id	sid	tid	token	lemma	pos
ijassa-659	1	1	adv	adv	PROPN
ijassa-659	1	2	syst	syst	PROPN
ijassa-659	1	3	sci	sci	PROPN
ijassa-659	1	4	appl	appl	PROPN
ijassa-659	1	5	2019	2019	NUM
ijassa-659	1	6	;	;	PUNCT
ijassa-659	1	7	04	04	NUM
ijassa-659	1	8	;	;	PUNCT
ijassa-659	1	9	25	25	NUM
ijassa-659	1	10	-	-	SYM
ijassa-659	1	11	44	44	NUM
ijassa-659	1	12	published	publish	VERB
ijassa-659	1	13	online	online	ADV
ijassa-659	1	14	at	at	ADP
ijassa-659	1	15	http://ijassa.ipu.ru/index.php/ijassa/article/view/659	http://ijassa.ipu.ru/index.php/ijassa/article/view/659	NOUN
ijassa-659	1	16	improving	improve	VERB
ijassa-659	1	17	semi	semi	ADJ
ijassa-659	1	18	-	-	ADJ
ijassa-659	1	19	supervised	supervised	ADJ
ijassa-659	1	20	clustering	clustering	ADJ
ijassa-659	1	21	algorithms	algorithm	NOUN
ijassa-659	1	22	with	with	ADP
ijassa-659	1	23	active	active	ADJ
ijassa-659	1	24	query	query	NOUN
ijassa-659	1	25	selection	selection	PROPN
ijassa-659	1	26	walid	walid	PROPN
ijassa-659	1	27	atwa*1	atwa*1	PROPN
ijassa-659	1	28	,	,	PUNCT
ijassa-659	1	29	mahmoud	mahmoud	PROPN
ijassa-659	1	30	emam2	emam2	NOUN
ijassa-659	1	31	1	1	X
ijassa-659	1	32	)	)	PUNCT
ijassa-659	1	33	computer	computer	NOUN
ijassa-659	1	34	science	science	NOUN
ijassa-659	1	35	department	department	PROPN
ijassa-659	1	36	,	,	PUNCT
ijassa-659	1	37	faculty	faculty	NOUN
ijassa-659	1	38	of	of	ADP
ijassa-659	1	39	computers	computer	NOUN
ijassa-659	1	40	and	and	CCONJ
ijassa-659	1	41	information	information	NOUN
ijassa-659	1	42	,	,	PUNCT
ijassa-659	1	43	menoufia	menoufia	PROPN
ijassa-659	1	44	university	university	PROPN
ijassa-659	1	45	,	,	PUNCT
ijassa-659	1	46	32511	32511	NUM
ijassa-659	1	47	,	,	PUNCT
ijassa-659	1	48	egypt	egypt	PROPN
ijassa-659	1	49	e	e	PROPN
ijassa-659	1	50	-	-	NOUN
ijassa-659	1	51	mail	mail	NOUN
ijassa-659	1	52	:	:	PUNCT
ijassa-659	1	53	walid.atwa@ci.menofia.edu.eg	walid.atwa@ci.menofia.edu.eg	VERB
ijassa-659	1	54	2	2	NUM
ijassa-659	1	55	)	)	PUNCT
ijassa-659	1	56	mathematics	mathematic	NOUN
ijassa-659	1	57	and	and	CCONJ
ijassa-659	1	58	computer	computer	NOUN
ijassa-659	1	59	science	science	PROPN
ijassa-659	1	60	department	department	PROPN
ijassa-659	1	61	,	,	PUNCT
ijassa-659	1	62	faculty	faculty	NOUN
ijassa-659	1	63	of	of	ADP
ijassa-659	1	64	science	science	NOUN
ijassa-659	1	65	,	,	PUNCT
ijassa-659	1	66	menoufia	menoufia	PROPN
ijassa-659	1	67	university	university	PROPN
ijassa-659	1	68	,	,	PUNCT
ijassa-659	1	69	32511	32511	NUM
ijassa-659	1	70	,	,	PUNCT
ijassa-659	1	71	egypt	egypt	PROPN
ijassa-659	1	72	e	e	PROPN
ijassa-659	1	73	-	-	NOUN
ijassa-659	1	74	mail	mail	NOUN
ijassa-659	1	75	:	:	PUNCT
ijassa-659	1	76	ma7moud_emam@yahoo.com	ma7moud_emam@yahoo.com	PROPN
ijassa-659	1	77	received	receive	VERB
ijassa-659	1	78	october	october	PROPN
ijassa-659	1	79	30	30	NUM
ijassa-659	1	80	,	,	PUNCT
ijassa-659	1	81	2018	2018	NUM
ijassa-659	1	82	;	;	PUNCT
ijassa-659	1	83	revised	revise	VERB
ijassa-659	1	84	april	april	PROPN
ijassa-659	1	85	3	3	NUM
ijassa-659	1	86	,	,	PUNCT
ijassa-659	1	87	2019	2019	NUM
ijassa-659	1	88	;	;	PUNCT
ijassa-659	1	89	published	publish	VERB
ijassa-659	1	90	december	december	PROPN
ijassa-659	1	91	31	31	NUM
ijassa-659	1	92	,	,	PUNCT
ijassa-659	1	93	2019	2019	NUM
ijassa-659	1	94	abstract	abstract	NOUN
ijassa-659	1	95	:	:	PUNCT
ijassa-659	1	96	semi	semi	ADJ
ijassa-659	1	97	-	-	ADJ
ijassa-659	1	98	supervised	supervised	ADJ
ijassa-659	1	99	clustering	clustering	NOUN
ijassa-659	1	100	algorithms	algorithms	NOUN
ijassa-659	1	101	use	use	VERB
ijassa-659	1	102	a	a	DET
ijassa-659	1	103	small	small	ADJ
ijassa-659	1	104	amount	amount	NOUN
ijassa-659	1	105	of	of	ADP
ijassa-659	1	106	supervised	supervised	ADJ
ijassa-659	1	107	data	datum	NOUN
ijassa-659	1	108	in	in	ADP
ijassa-659	1	109	the	the	DET
ijassa-659	1	110	form	form	NOUN
ijassa-659	1	111	of	of	ADP
ijassa-659	1	112	pairwise	pairwise	NOUN
ijassa-659	1	113	constraints	constraint	NOUN
ijassa-659	1	114	to	to	PART
ijassa-659	1	115	improve	improve	VERB
ijassa-659	1	116	the	the	DET
ijassa-659	1	117	clustering	clustering	ADJ
ijassa-659	1	118	performance	performance	NOUN
ijassa-659	1	119	.	.	PUNCT
ijassa-659	2	1	however	however	ADV
ijassa-659	2	2	,	,	PUNCT
ijassa-659	2	3	most	most	ADJ
ijassa-659	2	4	current	current	ADJ
ijassa-659	2	5	algorithms	algorithm	NOUN
ijassa-659	2	6	are	be	AUX
ijassa-659	2	7	passive	passive	ADJ
ijassa-659	2	8	in	in	ADP
ijassa-659	2	9	the	the	DET
ijassa-659	2	10	sense	sense	NOUN
ijassa-659	2	11	that	that	SCONJ
ijassa-659	2	12	the	the	DET
ijassa-659	2	13	pairwise	pairwise	NOUN
ijassa-659	2	14	constraints	constraint	NOUN
ijassa-659	2	15	are	be	AUX
ijassa-659	2	16	provided	provide	VERB
ijassa-659	2	17	beforehand	beforehand	ADV
ijassa-659	2	18	and	and	CCONJ
ijassa-659	2	19	selected	select	VERB
ijassa-659	2	20	randomly	randomly	ADV
ijassa-659	2	21	.	.	PUNCT
ijassa-659	3	1	this	this	PRON
ijassa-659	3	2	may	may	AUX
ijassa-659	3	3	lead	lead	VERB
ijassa-659	3	4	to	to	ADP
ijassa-659	3	5	the	the	DET
ijassa-659	3	6	use	use	NOUN
ijassa-659	3	7	of	of	ADP
ijassa-659	3	8	constraints	constraint	NOUN
ijassa-659	3	9	that	that	PRON
ijassa-659	3	10	are	be	AUX
ijassa-659	3	11	redundant	redundant	ADJ
ijassa-659	3	12	,	,	PUNCT
ijassa-659	3	13	unnecessary	unnecessary	ADJ
ijassa-659	3	14	,	,	PUNCT
ijassa-659	3	15	or	or	CCONJ
ijassa-659	3	16	even	even	ADV
ijassa-659	3	17	harmful	harmful	ADJ
ijassa-659	3	18	to	to	ADP
ijassa-659	3	19	the	the	DET
ijassa-659	3	20	clustering	clustering	ADJ
ijassa-659	3	21	results	result	NOUN
ijassa-659	3	22	.	.	PUNCT
ijassa-659	4	1	in	in	ADP
ijassa-659	4	2	this	this	DET
ijassa-659	4	3	paper	paper	NOUN
ijassa-659	4	4	,	,	PUNCT
ijassa-659	4	5	addressing	address	VERB
ijassa-659	4	6	the	the	DET
ijassa-659	4	7	problem	problem	NOUN
ijassa-659	4	8	of	of	ADP
ijassa-659	4	9	constraint	constraint	NOUN
ijassa-659	4	10	selection	selection	NOUN
ijassa-659	4	11	to	to	PART
ijassa-659	4	12	improve	improve	VERB
ijassa-659	4	13	the	the	DET
ijassa-659	4	14	performance	performance	NOUN
ijassa-659	4	15	of	of	ADP
ijassa-659	4	16	semisupervised	semisupervised	ADJ
ijassa-659	4	17	clustering	clustering	ADJ
ijassa-659	4	18	algorithms	algorithm	NOUN
ijassa-659	4	19	.	.	PUNCT
ijassa-659	5	1	based	base	VERB
ijassa-659	5	2	on	on	ADP
ijassa-659	5	3	the	the	DET
ijassa-659	5	4	concepts	concept	NOUN
ijassa-659	5	5	of	of	ADP
ijassa-659	5	6	maximum	maximum	ADJ
ijassa-659	5	7	mean	mean	NOUN
ijassa-659	5	8	discrepancy	discrepancy	NOUN
ijassa-659	5	9	,	,	PUNCT
ijassa-659	5	10	proposed	propose	VERB
ijassa-659	5	11	method	method	NOUN
ijassa-659	5	12	selects	select	VERB
ijassa-659	5	13	a	a	DET
ijassa-659	5	14	batch	batch	NOUN
ijassa-659	5	15	of	of	ADP
ijassa-659	5	16	most	most	ADV
ijassa-659	5	17	informative	informative	ADJ
ijassa-659	5	18	instances	instance	NOUN
ijassa-659	5	19	that	that	PRON
ijassa-659	5	20	minimize	minimize	VERB
ijassa-659	5	21	the	the	DET
ijassa-659	5	22	difference	difference	NOUN
ijassa-659	5	23	in	in	ADP
ijassa-659	5	24	distribution	distribution	NOUN
ijassa-659	5	25	between	between	ADP
ijassa-659	5	26	the	the	DET
ijassa-659	5	27	labeled	label	VERB
ijassa-659	5	28	and	and	CCONJ
ijassa-659	5	29	unlabeled	unlabeled	ADJ
ijassa-659	5	30	data	datum	NOUN
ijassa-659	5	31	.	.	PUNCT
ijassa-659	6	1	then	then	ADV
ijassa-659	6	2	,	,	PUNCT
ijassa-659	6	3	querying	query	VERB
ijassa-659	6	4	these	these	DET
ijassa-659	6	5	instances	instance	NOUN
ijassa-659	6	6	with	with	ADP
ijassa-659	6	7	the	the	DET
ijassa-659	6	8	existing	exist	VERB
ijassa-659	6	9	neighborhoods	neighborhood	NOUN
ijassa-659	6	10	to	to	PART
ijassa-659	6	11	determine	determine	VERB
ijassa-659	6	12	which	which	DET
ijassa-659	6	13	neighborhood	neighborhood	NOUN
ijassa-659	6	14	they	they	PRON
ijassa-659	6	15	belong	belong	VERB
ijassa-659	6	16	.	.	PUNCT
ijassa-659	7	1	the	the	DET
ijassa-659	7	2	experimental	experimental	ADJ
ijassa-659	7	3	results	result	NOUN
ijassa-659	7	4	with	with	ADP
ijassa-659	7	5	state	state	NOUN
ijassa-659	7	6	-	-	PUNCT
ijassa-659	7	7	of	of	ADP
ijassa-659	7	8	-	-	PUNCT
ijassa-659	7	9	the	the	DET
ijassa-659	7	10	-	-	PUNCT
ijassa-659	7	11	art	art	NOUN
ijassa-659	7	12	methods	method	NOUN
ijassa-659	7	13	on	on	ADP
ijassa-659	7	14	different	different	ADJ
ijassa-659	7	15	real	real	ADJ
ijassa-659	7	16	-	-	PUNCT
ijassa-659	7	17	world	world	NOUN
ijassa-659	7	18	dataset	dataset	NOUN
ijassa-659	7	19	demonstrate	demonstrate	VERB
ijassa-659	7	20	the	the	DET
ijassa-659	7	21	effectiveness	effectiveness	NOUN
ijassa-659	7	22	and	and	CCONJ
ijassa-659	7	23	efficiency	efficiency	NOUN
ijassa-659	7	24	of	of	ADP
ijassa-659	7	25	the	the	DET
ijassa-659	7	26	proposed	propose	VERB
ijassa-659	7	27	method	method	NOUN
ijassa-659	7	28	.	.	PUNCT
ijassa-659	8	1	keywords	keyword	NOUN
ijassa-659	8	2	:	:	PUNCT
ijassa-659	8	3	semi	semi	ADJ
ijassa-659	8	4	-	-	ADJ
ijassa-659	8	5	supervised	supervised	ADJ
ijassa-659	8	6	clustering	clustering	NOUN
ijassa-659	8	7	,	,	PUNCT
ijassa-659	8	8	active	active	ADJ
ijassa-659	8	9	learning	learning	NOUN
ijassa-659	8	10	,	,	PUNCT
ijassa-659	8	11	pairwise	pairwise	NOUN
ijassa-659	8	12	constraints	constraint	NOUN
ijassa-659	8	13	.	.	PUNCT
ijassa-659	9	1	1	1	X
ijassa-659	9	2	.	.	X
ijassa-659	9	3	introduction	introduction	NOUN
ijassa-659	9	4	recently	recently	ADV
ijassa-659	9	5	,	,	PUNCT
ijassa-659	9	6	semi	semi	ADJ
ijassa-659	9	7	-	-	ADJ
ijassa-659	9	8	supervised	supervised	ADJ
ijassa-659	9	9	clustering	clustering	NOUN
ijassa-659	9	10	(	(	PUNCT
ijassa-659	9	11	also	also	ADV
ijassa-659	9	12	known	know	VERB
ijassa-659	9	13	as	as	ADP
ijassa-659	9	14	constraint	constraint	NOUN
ijassa-659	9	15	-	-	PUNCT
ijassa-659	9	16	based	base	VERB
ijassa-659	9	17	clustering	clustering	NOUN
ijassa-659	9	18	)	)	PUNCT
ijassa-659	9	19	algorithms	algorithm	NOUN
ijassa-659	9	20	have	have	AUX
ijassa-659	9	21	become	become	VERB
ijassa-659	9	22	a	a	DET
ijassa-659	9	23	topic	topic	NOUN
ijassa-659	9	24	of	of	ADP
ijassa-659	9	25	significant	significant	ADJ
ijassa-659	9	26	interest	interest	NOUN
ijassa-659	9	27	for	for	ADP
ijassa-659	9	28	many	many	ADJ
ijassa-659	9	29	researchers	researcher	NOUN
ijassa-659	9	30	.	.	PUNCT
ijassa-659	10	1	these	these	DET
ijassa-659	10	2	algorithms	algorithm	NOUN
ijassa-659	10	3	aim	aim	VERB
ijassa-659	10	4	to	to	PART
ijassa-659	10	5	improve	improve	VERB
ijassa-659	10	6	the	the	DET
ijassa-659	10	7	clustering	clustering	ADJ
ijassa-659	10	8	performance	performance	NOUN
ijassa-659	10	9	with	with	ADP
ijassa-659	10	10	the	the	DET
ijassa-659	10	11	help	help	NOUN
ijassa-659	10	12	of	of	ADP
ijassa-659	10	13	user	user	NOUN
ijassa-659	10	14	-	-	PUNCT
ijassa-659	10	15	provided	provide	VERB
ijassa-659	10	16	side	side	NOUN
ijassa-659	10	17	information	information	NOUN
ijassa-659	10	18	.	.	PUNCT
ijassa-659	11	1	there	there	PRON
ijassa-659	11	2	are	be	VERB
ijassa-659	11	3	several	several	ADJ
ijassa-659	11	4	types	type	NOUN
