN
fcis-6909	160	34	best	well	ADV
fcis-6909	160	35	,	,	PUNCT
fcis-6909	160	36	i.e.	i.e.	X
fcis-6909	160	37	,	,	PUNCT
fcis-6909	160	38	the	the	DET
fcis-6909	160	39	first	first	ADJ
fcis-6909	160	40	category	category	NOUN
fcis-6909	160	41	characteristic	characteristic	ADJ
fcis-6909	160	42	value	value	NOUN
fcis-6909	160	43	of	of	ADP
fcis-6909	160	44	all	all	DET
fcis-6909	160	45	items	item	NOUN
fcis-6909	160	46	is	be	AUX
fcis-6909	160	47	set	set	VERB
fcis-6909	160	48	to	to	ADP
fcis-6909	160	49	1	1	NUM
fcis-6909	160	50	.	.	PUNCT
fcis-6909	161	1	however	however	ADV
fcis-6909	161	2	,	,	PUNCT
fcis-6909	161	3	it	it	PRON
fcis-6909	161	4	will	will	AUX
fcis-6909	161	5	not	not	PART
fcis-6909	161	6	be	be	AUX
fcis-6909	161	7	set	set	VERB
fcis-6909	161	8	this	this	DET
fcis-6909	161	9	way	way	NOUN
fcis-6909	161	10	in	in	ADP
fcis-6909	161	11	practice	practice	NOUN
fcis-6909	161	12	,	,	PUNCT
fcis-6909	161	13	and	and	CCONJ
fcis-6909	161	14	focusing	focus	VERB
fcis-6909	161	15	only	only	ADV
fcis-6909	161	16	on	on	ADP
fcis-6909	161	17	the	the	DET
fcis-6909	161	18	first	first	ADJ
fcis-6909	161	19	category	category	NOUN
fcis-6909	161	20	will	will	AUX
fcis-6909	161	21	lose	lose	VERB
fcis-6909	161	22	the	the	DET
fcis-6909	161	23	original	original	ADJ
fcis-6909	161	24	meaning	meaning	NOUN
fcis-6909	161	25	of	of	ADP
fcis-6909	161	26	multiple	multiple	ADJ
fcis-6909	161	27	categories	category	NOUN
fcis-6909	161	28	.	.	PUNCT
fcis-6909	162	1	option	option	NOUN
fcis-6909	162	2	3	3	NUM
fcis-6909	162	3	and	and	CCONJ
fcis-6909	162	4	option	option	NOUN
fcis-6909	162	5	4	4	NUM
fcis-6909	162	6	have	have	VERB
fcis-6909	162	7	their	their	PRON
fcis-6909	162	8	advantages	advantage	NOUN
fcis-6909	162	9	and	and	CCONJ
fcis-6909	162	10	disadvantages	disadvantage	NOUN
fcis-6909	162	11	,	,	PUNCT
fcis-6909	162	12	but	but	CCONJ
fcis-6909	162	13	most	most	ADJ
fcis-6909	162	14	of	of	ADP
fcis-6909	162	15	the	the	DET
fcis-6909	162	16	indicators	indicator	NOUN
fcis-6909	162	17	in	in	ADP
fcis-6909	162	18	option	option	NOUN
fcis-6909	162	19	3	3	NUM
fcis-6909	162	20	are	be	AUX
fcis-6909	162	21	better	well	ADJ
fcis-6909	162	22	than	than	ADP
fcis-6909	162	23	option	option	NOUN
fcis-6909	162	24	4	4	NUM
fcis-6909	162	25	,	,	PUNCT
fcis-6909	162	26	and	and	CCONJ
fcis-6909	162	27	option	option	NOUN
fcis-6909	162	28	3	3	NUM
fcis-6909	162	29	will	will	AUX
fcis-6909	162	30	focus	focus	VERB
fcis-6909	162	31	on	on	ADP
fcis-6909	162	32	more	more	ADJ
fcis-6909	162	33	classes	class	NOUN
fcis-6909	162	34	,	,	PUNCT
fcis-6909	162	35	which	which	PRON
fcis-6909	162	36	helps	help	VERB
fcis-6909	162	37	to	to	PART
fcis-6909	162	38	discover	discover	VERB
fcis-6909	162	39	the	the	DET
fcis-6909	162	40	causality	causality	NOUN
fcis-6909	162	41	,	,	PUNCT
fcis-6909	162	42	so	so	ADV
fcis-6909	162	43	option	option	NOUN
fcis-6909	162	44	3	3	NUM
fcis-6909	162	45	is	be	AUX
fcis-6909	162	46	chosen	choose	VERB
fcis-6909	162	47	as	as	ADP
fcis-6909	162	48	the	the	DET
fcis-6909	162	49	final	final	ADJ
fcis-6909	162	50	setting	setting	NOUN
fcis-6909	162	51	in	in	ADP
fcis-6909	162	52	this	this	DET
fcis-6909	162	53	work	work	NOUN
fcis-6909	162	54	.	.	PUNCT
fcis-6909	163	1	14	14	NUM
fcis-6909	163	2	4.5	4.5	NUM
fcis-6909	163	3	.	.	PUNCT
fcis-6909	164	1	ablation	ablation	NOUN
fcis-6909	164	2	experiment	experiment	NOUN
fcis-6909	164	3	(	(	PUNCT
fcis-6909	164	4	rq3	rq3	NOUN
fcis-6909	164	5	)	)	PUNCT
fcis-6909	164	6	ablation	ablation	NOUN
fcis-6909	164	7	experiments	experiment	NOUN
fcis-6909	164	8	are	be	AUX
fcis-6909	164	9	performed	perform	VERB
fcis-6909	164	10	by	by	ADP
fcis-6909	164	11	partially	partially	ADV
fcis-6909	164	12	or	or	CCONJ
fcis-6909	164	13	completely	completely	ADV
fcis-6909	164	14	destroying	destroy	VERB
fcis-6909	164	15	specific	specific	ADJ
fcis-6909	164	16	components	component	NOUN
fcis-6909	164	17	in	in	ADP
fcis-6909	164	18	order	order	NOUN
fcis-6909	164	19	to	to	PART
fcis-6909	164	20	assess	assess	VERB
fcis-6909	164	21	their	their	PRON
fcis-6909	164	22	contribution	contribution	NOUN
fcis-6909	164	23	to	to	ADP
fcis-6909	164	24	the	the	DET
fcis-6909	164	25	overall	overall	ADJ
fcis-6909	164	26	system	system	NOUN
fcis-6909	164	27	.	.	PUNCT
fcis-6909	165	1	in	in	ADP
fcis-6909	165	2	the	the	DET
fcis-6909	165	3	case	case	NOUN
fcis-6909	165	4	of	of	ADP
fcis-6909	165	5	the	the	DET
fcis-6909	165	6	present	present	ADJ
fcis-6909	165	7	algorithm	algorithm	NOUN
fcis-6909	165	8	,	,	PUNCT
fcis-6909	165	9	it	it	PRON
fcis-6909	165	10	mainly	mainly	ADV
fcis-6909	165	11	includes	include	VERB
fcis-6909	165	12	a	a	DET
fcis-6909	165	13	denoising	denoise	VERB
fcis-6909	165	14	computational	computational	ADJ
fcis-6909	165	15	approach	approach	NOUN
fcis-6909	165	16	for	for	ADP
fcis-6909	165	17	false	false	ADJ
fcis-6909	165	18	positive	positive	ADJ
fcis-6909	165	19	feedback	feedback	NOUN
fcis-6909	165	20	and	and	CCONJ
fcis-6909	165	21	a	a	DET
fcis-6909	165	22	causal	causal	ADJ
fcis-6909	165	23	inference	inference	NOUN
fcis-6909	165	24	for	for	ADP
fcis-6909	165	25	bias	bias	NOUN
fcis-6909	165	26	amplification	amplification	NOUN
fcis-6909	165	27	,	,	PUNCT
fcis-6909	165	28	so	so	ADV