ijassa-659	11	5	of	of	ADP
ijassa-659	11	6	side	side	NOUN
ijassa-659	11	7	information	information	NOUN
ijassa-659	11	8	but	but	CCONJ
ijassa-659	11	9	pairwise	pairwise	NOUN
ijassa-659	11	10	constraints	constraint	NOUN
ijassa-659	11	11	are	be	AUX
ijassa-659	11	12	the	the	DET
ijassa-659	11	13	most	most	ADV
ijassa-659	11	14	used	used	ADJ
ijassa-659	11	15	one	one	NUM
ijassa-659	11	16	.	.	PUNCT
ijassa-659	12	1	there	there	PRON
ijassa-659	12	2	are	be	VERB
ijassa-659	12	3	two	two	NUM
ijassa-659	12	4	types	type	NOUN
ijassa-659	12	5	of	of	ADP
ijassa-659	12	6	pairwise	pairwise	NOUN
ijassa-659	12	7	constraints	constraint	NOUN
ijassa-659	12	8	:	:	PUNCT
ijassa-659	12	9	must	must	AUX
ijassa-659	12	10	-	-	PUNCT
ijassa-659	12	11	link	link	VERB
ijassa-659	12	12	(	(	PUNCT
ijassa-659	12	13	ml	ml	NOUN
ijassa-659	12	14	)	)	PUNCT
ijassa-659	12	15	and	and	CCONJ
ijassa-659	12	16	cannot	cannot	NOUN
ijassa-659	12	17	-	-	PUNCT
ijassa-659	12	18	link	link	NOUN
ijassa-659	12	19	(	(	PUNCT
ijassa-659	12	20	cl	cl	NOUN
ijassa-659	12	21	)	)	PUNCT
ijassa-659	12	22	constraints	constraint	NOUN
ijassa-659	12	23	.	.	PUNCT
ijassa-659	13	1	the	the	DET
ijassa-659	13	2	must	must	AUX
ijassa-659	13	3	-	-	PUNCT
ijassa-659	13	4	link	link	NOUN
ijassa-659	13	5	constraint	constraint	NOUN
ijassa-659	13	6	indicates	indicate	VERB
ijassa-659	13	7	that	that	SCONJ
ijassa-659	13	8	the	the	DET
ijassa-659	13	9	two	two	NUM
ijassa-659	13	10	objects	object	NOUN
ijassa-659	13	11	must	must	AUX
ijassa-659	13	12	be	be	AUX
ijassa-659	13	13	grouped	group	VERB
ijassa-659	13	14	into	into	ADP
ijassa-659	13	15	the	the	DET
ijassa-659	13	16	same	same	ADJ
ijassa-659	13	17	cluster	cluster	NOUN
ijassa-659	13	18	while	while	SCONJ
ijassa-659	13	19	the	the	DET
ijassa-659	13	20	cannot	cannot	NOUN
ijassa-659	13	21	-	-	PUNCT
ijassa-659	13	22	link	link	NOUN
ijassa-659	13	23	constraint	constraint	NOUN
ijassa-659	13	24	indicates	indicate	VERB
ijassa-659	13	25	that	that	SCONJ
ijassa-659	13	26	the	the	DET
ijassa-659	13	27	two	two	NUM
ijassa-659	13	28	objects	object	NOUN
ijassa-659	13	29	must	must	AUX
ijassa-659	13	30	be	be	AUX
ijassa-659	13	31	in	in	ADP
ijassa-659	13	32	different	different	ADJ
ijassa-659	13	33	clusters	cluster	NOUN
ijassa-659	13	34	.	.	PUNCT
ijassa-659	14	1	the	the	DET
ijassa-659	14	2	existing	exist	VERB
ijassa-659	14	3	constraint	constraint	NOUN
ijassa-659	14	4	-	-	PUNCT
ijassa-659	14	5	based	base	VERB
ijassa-659	14	6	clustering	clustering	ADJ
ijassa-659	14	7	algorithms	algorithm	NOUN
ijassa-659	14	8	assumed	assume	VERB
ijassa-659	14	9	that	that	SCONJ
ijassa-659	14	10	they	they	PRON
ijassa-659	14	11	can	can	AUX
ijassa-659	14	12	improve	improve	VERB
ijassa-659	14	13	the	the	DET
ijassa-659	14	14	clustering	clustering	ADJ
ijassa-659	14	15	performance	performance	NOUN
ijassa-659	14	16	with	with	ADP
ijassa-659	14	17	a	a	DET
ijassa-659	14	18	suitable	suitable	ADJ
ijassa-659	14	19	passively	passively	ADV
ijassa-659	14	20	chosen	choose	VERB
ijassa-659	14	21	set	set	NOUN
ijassa-659	14	22	of	of	ADP
ijassa-659	14	23	constraints	constraint	NOUN
ijassa-659	14	24	[	[	X
ijassa-659	14	25	1	1	NUM
ijassa-659	14	26	,	,	PUNCT
ijassa-659	14	27	2	2	NUM
ijassa-659	14	28	]	]	PUNCT
ijassa-659	14	29	.	.	PUNCT
ijassa-659	15	1	however	however	ADV
ijassa-659	15	2	,	,	PUNCT
ijassa-659	15	3	if	if	SCONJ
ijassa-659	15	4	the	the	DET
ijassa-659	15	5	constraints	constraint	NOUN
ijassa-659	15	6	are	be	AUX
ijassa-659	15	7	selected	select	VERB
ijassa-659	15	8	improperly	improperly	ADV
ijassa-659	15	9	,	,	PUNCT
ijassa-659	15	10	they	they	PRON
ijassa-659	15	11	may	may	AUX
ijassa-659	15	12	also	also	ADV
ijassa-659	15	13	degrade	degrade	VERB
ijassa-659	15	14	the	the	DET
ijassa-659	15	15	clustering	clustering	ADJ
ijassa-659	15	16	performance	performance	NOUN
ijassa-659	15	17	[	[	X
ijassa-659	15	18	3	3	NUM
ijassa-659	15	19	,	,	PUNCT
ijassa-659	15	20	4	4	NUM
ijassa-659	15	21	]	]	PUNCT
ijassa-659	15	22	.	.	PUNCT
ijassa-659	16	1	moreover	moreover	ADV
ijassa-659	16	2	,	,	PUNCT
ijassa-659	16	3	selecting	select	VERB
ijassa-659	16	4	constraints	constraint	NOUN
ijassa-659	16	5	typically	typically	ADV
ijassa-659	16	6	requires	require	VERB
ijassa-659	16	7	a	a	DET
ijassa-659	16	8	user	user	NOUN
ijassa-659	16	9	to	to	PART
ijassa-659	16	10	manually	manually	ADV
ijassa-659	16	11	inspect	inspect	VERB
ijassa-659	16	12	the	the	DET
ijassa-659	16	13	data	datum	NOUN
ijassa-659	16	14	instances	instance	NOUN
ijassa-659	16	15	that	that	PRON
ijassa-659	16	16	can	can	AUX
ijassa-659	16	17	be	be	AUX
ijassa-659	16	18	time	time	NOUN
ijassa-659	16	19	consuming	consume	VERB
ijassa-659	16	20	and	and	CCONJ
ijassa-659	16	21	costly	costly	ADJ
ijassa-659	16	22	.	.	PUNCT
ijassa-659	17	1	for	for	ADP
ijassa-659	17	2	those	those	DET
ijassa-659	17	3	reasons	reason	NOUN
ijassa-659	17	4	,	,	PUNCT
ijassa-659	17	5	the	the	DET
ijassa-659	17	6	proposed	propose	VERB
ijassa-659	17	7	method	method	NOUN
ijassa-659	17	8	optimize	optimize	VERB
ijassa-659	17	9	the	the	DET
ijassa-659	17	10	selection	selection	NOUN
ijassa-659	17	11	of	of	ADP
ijassa-659	17	12	the	the	DET
ijassa-659	17	13	constraints	constraint	NOUN
ijassa-659	17	14	to	to	PART
ijassa-659	17	15	improve	improve	VERB
ijassa-659	17	16	the	the	DET
ijassa-659	17	17	performance	performance	NOUN
ijassa-659	17	18	of	of	ADP
ijassa-659	17	19	semi	semi	ADJ
ijassa-659	17	20	-	-	ADJ
ijassa-659	17	21	supervised	supervised	ADJ
ijassa-659	17	22	clustering	clustering	ADJ
ijassa-659	17	23	algorithms	algorithm	NOUN
ijassa-659	17	24	.	.	PUNCT
ijassa-659	18	1	*	*	PUNCT
ijassa-659	18	2	corresponding	correspond	VERB
ijassa-659	18	3	author	author	NOUN
ijassa-659	18	4	:	:	PUNCT
ijassa-659	18	5	walid.atwa@ci.menofia.edu.eg	walid.atwa@ci.menofia.edu.eg	VERB
ijassa-659	18	6	mailto:walid.atwa@ci.menofia.edu.eg	mailto:walid.atwa@ci.menofia.edu.eg	X
ijassa-659	18	7	mailto:ma7moud_emam@yahoo.com	mailto:ma7moud_emam@yahoo.com	PUNCT
ijassa-659	18	8	26	26	NUM
ijassa-659	18	9	w.	w.	NOUN
ijassa-659	18	10	atwa	atwa	PROPN
ijassa-659	18	11	,	,	PUNCT
ijassa-659	18	12	m.	m.	NOUN
ijassa-659	18	13	emam	emam	PROPN
ijassa-659	18	14	copyright	copyright	NOUN
ijassa-659	18	15	©	©	PROPN
ijassa-659	18	16	2019	2019	NUM
ijassa-659	18	17	assa	assa	PROPN
ijassa-659	18	18	adv	adv	PROPN
ijassa-659	18	19	.	.	PUNCT
ijassa-659	19	1	in	in	ADP
ijassa-659	19	2	systems	system	NOUN
ijassa-659	19	3	science	science	NOUN
ijassa-659	19	4	and	and	CCONJ
ijassa-659	19	5	appl	appl	NOUN
ijassa-659	19	6	.	.	PUNCT
ijassa-659	20	1	(	(	PUNCT
ijassa-659	20	2	2019	2019	NUM
ijassa-659	20	3	)	)	PUNCT
ijassa-659	20	4	active	active	ADJ
ijassa-659	20	5	learning	learning	NOUN
ijassa-659	20	6	is	be	AUX
ijassa-659	20	7	well	well	ADV
ijassa-659	20	8	motivated	motivated	ADJ
ijassa-659	20	9	in	in	ADP
ijassa-659	20	10	many	many	ADJ
ijassa-659	20	11	supervised	supervised	ADJ
ijassa-659	20	12	learning	learn	VERB
ijassa-659	20	13	scenarios	scenario	NOUN
ijassa-659	20	14	where	where	SCONJ
ijassa-659	20	15	unlabeled	unlabeled	ADJ
ijassa-659	20	16	instances	instance	NOUN
ijassa-659	20	17	are	be	AUX
ijassa-659	20	18	abundant	abundant	ADJ
ijassa-659	20	19	and	and	CCONJ
ijassa-659	20	20	easy	easy	ADJ
ijassa-659	20	21	to	to	PART
ijassa-659	20	22	retrieve	retrieve	VERB
ijassa-659	20	23	but	but	CCONJ
ijassa-659	20	24	labeled	label	VERB
ijassa-659	20	25	instances	instance	NOUN
ijassa-659	20	26	are	be	AUX
ijassa-659	20	27	difficult	difficult	ADJ
ijassa-659	20	28	,	,	PUNCT
ijassa-659	20	29	time	time	NOUN
ijassa-659	20	30	-	-	PUNCT
ijassa-659	20	31	consuming	consume	VERB
ijassa-659	20	32	and	and	CCONJ
ijassa-659	20	33	expensive	expensive	ADJ
ijassa-659	20	34	to	to	PART
ijassa-659	20	35	obtain	obtain	VERB
ijassa-659	20	36	.	.	PUNCT
ijassa-659	21	1	for	for	ADP
ijassa-659	21	2	example	example	NOUN
ijassa-659	21	3	,	,	PUNCT
ijassa-659	21	4	it	it	PRON
ijassa-659	21	5	is	be	AUX
ijassa-659	21	6	easy	easy	ADJ
ijassa-659	21	7	to	to	PART
ijassa-659	21	8	gather	gather	VERB
ijassa-659	21	9	large	large	ADJ
ijassa-659	21	10	amounts	amount	NOUN
ijassa-659	21	11	of	of	ADP
ijassa-659	21	12	unlabeled	unlabeled	ADJ
ijassa-659	21	13	documents	document	NOUN
ijassa-659	21	14	or	or	CCONJ
ijassa-659	21	15	images	image	NOUN
ijassa-659	21	16	from	from	ADP
ijassa-659	21	17	the	the	DET
ijassa-659	21	18	internet	internet	NOUN
ijassa-659	21	19	,	,	PUNCT
ijassa-659	21	20	whereas	whereas	SCONJ
ijassa-659	21	21	querying	query	VERB
ijassa-659	21	22	them	they	PRON
ijassa-659	21	23	requires	require	VERB
ijassa-659	21	24	manual	manual	ADJ
ijassa-659	21	25	effort	effort	NOUN
ijassa-659	21	26	from	from	ADP
ijassa-659	21	27	experienced	experienced	ADJ
ijassa-659	21	28	human	human	ADJ
ijassa-659	21	29	annotators	annotator	NOUN
ijassa-659	21	30	.	.	PUNCT
ijassa-659	22	1	hence	hence	ADV
ijassa-659	22	2	there	there	PRON
ijassa-659	22	3	is	be	VERB
ijassa-659	22	4	a	a	DET
ijassa-659	22	5	need	need	NOUN
ijassa-659	22	6	to	to	PART
ijassa-659	22	7	select	select	VERB
ijassa-659	22	8	an	an	DET
ijassa-659	22	9	optimal	optimal	ADJ
ijassa-659	22	10	set	set	NOUN
ijassa-659	22	11	of	of	ADP
ijassa-659	22	12	instances	instance	NOUN
ijassa-659	22	13	from	from	ADP
ijassa-659	22	14	the	the	DET
ijassa-659	22	15	pool	pool	NOUN
ijassa-659	22	16	of	of	ADP
ijassa-659	22	17	unlabeled	unlabeled	ADJ
ijassa-659	22	18	data	datum	NOUN
ijassa-659	22	19	for	for	ADP
ijassa-659	22	20	querying	query	VERB
ijassa-659	22	21	.	.	PUNCT
ijassa-659	23	1	randomly	randomly	ADJ
ijassa-659	23	2	selection	selection	NOUN
ijassa-659	23	3	of	of	ADP
ijassa-659	23	4	unlabeled	unlabele	VERB
ijassa-659	23	5	instances	instance	NOUN
ijassa-659	23	6	for	for	ADP
ijassa-659	23	7	querying	query	VERB
ijassa-659	23	8	is	be	AUX
ijassa-659	23	9	inefficient	inefficient	ADJ
ijassa-659	23	10	in	in	ADP
ijassa-659	23	11	many	many	ADJ
ijassa-659	23	12	situations	situation	NOUN
ijassa-659	23	13	,	,	PUNCT
ijassa-659	23	14	since	since	SCONJ
ijassa-659	23	15	non	non	ADJ
ijassa-659	23	16	-	-	ADJ
ijassa-659	23	17	informative	informative	ADJ
ijassa-659	23	18	or	or	CCONJ
ijassa-659	23	19	redundant	redundant	ADJ
ijassa-659	23	20	instances	instance	NOUN
ijassa-659	23	21	might	might	AUX
ijassa-659	23	22	be	be	AUX
ijassa-659	23	23	selected	select	VERB
ijassa-659	23	24	.	.	PUNCT
ijassa-659	24	1	active	active	ADJ
ijassa-659	24	2	learning	learn	VERB
ijassa-659	24	3	algorithms	algorithm	NOUN
ijassa-659	24	4	select	select	VERB
ijassa-659	24	5	the	the	DET
ijassa-659	24	6	most	most	ADV
ijassa-659	24	7	informative	informative	ADJ
ijassa-659	24	8	unlabeled	unlabele	VERB
ijassa-659	24	9	instances	instance	NOUN
ijassa-659	24	10	from	from	ADP
ijassa-659	24	11	enormous	enormous	ADJ
ijassa-659	24	12	amount	amount	NOUN
ijassa-659	24	13	of	of	ADP
ijassa-659	24	14	unlabeled	unlabeled	ADJ
ijassa-659	24	15	data	datum	NOUN
ijassa-659	24	16	for	for	ADP
ijassa-659	24	17	querying	query	VERB
ijassa-659	24	18	.	.	PUNCT