fcis-6909	165	29	two	two	NUM
fcis-6909	165	30	different	different	ADJ
fcis-6909	165	31	ablation	ablation	NOUN
fcis-6909	165	32	algorithms	algorithm	NOUN
fcis-6909	165	33	are	be	AUX
fcis-6909	165	34	designed	design	VERB
fcis-6909	165	35	for	for	ADP
fcis-6909	165	36	these	these	DET
fcis-6909	165	37	two	two	NUM
fcis-6909	165	38	components	component	NOUN
fcis-6909	165	39	,	,	PUNCT
fcis-6909	165	40	and	and	CCONJ
fcis-6909	165	41	the	the	DET
fcis-6909	165	42	backbone	backbone	NOUN
fcis-6909	165	43	model	model	NOUN
fcis-6909	165	44	for	for	ADP
fcis-6909	165	45	both	both	DET
fcis-6909	165	46	applications	application	NOUN
fcis-6909	165	47	is	be	AUX
fcis-6909	165	48	chosen	choose	VERB
fcis-6909	165	49	to	to	PART
fcis-6909	165	50	be	be	AUX
fcis-6909	165	51	fm	fm	PROPN
fcis-6909	165	52	.	.	PUNCT
fcis-6909	166	1	fm	fm	PROPN
fcis-6909	166	2	:	:	PUNCT
fcis-6909	166	3	as	as	SCONJ
fcis-6909	166	4	baseline	baseline	NOUN
fcis-6909	166	5	.	.	PUNCT
fcis-6909	167	1	bdcr	bdcr	NOUN
fcis-6909	167	2	/	/	PUNCT
fcis-6909	168	1	c	c	NOUN
fcis-6909	168	2	:	:	PUNCT
fcis-6909	168	3	after	after	ADP
fcis-6909	168	4	data	datum	NOUN
fcis-6909	168	5	denoising	denoising	NOUN
fcis-6909	168	6	,	,	PUNCT
fcis-6909	168	7	the	the	DET
fcis-6909	168	8	causal	causal	ADJ
fcis-6909	168	9	model	model	NOUN
fcis-6909	168	10	is	be	AUX
fcis-6909	168	11	not	not	PART
fcis-6909	168	12	considered	consider	VERB
fcis-6909	168	13	and	and	CCONJ
fcis-6909	168	14	sent	send	VERB
fcis-6909	168	15	back	back	ADV
fcis-6909	168	16	to	to	ADP
fcis-6909	168	17	fm	fm	NOUN
fcis-6909	168	18	directly	directly	ADV
fcis-6909	168	19	for	for	ADP
fcis-6909	168	20	calculation	calculation	NOUN
fcis-6909	168	21	.	.	PUNCT
fcis-6909	169	1	bdcr	bdcr	NOUN
fcis-6909	169	2	/	/	PROPN
fcis-6909	170	1	d	d	NOUN
fcis-6909	170	2	:	:	PUNCT
fcis-6909	170	3	on	on	ADP
fcis-6909	170	4	the	the	DET
fcis-6909	170	5	basis	basis	NOUN
fcis-6909	170	6	of	of	ADP
fcis-6909	170	7	bdcr	bdcr	NOUN
fcis-6909	170	8	without	without	ADP
fcis-6909	170	9	considering	consider	VERB
fcis-6909	170	10	the	the	DET
fcis-6909	170	11	effect	effect	NOUN
fcis-6909	170	12	of	of	ADP
fcis-6909	170	13	denoising	denoise	VERB
fcis-6909	170	14	on	on	ADP
fcis-6909	170	15	data	datum	NOUN
fcis-6909	170	16	purity	purity	NOUN
fcis-6909	170	17	,	,	PUNCT
fcis-6909	170	18	this	this	DET
fcis-6909	170	19	algorithm	algorithm	NOUN
fcis-6909	170	20	does	do	AUX
fcis-6909	170	21	not	not	PART
fcis-6909	170	22	truncate	truncate	VERB
fcis-6909	170	23	according	accord	VERB
fcis-6909	170	24	to	to	ADP
fcis-6909	170	25	the	the	DET
fcis-6909	170	26	loss	loss	NOUN
fcis-6909	170	27	value	value	NOUN
fcis-6909	170	28	after	after	ADP
fcis-6909	170	29	entering	enter	VERB
fcis-6909	170	30	the	the	DET
fcis-6909	170	31	fm	fm	PROPN
fcis-6909	170	32	model	model	NOUN
fcis-6909	170	33	in	in	ADP
fcis-6909	170	34	the	the	DET
fcis-6909	170	35	calculation	calculation	NOUN
fcis-6909	170	36	,	,	PUNCT
fcis-6909	170	37	and	and	CCONJ
fcis-6909	170	38	goes	go	VERB
fcis-6909	170	39	directly	directly	ADV
fcis-6909	170	40	to	to	ADP
fcis-6909	170	41	the	the	DET
fcis-6909	170	42	causal	causal	ADJ
fcis-6909	170	43	calculation	calculation	NOUN
fcis-6909	170	44	.	.	PUNCT
fcis-6909	171	1	table	table	NOUN
fcis-6909	171	2	5	5	NUM
fcis-6909	171	3	.	.	PUNCT
fcis-6909	171	4	comparison	comparison	NOUN
fcis-6909	171	5	of	of	ADP
fcis-6909	171	6	bdcr	bdcr	NOUN
fcis-6909	171	7	ablation	ablation	NOUN
fcis-6909	171	8	models	model	NOUN
fcis-6909	171	9	ml-1	ml-1	PART
fcis-6909	171	10	m	m	VERB
fcis-6909	171	11	amazon	amazon	NOUN
fcis-6909	171	12	-	-	PUNCT
fcis-6909	171	13	book	book	NOUN
fcis-6909	171	14	r@10	r@10	NOUN
fcis-6909	171	15	r@20	r@20	NUM
fcis-6909	171	16	n@10	n@10	NUM
fcis-6909	171	17	n@20	n@20	X
fcis-6909	171	18	r@10	r@10	NOUN
fcis-6909	171	19	r@20	r@20	NUM
fcis-6909	171	20	n@10	n@10	NUM
fcis-6909	171	21	n@20	n@20	INTJ
fcis-6909	171	22	fm	fm	PROPN
fcis-6909	171	23	0.0676	0.0676	NUM
fcis-6909	171	24	0.1162	0.1162	NUM
fcis-6909	171	25	0.0566	0.0566	NUM
fcis-6909	171	26	0.0716	0.0716	NUM
fcis-6909	171	27	0.0231	0.0231	NUM
fcis-6909	171	28	0.0409	0.0409	NUM
fcis-6909	171	29	0.0148	0.0148	NUM
fcis-6909	171	30	0.0206	0.0206	NUM
fcis-6909	171	31	bdcr	bdcr	NOUN
fcis-6909	171	32	/	/	SYM
fcis-6909	171	33	c	c	NOUN
fcis-6909	171	34	0.0688	0.0688	NUM
fcis-6909	171	35	0.1205	0.1205	NUM
fcis-6909	171	36	0.0572	0.0572	NUM
fcis-6909	171	37	0.073	0.073	NUM
fcis-6909	171	38	0.0226	0.0226	NUM
fcis-6909	171	39	0.0405	0.0405	NUM
fcis-6909	171	40	0.0148	0.0148	NUM
fcis-6909	171	41	0.0207	0.0207	NUM
fcis-6909	171	42	bdcr	bdcr	NOUN
fcis-6909	171	43	/	/	PUNCT
fcis-6909	172	1	d	d	PROPN
fcis-6909	172	2	0.0889	0.0889	NUM
fcis-6909	172	3	0.1478	0.1478	NUM
fcis-6909	172	4	0.0653	0.0653	NUM
fcis-6909	172	5	0.0842	0.0842	NUM
fcis-6909	172	6	0.0273	0.0273	NUM
fcis-6909	172	7	0.0469	0.0469	NUM
fcis-6909	172	8	0.0177	0.0177	NUM
fcis-6909	172	9	0.0243	0.0243	NUM
fcis-6909	172	10	bdcr	bdcr	VERB
fcis-6909	172	11	0.0917	0.0917	NUM
fcis-6909	172	12	0.1489	0.1489	NUM
fcis-6909	172	13	0.0656	0.0656	NUM
fcis-6909	172	14	0.0841	0.0841	NUM
fcis-6909	173	1	0.0273	0.0273	NUM
fcis-6909	173	2	0.0477	0.0477	NUM
fcis-6909	173	3	0.0174	0.0174	NUM
fcis-6909	173	4	0.0242	0.0242	NUM
fcis-6909	173	5	from	from	ADP
fcis-6909	173	6	the	the	DET
fcis-6909	173	7	table	table	NOUN
fcis-6909	173	8	5	5	NUM
fcis-6909	173	9	,	,	PUNCT
fcis-6909	173	10	it	it	PRON
fcis-6909	173	11	can	can	AUX
fcis-6909	173	12	be	be	AUX
fcis-6909	173	13	found	find	VERB
fcis-6909	173	14	that	that	SCONJ
fcis-6909	173	15	both	both	DET
fcis-6909	173	16	ablation	ablation	NOUN
fcis-6909	173	17	algorithms	algorithm	NOUN
fcis-6909	173	18	are	be	AUX
fcis-6909	173	19	improved	improve	VERB
fcis-6909	173	20	compared	compare	VERB
fcis-6909	173	21	to	to	ADP
fcis-6909	173	22	fm	fm	PROPN
fcis-6909	173	23	,	,	PUNCT
fcis-6909	173	24	and	and	CCONJ
fcis-6909	173	25	the	the	DET
fcis-6909	173	26	biggest	big	ADJ
fcis-6909	173	27	improvement	improvement	NOUN
fcis-6909	173	28	is	be	AUX
fcis-6909	173	29	bdcr	bdcr	NOUN
fcis-6909	173	30	/	/	SYM
fcis-6909	173	31	d	d	NOUN
fcis-6909	173	32	,	,	PUNCT
fcis-6909	173	33	i.e.	i.e.	X
fcis-6909	173	34	,	,	PUNCT
fcis-6909	173	35	the	the	DET
fcis-6909	173	36	elimination	elimination	NOUN
fcis-6909	173	37	of	of	ADP
fcis-6909	173	38	the	the	DET
fcis-6909	173	39	denoising	denoising	NOUN
fcis-6909	173	40	module	module	NOUN
fcis-6909	173	41	does	do	AUX
fcis-6909	173	42	not	not	PART
fcis-6909	173	43	have	have	VERB
fcis-6909	173	44	as	as	ADV
fcis-6909	173	45	big	big	ADJ
fcis-6909	173	46	an	an	DET
fcis-6909	173	47	effect	effect	NOUN
fcis-6909	173	48	on	on	ADP
fcis-6909	173	49	the	the	DET
fcis-6909	173	50	results	result	NOUN
fcis-6909	173	51	as	as	ADP
fcis-6909	173	52	the	the	DET
fcis-6909	173	53	causal	causal	NOUN
fcis-6909	173	54	module	module	NOUN
fcis-6909	173	55	,	,	PUNCT
fcis-6909	173	56	which	which	PRON
fcis-6909	173	57	indicates	indicate	VERB
fcis-6909	173	58	the	the	DET
fcis-6909	173	59	importance	importance	NOUN
fcis-6909	173	60	of	of	ADP
fcis-6909	173	61	causal	causal	ADJ
fcis-6909	173	62	reasoning	reasoning	NOUN
fcis-6909	173	63	from	from	ADP
fcis-6909	173	64	the	the	DET
fcis-6909	173	65	side	side	NOUN
fcis-6909	173	66	.	.	PUNCT
fcis-6909	174	1	despite	despite	SCONJ
fcis-6909	174	2	the	the	DET
fcis-6909	174	3	relatively	relatively	ADV
fcis-6909	174	4	highperformance	highperformance	NOUN
fcis-6909	174	5	improvement	improvement	NOUN
fcis-6909	174	6	of	of	ADP
fcis-6909	174	7	the	the	DET
fcis-6909	174	8	causal	causal	NOUN
fcis-6909	174	9	module	module	NOUN
fcis-6909	174	10	,	,	PUNCT
fcis-6909	174	11	the	the	DET
fcis-6909	174	12	experimental	experimental	ADJ
fcis-6909	174	13	results	result	NOUN
fcis-6909	174	14	were	be	AUX
fcis-6909	174	15	greatly	greatly	ADV
fcis-6909	174	16	improved	improve	VERB
fcis-6909	174	17	by	by	ADP
fcis-6909	174	18	adding	add	VERB
fcis-6909	174	19	denoising	denoising	NOUN
fcis-6909	174	20	,	,	PUNCT
fcis-6909	174	21	which	which	PRON
fcis-6909	174	22	also	also	ADV
fcis-6909	174	23	indicates	indicate	VERB
fcis-6909	174	24	that	that	SCONJ
fcis-6909	174	25	noise	noise	NOUN
fcis-6909	174	26	in	in	ADP
fcis-6909	174	27	the	the	DET
fcis-6909	174	28	data	data	NOUN
fcis-6909	174	29	has	have	VERB
fcis-6909	174	30	a	a	DET
fcis-6909	174	31	significant	significant	ADJ
fcis-6909	174	32	negative	negative	ADJ
fcis-6909	174	33	effect	effect	NOUN
fcis-6909	174	34	on	on	ADP
fcis-6909	174	35	the	the	DET
fcis-6909	174	36	recommendation	recommendation	NOUN
fcis-6909	174	37	.	.	PUNCT
fcis-6909	175	1	5	5	X
fcis-6909	175	2	.	.	X
fcis-6909	175	3	conclusion	conclusion	NOUN
fcis-6909	175	4	the	the	DET
fcis-6909	175	5	traditional	traditional	ADJ
fcis-6909	175	6	causal	causal	ADJ
fcis-6909	175	7	inference	inference	NOUN
fcis-6909	175	8	algorithm	algorithm	NOUN
fcis-6909	175	9	is	be	AUX
fcis-6909	175	10	improved	improve	VERB
fcis-6909	175	11	and	and	CCONJ
fcis-6909	175	12	the	the	DET
fcis-6909	175	13	bdcr	bdcr	NOUN
fcis-6909	175	14	algorithm	algorithm	NOUN
fcis-6909	175	15	is	be	AUX
fcis-6909	175	16	proposed	propose	VERB
fcis-6909	175	17	.	.	PUNCT
fcis-6909	176	1	firstly	firstly	ADV
fcis-6909	176	2	,	,	PUNCT
fcis-6909	176	3	an	an	DET
fcis-6909	176	4	implicit	implicit	ADJ
fcis-6909	176	5	feedback	feedback	NOUN
fcis-6909	176	6	to	to	PART
fcis-6909	176	7	display	display	VERB
fcis-6909	176	8	feedback	feedback	NOUN
fcis-6909	176	9	conversion	conversion	NOUN
fcis-6909	176	10	is	be	AUX
fcis-6909	176	11	used	use	VERB
fcis-6909	176	12	to	to	PART
fcis-6909	176	13	take	take	VERB
fcis-6909	176	14	an	an	DET
fcis-6909	176	15	eigenvalue	eigenvalue	ADJ
fcis-6909	176	16	track	track	NOUN
fcis-6909	176	17	for	for	ADP
fcis-6909	176	18	embedding	embed	VERB
fcis-6909	176	19	to	to	PART
fcis-6909	176	20	make	make	VERB
fcis-6909	176	21	it	it	PRON
fcis-6909	176	22	conform	conform	VERB
fcis-6909	176	23	to	to	ADP
fcis-6909	176	24	the	the	DET
fcis-6909	176	25	eigenvalue	eigenvalue	ADJ
fcis-6909	176	26	distribution	distribution	NOUN
fcis-6909	176	27	in	in	ADP
fcis-6909	176	28	realistic	realistic	ADJ
fcis-6909	176	29	situations	situation	NOUN
fcis-6909	176	30	;	;	PUNCT
fcis-6909	176	31	secondly	secondly	ADV
fcis-6909	176	32	,	,	PUNCT
fcis-6909	176	33	a	a	DET
fcis-6909	176	34	fusion	fusion	NOUN
fcis-6909	176	35	of	of	ADP
fcis-6909	176	36	denoising	denoising	NOUN