ijassa-659	25	1	specifically	specifically	ADV
ijassa-659	25	2	,	,	PUNCT
ijassa-659	25	3	the	the	DET
ijassa-659	25	4	goal	goal	NOUN
ijassa-659	25	5	of	of	ADP
ijassa-659	25	6	active	active	ADJ
ijassa-659	25	7	learning	learning	NOUN
ijassa-659	25	8	is	be	AUX
ijassa-659	25	9	to	to	PART
ijassa-659	25	10	query	query	VERB
ijassa-659	25	11	data	datum	NOUN
ijassa-659	25	12	as	as	ADV
ijassa-659	25	13	little	little	ADJ
ijassa-659	25	14	as	as	ADP
ijassa-659	25	15	possible	possible	ADJ
ijassa-659	25	16	to	to	PART
ijassa-659	25	17	achieve	achieve	VERB
ijassa-659	25	18	a	a	DET
ijassa-659	25	19	certain	certain	ADJ
ijassa-659	25	20	performance	performance	NOUN
ijassa-659	25	21	,	,	PUNCT
ijassa-659	25	22	thus	thus	ADV
ijassa-659	25	23	saving	save	VERB
ijassa-659	25	24	considerable	considerable	ADJ
ijassa-659	25	25	cost	cost	NOUN
ijassa-659	25	26	for	for	ADP
ijassa-659	25	27	generating	generate	VERB
ijassa-659	25	28	good	good	ADJ
ijassa-659	25	29	queries	query	NOUN
ijassa-659	25	30	.	.	PUNCT
ijassa-659	26	1	in	in	ADP
ijassa-659	26	2	this	this	DET
ijassa-659	26	3	paper	paper	NOUN
ijassa-659	26	4	,	,	PUNCT
ijassa-659	26	5	the	the	DET
ijassa-659	26	6	active	active	ADJ
ijassa-659	26	7	learning	learning	NOUN
ijassa-659	26	8	is	be	AUX
ijassa-659	26	9	applied	apply	VERB
ijassa-659	26	10	in	in	ADP
ijassa-659	26	11	an	an	DET
ijassa-659	26	12	iterative	iterative	NOUN
ijassa-659	26	13	manner	manner	NOUN
ijassa-659	26	14	.	.	PUNCT
ijassa-659	27	1	in	in	ADP
ijassa-659	27	2	each	each	DET
ijassa-659	27	3	iteration	iteration	NOUN
ijassa-659	27	4	,	,	PUNCT
ijassa-659	27	5	set	set	VERB
ijassa-659	27	6	of	of	ADP
ijassa-659	27	7	queries	query	NOUN
ijassa-659	27	8	are	be	AUX
ijassa-659	27	9	selected	select	VERB
ijassa-659	27	10	and	and	CCONJ
ijassa-659	27	11	queried	query	VERB
ijassa-659	27	12	with	with	ADP
ijassa-659	27	13	the	the	DET
ijassa-659	27	14	existing	exist	VERB
ijassa-659	27	15	neighborhoods	neighborhood	NOUN
ijassa-659	27	16	to	to	PART
ijassa-659	27	17	improve	improve	VERB
ijassa-659	27	18	the	the	DET
ijassa-659	27	19	clustering	clustering	ADJ
ijassa-659	27	20	results	result	NOUN
ijassa-659	27	21	.	.	PUNCT
ijassa-659	28	1	specifically	specifically	ADV
ijassa-659	28	2	,	,	PUNCT
ijassa-659	28	3	selecting	select	VERB
ijassa-659	28	4	a	a	DET
ijassa-659	28	5	batch	batch	NOUN
ijassa-659	28	6	of	of	ADP
ijassa-659	28	7	informative	informative	ADJ
ijassa-659	28	8	query	query	NOUN
ijassa-659	28	9	instances	instance	VERB
ijassa-659	28	10	such	such	ADJ
ijassa-659	28	11	that	that	SCONJ
ijassa-659	28	12	the	the	DET
ijassa-659	28	13	distribution	distribution	NOUN
ijassa-659	28	14	represented	represent	VERB
ijassa-659	28	15	by	by	ADP
ijassa-659	28	16	the	the	DET
ijassa-659	28	17	selected	select	VERB
ijassa-659	28	18	query	query	NOUN
ijassa-659	28	19	set	set	VERB
ijassa-659	28	20	and	and	CCONJ
ijassa-659	28	21	the	the	DET
ijassa-659	28	22	available	available	ADJ
ijassa-659	28	23	labeled	label	VERB
ijassa-659	28	24	data	datum	NOUN
ijassa-659	28	25	is	be	AUX
ijassa-659	28	26	closest	close	ADJ
ijassa-659	28	27	to	to	ADP
ijassa-659	28	28	the	the	DET
ijassa-659	28	29	distribution	distribution	NOUN
ijassa-659	28	30	represented	represent	VERB
ijassa-659	28	31	by	by	ADP
ijassa-659	28	32	the	the	DET
ijassa-659	28	33	unlabeled	unlabeled	ADJ
ijassa-659	28	34	data	datum	NOUN
ijassa-659	28	35	.	.	PUNCT
ijassa-659	29	1	in	in	ADP
ijassa-659	29	2	other	other	ADJ
ijassa-659	29	3	words	word	NOUN
ijassa-659	29	4	,	,	PUNCT
ijassa-659	29	5	the	the	DET
ijassa-659	29	6	proposed	propose	VERB
ijassa-659	29	7	method	method	NOUN
ijassa-659	29	8	select	select	VERB
ijassa-659	29	9	a	a	DET
ijassa-659	29	10	set	set	NOUN
ijassa-659	29	11	of	of	ADP
ijassa-659	29	12	samples	sample	NOUN
ijassa-659	29	13	s	s	VERB
ijassa-659	29	14	from	from	ADP
ijassa-659	29	15	the	the	DET
ijassa-659	29	16	unlabeled	unlabeled	ADJ
ijassa-659	29	17	data	datum	NOUN
ijassa-659	29	18	,	,	PUNCT
ijassa-659	29	19	denoted	denote	VERB
ijassa-659	29	20	by	by	ADP
ijassa-659	29	21	du	du	PROPN
ijassa-659	29	22	,	,	PUNCT
ijassa-659	29	23	such	such	ADJ
ijassa-659	29	24	that	that	SCONJ
ijassa-659	29	25	the	the	DET
ijassa-659	29	26	probability	probability	NOUN
ijassa-659	29	27	distributions	distribution	NOUN
ijassa-659	29	28	represented	represent	VERB
ijassa-659	29	29	by	by	ADP
ijassa-659	29	30	dl	dl	PROPN
ijassa-659	29	31	∪	∪	PROPN
ijassa-659	29	32	s	s	PROPN
ijassa-659	29	33	and	and	CCONJ
ijassa-659	29	34	du	du	PROPN
ijassa-659	29	35	\	\	PROPN
ijassa-659	29	36	s	s	PROPN
ijassa-659	29	37	,	,	PUNCT
ijassa-659	29	38	where	where	SCONJ
ijassa-659	29	39	dl	dl	PROPN
ijassa-659	29	40	is	be	AUX
ijassa-659	29	41	the	the	DET
ijassa-659	29	42	set	set	NOUN
ijassa-659	29	43	of	of	ADP
ijassa-659	29	44	available	available	ADJ
ijassa-659	29	45	labeled	label	VERB
ijassa-659	29	46	data	datum	NOUN
ijassa-659	29	47	,	,	PUNCT
ijassa-659	29	48	are	be	AUX
ijassa-659	29	49	similar	similar	ADJ
ijassa-659	29	50	to	to	ADP
ijassa-659	29	51	each	each	DET
ijassa-659	29	52	other	other	ADJ
ijassa-659	29	53	.	.	PUNCT
ijassa-659	30	1	then	then	ADV
ijassa-659	30	2	measuring	measure	VERB
ijassa-659	30	3	the	the	DET
ijassa-659	30	4	difference	difference	NOUN
ijassa-659	30	5	in	in	ADP
ijassa-659	30	6	the	the	DET
ijassa-659	30	7	probability	probability	NOUN
ijassa-659	30	8	distribution	distribution	NOUN
ijassa-659	30	9	between	between	ADP
ijassa-659	30	10	the	the	DET
ijassa-659	30	11	two	two	NUM
ijassa-659	30	12	sets	set	NOUN
ijassa-659	30	13	of	of	ADP
ijassa-659	30	14	data	datum	NOUN
ijassa-659	30	15	using	use	VERB
ijassa-659	30	16	the	the	DET
ijassa-659	30	17	maximum	maximum	ADJ
ijassa-659	30	18	mean	mean	NOUN
ijassa-659	30	19	discrepancy	discrepancy	NOUN
ijassa-659	30	20	(	(	PUNCT
ijassa-659	30	21	mmd	mmd	X
ijassa-659	30	22	)	)	PUNCT
ijassa-659	31	1	[	[	X
ijassa-659	31	2	5	5	NUM
ijassa-659	31	3	,	,	PUNCT
ijassa-659	31	4	6	6	NUM
ijassa-659	31	5	]	]	PUNCT
ijassa-659	31	6	.	.	PUNCT
ijassa-659	32	1	maximum	maximum	ADJ
ijassa-659	32	2	mean	mean	NOUN
ijassa-659	32	3	discrepancy	discrepancy	NOUN
ijassa-659	32	4	is	be	AUX
ijassa-659	32	5	a	a	DET
ijassa-659	32	6	statistical	statistical	ADJ
ijassa-659	32	7	test	test	NOUN
ijassa-659	32	8	based	base	VERB
ijassa-659	32	9	on	on	ADP
ijassa-659	32	10	the	the	DET
ijassa-659	32	11	fact	fact	NOUN
ijassa-659	32	12	that	that	SCONJ
ijassa-659	32	13	two	two	NUM
ijassa-659	32	14	distributions	distribution	NOUN
ijassa-659	32	15	are	be	AUX
ijassa-659	32	16	different	different	ADJ
ijassa-659	32	17	if	if	SCONJ
ijassa-659	32	18	and	and	CCONJ
ijassa-659	32	19	only	only	ADV
ijassa-659	32	20	if	if	SCONJ
ijassa-659	32	21	there	there	PRON
ijassa-659	32	22	exists	exist	VERB
ijassa-659	32	23	at	at	ADV
ijassa-659	32	24	least	least	ADV
ijassa-659	32	25	one	one	NUM
ijassa-659	32	26	function	function	NOUN
ijassa-659	32	27	in	in	ADP
ijassa-659	32	28	reproducing	reproduce	VERB
ijassa-659	32	29	kernel	kernel	PROPN
ijassa-659	32	30	hilbert	hilbert	PROPN
ijassa-659	32	31	space	space	NOUN
ijassa-659	32	32	(	(	PUNCT
ijassa-659	32	33	rkhs	rkh	NOUN
ijassa-659	32	34	)	)	PUNCT
ijassa-659	33	1	[	[	X
ijassa-659	33	2	6	6	NUM
ijassa-659	33	3	]	]	PUNCT
ijassa-659	33	4	having	have	VERB
ijassa-659	33	5	different	different	ADJ
ijassa-659	33	6	expectations	expectation	NOUN
ijassa-659	33	7	on	on	ADP
ijassa-659	33	8	the	the	DET
ijassa-659	33	9	two	two	NUM
ijassa-659	33	10	distributions	distribution	NOUN
ijassa-659	33	11	.	.	PUNCT
ijassa-659	34	1	once	once	SCONJ
ijassa-659	34	2	the	the	DET
ijassa-659	34	3	batch	batch	NOUN
ijassa-659	34	4	of	of	ADP
ijassa-659	34	5	informative	informative	ADJ
ijassa-659	34	6	query	query	NOUN
ijassa-659	34	7	instances	instance	NOUN
ijassa-659	34	8	are	be	AUX
ijassa-659	34	9	selected	select	VERB
ijassa-659	34	10	,	,	PUNCT
ijassa-659	34	11	the	the	DET
ijassa-659	34	12	proposed	propose	VERB
ijassa-659	34	13	method	method	NOUN
ijassa-659	34	14	query	query	VERB
ijassa-659	34	15	them	they	PRON
ijassa-659	34	16	with	with	ADP
ijassa-659	34	17	the	the	DET
ijassa-659	34	18	existing	exist	VERB
ijassa-659	34	19	neighborhoods	neighborhood	NOUN
ijassa-659	34	20	to	to	PART
ijassa-659	34	21	determine	determine	VERB
ijassa-659	34	22	which	which	DET
ijassa-659	34	23	neighborhood	neighborhood	NOUN
ijassa-659	34	24	they	they	PRON
ijassa-659	34	25	belong	belong	VERB
ijassa-659	34	26	.	.	PUNCT
ijassa-659	35	1	a	a	DET
ijassa-659	35	2	neighborhood	neighborhood	NOUN
ijassa-659	35	3	contains	contain	VERB
ijassa-659	35	4	a	a	DET
ijassa-659	35	5	set	set	NOUN
ijassa-659	35	6	of	of	ADP
ijassa-659	35	7	data	datum	NOUN
ijassa-659	35	8	instances	instance	NOUN
ijassa-659	35	9	that	that	PRON
ijassa-659	35	10	are	be	AUX
ijassa-659	35	11	known	know	VERB
ijassa-659	35	12	to	to	PART
ijassa-659	35	13	belong	belong	VERB
ijassa-659	35	14	to	to	ADP
ijassa-659	35	15	the	the	DET
ijassa-659	35	16	same	same	ADJ
ijassa-659	35	17	cluster	cluster	NOUN
ijassa-659	35	18	(	(	PUNCT
ijassa-659	35	19	i.e.	i.e.	X
ijassa-659	35	20	,	,	PUNCT
ijassa-659	35	21	connected	connect	VERB
ijassa-659	35	22	by	by	ADP
ijassa-659	35	23	must	must	AUX
ijassa-659	35	24	-	-	PUNCT
ijassa-659	35	25	link	link	NOUN
ijassa-659	35	26	constraints	constraint	NOUN
ijassa-659	35	27	)	)	PUNCT
ijassa-659	35	28	and	and	CCONJ
ijassa-659	35	29	different	different	ADJ
ijassa-659	35	30	neighborhoods	neighborhood	NOUN
ijassa-659	35	31	are	be	AUX
ijassa-659	35	32	known	know	VERB
ijassa-659	35	33	to	to	PART
ijassa-659	35	34	belong	belong	VERB
ijassa-659	35	35	to	to	ADP
ijassa-659	35	36	different	different	ADJ
ijassa-659	35	37	clusters	cluster	NOUN
ijassa-659	35	38	(	(	PUNCT
ijassa-659	35	39	i.e.	i.e.	X
ijassa-659	35	40	,	,	PUNCT
ijassa-659	35	41	connected	connect	VERB
ijassa-659	35	42	by	by	ADP
ijassa-659	35	43	cannot	cannot	NOUN
ijassa-659	35	44	-	-	PUNCT
ijassa-659	35	45	link	link	NOUN
ijassa-659	35	46	constraints	constraint	NOUN
ijassa-659	35	47	)	)	PUNCT
ijassa-659	35	48	.	.	PUNCT
ijassa-659	36	1	well	well	ADV
ijassa-659	36	2	-	-	PUNCT
ijassa-659	36	3	formed	form	VERB
ijassa-659	36	4	neighborhoods	neighborhood	NOUN
ijassa-659	36	5	can	can	AUX
ijassa-659	36	6	provide	provide	VERB
ijassa-659	36	7	valuable	valuable	ADJ
ijassa-659	36	8	information	information	NOUN
ijassa-659	36	9	regarding	regard	VERB
ijassa-659	36	10	what	what	PRON
ijassa-659	36	11	the	the	DET
ijassa-659	36	12	underlying	underlie	VERB
ijassa-659	36	13	clusters	cluster	NOUN
ijassa-659	36	14	look	look	VERB
ijassa-659	36	15	like	like	ADP
ijassa-659	36	16	.	.	PUNCT
ijassa-659	37	1	we	we	PRON
ijassa-659	37	2	empirically	empirically	ADV
ijassa-659	37	3	evaluate	evaluate	VERB
ijassa-659	37	4	the	the	DET
ijassa-659	37	5	proposed	propose	VERB
ijassa-659	37	6	method	method	NOUN
ijassa-659	37	7	with	with	ADP
ijassa-659	37	8	baseline	baseline	NOUN
ijassa-659	37	9	and	and	CCONJ
ijassa-659	37	10	state	state	NOUN
ijassa-659	37	11	-	-	PUNCT