fcis-6909	176	37	and	and	CCONJ
fcis-6909	176	38	causal	causal	ADJ
fcis-6909	176	39	inference	inference	NOUN
fcis-6909	176	40	is	be	AUX
fcis-6909	176	41	used	use	VERB
fcis-6909	176	42	to	to	PART
fcis-6909	176	43	remove	remove	VERB
fcis-6909	176	44	the	the	DET
fcis-6909	176	45	implied	imply	VERB
fcis-6909	176	46	false	false	ADJ
fcis-6909	176	47	positive	positive	ADJ
fcis-6909	176	48	feedback	feedback	NOUN
fcis-6909	176	49	in	in	ADP
fcis-6909	176	50	the	the	DET
fcis-6909	176	51	dataset	dataset	NOUN
fcis-6909	176	52	and	and	CCONJ
fcis-6909	176	53	the	the	DET
fcis-6909	176	54	bias	bias	NOUN
fcis-6909	176	55	amplification	amplification	NOUN
fcis-6909	176	56	in	in	ADP
fcis-6909	176	57	the	the	DET
fcis-6909	176	58	recommendation	recommendation	NOUN
fcis-6909	176	59	process	process	NOUN
fcis-6909	176	60	,	,	PUNCT
fcis-6909	176	61	respectively	respectively	ADV
fcis-6909	176	62	;	;	PUNCT
fcis-6909	176	63	subsequently	subsequently	ADV
fcis-6909	176	64	,	,	PUNCT
fcis-6909	176	65	the	the	DET
fcis-6909	176	66	traditional	traditional	ADJ
fcis-6909	176	67	recommendation	recommendation	NOUN
fcis-6909	176	68	and	and	CCONJ
fcis-6909	176	69	the	the	DET
fcis-6909	176	70	proposed	propose	VERB
fcis-6909	176	71	algorithm	algorithm	NOUN
fcis-6909	176	72	are	be	AUX
fcis-6909	176	73	dynamically	dynamically	ADV
fcis-6909	176	74	combined	combine	VERB
fcis-6909	176	75	by	by	ADP
fcis-6909	176	76	measuring	measure	VERB
fcis-6909	176	77	the	the	DET
fcis-6909	176	78	user	user	NOUN
fcis-6909	176	79	interest	interest	NOUN
fcis-6909	176	80	drift	drift	NOUN
fcis-6909	176	81	;	;	PUNCT
fcis-6909	176	82	finally	finally	ADV
fcis-6909	176	83	,	,	PUNCT
fcis-6909	176	84	the	the	DET
fcis-6909	176	85	bdcr	bdcr	NOUN
fcis-6909	176	86	algorithm	algorithm	NOUN
fcis-6909	176	87	is	be	AUX
fcis-6909	176	88	tested	test	VERB
fcis-6909	176	89	on	on	ADP
fcis-6909	176	90	two	two	NUM
fcis-6909	176	91	finally	finally	ADV
fcis-6909	176	92	,	,	PUNCT
fcis-6909	176	93	the	the	DET
fcis-6909	176	94	superiority	superiority	NOUN
fcis-6909	176	95	of	of	ADP
fcis-6909	176	96	bdcr	bdcr	NOUN
fcis-6909	176	97	is	be	AUX
fcis-6909	176	98	verified	verify	VERB
fcis-6909	176	99	by	by	ADP
fcis-6909	176	100	comparing	compare	VERB
fcis-6909	176	101	it	it	PRON
fcis-6909	176	102	with	with	ADP
fcis-6909	176	103	five	five	NUM
fcis-6909	176	104	classical	classical	ADJ
fcis-6909	176	105	algorithms	algorithm	NOUN
fcis-6909	176	106	on	on	ADP
fcis-6909	176	107	two	two	NUM
fcis-6909	176	108	open	open	ADJ
fcis-6909	176	109	datasets	dataset	NOUN
fcis-6909	176	110	.	.	PUNCT
fcis-6909	177	1	references	reference	NOUN
fcis-6909	177	2	[	[	X
fcis-6909	177	3	1	1	NUM
fcis-6909	177	4	]	]	X
fcis-6909	177	5	dong	dong	PROPN
fcis-6909	177	6	z	z	PROPN
fcis-6909	177	7	,	,	PUNCT
fcis-6909	177	8	wang	wang	PROPN
fcis-6909	177	9	z	z	PROPN
fcis-6909	177	10	,	,	PUNCT
fcis-6909	177	11	xu	xu	PROPN
fcis-6909	177	12	j	j	PROPN
fcis-6909	177	13	,	,	PUNCT
fcis-6909	177	14	et	et	PROPN
fcis-6909	177	15	al	al	PROPN
fcis-6909	177	16	.	.	PUNCT
fcis-6909	178	1	a	a	DET
fcis-6909	178	2	brief	brief	ADJ
fcis-6909	178	3	history	history	NOUN
fcis-6909	178	4	of	of	ADP
fcis-6909	178	5	recommender	recommender	NOUN
fcis-6909	178	6	systems[z	systems[z	PROPN
fcis-6909	178	7	]	]	PUNCT
fcis-6909	178	8	.	.	PUNCT
fcis-6909	179	1	ithaca	ithaca	PROPN
fcis-6909	179	2	:	:	PUNCT
fcis-6909	179	3	cornell	cornell	PROPN
fcis-6909	179	4	university	university	PROPN
fcis-6909	179	5	library	library	NOUN
fcis-6909	179	6	,	,	PUNCT
fcis-6909	179	7	arxiv.org	arxiv.org	NOUN
fcis-6909	179	8	,	,	PUNCT
fcis-6909	179	9	2022	2022	NUM
fcis-6909	179	10	.	.	PUNCT
fcis-6909	180	1	[	[	X
fcis-6909	180	2	2	2	NUM
fcis-6909	180	3	]	]	X
fcis-6909	180	4	chen	chen	PROPN
fcis-6909	180	5	j	j	PROPN
fcis-6909	180	6	,	,	PUNCT
fcis-6909	180	7	dong	dong	PROPN
fcis-6909	180	8	h	h	PROPN
fcis-6909	180	9	,	,	PUNCT
fcis-6909	180	10	wang	wang	PROPN
fcis-6909	180	11	x	x	PROPN
fcis-6909	180	12	,	,	PUNCT
fcis-6909	180	13	et	et	PROPN
fcis-6909	180	14	al	al	PROPN
fcis-6909	180	15	.	.	PROPN
fcis-6909	180	16	bias	bias	PROPN
fcis-6909	180	17	and	and	CCONJ
fcis-6909	180	18	debias	debia	NOUN
fcis-6909	180	19	in	in	ADP
fcis-6909	180	20	recommender	recommender	NOUN
fcis-6909	180	21	system	system	NOUN
fcis-6909	180	22	:	:	PUNCT
fcis-6909	180	23	a	a	DET
fcis-6909	180	24	survey	survey	NOUN
fcis-6909	180	25	and	and	CCONJ
fcis-6909	180	26	future	future	ADJ
fcis-6909	180	27	directions[j	directions[j	PROPN
fcis-6909	180	28	]	]	PUNCT
fcis-6909	180	29	.	.	PUNCT
fcis-6909	181	1	2020	2020	NUM
fcis-6909	181	2	.	.	PUNCT
fcis-6909	182	1	[	[	X
fcis-6909	182	2	3	3	NUM
fcis-6909	182	3	]	]	X
fcis-6909	182	4	sato	sato	PROPN
fcis-6909	182	5	m	m	PRON
fcis-6909	182	6	,	,	PUNCT
fcis-6909	182	7	takemori	takemori	PROPN
fcis-6909	182	8	s	s	PROPN
fcis-6909	182	9	,	,	PUNCT
fcis-6909	182	10	singh	singh	PROPN
fcis-6909	182	11	j	j	PROPN
fcis-6909	182	12	,	,	PUNCT
fcis-6909	182	13	et	et	PROPN
fcis-6909	182	14	al	al	PROPN
fcis-6909	182	15	.	.	PROPN
fcis-6909	182	16	unbiased	unbiased	ADJ
fcis-6909	182	17	learning	learning	NOUN
fcis-6909	182	18	for	for	ADP
fcis-6909	182	19	the	the	DET
fcis-6909	182	20	causal	causal	ADJ