ijassa-659	37	12	of	of	ADP
ijassa-659	37	13	-	-	PUNCT
ijassa-659	37	14	the	the	DET
ijassa-659	37	15	-	-	PUNCT
ijassa-659	37	16	art	art	NOUN
ijassa-659	37	17	methods	method	NOUN
ijassa-659	37	18	on	on	ADP
ijassa-659	37	19	uci	uci	PROPN
ijassa-659	37	20	real	real	ADJ
ijassa-659	37	21	datasets	dataset	NOUN
ijassa-659	37	22	.	.	PUNCT
ijassa-659	38	1	the	the	DET
ijassa-659	38	2	evaluation	evaluation	NOUN
ijassa-659	38	3	results	result	NOUN
ijassa-659	38	4	demonstrate	demonstrate	VERB
ijassa-659	38	5	that	that	SCONJ
ijassa-659	38	6	the	the	DET
ijassa-659	38	7	proposed	propose	VERB
ijassa-659	38	8	method	method	NOUN
ijassa-659	38	9	achieves	achieve	VERB
ijassa-659	38	10	consistent	consistent	ADJ
ijassa-659	38	11	improvements	improvement	NOUN
ijassa-659	38	12	over	over	ADP
ijassa-659	38	13	the	the	DET
ijassa-659	38	14	baseline	baseline	ADJ
ijassa-659	38	15	methods	method	NOUN
ijassa-659	38	16	.	.	PUNCT
ijassa-659	39	1	the	the	DET
ijassa-659	39	2	remainder	remainder	NOUN
ijassa-659	39	3	of	of	ADP
ijassa-659	39	4	the	the	DET
ijassa-659	39	5	paper	paper	NOUN
ijassa-659	39	6	is	be	AUX
ijassa-659	39	7	organized	organize	VERB
ijassa-659	39	8	as	as	SCONJ
ijassa-659	39	9	follows	follow	VERB
ijassa-659	39	10	.	.	PUNCT
ijassa-659	40	1	section	section	NOUN
ijassa-659	40	2	2	2	NUM
ijassa-659	40	3	presents	present	VERB
ijassa-659	40	4	a	a	DET
ijassa-659	40	5	brief	brief	ADJ
ijassa-659	40	6	review	review	NOUN
ijassa-659	40	7	of	of	ADP
ijassa-659	40	8	the	the	DET
ijassa-659	40	9	related	related	ADJ
ijassa-659	40	10	work	work	NOUN
ijassa-659	40	11	.	.	PUNCT
ijassa-659	41	1	section	section	NOUN
ijassa-659	41	2	3	3	NUM
ijassa-659	41	3	introduces	introduce	NOUN
ijassa-659	41	4	the	the	DET
ijassa-659	41	5	proposed	propose	VERB
ijassa-659	41	6	method	method	NOUN
ijassa-659	41	7	.	.	PUNCT
ijassa-659	42	1	experimental	experimental	ADJ
ijassa-659	42	2	results	result	NOUN
ijassa-659	42	3	are	be	AUX
ijassa-659	42	4	presented	present	VERB
ijassa-659	42	5	in	in	ADP
ijassa-659	42	6	section	section	NOUN
ijassa-659	42	7	4	4	NUM
ijassa-659	42	8	.	.	PUNCT
ijassa-659	43	1	finally	finally	ADV
ijassa-659	43	2	,	,	PUNCT
ijassa-659	43	3	conclude	conclude	VERB
ijassa-659	43	4	the	the	DET
ijassa-659	43	5	paper	paper	NOUN
ijassa-659	43	6	and	and	CCONJ
ijassa-659	43	7	discuss	discuss	VERB
ijassa-659	43	8	future	future	ADJ
ijassa-659	43	9	directions	direction	NOUN
ijassa-659	43	10	in	in	ADP
ijassa-659	43	11	section	section	NOUN
ijassa-659	43	12	5	5	NUM
ijassa-659	43	13	.	.	NOUN
ijassa-659	43	14	2	2	NUM
ijassa-659	43	15	.	.	X
ijassa-659	43	16	related	relate	VERB
ijassa-659	43	17	work	work	NOUN
ijassa-659	43	18	active	active	ADJ
ijassa-659	43	19	learning	learning	NOUN
ijassa-659	43	20	has	have	VERB
ijassa-659	43	21	a	a	DET
ijassa-659	43	22	long	long	ADJ
ijassa-659	43	23	history	history	NOUN
ijassa-659	43	24	in	in	ADP
ijassa-659	43	25	supervised	supervised	ADJ
ijassa-659	43	26	learning	learn	VERB
ijassa-659	43	27	algorithms	algorithm	NOUN
ijassa-659	43	28	[	[	X
ijassa-659	43	29	7	7	NUM
ijassa-659	43	30	,	,	PUNCT
ijassa-659	43	31	8	8	NUM
ijassa-659	43	32	,	,	PUNCT
ijassa-659	43	33	9	9	NUM
ijassa-659	43	34	,	,	PUNCT
ijassa-659	43	35	10	10	NUM
ijassa-659	43	36	,	,	PUNCT
ijassa-659	43	37	11	11	NUM
ijassa-659	43	38	]	]	PUNCT
ijassa-659	43	39	.	.	PUNCT
ijassa-659	44	1	recently	recently	ADV
ijassa-659	44	2	,	,	PUNCT
ijassa-659	44	3	few	few	ADJ
ijassa-659	44	4	studies	study	NOUN
ijassa-659	44	5	reported	report	VERB
ijassa-659	44	6	the	the	DET
ijassa-659	44	7	result	result	NOUN
ijassa-659	44	8	of	of	ADP
ijassa-659	44	9	using	use	VERB
ijassa-659	44	10	active	active	ADJ
ijassa-659	44	11	learning	learning	NOUN
ijassa-659	44	12	in	in	ADP
ijassa-659	44	13	constrained	constrain	VERB
ijassa-659	44	14	-	-	PUNCT
ijassa-659	44	15	based	base	VERB
ijassa-659	44	16	clustering	clustering	ADJ
ijassa-659	44	17	problems	problem	NOUN
ijassa-659	44	18	.	.	PUNCT
ijassa-659	45	1	the	the	DET
ijassa-659	45	2	first	first	ADJ
ijassa-659	45	3	study	study	NOUN
ijassa-659	45	4	was	be	AUX
ijassa-659	45	5	conducted	conduct	VERB
ijassa-659	45	6	by	by	ADP
ijassa-659	45	7	basu	basu	PROPN
ijassa-659	45	8	et	et	PROPN
ijassa-659	45	9	al	al	PROPN
ijassa-659	45	10	.	.	PUNCT
ijassa-659	46	1	[	[	X
ijassa-659	46	2	1	1	X
ijassa-659	46	3	]	]	PUNCT
ijassa-659	46	4	that	that	PRON
ijassa-659	46	5	proposed	propose	VERB
ijassa-659	46	6	an	an	DET
ijassa-659	46	7	active	active	ADJ
ijassa-659	46	8	k	k	ADJ
ijassa-659	46	9	-	-	PUNCT
ijassa-659	46	10	means	means	NOUN
ijassa-659	46	11	clustering	clustering	NOUN
ijassa-659	46	12	using	use	VERB
ijassa-659	46	13	the	the	DET
ijassa-659	46	14	farthest	farth	ADJ
ijassa-659	46	15	-	-	PUNCT
ijassa-659	46	16	first	first	ADJ
ijassa-659	46	17	strategy	strategy	NOUN
ijassa-659	46	18	that	that	PRON
ijassa-659	46	19	has	have	AUX
ijassa-659	46	20	two	two	NUM
ijassa-659	46	21	-	-	PUNCT
ijassa-659	46	22	phases	phase	NOUN
ijassa-659	46	23	(	(	PUNCT
ijassa-659	46	24	explore	explore	VERB
ijassa-659	46	25	and	and	CCONJ
ijassa-659	46	26	consolidate	consolidate	VERB
ijassa-659	46	27	)	)	PUNCT
ijassa-659	46	28	.	.	PUNCT
ijassa-659	47	1	the	the	DET
ijassa-659	47	2	first	first	ADJ
ijassa-659	47	3	phase	phase	NOUN
ijassa-659	47	4	(	(	PUNCT
ijassa-659	47	5	explore	explore	NOUN
ijassa-659	47	6	)	)	PUNCT
ijassa-659	47	7	uses	use	VERB
ijassa-659	47	8	the	the	DET
ijassa-659	47	9	farthest	farth	ADJ
ijassa-659	47	10	-	-	PUNCT
ijassa-659	47	11	first	first	ADJ
ijassa-659	47	12	scheme	scheme	NOUN
ijassa-659	47	13	to	to	PART
ijassa-659	47	14	form	form	VERB
ijassa-659	47	15	appropriate	appropriate	ADJ
ijassa-659	47	16	queries	query	NOUN
ijassa-659	47	17	for	for	ADP
ijassa-659	47	18	getting	get	VERB
ijassa-659	47	19	the	the	DET
ijassa-659	47	20	required	require	VERB
ijassa-659	47	21	pairwise	pairwise	NOUN
ijassa-659	47	22	disjoint	disjoint	NOUN
ijassa-659	47	23	neighborhoods	neighborhood	NOUN
ijassa-659	47	24	.	.	PUNCT
ijassa-659	48	1	at	at	ADP
ijassa-659	48	2	the	the	DET
ijassa-659	48	3	end	end	NOUN
ijassa-659	48	4	of	of	ADP
ijassa-659	48	5	explore	explore	NOUN
ijassa-659	48	6	,	,	PUNCT
ijassa-659	48	7	at	at	ADV
ijassa-659	48	8	least	least	ADV
ijassa-659	48	9	one	one	NUM
ijassa-659	48	10	point	point	NOUN
ijassa-659	48	11	has	have	AUX
ijassa-659	48	12	been	be	AUX
ijassa-659	48	13	obtained	obtain	VERB
ijassa-659	48	14	per	per	ADP
ijassa-659	48	15	cluster	cluster	NOUN
ijassa-659	48	16	.	.	PUNCT
ijassa-659	49	1	the	the	DET
ijassa-659	49	2	second	second	ADJ
ijassa-659	49	3	phase	phase	NOUN
ijassa-659	49	4	(	(	PUNCT
ijassa-659	49	5	consolidate	consolidate	VERB
ijassa-659	49	6	)	)	PUNCT
ijassa-659	49	7	iteratively	iteratively	ADV
ijassa-659	49	8	expands	expand	VERB
ijassa-659	49	9	the	the	DET
ijassa-659	49	10	neighborhoods	neighborhood	NOUN
ijassa-659	49	11	.	.	PUNCT
ijassa-659	50	1	where	where	SCONJ
ijassa-659	50	2	in	in	ADP
ijassa-659	50	3	each	each	DET
ijassa-659	50	4	iteration	iteration	NOUN
ijassa-659	50	5	it	it	PRON
ijassa-659	50	6	selects	select	VERB
ijassa-659	50	7	a	a	DET
ijassa-659	50	8	random	random	ADJ
ijassa-659	50	9	point	point	NOUN
ijassa-659	50	10	outside	outside	ADP
ijassa-659	50	11	any	any	DET
ijassa-659	50	12	neighborhood	neighborhood	NOUN
ijassa-659	50	13	and	and	CCONJ
ijassa-659	50	14	queries	query	VERB
ijassa-659	50	15	it	it	PRON
ijassa-659	50	16	against	against	ADP
ijassa-659	50	17	the	the	DET
ijassa-659	50	18	existing	exist	VERB
ijassa-659	50	19	neighborhoods	neighborhood	NOUN
ijassa-659	50	20	until	until	SCONJ
ijassa-659	50	21	a	a	DET
ijassa-659	50	22	must	must	AUX
ijassa-659	50	23	-	-	PUNCT
ijassa-659	50	24	link	link	NOUN
ijassa-659	50	25	is	be	AUX
ijassa-659	50	26	found	find	VERB
ijassa-659	50	27	.	.	PUNCT
ijassa-659	51	1	an	an	DET
ijassa-659	51	2	improvement	improvement	NOUN
ijassa-659	51	3	version	version	NOUN
ijassa-659	51	4	called	call	VERB
ijassa-659	51	5	min	min	PROPN
ijassa-659	51	6	-	-	ADJ
ijassa-659	51	7	max	max	PROPN
ijassa-659	51	8	approach	approach	NOUN
ijassa-659	51	9	[	[	X
ijassa-659	51	10	12	12	NUM
ijassa-659	51	11	]	]	PUNCT
ijassa-659	51	12	,	,	PUNCT
ijassa-659	51	13	which	which	PRON
ijassa-659	51	14	modifies	modify	VERB
ijassa-659	51	15	the	the	DET
ijassa-659	51	16	consolidate	consolidate	ADJ
ijassa-659	51	17	phase	phase	NOUN
ijassa-659	51	18	by	by	ADP
ijassa-659	51	19	selecting	select	VERB
ijassa-659	51	20	the	the	DET
ijassa-659	51	21	most	most	ADV
ijassa-659	51	22	uncertain	uncertain	ADJ
ijassa-659	51	23	point	point	NOUN
ijassa-659	51	24	to	to	ADP
ijassa-659	51	25	query	query	NOUN
ijassa-659	51	26	,	,	PUNCT
ijassa-659	51	27	instead	instead	ADV
ijassa-659	51	28	of	of	ADP
ijassa-659	51	29	selecting	select	VERB
ijassa-659	51	30	the	the	DET
ijassa-659	51	31	point	point	NOUN
ijassa-659	51	32	improving	improve	VERB
ijassa-659	51	33	semi	semi	ADJ
ijassa-659	51	34	-	-	ADJ
ijassa-659	51	35	supervised	supervised	ADJ
ijassa-659	51	36	clustering	clustering	ADJ
ijassa-659	51	37	algorithms	algorithm	NOUN
ijassa-659	51	38	with	with	ADP
ijassa-659	51	39	active	active	ADJ
ijassa-659	51	40	query	query	NOUN
ijassa-659	51	41	selection	selection	NOUN
ijassa-659	51	42	27	27	NUM
ijassa-659	51	43	copyright	copyright	NOUN
ijassa-659	51	44	©	©	PROPN
ijassa-659	51	45	2019	2019	NUM
ijassa-659	51	46	assa	assa	PROPN
ijassa-659	51	47	adv	adv	PROPN
ijassa-659	51	48	.	.	PUNCT
ijassa-659	52	1	in	in	ADP
ijassa-659	52	2	systems	system	NOUN
ijassa-659	52	3	science	science	NOUN
ijassa-659	52	4	and	and	CCONJ
ijassa-659	52	5	appl	appl	NOUN
ijassa-659	52	6	.	.	PUNCT
ijassa-659	53	1	(	(	PUNCT
ijassa-659	53	2	2019	2019	NUM
ijassa-659	53	3	)	)	PUNCT
ijassa-659	53	4	randomly	randomly	ADV
ijassa-659	53	5	.	.	PUNCT
ijassa-659	54	1	the	the	DET
ijassa-659	54	2	idea	idea	NOUN
ijassa-659	54	3	is	be	AUX
ijassa-659	54	4	to	to	PART
ijassa-659	54	5	select	select	VERB
ijassa-659	54	6	the	the	DET
ijassa-659	54	7	data	datum	NOUN
ijassa-659	54	8	point	point	NOUN
ijassa-659	54	9	whose	whose	DET
ijassa-659	54	10	largest	large	ADJ
ijassa-659	54	11	similarity	similarity	NOUN
ijassa-659	54	12	to	to	ADP
ijassa-659	54	13	the	the	DET
ijassa-659	54	14	skeleton	skeleton	NOUN
ijassa-659	54	15	is	be	AUX
ijassa-659	54	16	the	the	DET
ijassa-659	54	17	smallest	small	ADJ
ijassa-659	54	18	.	.	PUNCT
ijassa-659	55	1	by	by	ADP
ijassa-659	55	2	this	this	DET
ijassa-659	55	3	way	way	NOUN
ijassa-659	55	4	,	,	PUNCT
ijassa-659	55	5	data	datum	NOUN
ijassa-659	55	6	points	point	NOUN
ijassa-659	55	7	with	with	ADP
ijassa-659	55	8	largest	large	ADJ
ijassa-659	55	9	uncertainty	uncertainty	NOUN
ijassa-659	55	10	in	in	ADP
ijassa-659	55	11	cluster	cluster	NOUN
ijassa-659	55	12	membership	membership	NOUN
ijassa-659	55	13	are	be	AUX
ijassa-659	55	14	chosen	choose	VERB
ijassa-659	55	15	first	first	ADV
ijassa-659	55	16	to	to	PART
ijassa-659	55	17	express	express	VERB
ijassa-659	55	18	the	the	DET
ijassa-659	55	19	user	user	NOUN