fcis-6909	182	21	effect	effect	NOUN
fcis-6909	182	22	of	of	ADP
fcis-6909	182	23	recommendation	recommendation	NOUN
fcis-6909	182	24	:	:	PUNCT
fcis-6909	182	25	proceedings	proceeding	NOUN
fcis-6909	182	26	of	of	ADP
fcis-6909	182	27	the	the	DET
fcis-6909	182	28	14th	14th	ADJ
fcis-6909	182	29	acm	acm	NOUN
fcis-6909	182	30	conference	conference	NOUN
fcis-6909	182	31	on	on	ADP
fcis-6909	182	32	recommender	recommender	PROPN
fcis-6909	182	33	systems[c	systems[c	PROPN
fcis-6909	182	34	]	]	X
fcis-6909	182	35	,	,	PUNCT
fcis-6909	182	36	2020	2020	NUM
fcis-6909	182	37	.	.	PUNCT
fcis-6909	183	1	[	[	X
fcis-6909	183	2	4	4	NUM
fcis-6909	183	3	]	]	X
fcis-6909	183	4	wei	wei	PROPN
fcis-6909	183	5	t	t	PROPN
fcis-6909	183	6	,	,	PUNCT
fcis-6909	183	7	feng	feng	PROPN
fcis-6909	183	8	f	f	PROPN
fcis-6909	183	9	,	,	PUNCT
fcis-6909	183	10	chen	chen	PROPN
fcis-6909	183	11	j	j	PROPN
fcis-6909	183	12	,	,	PUNCT
fcis-6909	183	13	et	et	PROPN
fcis-6909	183	14	al	al	PROPN
fcis-6909	183	15	.	.	PROPN
fcis-6909	183	16	model	model	ADJ
fcis-6909	183	17	-	-	ADJ
fcis-6909	183	18	agnostic	agnostic	ADJ
fcis-6909	183	19	counterfactual	counterfactual	ADJ
fcis-6909	183	20	reasoning	reasoning	NOUN
fcis-6909	183	21	for	for	ADP
fcis-6909	183	22	eliminating	eliminate	VERB
fcis-6909	183	23	popularity	popularity	NOUN
fcis-6909	183	24	bias	bias	NOUN
fcis-6909	183	25	in	in	ADP
fcis-6909	183	26	recommender	recommender	NOUN
fcis-6909	183	27	system	system	NOUN
fcis-6909	183	28	,	,	PUNCT
fcis-6909	183	29	ithaca	ithaca	PROPN
fcis-6909	183	30	,	,	PUNCT
fcis-6909	183	31	2021	2021	NUM
fcis-6909	183	32	.	.	PUNCT
fcis-6909	184	1	cornell	cornell	PROPN
fcis-6909	184	2	university	university	PROPN
fcis-6909	184	3	library	library	NOUN
fcis-6909	184	4	,	,	PUNCT
fcis-6909	184	5	arxiv.org	arxiv.org	NOUN
fcis-6909	184	6	,	,	PUNCT
fcis-6909	184	7	2021;2020	2021;2020	NUM
fcis-6909	184	8	.	.	PUNCT
fcis-6909	185	1	[	[	X
fcis-6909	185	2	5	5	NUM
fcis-6909	185	3	]	]	X
fcis-6909	185	4	tan	tan	PROPN
fcis-6909	185	5	j	j	PROPN
fcis-6909	185	6	,	,	PUNCT
fcis-6909	185	7	xu	xu	PROPN
fcis-6909	185	8	s	s	PROPN
fcis-6909	185	9	,	,	PUNCT
fcis-6909	185	10	ge	ge	PROPN
fcis-6909	185	11	y	y	PROPN
fcis-6909	185	12	,	,	PUNCT
fcis-6909	185	13	et	et	PROPN
fcis-6909	185	14	al	al	PROPN
fcis-6909	185	15	.	.	PUNCT
fcis-6909	185	16	counterfactual	counterfactual	ADJ
fcis-6909	185	17	explainable	explainable	ADJ
fcis-6909	185	18	recommendation	recommendation	NOUN
fcis-6909	185	19	:	:	PUNCT
fcis-6909	185	20	proceedings	proceeding	NOUN
fcis-6909	185	21	of	of	ADP
fcis-6909	185	22	the	the	DET
fcis-6909	185	23	30th	30th	ADJ
fcis-6909	185	24	acm	acm	PROPN
fcis-6909	185	25	international	international	ADJ
fcis-6909	185	26	conference	conference	NOUN
fcis-6909	185	27	on	on	ADP
fcis-6909	185	28	information	information	NOUN
fcis-6909	185	29	&	&	CCONJ
fcis-6909	185	30	knowledge	knowledge	PROPN
fcis-6909	185	31	management[c	management[c	PROPN
fcis-6909	185	32	]	]	PUNCT
fcis-6909	185	33	,	,	PUNCT
fcis-6909	185	34	2021	2021	NUM
fcis-6909	185	35	.	.	PUNCT
fcis-6909	186	1	[	[	X
fcis-6909	186	2	6	6	X
fcis-6909	186	3	]	]	X
fcis-6909	186	4	rubin	rubin	PROPN
fcis-6909	186	5	d	d	PROPN
fcis-6909	186	6	b.	b.	PROPN
fcis-6909	186	7	estimating	estimate	VERB
fcis-6909	186	8	causal	causal	ADJ
fcis-6909	186	9	effects	effect	NOUN
fcis-6909	186	10	of	of	ADP
fcis-6909	186	11	treatments	treatment	NOUN
fcis-6909	186	12	in	in	ADP
fcis-6909	186	13	randomized	randomized	ADJ
fcis-6909	186	14	and	and	CCONJ
fcis-6909	186	15	nonrandomized	nonrandomize	VERB
fcis-6909	186	16	studies.[j	studies.[j	PROPN
fcis-6909	186	17	]	]	PUNCT
fcis-6909	186	18	.	.	PUNCT
fcis-6909	187	1	journal	journal	PROPN
fcis-6909	187	2	of	of	ADP
fcis-6909	187	3	educational	educational	ADJ
fcis-6909	187	4	psychology	psychology	NOUN
fcis-6909	187	5	,	,	PUNCT
fcis-6909	187	6	1974,66(5	1974,66(5	PROPN
fcis-6909	187	7	):	):	PUNCT
fcis-6909	187	8	688	688	NUM
fcis-6909	187	9	.	.	PUNCT
fcis-6909	188	1	[	[	X
fcis-6909	188	2	7	7	X
fcis-6909	188	3	]	]	PUNCT
fcis-6909	188	4	pearl	pearl	NOUN
fcis-6909	188	5	j.	j.	PROPN
fcis-6909	188	6	causality[m	causality[m	PROPN
fcis-6909	188	7	]	]	PUNCT
fcis-6909	188	8	.	.	PUNCT
fcis-6909	189	1	cambridge	cambridge	PROPN
fcis-6909	189	2	university	university	PROPN
fcis-6909	189	3	press	press	NOUN
fcis-6909	189	4	,	,	PUNCT
fcis-6909	189	5	2009	2009	NUM
fcis-6909	189	6	.	.	PUNCT
fcis-6909	190	1	[	[	X
fcis-6909	190	2	8	8	NUM
fcis-6909	190	3	]	]	SYM
fcis-6909	190	4	pearl	pearl	NOUN
fcis-6909	190	5	j.	j.	PROPN
fcis-6909	190	6	causal	causal	PROPN
fcis-6909	190	7	inference	inference	PROPN
fcis-6909	190	8	in	in	ADP
fcis-6909	190	9	statistics	statistic	NOUN
fcis-6909	190	10	:	:	PUNCT
fcis-6909	190	11	an	an	DET
fcis-6909	190	12	overview[j	overview[j	PROPN
fcis-6909	190	13	]	]	PUNCT
fcis-6909	190	14	.	.	PUNCT
fcis-6909	191	1	2009	2009	NUM
fcis-6909	191	2	.	.	PUNCT
fcis-6909	192	1	[	[	X
fcis-6909	192	2	9	9	NUM
fcis-6909	192	3	]	]	PUNCT
fcis-6909	192	4	ball	ball	NOUN
fcis-6909	192	5	g	g	PROPN
fcis-6909	192	6	,	,	PUNCT
fcis-6909	192	7	breese	breese	PROPN
fcis-6909	192	8	j.	j.	PROPN
fcis-6909	192	9	emotion	emotion	PROPN
fcis-6909	192	10	and	and	CCONJ
fcis-6909	192	11	personality	personality	NOUN
fcis-6909	192	12	in	in	ADP
fcis-6909	192	13	a	a	DET
fcis-6909	192	14	conversational	conversational	ADJ
fcis-6909	192	15	character	character	NOUN
fcis-6909	192	16	:	:	PUNCT
fcis-6909	192	17	proceedings	proceeding	NOUN
fcis-6909	192	18	of	of	ADP
fcis-6909	192	19	the	the	DET
fcis-6909	192	20	workshop	workshop	NOUN
fcis-6909	192	21	on	on	ADP
fcis-6909	192	22	embodied	embody	VERB
fcis-6909	192	23	conversational	conversational	ADJ
fcis-6909	192	24	characters[c	characters[c	PROPN
fcis-6909	192	25	]	]	PUNCT
fcis-6909	192	26	,	,	PUNCT
fcis-6909	192	27	1998	1998	NUM
fcis-6909	192	28	.	.	PUNCT
fcis-6909	193	1	[	[	X
fcis-6909	193	2	10	10	NUM
fcis-6909	193	3	]	]	X
fcis-6909	193	4	zhang	zhang	PROPN
fcis-6909	193	5	w	w	PROPN
fcis-6909	193	6	,	,	PUNCT
fcis-6909	193	7	bao	bao	PROPN
fcis-6909	193	8	w	w	PROPN
fcis-6909	193	9	,	,	PUNCT
fcis-6909	193	10	liu	liu	PROPN
fcis-6909	193	11	x	x	PROPN
fcis-6909	193	12	,	,	PUNCT
fcis-6909	193	13	et	et	PROPN
fcis-6909	193	14	al	al	PROPN
fcis-6909	193	15	.	.	PUNCT
fcis-6909	194	1	large	large	ADJ
fcis-6909	194	2	-	-	PUNCT
fcis-6909	194	3	scale	scale	NOUN
fcis-6909	194	4	causal	causal	ADJ
fcis-6909	194	5	approaches	approach	NOUN
fcis-6909	194	6	to	to	ADP
fcis-6909	194	7	debiasing	debiase	VERB
fcis-6909	194	8	post	post	ADJ
fcis-6909	194	9	-	-	ADJ
fcis-6909	194	10	click	click	ADJ
fcis-6909	194	11	conversion	conversion	NOUN
fcis-6909	194	12	rate	rate	NOUN
fcis-6909	194	13	estimation	estimation	NOUN
fcis-6909	194	14	with	with	ADP
fcis-6909	194	15	multi	multi	ADJ
fcis-6909	194	16	-	-	NOUN
fcis-6909	194	17	task	task	ADJ
fcis-6909	194	18	learning	learning	NOUN
fcis-6909	194	19	:	:	PUNCT
fcis-6909	194	20	proceedings	proceeding	NOUN
fcis-6909	194	21	of	of	ADP
fcis-6909	194	22	the	the	DET
fcis-6909	194	23	web	web	NOUN
fcis-6909	194	24	conference	conference	NOUN
fcis-6909	194	25	2020[c	2020[c	NOUN
fcis-6909	194	26	]	]	X
fcis-6909	194	27	,	,	PUNCT
fcis-6909	194	28	2020	2020	NUM
fcis-6909	194	29	.	.	PUNCT
fcis-6909	195	1	[	[	X
fcis-6909	195	2	11	11	NUM
fcis-6909	195	3	]	]	X
fcis-6909	195	4	wang	wang	PROPN
fcis-6909	195	5	z	z	PROPN
fcis-6909	195	6	,	,	PUNCT
fcis-6909	195	7	shen	shen	PROPN
fcis-6909	195	8	s	s	PROPN
fcis-6909	195	9	,	,	PUNCT
fcis-6909	195	10	wang	wang	PROPN
fcis-6909	195	11	z	z	PROPN
fcis-6909	195	12	,	,	PUNCT
fcis-6909	195	13	et	et	PROPN
fcis-6909	195	14	al	al	PROPN
fcis-6909	195	15	.	.	PROPN
fcis-6909	195	16	unbiased	unbiased	ADJ
fcis-6909	195	17	sequential	sequential	ADJ
fcis-6909	195	18	recommendation	recommendation	NOUN
fcis-6909	195	19	with	with	ADP
fcis-6909	195	20	latent	latent	NOUN
fcis-6909	195	21	confounders	confounder	NOUN
fcis-6909	195	22	:	:	PUNCT
fcis-6909	195	23	proceedings	proceeding	NOUN
fcis-6909	195	24	of	of	ADP
fcis-6909	195	25	the	the	DET
fcis-6909	195	26	acm	acm	PROPN
fcis-6909	195	27	web	web	NOUN
fcis-6909	195	28	conference	conference	NOUN
fcis-6909	195	29	2022[c	2022[c	PROPN
fcis-6909	195	30	]	]	X
fcis-6909	195	31	,	,	PUNCT
fcis-6909	195	32	2022	2022	NUM
fcis-6909	195	33	.	.	PUNCT
fcis-6909	196	1	[	[	X
fcis-6909	196	2	12	12	NUM
fcis-6909	196	3	]	]	X
fcis-6909	196	4	cosley	cosley	PROPN
fcis-6909	196	5	d	d	PROPN
fcis-6909	196	6	,	,	PUNCT
fcis-6909	196	7	lam	lam	PROPN
fcis-6909	196	8	s	s	PROPN
fcis-6909	196	9	k	k	PROPN
fcis-6909	196	10	,	,	PUNCT
fcis-6909	196	11	albert	albert	PROPN
fcis-6909	196	12	i	i	PROPN
fcis-6909	196	13	,	,	PUNCT
fcis-6909	196	14	et	et	PROPN
fcis-6909	196	15	al	al	PROPN
fcis-6909	196	16	.	.	PROPN
fcis-6909	196	17	is	be	AUX
fcis-6909	196	18	seeing	see	VERB
fcis-6909	196	19	believing	believe	VERB
fcis-6909	196	20	?	?	PUNCT
fcis-6909	197	1	how	how	SCONJ
fcis-6909	197	2	recommender	recommender	NOUN
fcis-6909	197	3	system	system	NOUN
fcis-6909	197	4	interfaces	interface	NOUN
fcis-6909	197	5	affect	affect	VERB
fcis-6909	197	6	users	user	NOUN
fcis-6909	197	7	'	'	PART
fcis-6909	197	8	opinions	opinion	NOUN
fcis-6909	197	9	:	:	PUNCT
fcis-6909	197	10	proceedings	proceeding	NOUN
fcis-6909	197	11	of	of	ADP
fcis-6909	197	12	the	the	DET
fcis-6909	197	13	sigchi	sigchi	PROPN
fcis-6909	197	14	conference	conference	NOUN
fcis-6909	197	15	on	on	ADP
fcis-6909	197	16	human	human	ADJ
fcis-6909	197	17	factors	factor	NOUN
fcis-6909	197	18	in	in	ADP
fcis-6909	197	19	computing	compute	VERB
fcis-6909	197	20	systems[c	systems[c	PROPN
fcis-6909	197	21	]	]	X
fcis-6909	197	22	,	,	PUNCT
fcis-6909	197	23	2003	2003	NUM
fcis-6909	197	24	.	.	PUNCT
fcis-6909	198	1	[	[	X
fcis-6909	198	2	13	13	NUM
fcis-6909	198	3	]	]	PUNCT
fcis-6909	198	4	amatriain	amatriain	NOUN
fcis-6909	198	5	x	x	X
fcis-6909	198	6	,	,	PUNCT
fcis-6909	198	7	pujol	pujol	NOUN
fcis-6909	198	8	j	j	PROPN
fcis-6909	198	9	m	m	PROPN
fcis-6909	198	10	,	,	PUNCT
fcis-6909	198	11	oliver	oliver	PROPN
fcis-6909	198	12	n.	n.	PROPN
fcis-6909	199	1	i	i	PRON
fcis-6909	199	2	like	like	VERB
fcis-6909	199	3	it	it	PRON
fcis-6909	199	4	...	...	PUNCT
fcis-6909	200	1	i	i	PRON
fcis-6909	200	2	like	like	VERB
fcis-6909	200	3	it	it	PRON
fcis-6909	200	4	not	not	PART
fcis-6909	200	5	:	:	PUNCT
fcis-6909	200	6	evaluating	evaluate	VERB
fcis-6909	200	7	user	user	NOUN