ijassa-659	55	20	queries	query	NOUN
ijassa-659	55	21	.	.	PUNCT
ijassa-659	56	1	however	however	ADV
ijassa-659	56	2	,	,	PUNCT
ijassa-659	56	3	both	both	DET
ijassa-659	56	4	previous	previous	ADJ
ijassa-659	56	5	methods	method	NOUN
ijassa-659	56	6	do	do	AUX
ijassa-659	56	7	not	not	PART
ijassa-659	56	8	work	work	VERB
ijassa-659	56	9	well	well	ADV
ijassa-659	56	10	in	in	ADP
ijassa-659	56	11	the	the	DET
ijassa-659	56	12	case	case	NOUN
ijassa-659	56	13	of	of	ADP
ijassa-659	56	14	a	a	DET
ijassa-659	56	15	dataset	dataset	NOUN
ijassa-659	56	16	with	with	ADP
ijassa-659	56	17	a	a	DET
ijassa-659	56	18	large	large	ADJ
ijassa-659	56	19	number	number	NOUN
ijassa-659	56	20	of	of	ADP
ijassa-659	56	21	clusters	cluster	NOUN
ijassa-659	56	22	or	or	CCONJ
ijassa-659	56	23	unbalanced	unbalanced	ADJ
ijassa-659	56	24	datasets	dataset	NOUN
ijassa-659	56	25	with	with	ADP
ijassa-659	56	26	small	small	ADJ
ijassa-659	56	27	clusters	cluster	NOUN
ijassa-659	56	28	.	.	PUNCT
ijassa-659	57	1	xu	xu	PROPN
ijassa-659	57	2	et	et	PROPN
ijassa-659	58	1	al	al	PROPN
ijassa-659	58	2	.	.	PUNCT
ijassa-659	59	1	[	[	X
ijassa-659	59	2	13	13	NUM
ijassa-659	59	3	]	]	PUNCT
ijassa-659	59	4	proposed	propose	VERB
ijassa-659	59	5	an	an	DET
ijassa-659	59	6	active	active	ADJ
ijassa-659	59	7	constrained	constrained	ADJ
ijassa-659	59	8	spectral	spectral	ADJ
ijassa-659	59	9	clustering	clustering	ADJ
ijassa-659	59	10	algorithm	algorithm	NOUN
ijassa-659	59	11	that	that	PRON
ijassa-659	59	12	examines	examine	VERB
ijassa-659	59	13	the	the	DET
ijassa-659	59	14	eigenvectors	eigenvector	NOUN
ijassa-659	59	15	to	to	PART
ijassa-659	59	16	identify	identify	VERB
ijassa-659	59	17	the	the	DET
ijassa-659	59	18	boundary	boundary	ADJ
ijassa-659	59	19	points	point	NOUN
ijassa-659	59	20	(	(	PUNCT
ijassa-659	59	21	of	of	ADP
ijassa-659	59	22	two	two	NUM
ijassa-659	59	23	classes	class	NOUN
ijassa-659	59	24	)	)	PUNCT
ijassa-659	59	25	and	and	CCONJ
ijassa-659	59	26	sparse	sparse	ADJ
ijassa-659	59	27	points	point	NOUN
ijassa-659	59	28	;	;	PUNCT
ijassa-659	59	29	then	then	ADV
ijassa-659	59	30	it	it	PRON
ijassa-659	59	31	queries	query	VERB
ijassa-659	59	32	the	the	DET
ijassa-659	59	33	oracle	oracle	NOUN
ijassa-659	59	34	for	for	ADP
ijassa-659	59	35	constraints	constraint	NOUN
ijassa-659	59	36	based	base	VERB
ijassa-659	59	37	on	on	ADP
ijassa-659	59	38	these	these	DET
ijassa-659	59	39	points	point	NOUN
ijassa-659	59	40	.	.	PUNCT
ijassa-659	60	1	it	it	PRON
ijassa-659	60	2	has	have	AUX
ijassa-659	60	3	shown	show	VERB
ijassa-659	60	4	limited	limited	ADJ
ijassa-659	60	5	applicability	applicability	NOUN
ijassa-659	60	6	because	because	SCONJ
ijassa-659	60	7	it	it	PRON
ijassa-659	60	8	requires	require	VERB
ijassa-659	60	9	many	many	ADJ
ijassa-659	60	10	queries	query	NOUN
ijassa-659	60	11	to	to	ADP
ijassa-659	60	12	the	the	DET
ijassa-659	60	13	oracle	oracle	NOUN
ijassa-659	60	14	and	and	CCONJ
ijassa-659	60	15	assumes	assume	VERB
ijassa-659	60	16	that	that	SCONJ
ijassa-659	60	17	errors	error	NOUN
ijassa-659	60	18	in	in	ADP
ijassa-659	60	19	the	the	DET
ijassa-659	60	20	clustering	clustering	ADJ
ijassa-659	60	21	result	result	NOUN
ijassa-659	60	22	only	only	ADV
ijassa-659	60	23	occur	occur	VERB
ijassa-659	60	24	on	on	ADP
ijassa-659	60	25	the	the	DET
ijassa-659	60	26	boundary	boundary	ADJ
ijassa-659	60	27	points	point	NOUN
ijassa-659	60	28	.	.	PUNCT
ijassa-659	61	1	wang	wang	PROPN
ijassa-659	61	2	et	et	PROPN
ijassa-659	61	3	al	al	PROPN
ijassa-659	61	4	.	.	PUNCT
ijassa-659	62	1	[	[	X
ijassa-659	62	2	14	14	NUM
ijassa-659	62	3	]	]	PUNCT
ijassa-659	62	4	presented	present	VERB
ijassa-659	62	5	another	another	DET
ijassa-659	62	6	spectral	spectral	ADJ
ijassa-659	62	7	active	active	ADJ
ijassa-659	62	8	clustering	clustering	NOUN
ijassa-659	62	9	technique	technique	NOUN
ijassa-659	62	10	that	that	PRON
ijassa-659	62	11	identifies	identify	VERB
ijassa-659	62	12	informative	informative	ADJ
ijassa-659	62	13	pairs	pair	NOUN
ijassa-659	62	14	according	accord	VERB
ijassa-659	62	15	to	to	ADP
ijassa-659	62	16	the	the	DET
ijassa-659	62	17	entropy	entropy	NOUN
ijassa-659	62	18	of	of	ADP
ijassa-659	62	19	the	the	DET
ijassa-659	62	20	pair	pair	NOUN
ijassa-659	62	21	example	example	NOUN
ijassa-659	62	22	.	.	PUNCT
ijassa-659	63	1	but	but	CCONJ
ijassa-659	63	2	,	,	PUNCT
ijassa-659	63	3	these	these	DET
ijassa-659	63	4	approaches	approach	NOUN
ijassa-659	63	5	have	have	AUX
ijassa-659	63	6	limited	limit	VERB
ijassa-659	63	7	their	their	PRON
ijassa-659	63	8	work	work	NOUN
ijassa-659	63	9	to	to	ADP
ijassa-659	63	10	two	two	NUM
ijassa-659	63	11	classes	class	NOUN
ijassa-659	63	12	,	,	PUNCT
ijassa-659	63	13	and	and	CCONJ
ijassa-659	63	14	the	the	DET
ijassa-659	63	15	direct	direct	ADJ
ijassa-659	63	16	generalizations	generalization	NOUN
ijassa-659	63	17	to	to	AUX
ijassa-659	63	18	multi	multi	ADJ
ijassa-659	63	19	-	-	NOUN
ijassa-659	63	20	classes	class	NOUN
ijassa-659	63	21	cases	case	NOUN
ijassa-659	63	22	are	be	AUX
ijassa-659	63	23	not	not	PART
ijassa-659	63	24	known	know	VERB
ijassa-659	63	25	.	.	PUNCT
ijassa-659	64	1	vu	vu	X
ijassa-659	64	2	et	et	PROPN
ijassa-659	64	3	al	al	PROPN
ijassa-659	64	4	.	.	PUNCT
ijassa-659	65	1	[	[	X
ijassa-659	65	2	15	15	NUM
ijassa-659	65	3	,	,	PUNCT
ijassa-659	65	4	16	16	NUM
ijassa-659	65	5	]	]	PUNCT
ijassa-659	65	6	proposed	propose	VERB
ijassa-659	65	7	an	an	DET
ijassa-659	65	8	active	active	ADJ
ijassa-659	65	9	query	query	NOUN
ijassa-659	65	10	selection	selection	NOUN
ijassa-659	65	11	method	method	NOUN
ijassa-659	65	12	based	base	VERB
ijassa-659	65	13	on	on	ADP
ijassa-659	65	14	a	a	DET
ijassa-659	65	15	constraint	constraint	NOUN
ijassa-659	65	16	utility	utility	NOUN
ijassa-659	65	17	function	function	NOUN
ijassa-659	65	18	called	call	VERB
ijassa-659	65	19	ability	ability	NOUN
ijassa-659	65	20	to	to	PART
ijassa-659	65	21	separate	separate	VERB
ijassa-659	65	22	between	between	ADP
ijassa-659	65	23	clusters	cluster	NOUN
ijassa-659	65	24	.	.	PUNCT
ijassa-659	66	1	this	this	DET
ijassa-659	66	2	method	method	NOUN
ijassa-659	66	3	relies	rely	VERB
ijassa-659	66	4	on	on	ADP
ijassa-659	66	5	two	two	NUM
ijassa-659	66	6	aspects	aspect	NOUN
ijassa-659	66	7	:	:	PUNCT
ijassa-659	66	8	(	(	PUNCT
ijassa-659	66	9	1	1	X
ijassa-659	66	10	)	)	PUNCT
ijassa-659	66	11	a	a	DET
ijassa-659	66	12	knearest	knearest	NOUN
ijassa-659	66	13	neighbors	neighbor	NOUN
ijassa-659	66	14	graph	graph	NOUN
ijassa-659	66	15	is	be	AUX
ijassa-659	66	16	used	use	VERB
ijassa-659	66	17	to	to	PART
ijassa-659	66	18	determine	determine	VERB
ijassa-659	66	19	the	the	DET
ijassa-659	66	20	best	good	ADJ
ijassa-659	66	21	candidate	candidate	NOUN
ijassa-659	66	22	queries	query	NOUN
ijassa-659	66	23	in	in	ADP
ijassa-659	66	24	the	the	DET
ijassa-659	66	25	sparse	sparse	ADJ
ijassa-659	66	26	regions	region	NOUN
ijassa-659	66	27	of	of	ADP
ijassa-659	66	28	the	the	DET
ijassa-659	66	29	dataset	dataset	NOUN
ijassa-659	66	30	between	between	ADP
ijassa-659	66	31	the	the	DET
ijassa-659	66	32	clusters	cluster	NOUN
ijassa-659	66	33	,	,	PUNCT
ijassa-659	66	34	and	and	CCONJ
ijassa-659	66	35	(	(	PUNCT
ijassa-659	66	36	2	2	X
ijassa-659	66	37	)	)	PUNCT
ijassa-659	66	38	a	a	DET
ijassa-659	66	39	propagation	propagation	NOUN
ijassa-659	66	40	procedure	procedure	NOUN
ijassa-659	66	41	allows	allow	VERB
ijassa-659	66	42	each	each	DET
ijassa-659	66	43	user	user	NOUN
ijassa-659	66	44	query	query	NOUN
ijassa-659	66	45	to	to	PART
ijassa-659	66	46	generate	generate	VERB
ijassa-659	66	47	several	several	ADJ
ijassa-659	66	48	constraints	constraint	NOUN
ijassa-659	66	49	which	which	PRON
ijassa-659	66	50	limits	limit	VERB
ijassa-659	66	51	the	the	DET
ijassa-659	66	52	user	user	NOUN
ijassa-659	66	53	intervention	intervention	NOUN
ijassa-659	66	54	.	.	PUNCT
ijassa-659	67	1	the	the	DET
ijassa-659	67	2	propagation	propagation	NOUN
ijassa-659	67	3	procedure	procedure	NOUN
ijassa-659	67	4	discovers	discover	VERB
ijassa-659	67	5	new	new	ADJ
ijassa-659	67	6	constraints	constraint	NOUN
ijassa-659	67	7	from	from	ADP
ijassa-659	67	8	the	the	DET
ijassa-659	67	9	information	information	NOUN
ijassa-659	67	10	stored	store	VERB
ijassa-659	67	11	in	in	ADP
ijassa-659	67	12	already	already	ADV
ijassa-659	67	13	chosen	choose	VERB
ijassa-659	67	14	constraints	constraint	NOUN
ijassa-659	67	15	using	use	VERB
ijassa-659	67	16	the	the	DET
ijassa-659	67	17	notion	notion	NOUN
ijassa-659	67	18	of	of	ADP
ijassa-659	67	19	strong	strong	ADJ
ijassa-659	67	20	paths	path	NOUN
ijassa-659	67	21	.	.	PUNCT
ijassa-659	68	1	subsequently	subsequently	ADV
ijassa-659	68	2	,	,	PUNCT
ijassa-659	68	3	the	the	DET
ijassa-659	68	4	size	size	NOUN
ijassa-659	68	5	of	of	ADP
ijassa-659	68	6	the	the	DET
ijassa-659	68	7	candidate	candidate	NOUN
ijassa-659	68	8	set	set	NOUN
ijassa-659	68	9	is	be	AUX
ijassa-659	68	10	reduced	reduce	VERB
ijassa-659	68	11	by	by	ADP
ijassa-659	68	12	a	a	DET
ijassa-659	68	13	refinement	refinement	NOUN
ijassa-659	68	14	procedure	procedure	NOUN
ijassa-659	68	15	that	that	PRON
ijassa-659	68	16	removes	remove	VERB
ijassa-659	68	17	constraints	constraint	NOUN
ijassa-659	68	18	between	between	ADP
ijassa-659	68	19	objects	object	NOUN
ijassa-659	68	20	that	that	PRON
ijassa-659	68	21	are	be	AUX
ijassa-659	68	22	likely	likely	ADJ
ijassa-659	68	23	to	to	PART
ijassa-659	68	24	be	be	AUX
ijassa-659	68	25	in	in	ADP
ijassa-659	68	26	the	the	DET
ijassa-659	68	27	same	same	ADJ
ijassa-659	68	28	cluster	cluster	NOUN
ijassa-659	68	29	.	.	PUNCT
ijassa-659	69	1	specifically	specifically	ADV
ijassa-659	69	2	,	,	PUNCT
ijassa-659	69	3	the	the	DET
ijassa-659	69	4	refinement	refinement	NOUN
ijassa-659	69	5	procedure	procedure	NOUN
ijassa-659	69	6	removes	remove	VERB
ijassa-659	69	7	candidate	candidate	NOUN
ijassa-659	69	8	constraints	constraint	NOUN
ijassa-659	69	9	that	that	PRON
ijassa-659	69	10	are	be	AUX
ijassa-659	69	11	linked	link	VERB
ijassa-659	69	12	by	by	ADP
ijassa-659	69	13	a	a	DET
ijassa-659	69	14	strong	strong	ADJ
ijassa-659	69	15	path	path	NOUN
ijassa-659	69	16	.	.	PUNCT
ijassa-659	70	1	recently	recently	ADV
ijassa-659	70	2	,	,	PUNCT
ijassa-659	70	3	xiong	xiong	PROPN
ijassa-659	70	4	et	et	PROPN
ijassa-659	70	5	al	al	PROPN
ijassa-659	70	6	.	.	PUNCT
ijassa-659	71	1	[	[	X
ijassa-659	71	2	17	17	NUM
ijassa-659	71	3	]	]	PUNCT
ijassa-659	71	4	proposed	propose	VERB
ijassa-659	71	5	an	an	DET
ijassa-659	71	6	active	active	ADJ
ijassa-659	71	7	learning	learning	NOUN
ijassa-659	71	8	method	method	NOUN
ijassa-659	71	9	based	base	VERB
ijassa-659	71	10	on	on	ADP
ijassa-659	71	11	the	the	DET
ijassa-659	71	12	classic	classic	ADJ
ijassa-659	71	13	uncertaintybased	uncertaintybase	VERB
ijassa-659	71	14	principle	principle	NOUN
ijassa-659	71	15	.	.	PUNCT
ijassa-659	72	1	they	they	PRON
ijassa-659	72	2	studied	study	VERB
ijassa-659	72	3	the	the	DET
ijassa-659	72	4	selection	selection	NOUN
ijassa-659	72	5	of	of	ADP