fcis-6909	200	8	ratings	rating	NOUN
fcis-6909	200	9	noise	noise	NOUN
fcis-6909	200	10	in	in	ADP
fcis-6909	200	11	recommender	recommender	NOUN
fcis-6909	200	12	systems	system	NOUN
fcis-6909	200	13	:	:	PUNCT
fcis-6909	200	14	user	user	NOUN
fcis-6909	200	15	modeling	modeling	NOUN
fcis-6909	200	16	,	,	PUNCT
fcis-6909	200	17	adaptation	adaptation	NOUN
fcis-6909	200	18	,	,	PUNCT
fcis-6909	200	19	and	and	CCONJ
fcis-6909	200	20	personalization	personalization	NOUN
fcis-6909	200	21	:	:	PUNCT
fcis-6909	200	22	17th	17th	ADJ
fcis-6909	200	23	international	international	ADJ
fcis-6909	200	24	conference	conference	NOUN
fcis-6909	200	25	,	,	PUNCT
fcis-6909	200	26	umap	umap	ADJ
fcis-6909	200	27	2009	2009	NUM
fcis-6909	200	28	,	,	PUNCT
fcis-6909	200	29	formerly	formerly	ADV
fcis-6909	200	30	um	um	INTJ
fcis-6909	200	31	and	and	CCONJ
fcis-6909	200	32	ah	ah	INTJ
fcis-6909	200	33	,	,	PUNCT
fcis-6909	200	34	trento	trento	PROPN
fcis-6909	200	35	,	,	PUNCT
fcis-6909	200	36	italy	italy	PROPN
fcis-6909	200	37	,	,	PUNCT
fcis-6909	200	38	june	june	PROPN
fcis-6909	200	39	22	22	NUM
fcis-6909	200	40	-	-	SYM
fcis-6909	200	41	26	26	NUM
fcis-6909	200	42	,	,	PUNCT
fcis-6909	200	43	2009	2009	NUM
fcis-6909	200	44	.	.	PUNCT
fcis-6909	201	1	proceedings	proceeding	NOUN
fcis-6909	201	2	17[c	17[c	NUM
fcis-6909	201	3	]	]	NOUN
fcis-6909	201	4	:	:	PUNCT
fcis-6909	201	5	springer	springer	NOUN
fcis-6909	201	6	,	,	PUNCT
fcis-6909	201	7	2009	2009	NUM
fcis-6909	201	8	.	.	PUNCT
fcis-6909	202	1	[	[	X
fcis-6909	202	2	14	14	NUM
fcis-6909	202	3	]	]	X
fcis-6909	202	4	lu	lu	PROPN
fcis-6909	202	5	h	h	NOUN
fcis-6909	202	6	,	,	PUNCT
fcis-6909	202	7	zhang	zhang	PROPN
fcis-6909	202	8	m	m	PROPN
fcis-6909	202	9	,	,	PUNCT
fcis-6909	202	10	ma	ma	PROPN
fcis-6909	202	11	s.	s.	PROPN
fcis-6909	202	12	between	between	ADP
fcis-6909	202	13	clicks	click	NOUN
fcis-6909	202	14	and	and	CCONJ
fcis-6909	202	15	satisfaction	satisfaction	NOUN
fcis-6909	202	16	:	:	PUNCT
fcis-6909	202	17	study	study	NOUN
fcis-6909	202	18	on	on	ADP
fcis-6909	202	19	multi	multi	ADJ
fcis-6909	202	20	-	-	ADJ
fcis-6909	202	21	phase	phase	NOUN
fcis-6909	202	22	user	user	NOUN
fcis-6909	202	23	preferences	preference	NOUN
fcis-6909	202	24	and	and	CCONJ
fcis-6909	202	25	satisfaction	satisfaction	NOUN
fcis-6909	202	26	for	for	ADP
fcis-6909	202	27	online	online	ADJ
fcis-6909	202	28	news	news	NOUN
fcis-6909	202	29	reading	reading	NOUN
fcis-6909	202	30	:	:	PUNCT
fcis-6909	202	31	the	the	DET
fcis-6909	202	32	41st	41st	PROPN
fcis-6909	202	33	international	international	PROPN
fcis-6909	202	34	acm	acm	PROPN
fcis-6909	202	35	sigir	sigir	PROPN
fcis-6909	202	36	conference	conference	PROPN
fcis-6909	202	37	on	on	ADP
fcis-6909	202	38	research	research	PROPN
fcis-6909	202	39	&	&	CCONJ
fcis-6909	202	40	development	development	NOUN
fcis-6909	202	41	in	in	ADP
fcis-6909	202	42	information	information	NOUN
fcis-6909	202	43	retrieval[c	retrieval[c	PROPN
fcis-6909	202	44	]	]	X
fcis-6909	202	45	,	,	PUNCT
fcis-6909	202	46	2018	2018	NUM
fcis-6909	202	47	.	.	PUNCT
fcis-6909	203	1	[	[	X
fcis-6909	203	2	15	15	NUM
fcis-6909	203	3	]	]	X
fcis-6909	203	4	wen	wen	PROPN
fcis-6909	203	5	h	h	PROPN
fcis-6909	203	6	,	,	PUNCT
fcis-6909	203	7	yang	yang	PROPN
fcis-6909	203	8	l	l	PROPN
fcis-6909	203	9	,	,	PUNCT
fcis-6909	203	10	estrin	estrin	PROPN
fcis-6909	203	11	d.	d.	NOUN
fcis-6909	203	12	leveraging	leverage	VERB
fcis-6909	203	13	post	post	ADJ
fcis-6909	203	14	-	-	ADJ
fcis-6909	203	15	click	click	ADJ
fcis-6909	203	16	feedback	feedback	NOUN
fcis-6909	203	17	for	for	ADP
fcis-6909	203	18	content	content	NOUN
fcis-6909	203	19	recommendations	recommendation	NOUN
fcis-6909	203	20	:	:	PUNCT
fcis-6909	203	21	proceedings	proceeding	NOUN
fcis-6909	203	22	of	of	ADP
fcis-6909	203	23	the	the	DET
fcis-6909	203	24	13th	13th	NOUN
fcis-6909	203	25	acm	acm	NOUN
fcis-6909	203	26	conference	conference	NOUN
fcis-6909	203	27	on	on	ADP
fcis-6909	203	28	recommender	recommender	PROPN
fcis-6909	203	29	systems[c	systems[c	PROPN
fcis-6909	203	30	]	]	X
fcis-6909	203	31	,	,	PUNCT
fcis-6909	203	32	2019	2019	NUM
fcis-6909	203	33	.	.	PUNCT
fcis-6909	204	1	[	[	X
fcis-6909	204	2	16	16	NUM
fcis-6909	204	3	]	]	X
fcis-6909	204	4	wang	wang	PROPN
fcis-6909	204	5	w	w	PROPN
fcis-6909	204	6	,	,	PUNCT
fcis-6909	204	7	feng	feng	PROPN
fcis-6909	204	8	f	f	AUX
fcis-6909	204	9	,	,	PUNCT
fcis-6909	204	10	he	he	PRON
fcis-6909	204	11	x	x	NOUN
fcis-6909	204	12	,	,	PUNCT
fcis-6909	204	13	et	et	PROPN
fcis-6909	204	14	al	al	PROPN
fcis-6909	204	15	.	.	PUNCT
fcis-6909	205	1	denoising	denoise	VERB
fcis-6909	205	2	implicit	implicit	ADJ
fcis-6909	205	3	feedback	feedback	NOUN
fcis-6909	205	4	for	for	ADP
fcis-6909	205	5	recommendation	recommendation	NOUN
fcis-6909	205	6	:	:	PUNCT
fcis-6909	205	7	wsdm'21	wsdm'21	NOUN
fcis-6909	205	8	:	:	PUNCT
fcis-6909	205	9	acm	acm	PROPN
fcis-6909	205	10	international	international	ADJ
fcis-6909	205	11	conference	conference	NOUN
fcis-6909	205	12	on	on	ADP
fcis-6909	205	13	web	web	NOUN
fcis-6909	205	14	search	search	NOUN
fcis-6909	205	15	and	and	CCONJ
fcis-6909	205	16	data	datum	NOUN
fcis-6909	205	17	mining[c	mining[c	PROPN
fcis-6909	205	18	]	]	PUNCT
fcis-6909	205	19	,	,	PUNCT
fcis-6909	205	20	virtual	virtual	ADJ
fcis-6909	205	21	,	,	PUNCT
fcis-6909	205	22	online	online	PROPN
fcis-6909	205	23	,	,	PUNCT
fcis-6909	205	24	israel	israel	PROPN
fcis-6909	205	25	:	:	PUNCT
fcis-6909	205	26	association	association	NOUN
fcis-6909	205	27	for	for	ADP
fcis-6909	205	28	computing	computing	NOUN
fcis-6909	205	29	machinery	machinery	PROPN
fcis-6909	205	30	,	,	PUNCT
fcis-6909	205	31	inc	inc	PROPN
fcis-6909	205	32	,	,	PUNCT
fcis-6909	205	33	2021	2021	NUM
fcis-6909	205	34	.	.	PUNCT
fcis-6909	206	1	[	[	X
fcis-6909	206	2	17	17	NUM
fcis-6909	206	3	]	]	X
fcis-6909	206	4	wang	wang	PROPN
fcis-6909	206	5	w	w	PROPN
fcis-6909	206	6	,	,	PUNCT
fcis-6909	206	7	feng	feng	PROPN
fcis-6909	206	8	f	f	AUX
fcis-6909	206	9	,	,	PUNCT
fcis-6909	206	10	he	he	PRON
fcis-6909	206	11	x	x	NOUN
fcis-6909	206	12	,	,	PUNCT
fcis-6909	206	13	et	et	PROPN
fcis-6909	206	14	al	al	PROPN
fcis-6909	206	15	.	.	PROPN
fcis-6909	206	16	deconfounded	deconfounde	VERB
fcis-6909	206	17	recommendation	recommendation	NOUN
fcis-6909	206	18	for	for	ADP
fcis-6909	206	19	alleviating	alleviate	VERB
fcis-6909	206	20	bias	bias	NOUN
fcis-6909	206	21	amplification	amplification	NOUN
fcis-6909	206	22	,	,	PUNCT
fcis-6909	206	23	virtual	virtual	ADJ
fcis-6909	206	24	,	,	PUNCT
fcis-6909	206	25	online	online	ADJ
fcis-6909	206	26	,	,	PUNCT
fcis-6909	206	27	singapore	singapore	PROPN
fcis-6909	206	28	:	:	PUNCT
fcis-6909	206	29	association	association	NOUN
fcis-6909	206	30	for	for	ADP
fcis-6909	206	31	computing	compute	VERB
fcis-6909	206	32	machinery	machinery	NOUN
fcis-6909	206	33	,	,	PUNCT
fcis-6909	206	34	2021	2021	NUM
fcis-6909	206	35	.	.	PUNCT
fcis-6909	207	1	[	[	X
fcis-6909	207	2	18	18	NUM
fcis-6909	207	3	]	]	X
fcis-6909	207	4	schnabel	schnabel	PROPN
fcis-6909	207	5	t	t	PROPN
fcis-6909	207	6	,	,	PUNCT
fcis-6909	207	7	swaminathan	swaminathan	ADV
fcis-6909	207	8	a	a	X
fcis-6909	207	9	,	,	PUNCT
fcis-6909	207	10	singh	singh	PROPN
fcis-6909	207	11	a	a	X
fcis-6909	207	12	,	,	PUNCT
fcis-6909	207	13	et	et	PROPN
fcis-6909	207	14	al	al	PROPN
fcis-6909	207	15	.	.	PUNCT
fcis-6909	208	1	recommendations	recommendation	NOUN
fcis-6909	208	2	as	as	ADP
fcis-6909	208	3	treatments	treatment	NOUN
fcis-6909	208	4	:	:	PUNCT
fcis-6909	208	5	debiasing	debiase	VERB
fcis-6909	208	6	learning	learning	NOUN
fcis-6909	208	7	and	and	CCONJ
fcis-6909	208	8	evaluation	evaluation	NOUN
fcis-6909	208	9	:	:	PUNCT
fcis-6909	208	10	icml	icml	VERB
fcis-6909	208	11	2016[c	2016[c	NUM
fcis-6909	208	12	]	]	SYM
fcis-6909	208	13	,	,	PUNCT
fcis-6909	208	14	2016	2016	NUM
fcis-6909	208	15	.	.	PUNCT
fcis-6909	209	1	15	15	NUM
fcis-6909	210	1	[	[	SYM
fcis-6909	210	2	19	19	NUM
fcis-6909	210	3	]	]	X
fcis-6909	210	4	zheng	zheng	PROPN
fcis-6909	210	5	y	y	PROPN
fcis-6909	210	6	,	,	PUNCT
fcis-6909	210	7	gao	gao	PROPN
fcis-6909	210	8	c	c	PROPN
fcis-6909	210	9	,	,	PUNCT
fcis-6909	210	10	li	li	PROPN
fcis-6909	210	11	x	x	PROPN
fcis-6909	210	12	,	,	PUNCT
fcis-6909	210	13	et	et	PROPN
fcis-6909	210	14	al	al	PROPN
fcis-6909	210	15	.	.	PUNCT
fcis-6909	211	1	disentangling	disentangle	VERB
fcis-6909	211	2	user	user	NOUN
fcis-6909	211	3	interest	interest	NOUN
fcis-6909	211	4	and	and	CCONJ
fcis-6909	211	5	conformity	conformity	NOUN
fcis-6909	211	6	for	for	ADP
fcis-6909	211	7	recommendation	recommendation	NOUN
fcis-6909	211	8	with	with	ADP
fcis-6909	211	9	causal	causal	NOUN
fcis-6909	211	10	embedding	embedding	NOUN
fcis-6909	211	11	,	,	PUNCT
fcis-6909	211	12	ljubljana	ljubljana	PROPN
fcis-6909	211	13	,	,	PUNCT
fcis-6909	211	14	slovenia	slovenia	PROPN
fcis-6909	211	15	:	:	PUNCT
fcis-6909	211	16	association	association	NOUN
fcis-6909	211	17	for	for	ADP
fcis-6909	211	18	computing	computing	NOUN
fcis-6909	211	19	machinery	machinery	PROPN
fcis-6909	211	20	,	,	PUNCT
fcis-6909	211	21	inc	inc	PROPN
fcis-6909	211	22	,	,	PUNCT
fcis-6909	211	23	2021	2021	NUM
fcis-6909	211	24	.	.	PUNCT
fcis-6909	212	1	[	[	X
fcis-6909	212	2	20	20	NUM
fcis-6909	212	3	]	]	X
fcis-6909	212	4	zhang	zhang	PROPN
fcis-6909	212	5	y	y	PROPN
fcis-6909	212	6	,	,	PUNCT
fcis-6909	212	7	feng	feng	PROPN
fcis-6909	212	8	f	f	AUX
fcis-6909	212	9	,	,	PUNCT
fcis-6909	212	10	he	he	PRON
fcis-6909	212	11	x	x	NOUN
fcis-6909	212	12	,	,	PUNCT
fcis-6909	212	13	et	et	PROPN
fcis-6909	212	14	al	al	PROPN
fcis-6909	212	15	.	.	PROPN
fcis-6909	212	16	causal	causal	PROPN
fcis-6909	212	17	intervention	intervention	NOUN
fcis-6909	212	18	for	for	ADP
fcis-6909	212	19	leveraging	leverage	VERB
fcis-6909	212	20	popularity	popularity	NOUN
fcis-6909	212	21	bias	bias	NOUN
fcis-6909	212	22	in	in	ADP
fcis-6909	212	23	recommendation	recommendation	NOUN
fcis-6909	212	24	,	,	PUNCT
fcis-6909	212	25	virtual	virtual	ADJ
fcis-6909	212	26	,	,	PUNCT
fcis-6909	212	27	online	online	ADJ
fcis-6909	212	28	,	,	PUNCT
fcis-6909	212	29	canada	canada	PROPN
fcis-6909	212	30	:	:	PUNCT
fcis-6909	212	31	association	association	NOUN
fcis-6909	212	32	for	for	ADP
fcis-6909	212	33	computing	computing	NOUN
fcis-6909	212	34	machinery	machinery	PROPN
fcis-6909	212	35	,	,	PUNCT
fcis-6909	212	36	inc	inc	PROPN
fcis-6909	212	37	,	,	PUNCT
fcis-6909	212	38	2021	2021	NUM
fcis-6909	212	39	.	.	PUNCT