ijassa-659	72	6	constraints	constraint	NOUN
ijassa-659	72	7	by	by	ADP
ijassa-659	72	8	selecting	select	VERB
ijassa-659	72	9	the	the	DET
ijassa-659	72	10	most	most	ADV
ijassa-659	72	11	informative	informative	ADJ
ijassa-659	72	12	instance	instance	NOUN
ijassa-659	72	13	to	to	PART
ijassa-659	72	14	form	form	VERB
ijassa-659	72	15	queries	query	NOUN
ijassa-659	72	16	accordingly	accordingly	ADV
ijassa-659	72	17	.	.	PUNCT
ijassa-659	73	1	the	the	DET
ijassa-659	73	2	responses	response	NOUN
ijassa-659	73	3	to	to	ADP
ijassa-659	73	4	the	the	DET
ijassa-659	73	5	queries	query	NOUN
ijassa-659	73	6	are	be	AUX
ijassa-659	73	7	then	then	ADV
ijassa-659	73	8	used	use	VERB
ijassa-659	73	9	to	to	PART
ijassa-659	73	10	improve	improve	VERB
ijassa-659	73	11	the	the	DET
ijassa-659	73	12	clustering	clustering	ADJ
ijassa-659	73	13	results	result	NOUN
ijassa-659	73	14	.	.	PUNCT
ijassa-659	74	1	however	however	ADV
ijassa-659	74	2	,	,	PUNCT
ijassa-659	74	3	this	this	DET
ijassa-659	74	4	method	method	NOUN
ijassa-659	74	5	selects	select	VERB
ijassa-659	74	6	only	only	ADV
ijassa-659	74	7	single	single	ADJ
ijassa-659	74	8	instance	instance	NOUN
ijassa-659	74	9	that	that	PRON
ijassa-659	74	10	can	can	AUX
ijassa-659	74	11	become	become	VERB
ijassa-659	74	12	very	very	ADV
ijassa-659	74	13	slow	slow	ADJ
ijassa-659	74	14	for	for	ADP
ijassa-659	74	15	retraining	retrain	VERB
ijassa-659	74	16	with	with	ADP
ijassa-659	74	17	each	each	DET
ijassa-659	74	18	single	single	ADJ
ijassa-659	74	19	instance	instance	NOUN
ijassa-659	74	20	being	be	AUX
ijassa-659	74	21	queried	query	VERB
ijassa-659	74	22	.	.	PUNCT
ijassa-659	75	1	furthermore	furthermore	ADV
ijassa-659	75	2	,	,	PUNCT
ijassa-659	75	3	if	if	SCONJ
ijassa-659	75	4	a	a	DET
ijassa-659	75	5	parallel	parallel	ADJ
ijassa-659	75	6	querying	query	VERB
ijassa-659	75	7	system	system	NOUN
ijassa-659	75	8	is	be	AUX
ijassa-659	75	9	available	available	ADJ
ijassa-659	75	10	,	,	PUNCT
ijassa-659	75	11	e.g.	e.g.	ADV
ijassa-659	75	12	,	,	PUNCT
ijassa-659	75	13	multiple	multiple	ADJ
ijassa-659	75	14	annotators	annotator	NOUN
ijassa-659	75	15	working	work	VERB
ijassa-659	75	16	in	in	ADP
ijassa-659	75	17	parallel	parallel	NOUN
ijassa-659	75	18	,	,	PUNCT
ijassa-659	75	19	these	these	DET
ijassa-659	75	20	methods	method	NOUN
ijassa-659	75	21	would	would	AUX
ijassa-659	75	22	not	not	PART
ijassa-659	75	23	be	be	AUX
ijassa-659	75	24	able	able	ADJ
ijassa-659	75	25	to	to	PART
ijassa-659	75	26	make	make	VERB
ijassa-659	75	27	the	the	DET
ijassa-659	75	28	effective	effective	ADJ
ijassa-659	75	29	use	use	NOUN
ijassa-659	75	30	of	of	ADP
ijassa-659	75	31	the	the	DET
ijassa-659	75	32	resources	resource	NOUN
ijassa-659	75	33	.	.	PUNCT
ijassa-659	76	1	li	li	PROPN
ijassa-659	76	2	et	et	PROPN
ijassa-659	76	3	al	al	PROPN
ijassa-659	76	4	.	.	PUNCT
ijassa-659	77	1	[	[	X
ijassa-659	77	2	18	18	NUM
ijassa-659	77	3	]	]	PUNCT
ijassa-659	77	4	proposed	propose	VERB
ijassa-659	77	5	an	an	DET
ijassa-659	77	6	active	active	ADJ
ijassa-659	77	7	learning	learning	NOUN
ijassa-659	77	8	method	method	NOUN
ijassa-659	77	9	that	that	PRON
ijassa-659	77	10	makes	make	VERB
ijassa-659	77	11	embeddings	embedding	NOUN
ijassa-659	77	12	of	of	ADP
ijassa-659	77	13	labeled	label	VERB
ijassa-659	77	14	examples	example	NOUN
ijassa-659	77	15	to	to	ADP
ijassa-659	77	16	those	those	PRON
ijassa-659	77	17	of	of	ADP
ijassa-659	77	18	unlabeled	unlabeled	ADJ
ijassa-659	77	19	ones	one	NOUN
ijassa-659	77	20	and	and	CCONJ
ijassa-659	77	21	back	back	ADV
ijassa-659	77	22	via	via	ADP
ijassa-659	77	23	deep	deep	ADJ
ijassa-659	77	24	neural	neural	ADJ
ijassa-659	77	25	networks	network	NOUN
ijassa-659	77	26	.	.	PUNCT
ijassa-659	78	1	the	the	DET
ijassa-659	78	2	active	active	ADJ
ijassa-659	78	3	scheme	scheme	NOUN
ijassa-659	78	4	makes	make	VERB
ijassa-659	78	5	association	association	NOUN
ijassa-659	78	6	cycles	cycle	NOUN
ijassa-659	78	7	that	that	PRON
ijassa-659	78	8	end	end	VERB
ijassa-659	78	9	up	up	ADP
ijassa-659	78	10	at	at	ADP
ijassa-659	78	11	the	the	DET
ijassa-659	78	12	same	same	ADJ
ijassa-659	78	13	class	class	NOUN
ijassa-659	78	14	from	from	ADP
ijassa-659	78	15	that	that	SCONJ
ijassa-659	78	16	the	the	DET
ijassa-659	78	17	association	association	NOUN
ijassa-659	78	18	was	be	AUX
ijassa-659	78	19	started	start	VERB
ijassa-659	78	20	,	,	PUNCT
ijassa-659	78	21	which	which	PRON
ijassa-659	78	22	considers	consider	VERB
ijassa-659	78	23	both	both	CCONJ
ijassa-659	78	24	the	the	DET
ijassa-659	78	25	informativeness	informativeness	NOUN
ijassa-659	78	26	and	and	CCONJ
ijassa-659	78	27	representativeness	representativeness	ADJ
ijassa-659	78	28	of	of	ADP
ijassa-659	78	29	examples	example	NOUN
ijassa-659	78	30	.	.	PUNCT
ijassa-659	79	1	the	the	DET
ijassa-659	79	2	above	above	ADV
ijassa-659	79	3	mentioned	mention	VERB
ijassa-659	79	4	methods	method	NOUN
ijassa-659	79	5	form	form	VERB
ijassa-659	79	6	arrange	arrange	NOUN
ijassa-659	79	7	of	of	ADP
ijassa-659	79	8	studies	study	NOUN
ijassa-659	79	9	performed	perform	VERB
ijassa-659	79	10	in	in	ADP
ijassa-659	79	11	active	active	ADJ
ijassa-659	79	12	selection	selection	NOUN
ijassa-659	79	13	of	of	ADP
ijassa-659	79	14	constraints	constraint	NOUN
ijassa-659	79	15	.	.	PUNCT
ijassa-659	80	1	each	each	DET
ijassa-659	80	2	method	method	NOUN
ijassa-659	80	3	considered	consider	VERB
ijassa-659	80	4	a	a	DET
ijassa-659	80	5	basic	basic	ADJ
ijassa-659	80	6	assumption	assumption	NOUN
ijassa-659	80	7	on	on	ADP
ijassa-659	80	8	utility	utility	NOUN
ijassa-659	80	9	of	of	ADP
ijassa-659	80	10	constraints	constraint	NOUN
ijassa-659	80	11	.	.	PUNCT
ijassa-659	81	1	their	their	PRON
ijassa-659	81	2	applicability	applicability	NOUN
ijassa-659	81	3	on	on	ADP
ijassa-659	81	4	a	a	DET
ijassa-659	81	5	specific	specific	ADJ
ijassa-659	81	6	problem	problem	NOUN
ijassa-659	81	7	is	be	AUX
ijassa-659	81	8	highly	highly	ADV
ijassa-659	81	9	dependent	dependent	ADJ
ijassa-659	81	10	on	on	ADP
ijassa-659	81	11	correlation	correlation	NOUN
ijassa-659	81	12	between	between	ADP
ijassa-659	81	13	their	their	PRON
ijassa-659	81	14	assumption	assumption	NOUN
ijassa-659	81	15	and	and	CCONJ
ijassa-659	81	16	the	the	DET
ijassa-659	81	17	actual	actual	ADJ
ijassa-659	81	18	structure	structure	NOUN
ijassa-659	81	19	of	of	ADP
ijassa-659	81	20	data	datum	NOUN
ijassa-659	81	21	.	.	PUNCT
ijassa-659	82	1	3	3	X
ijassa-659	82	2	.	.	X
ijassa-659	82	3	active	active	ADJ
ijassa-659	82	4	constraint	constraint	NOUN
ijassa-659	82	5	selection	selection	NOUN
ijassa-659	82	6	method	method	NOUN
ijassa-659	82	7	semi	semi	ADJ
ijassa-659	82	8	-	-	ADJ
ijassa-659	82	9	supervised	supervised	ADJ
ijassa-659	82	10	clustering	clustering	ADJ
ijassa-659	82	11	algorithms	algorithm	NOUN
ijassa-659	82	12	attempt	attempt	VERB
ijassa-659	82	13	to	to	PART
ijassa-659	82	14	partition	partition	VERB
ijassa-659	82	15	the	the	DET
ijassa-659	82	16	unlabeled	unlabele	VERB
ijassa-659	82	17	data	datum	NOUN
ijassa-659	82	18	into	into	ADP
ijassa-659	82	19	a	a	DET
ijassa-659	82	20	set	set	NOUN
ijassa-659	82	21	of	of	ADP
ijassa-659	82	22	clusters	cluster	NOUN
ijassa-659	82	23	with	with	ADP
ijassa-659	82	24	the	the	DET
ijassa-659	82	25	help	help	NOUN
ijassa-659	82	26	of	of	ADP
ijassa-659	82	27	a	a	DET
ijassa-659	82	28	small	small	ADJ
ijassa-659	82	29	amount	amount	NOUN
ijassa-659	82	30	of	of	ADP
ijassa-659	82	31	pairwise	pairwise	NOUN
ijassa-659	82	32	constraints	constraint	NOUN
ijassa-659	82	33	(	(	PUNCT
ijassa-659	82	34	must	must	AUX
ijassa-659	82	35	-	-	PUNCT
ijassa-659	82	36	link	link	VERB
ijassa-659	82	37	and	and	CCONJ
ijassa-659	82	38	cannot	cannot	NOUN
ijassa-659	82	39	-	-	PUNCT
ijassa-659	82	40	link	link	NOUN
ijassa-659	82	41	)	)	PUNCT
ijassa-659	82	42	.	.	PUNCT
ijassa-659	83	1	in	in	ADP
ijassa-659	83	2	this	this	DET
ijassa-659	83	3	section	section	NOUN
ijassa-659	83	4	,	,	PUNCT
ijassa-659	83	5	addressing	address	VERB
ijassa-659	83	6	the	the	DET
ijassa-659	83	7	problem	problem	NOUN
ijassa-659	83	8	of	of	ADP
ijassa-659	83	9	how	how	SCONJ
ijassa-659	83	10	to	to	PART
ijassa-659	83	11	effectively	effectively	ADV
ijassa-659	83	12	choose	choose	VERB
ijassa-659	83	13	pairwise	pairwise	NOUN
ijassa-659	83	14	constraints	constraint	NOUN
ijassa-659	83	15	to	to	PART
ijassa-659	83	16	produce	produce	VERB
ijassa-659	83	17	accurate	accurate	ADJ
ijassa-659	83	18	clustering	clustering	NOUN
ijassa-659	83	19	results	result	NOUN
ijassa-659	83	20	.	.	PUNCT
ijassa-659	84	1	the	the	DET
ijassa-659	84	2	proposed	propose	VERB
ijassa-659	84	3	method	method	NOUN
ijassa-659	84	4	first	first	ADV
ijassa-659	84	5	selects	select	VERB
ijassa-659	84	6	a	a	DET
ijassa-659	84	7	batch	batch	NOUN
ijassa-659	84	8	of	of	ADP
ijassa-659	84	9	most	most	ADV
ijassa-659	84	10	informative	informative	ADJ
ijassa-659	84	11	instances	instance	NOUN
ijassa-659	84	12	that	that	PRON
ijassa-659	84	13	minimize	minimize	VERB
ijassa-659	84	14	the	the	DET
ijassa-659	84	15	difference	difference	NOUN
ijassa-659	84	16	in	in	ADP
ijassa-659	84	17	distribution	distribution	NOUN
ijassa-659	84	18	between	between	ADP
ijassa-659	84	19	the	the	DET
ijassa-659	84	20	labeled	label	VERB
ijassa-659	84	21	and	and	CCONJ
ijassa-659	84	22	unlabeled	unlabeled	ADJ
ijassa-659	84	23	data	datum	NOUN
ijassa-659	84	24	.	.	PUNCT
ijassa-659	85	1	then	then	ADV
ijassa-659	85	2	,	,	PUNCT
ijassa-659	85	3	querying	query	VERB
ijassa-659	85	4	these	these	DET
ijassa-659	85	5	instances	instance	NOUN
ijassa-659	85	6	with	with	ADP
ijassa-659	85	7	the	the	DET
ijassa-659	85	8	existing	exist	VERB
ijassa-659	85	9	neighborhoods	neighborhood	NOUN
ijassa-659	85	10	to	to	PART
ijassa-659	85	11	determine	determine	VERB
ijassa-659	85	12	which	which	DET
ijassa-659	85	13	neighborhood	neighborhood	NOUN
ijassa-659	85	14	they	they	PRON
ijassa-659	85	15	belong	belong	VERB
ijassa-659	85	16	.	.	PUNCT
ijassa-659	86	1	28	28	NUM
ijassa-659	86	2	w.	w.	PROPN
ijassa-659	86	3	atwa	atwa	PROPN
ijassa-659	86	4	,	,	PUNCT
ijassa-659	86	5	m.	m.	NOUN
ijassa-659	86	6	emam	emam	PROPN
ijassa-659	86	7	copyright	copyright	NOUN
ijassa-659	86	8	©	©	PROPN
ijassa-659	86	9	2019	2019	NUM
ijassa-659	86	10	assa	assa	PROPN
ijassa-659	86	11	adv	adv	PROPN
ijassa-659	86	12	.	.	PUNCT
ijassa-659	87	1	in	in	ADP
ijassa-659	87	2	systems	system	NOUN
ijassa-659	87	3	science	science	NOUN
ijassa-659	87	4	and	and	CCONJ
ijassa-659	87	5	appl	appl	NOUN
ijassa-659	87	6	.	.	PUNCT
ijassa-659	88	1	(	(	PUNCT
ijassa-659	88	2	2019	2019	NUM
ijassa-659	88	3	)	)	PUNCT
ijassa-659	88	4	3.1	3.1	NUM
ijassa-659	88	5	batch	batch	NOUN
ijassa-659	88	6	instances	instance	NOUN
ijassa-659	88	7	selection	selection	NOUN
ijassa-659	88	8	traditional	traditional	ADJ
ijassa-659	88	9	data	datum	NOUN
ijassa-659	88	10	mining	mining	NOUN
ijassa-659	88	11	and	and	CCONJ
ijassa-659	88	12	machine	machine	NOUN
ijassa-659	88	13	learning	learning	NOUN
ijassa-659	88	14	algorithms	algorithm	NOUN
ijassa-659	88	15	are	be	AUX
ijassa-659	88	16	based	base	VERB
ijassa-659	88	17	on	on	ADP
ijassa-659	88	18	the	the	DET
ijassa-659	88	19	assumption	assumption	NOUN
ijassa-659	88	20	that	that	SCONJ
ijassa-659	88	21	the	the	DET
ijassa-659	88	22	training	training	NOUN
ijassa-659	88	23	data	datum	NOUN
ijassa-659	88	24	(	(	PUNCT
ijassa-659	88	25	x	x	X
ijassa-659	88	26	,	,	PUNCT
ijassa-659	88	27	y	y	NOUN
ijassa-659	88	28	)	)	PUNCT
ijassa-659	88	29	represents	represent	VERB
ijassa-659	88	30	the	the	DET
ijassa-659	88	31	true	true	ADJ
ijassa-659	88	32	underlying	underlying	ADJ
ijassa-659	88	33	distributions	distribution	NOUN
ijassa-659	88	34	of	of	ADP
ijassa-659	88	35	x	x	X
ijassa-659	88	36	and	and	CCONJ
ijassa-659	88	37	y	y	PROPN
ijassa-659	88	38	where	where	SCONJ
ijassa-659	88	39	x={x1	x={x1	PROPN
ijassa-659	88	40	,	,	PUNCT
ijassa-659	88	41	x2	x2	PROPN
ijassa-659	88	42	,	,	PUNCT
ijassa-659	88	43	…	…	PUNCT
ijassa-659	88	44	,	,	PUNCT
ijassa-659	88	45	xn	xn	PRON
ijassa-659	88	46	}	}	PUNCT
ijassa-659	88	47	is	be	AUX
ijassa-659	88	48	the	the	DET
ijassa-659	88	49	training	training	NOUN
ijassa-659	88	50	data	datum	NOUN
ijassa-659	88	51	and	and	CCONJ
ijassa-659	88	52	their	their	PRON
ijassa-659	88	53	corresponding	corresponding	ADJ
ijassa-659	88	54	labels	label	NOUN
ijassa-659	88	55	y={y1	y={y1	NOUN
ijassa-659	88	56	,	,	PUNCT
ijassa-659	88	57	y2	y2	PROPN
ijassa-659	88	58	,	,	PUNCT
ijassa-659	88	59	…	…	PUNCT
ijassa-659	88	60	,	,	PUNCT
ijassa-659	88	61	yn	yn	NOUN
ijassa-659	88	62	}	}	PUNCT
ijassa-659	88	63	.	.	PUNCT
ijassa-659	89	1	hence	hence	ADV
ijassa-659	89	2	a	a	DET
ijassa-659	89	3	model	model	NOUN
ijassa-659	89	4	learned	learn	VERB
ijassa-659	89	5	on	on	ADP
ijassa-659	89	6	this	this	DET
ijassa-659	89	7	data	data	NOUN
ijassa-659	89	8	works	work	VERB
ijassa-659	89	9	well	well	ADV
ijassa-659	89	10	for	for	ADP
ijassa-659	89	11	the	the	DET
ijassa-659	89	12	test	test	NOUN
ijassa-659	89	13	data	datum	NOUN
ijassa-659	89	14	(	(	PUNCT
ijassa-659	89	15	xtest	xtest	ADJ
ijassa-659	89	16	,	,	PUNCT
ijassa-659	89	17	ytest	yt	ADJ
ijassa-659	89	18	)	)	PUNCT
ijassa-659	89	19	which	which	PRON
ijassa-659	89	20	is	be	AUX
ijassa-659	89	21	also	also	ADV
ijassa-659	89	22	drawn	draw	VERB
ijassa-659	89	23	independently	independently	ADV
ijassa-659	89	24	and	and	CCONJ
ijassa-659	89	25	identically	identically	ADV
ijassa-659	89	26	distributed	distribute	VERB
ijassa-659	89	27	from	from	ADP
ijassa-659	89	28	the	the	DET
ijassa-659	89	29	same	same	ADJ
ijassa-659	89	30	distribution	distribution	NOUN
ijassa-659	89	31	[	[	X
ijassa-659	89	32	5	5	NUM
ijassa-659	89	33	]	]	PUNCT
ijassa-659	89	34	.	.	PUNCT
ijassa-659	90	1	thus	thus	ADV
ijassa-659	90	2	,	,	PUNCT
ijassa-659	90	3	a	a	DET
ijassa-659	90	4	batch	batch	NOUN
ijassa-659	90	5	of	of	ADP
ijassa-659	90	6	query	query	NOUN
ijassa-659	90	7	instances	instance	NOUN
ijassa-659	90	8	can	can	AUX
ijassa-659	90	9	be	be	AUX
ijassa-659	90	10	selected	select	VERB
ijassa-659	90	11	from	from	ADP
ijassa-659	90	12	unlabeled	unlabeled	ADJ
ijassa-659	90	13	data	datum	NOUN
ijassa-659	90	14	such	such	ADJ
ijassa-659	90	15	that	that	SCONJ
ijassa-659	90	16	the	the	DET
ijassa-659	90	17	distribution	distribution	NOUN
ijassa-659	90	18	represented	represent	VERB
ijassa-659	90	19	by	by	ADP
ijassa-659	90	20	the	the	DET
ijassa-659	90	21	queried	query	VERB
ijassa-659	90	22	and	and	CCONJ
ijassa-659	90	23	labeled	label	VERB
ijassa-659	90	24	data	datum	NOUN
ijassa-659	90	25	is	be	AUX
ijassa-659	90	26	similar	similar	ADJ
ijassa-659	90	27	to	to	ADP
ijassa-659	90	28	the	the	DET
ijassa-659	90	29	probability	probability	NOUN
ijassa-659	90	30	distribution	distribution	NOUN
ijassa-659	90	31	of	of	ADP
ijassa-659	90	32	the	the	DET
ijassa-659	90	33	unlabeled	unlabele	VERB
ijassa-659	90	34	data	datum	NOUN
ijassa-659	90	35	set	set	VERB
ijassa-659	90	36	.	.	PUNCT
ijassa-659	91	1	in	in	ADP
ijassa-659	91	2	other	other	ADJ
ijassa-659	91	3	words	word	NOUN
ijassa-659	91	4	,	,	PUNCT
ijassa-659	91	5	selecting	select	VERB
ijassa-659	91	6	a	a	DET
ijassa-659	91	7	batch	batch	NOUN
ijassa-659	91	8	of	of	ADP
ijassa-659	91	9	instances	instance	NOUN
ijassa-659	91	10	s	s	VERB
ijassa-659	91	11	from	from	ADP
ijassa-659	91	12	the	the	DET
ijassa-659	91	13	unlabeled	unlabeled	ADJ
ijassa-659	91	14	data	datum	NOUN
ijassa-659	91	15	(	(	PUNCT
ijassa-659	91	16	denoted	denote	VERB
ijassa-659	91	17	by	by	ADP
ijassa-659	91	18	du	du	PROPN
ijassa-659	91	19	)	)	PUNCT
ijassa-659	91	20	such	such	ADJ
ijassa-659	91	21	that	that	SCONJ
ijassa-659	91	22	the	the	DET
ijassa-659	91	23	joint	joint	ADJ
ijassa-659	91	24	probability	probability	NOUN
ijassa-659	91	25	distribution	distribution	NOUN
ijassa-659	91	26	represented	represent	VERB
ijassa-659	91	27	by	by	ADP
ijassa-659	91	28	dl	dl	PROPN
ijassa-659	91	29	∪	∪	PROPN
ijassa-659	91	30	s	s	PROPN
ijassa-659	91	31	and	and	CCONJ
ijassa-659	91	32	du	du	PROPN
ijassa-659	91	33	\	\	PROPN
ijassa-659	91	34	s	s	NOUN
ijassa-659	91	35	are	be	AUX
ijassa-659	91	36	similar	similar	ADJ
ijassa-659	91	37	to	to	ADP
ijassa-659	91	38	each	each	DET
ijassa-659	91	39	other	other	ADJ
ijassa-659	91	40	,	,	PUNCT
ijassa-659	91	41	where	where	SCONJ
ijassa-659	91	42	dl	dl	PROPN
ijassa-659	91	43	is	be	AUX
ijassa-659	91	44	set	set	VERB
ijassa-659	91	45	of	of	ADP
ijassa-659	91	46	available	available	ADJ
ijassa-659	91	47	labeled	label	VERB
ijassa-659	91	48	data	datum	NOUN
ijassa-659	91	49	.	.	PUNCT
ijassa-659	92	1	this	this	DET
ijassa-659	92	2	function	function	NOUN
ijassa-659	92	3	is	be	AUX
ijassa-659	92	4	summarized	summarize	VERB
ijassa-659	92	5	in	in	ADP
ijassa-659	92	6	algorithm	algorithm	NOUN
ijassa-659	92	7	1	1	NUM
ijassa-659	92	8	.	.	PUNCT
ijassa-659	92	9	to	to	PART
ijassa-659	92	10	measure	measure	VERB
ijassa-659	92	11	the	the	DET
ijassa-659	92	12	difference	difference	NOUN
ijassa-659	92	13	between	between	ADP
ijassa-659	92	14	two	two	NUM
ijassa-659	92	15	distributions	distribution	NOUN
ijassa-659	92	16	,	,	PUNCT
ijassa-659	92	17	maximum	maximum	ADJ
ijassa-659	92	18	mean	mean	NOUN
ijassa-659	92	19	discrepancy	discrepancy	NOUN
ijassa-659	92	20	(	(	PUNCT
ijassa-659	92	21	mmd	mmd	PROPN
ijassa-659	92	22	)	)	PUNCT
ijassa-659	92	23	has	have	AUX
ijassa-659	92	24	been	be	AUX
ijassa-659	92	25	shown	show	VERB
ijassa-659	92	26	to	to	PART
ijassa-659	92	27	be	be	AUX
ijassa-659	92	28	an	an	DET
ijassa-659	92	29	effective	effective	ADJ
ijassa-659	92	30	measure	measure	NOUN
ijassa-659	92	31	of	of	ADP
ijassa-659	92	32	the	the	DET
ijassa-659	92	33	difference	difference	NOUN
ijassa-659	92	34	in	in	ADP
ijassa-659	92	35	their	their	PRON
ijassa-659	92	36	marginal	marginal	ADJ
ijassa-659	92	37	probability	probability	NOUN
ijassa-659	92	38	distributions	distribution	NOUN
ijassa-659	92	39	[	[	X
ijassa-659	92	40	5	5	NUM
ijassa-659	92	41	,	,	PUNCT
ijassa-659	92	42	6	6	NUM
ijassa-659	92	43	]	]	PUNCT
ijassa-659	92	44	.	.	PUNCT
ijassa-659	93	1	the	the	DET
ijassa-659	93	2	principal	principal	NOUN
ijassa-659	93	3	underlying	underlie	VERB
ijassa-659	93	4	the	the	DET
ijassa-659	93	5	maximum	maximum	ADJ
ijassa-659	93	6	mean	mean	NOUN
ijassa-659	93	7	discrepancy	discrepancy	NOUN
ijassa-659	93	8	is	be	AUX
ijassa-659	93	9	to	to	PART
ijassa-659	93	10	find	find	VERB
ijassa-659	93	11	a	a	DET
ijassa-659	93	12	function	function	NOUN
ijassa-659	93	13	that	that	PRON
ijassa-659	93	14	assumes	assume	VERB
ijassa-659	93	15	different	different	ADJ
ijassa-659	93	16	expectations	expectation	NOUN
ijassa-659	93	17	on	on	ADP
ijassa-659	93	18	two	two	NUM
ijassa-659	93	19	different	different	ADJ
ijassa-659	93	20	distributions	distribution	NOUN
ijassa-659	93	21	so	so	SCONJ
ijassa-659	93	22	that	that	SCONJ
ijassa-659	93	23	when	when	SCONJ
ijassa-659	93	24	evaluated	evaluate	VERB
ijassa-659	93	25	empirically	empirically	ADV
ijassa-659	93	26	on	on	ADP
ijassa-659	93	27	samples	sample	NOUN
ijassa-659	93	28	drawn	draw	VERB
ijassa-659	93	29	from	from	ADP
ijassa-659	93	30	the	the	DET
ijassa-659	93	31	different	different	ADJ
ijassa-659	93	32	distributions	distribution	NOUN
ijassa-659	93	33	it	it	PRON
ijassa-659	93	34	would	would	AUX
ijassa-659	93	35	tell	tell	VERB
ijassa-659	93	36	us	we	PRON
ijassa-659	93	37	whether	whether	SCONJ
ijassa-659	93	38	the	the	DET
ijassa-659	93	39	distributions	distribution	NOUN
ijassa-659	93	40	are	be	AUX
ijassa-659	93	41	similar	similar	ADJ
ijassa-659	93	42	or	or	CCONJ
ijassa-659	93	43	not	not	PART
ijassa-659	93	44	.	.	PUNCT
ijassa-659	94	1	let	let	VERB
ijassa-659	94	2	ℱ	ℱ	PRON
ijassa-659	94	3	be	be	AUX
ijassa-659	94	4	a	a	DET
ijassa-659	94	5	class	class	NOUN
ijassa-659	94	6	of	of	ADP
ijassa-659	94	7	functions𝑓	functions𝑓	NOUN
ijassa-659	94	8	:	:	PUNCT
ijassa-659	94	9	𝒳	𝒳	PROPN
ijassa-659	94	10	→	→	SYM
ijassa-659	94	11	ℝ	ℝ	PROPN
ijassa-659	94	12	..	..	PUNCT
ijassa-659	94	13	let	let	VERB
ijassa-659	94	14	p	p	NOUN
ijassa-659	94	15	and	and	CCONJ
ijassa-659	94	16	q	q	AUX
ijassa-659	94	17	be	be	AUX
ijassa-659	94	18	probability	probability	NOUN
ijassa-659	94	19	distributions	distribution	NOUN
ijassa-659	94	20	defined	define	VERB
ijassa-659	94	21	on	on	ADP
ijassa-659	94	22	a	a	DET
ijassa-659	94	23	domain	domain	NOUN
ijassa-659	94	24	𝒳	𝒳	PROPN
ijassa-659	94	25	,	,	PUNCT
ijassa-659	94	26	and	and	CCONJ
ijassa-659	94	27	let	let	VERB
ijassa-659	94	28	x	x	PUNCT
ijassa-659	94	29	=	=	SYM
ijassa-659	94	30	(	(	PUNCT
ijassa-659	94	31	x1	x1	PROPN
ijassa-659	94	32	,	,	PUNCT
ijassa-659	94	33	.	.	PUNCT
ijassa-659	94	34	.	.	PUNCT
ijassa-659	95	1	.	.	PUNCT
ijassa-659	96	1	,	,	PUNCT
ijassa-659	96	2	xm	xm	PROPN
ijassa-659	96	3	)	)	PUNCT
ijassa-659	96	4	and	and	CCONJ
ijassa-659	96	5	z	z	NOUN
ijassa-659	96	6	=	=	SYM
ijassa-659	96	7	(	(	PUNCT
ijassa-659	96	8	z1	z1	PROPN
ijassa-659	96	9	,	,	PUNCT
ijassa-659	96	10	.	.	PUNCT
ijassa-659	96	11	.	.	PUNCT
ijassa-659	96	12	.	.	PUNCT
ijassa-659	97	1	,	,	PUNCT
ijassa-659	97	2	zn	zn	AUX
ijassa-659	97	3	)	)	PUNCT
ijassa-659	97	4	be	be	VERB
ijassa-659	97	5	samples	sample	NOUN
ijassa-659	97	6	composed	compose	VERB
ijassa-659	97	7	of	of	ADP
ijassa-659	97	8	independent	independent	ADJ
ijassa-659	97	9	and	and	CCONJ
ijassa-659	97	10	identically	identically	ADV
ijassa-659	97	11	distributed	distribute	VERB
ijassa-659	97	12	observations	observation	NOUN
ijassa-659	97	13	drawn	draw	VERB
ijassa-659	97	14	from	from	ADP
ijassa-659	97	15	p	p	PROPN
ijassa-659	97	16	and	and	CCONJ
ijassa-659	97	17	q	q	NOUN
ijassa-659	97	18	,	,	PUNCT
ijassa-659	97	19	respectively	respectively	ADV
ijassa-659	97	20	.	.	PUNCT
ijassa-659	98	1	the	the	DET
ijassa-659	98	2	maximum	maximum	ADJ
ijassa-659	98	3	mean	mean	NOUN
ijassa-659	98	4	discrepancy	discrepancy	NOUN
ijassa-659	98	5	(	(	PUNCT
ijassa-659	98	6	mmd	mmd	X
ijassa-659	98	7	)	)	PUNCT
ijassa-659	99	1	[	[	X
ijassa-659	99	2	5	5	NUM
ijassa-659	99	3	,	,	PUNCT
ijassa-659	99	4	6	6	NUM
ijassa-659	99	5	]	]	PUNCT
ijassa-659	99	6	and	and	CCONJ
ijassa-659	99	7	its	its	PRON
ijassa-659	99	8	empirical	empirical	ADJ
ijassa-659	99	9	estimate	estimate	NOUN
ijassa-659	99	10	are	be	AUX
ijassa-659	99	11	defined	define	VERB
ijassa-659	99	12	as	as	ADP
ijassa-659	99	13	:	:	PUNCT
ijassa-659	99	14	mmd[ℱ	mmd[ℱ	X
ijassa-659	99	15	,	,	PUNCT
ijassa-659	99	16	𝑝	𝑝	NOUN
ijassa-659	99	17	,	,	PUNCT
ijassa-659	99	18	𝑞	𝑞	X
ijassa-659	99	19	]	]	X
ijassa-659	99	20	≔	≔	NOUN
ijassa-659	99	21	sup	sup	NOUN
ijassa-659	99	22	𝑓∈ℱ	𝑓∈ℱ	PROPN
ijassa-659	99	23	(	(	PUNCT
ijassa-659	99	24	𝐸𝑝[𝑓(𝑥	𝐸𝑝[𝑓(𝑥	PROPN
ijassa-659	99	25	)	)	PUNCT
ijassa-659	99	26	]	]	PUNCT
ijassa-659	99	27	−	−	PUNCT
ijassa-659	99	28	𝐸𝑞[𝑓(𝑧	𝐸𝑞[𝑓(𝑧	VERB
ijassa-659	99	29	)	)	PUNCT
ijassa-659	99	30	]	]	PUNCT
ijassa-659	99	31	)	)	PUNCT
ijassa-659	99	32	(	(	PUNCT
ijassa-659	99	33	1	1	X
ijassa-659	99	34	)	)	PUNCT
ijassa-659	99	35	mmd[ℱ	mmd[ℱ	PROPN
ijassa-659	99	36	,	,	PUNCT
ijassa-659	99	37	𝑋	𝑋	PROPN
ijassa-659	99	38	,	,	PUNCT
ijassa-659	99	39	𝑍	𝑍	PROPN
ijassa-659	99	40	]	]	PUNCT
ijassa-659	99	41	≔	≔	NOUN
ijassa-659	99	42	sup	sup	NOUN
ijassa-659	99	43	𝑓∈ℱ	𝑓∈ℱ	PROPN
ijassa-659	99	44	(	(	PUNCT
ijassa-659	99	45	1	1	NUM
ijassa-659	99	46	𝑚	𝑚	NOUN
ijassa-659	99	47	∑	∑	NOUN
ijassa-659	99	48	𝑓𝑚	𝑓𝑚	PROPN
ijassa-659	99	49	𝑖=1	𝑖=1	PROPN
ijassa-659	99	50	(	(	PUNCT
ijassa-659	99	51	𝑥𝑖	𝑥𝑖	NOUN
ijassa-659	99	52	)	)	PUNCT
ijassa-659	99	53	−	−	PROPN
ijassa-659	99	54	1	1	NUM
ijassa-659	99	55	𝑛	𝑛	PROPN
ijassa-659	99	56	∑	∑	PROPN
ijassa-659	99	57	𝑓𝑛	𝑓𝑛	PROPN
ijassa-659	99	58	𝑖=1	𝑖=1	PROPN
ijassa-659	99	59	(	(	PUNCT
ijassa-659	99	60	𝑧𝑖	𝑧𝑖	NOUN
ijassa-659	99	61	)	)	PUNCT
ijassa-659	99	62	)	)	PUNCT
ijassa-659	100	1	(	(	PUNCT
ijassa-659	100	2	2	2	X
ijassa-659	100	3	)	)	PUNCT
ijassa-659	100	4	there	there	PRON
ijassa-659	100	5	is	be	VERB
ijassa-659	100	6	a	a	DET
ijassa-659	100	7	class	class	NOUN
ijassa-659	100	8	of	of	ADP
ijassa-659	100	9	functions	function	NOUN
ijassa-659	100	10	for	for	ADP
ijassa-659	100	11	which	which	PRON
ijassa-659	100	12	mmd	mmd	NOUN
ijassa-659	100	13	may	may	AUX
ijassa-659	100	14	easily	easily	ADV
ijassa-659	100	15	be	be	AUX
ijassa-659	100	16	computed	compute	VERB
ijassa-659	100	17	,	,	PUNCT
ijassa-659	100	18	while	while	SCONJ
ijassa-659	100	19	retaining	retain	VERB
ijassa-659	100	20	the	the	DET
ijassa-659	100	21	ability	ability	NOUN
ijassa-659	100	22	to	to	PART
ijassa-659	100	23	detect	detect	VERB
ijassa-659	100	24	all	all	DET
ijassa-659	100	25	discrepancies	discrepancy	NOUN
ijassa-659	100	26	between	between	ADP
ijassa-659	100	27	p	p	NOUN
ijassa-659	100	28	and	and	CCONJ
ijassa-659	100	29	q	q	NOUN
ijassa-659	100	30	without	without	ADP
ijassa-659	100	31	making	make	VERB
ijassa-659	100	32	any	any	DET
ijassa-659	100	33	simplifying	simplifying	NOUN
ijassa-659	100	34	assumptions	assumption	NOUN
ijassa-659	100	35	.	.	PUNCT
ijassa-659	101	1	let	let	VERB
ijassa-659	101	2	ℋ	ℋ	PRON
ijassa-659	101	3	be	be	AUX
ijassa-659	101	4	a	a	DET
ijassa-659	101	5	complete	complete	ADJ
ijassa-659	101	6	inner	inner	ADJ
ijassa-659	101	7	product	product	NOUN
ijassa-659	101	8	space	space	NOUN
ijassa-659	101	9	(	(	PUNCT
ijassa-659	101	10	i.e.	i.e.	X
ijassa-659	101	11	,	,	PUNCT
ijassa-659	101	12	a	a	DET
ijassa-659	101	13	hilbert	hilbert	NOUN
ijassa-659	101	14	space	space	NOUN
ijassa-659	101	15	)	)	PUNCT
ijassa-659	101	16	of	of	ADP
ijassa-659	101	17	functions𝑓	functions𝑓	NOUN
ijassa-659	101	18	:	:	PUNCT
ijassa-659	101	19	𝒳	𝒳	PROPN
ijassa-659	101	20	→	→	SYM
ijassa-659	101	21	ℝ	ℝ	PROPN
ijassa-659	101	22	,	,	PUNCT
ijassa-659	101	23	where	where	SCONJ
ijassa-659	101	24	𝒳	𝒳	PROPN
ijassa-659	101	25	is	be	AUX
ijassa-659	101	26	a	a	DET
ijassa-659	101	27	nonempty	nonempty	ADJ
ijassa-659	101	28	compact	compact	ADJ
ijassa-659	101	29	set	set	NOUN
ijassa-659	101	30	.	.	PUNCT
ijassa-659	102	1	then	then	ADV
ijassa-659	102	2	ℋ	ℋ	PROPN
ijassa-659	102	3	is	be	AUX
ijassa-659	102	4	termed	term	VERB
ijassa-659	102	5	a	a	DET
ijassa-659	102	6	reproducing	reproduce	VERB
ijassa-659	102	7	kernel	kernel	NOUN
ijassa-659	102	8	hilbert	hilbert	PROPN
ijassa-659	102	9	space	space	NOUN
ijassa-659	102	10	if	if	SCONJ
ijassa-659	102	11	for	for	ADP
ijassa-659	102	12	all	all	DET
ijassa-659	102	13	x	x	SYM
ijassa-659	102	14	∈	∈	PROPN
ijassa-659	102	15	𝒳	𝒳	PROPN
ijassa-659	102	16	,	,	PUNCT
ijassa-659	102	17	the	the	DET
ijassa-659	102	18	linear	linear	ADJ
ijassa-659	102	19	point	point	NOUN
ijassa-659	102	20	evaluation	evaluation	NOUN
ijassa-659	102	21	functional	functional	ADJ
ijassa-659	102	22	mapping	mapping	NOUN
ijassa-659	102	23	f	f	PROPN
ijassa-659	102	24	→	→	SYM
ijassa-659	102	25	f(x	f(x	PROPN
ijassa-659	102	26	)	)	PUNCT
ijassa-659	102	27	exists	exist	VERB
ijassa-659	102	28	and	and	CCONJ
ijassa-659	102	29	continuous	continuous	ADJ
ijassa-659	102	30	.	.	PUNCT
ijassa-659	103	1	in	in	ADP
ijassa-659	103	2	this	this	DET
ijassa-659	103	3	case	case	NOUN
ijassa-659	103	4	,	,	PUNCT
ijassa-659	103	5	f(x	f(x	PROPN
ijassa-659	103	6	)	)	PUNCT
ijassa-659	103	7	can	can	AUX
ijassa-659	103	8	be	be	AUX
ijassa-659	103	9	expressed	express	VERB
ijassa-659	103	10	as	as	ADP
ijassa-659	103	11	an	an	DET
ijassa-659	103	12	inner	inner	ADJ
ijassa-659	103	13	product	product	NOUN
ijassa-659	103	14	via	via	ADP
ijassa-659	103	15	𝑓(𝑥	𝑓(𝑥	NOUN
ijassa-659	103	16	)	)	PUNCT
ijassa-659	103	17	=	=	PUNCT
ijassa-659	104	1	〈	〈	PROPN
ijassa-659	104	2	𝑓𝜙(𝑥)〉ℋ	𝑓𝜙(𝑥)〉ℋ	NOUN
ijassa-659	104	3	(	(	PUNCT
ijassa-659	104	4	3	3	X
ijassa-659	104	5	)	)	PUNCT
ijassa-659	104	6	where	where	SCONJ
ijassa-659	104	7	𝜙	𝜙	NOUN
ijassa-659	104	8	:	:	PUNCT
ijassa-659	104	9	𝒳	𝒳	PROPN
ijassa-659	104	10	→	→	SYM
ijassa-659	104	11	ℋ	ℋ	PROPN
ijassa-659	104	12	is	be	AUX
ijassa-659	104	13	known	know	VERB
ijassa-659	104	14	as	as	ADP
ijassa-659	104	15	the	the	DET
ijassa-659	104	16	feature	feature	NOUN
ijassa-659	104	17	space	space	NOUN
ijassa-659	104	18	map	map	NOUN
ijassa-659	104	19	from	from	ADP
ijassa-659	104	20	𝒳	𝒳	PRON
ijassa-659	104	21	to	to	ADP
ijassa-659	104	22	ℋ.	ℋ.	PROPN
ijassa-659	104	23	when	when	SCONJ
ijassa-659	104	24	ℱ	ℱ	PROPN
ijassa-659	104	25	is	be	AUX
ijassa-659	104	26	the	the	DET
ijassa-659	104	27	unit	unit	NOUN
ijassa-659	104	28	ball	ball	NOUN
ijassa-659	104	29	in	in	ADP
ijassa-659	104	30	a	a	DET
ijassa-659	104	31	characteristic	characteristic	ADJ
ijassa-659	104	32	rkhs	rkh	NOUN
ijassa-659	105	1	[	[	X
ijassa-659	105	2	6	6	NUM
ijassa-659	105	3	]	]	PUNCT
ijassa-659	105	4	,	,	PUNCT
ijassa-659	105	5	mmd	mmd	PROPN
ijassa-659	105	6	is	be	AUX
ijassa-659	105	7	defined	define	VERB
ijassa-659	105	8	as	as	ADP
ijassa-659	105	9	the	the	DET
ijassa-659	105	10	difference	difference	NOUN
ijassa-659	105	11	between	between	ADP
ijassa-659	105	12	the	the	DET
ijassa-659	105	13	means	mean	NOUN
ijassa-659	105	14	of	of	ADP
ijassa-659	105	15	two	two	NUM
ijassa-659	105	16	distributions	distribution	NOUN
ijassa-659	105	17	after	after	ADP
ijassa-659	105	18	mapping	map	VERB
ijassa-659	105	19	onto	onto	ADP
ijassa-659	105	20	the	the	DET
ijassa-659	105	21	characteristic	characteristic	ADJ
ijassa-659	105	22	rkhs	rkh	NOUN
ijassa-659	105	23	.	.	PUNCT
ijassa-659	106	1	an	an	DET
ijassa-659	106	2	empirical	empirical	ADJ
ijassa-659	106	3	estimate	estimate	NOUN
ijassa-659	106	4	of	of	ADP
ijassa-659	106	5	mmd	mmd	PROPN
ijassa-659	106	6	is	be	AUX
ijassa-659	106	7	then	then	ADV
ijassa-659	106	8	obtained	obtain	VERB
ijassa-659	106	9	as	as	ADP
ijassa-659	106	10	follows	follow	VERB
ijassa-659	106	11	:	:	PUNCT
ijassa-659	106	12	mmd[𝜙	mmd[𝜙	NOUN
ijassa-659	106	13	,	,	PUNCT
ijassa-659	106	14	𝑋	𝑋	PROPN
ijassa-659	106	15	,	,	PUNCT
ijassa-659	106	16	𝑍	𝑍	PROPN
ijassa-659	106	17	]	]	PUNCT
ijassa-659	106	18	≔	≔	NOUN
ijassa-659	106	19	‖	‖	NOUN
ijassa-659	106	20	1	1	NUM
ijassa-659	106	21	𝑚	𝑚	PRON
ijassa-659	106	22	∑	∑	PUNCT
ijassa-659	106	23	φ(𝑥𝑖	φ(𝑥𝑖	ADJ
ijassa-659	106	24	)	)	PUNCT
ijassa-659	106	25	−	−	PROPN
ijassa-659	106	26	1	1	NUM
ijassa-659	106	27	𝑛	𝑛	PROPN
ijassa-659	106	28	∑	∑	PUNCT
ijassa-659	106	29	φ(𝑧𝑖	φ(𝑧𝑖	PROPN
ijassa-659	106	30	)	)	PUNCT
ijassa-659	107	1	𝑛	𝑛	PRON
ijassa-659	107	2	𝑖=1	𝑖=1	PUNCT
ijassa-659	107	3	𝑚	𝑚	X
ijassa-659	107	4	𝑖=1	𝑖=1	PUNCT
ijassa-659	107	5	‖	‖	PROPN
ijassa-659	107	6	ℋ	ℋ	PROPN
ijassa-659	107	7	2	2	NUM
ijassa-659	107	8	(	(	PUNCT
ijassa-659	107	9	4	4	NUM
ijassa-659	107	10	)	)	PUNCT
ijassa-659	107	11	now	now	ADV
ijassa-659	107	12	,	,	PUNCT
ijassa-659	107	13	let	let	VERB
ijassa-659	107	14	us	we	PRON
ijassa-659	107	15	assume	assume	VERB
ijassa-659	107	16	that	that	SCONJ
ijassa-659	107	17	we	we	PRON
ijassa-659	107	18	have	have	VERB
ijassa-659	107	19	u	u	NOUN
ijassa-659	107	20	instances	instance	NOUN
ijassa-659	107	21	of	of	ADP
ijassa-659	107	22	unlabeled	unlabeled	ADJ
ijassa-659	107	23	data	datum	NOUN
ijassa-659	107	24	du	du	PROPN
ijassa-659	107	25	and	and	CCONJ
ijassa-659	107	26	l	l	NOUN
ijassa-659	107	27	instances	instance	NOUN
ijassa-659	107	28	of	of	ADP
ijassa-659	107	29	labeled	label	VERB
ijassa-659	107	30	data	datum	NOUN
ijassa-659	108	1	dl	dl	PROPN
ijassa-659	109	1	and	and	CCONJ
ijassa-659	109	2	we	we	PRON
ijassa-659	109	3	would	would	AUX
ijassa-659	109	4	like	like	VERB
ijassa-659	109	5	to	to	PART
ijassa-659	109	6	select	select	VERB
ijassa-659	109	7	a	a	DET
ijassa-659	109	8	batch	batch	NOUN
ijassa-659	109	9	s	s	NOUN
ijassa-659	109	10	of	of	ADP
ijassa-659	109	11	b	b	NOUN
ijassa-659	109	12	instances	instance	NOUN
ijassa-659	109	13	such	such	ADJ
ijassa-659	109	14	that	that	SCONJ
ijassa-659	109	15	the	the	DET
ijassa-659	109	16	distribution	distribution	NOUN
ijassa-659	109	17	of	of	ADP
ijassa-659	109	18	dl	dl	PROPN
ijassa-659	109	19	∪	∪	PROPN
ijassa-659	109	20	s	s	NOUN
ijassa-659	109	21	is	be	AUX
ijassa-659	109	22	similar	similar	ADJ
ijassa-659	109	23	to	to	ADP
ijassa-659	109	24	the	the	DET
ijassa-659	109	25	distribution	distribution	NOUN
ijassa-659	109	26	of	of	ADP
ijassa-659	109	27	du	du	PROPN
ijassa-659	109	28	\	\	PROPN
ijassa-659	109	29	s.	s.	PROPN
ijassa-659	109	30	thus	thus	ADV
ijassa-659	109	31	,	,	PUNCT
ijassa-659	109	32	the	the	DET
ijassa-659	109	33	mmd	mmd	NOUN
ijassa-659	109	34	between	between	ADP
ijassa-659	109	35	the	the	DET
ijassa-659	109	36	sets	set	NOUN
ijassa-659	109	37	dl	dl	PROPN
ijassa-659	109	38	∪	∪	PROPN
ijassa-659	109	39	s	s	PROPN
ijassa-659	109	40	and	and	CCONJ
ijassa-659	109	41	du	du	PROPN
ijassa-659	109	42	\	\	PROPN
ijassa-659	109	43	s	s	PART
ijassa-659	109	44	is	be	AUX
ijassa-659	109	45	defined	define	VERB
ijassa-659	109	46	by	by	ADP
ijassa-659	109	47	f(s	f(	NOUN
ijassa-659	109	48	)	)	PUNCT
ijassa-659	109	49	,	,	PUNCT
ijassa-659	109	50	can	can	AUX
ijassa-659	109	51	be	be	AUX
ijassa-659	109	52	computed	compute	VERB
ijassa-659	109	53	using	use	VERB
ijassa-659	109	54	the	the	DET
ijassa-659	109	55	expression	expression	NOUN
ijassa-659	109	56	in	in	ADP
ijassa-659	109	57	equation	equation	NOUN
ijassa-659	109	58	(	(	PUNCT
ijassa-659	109	59	4	4	NUM
ijassa-659	109	60	)	)	PUNCT
ijassa-659	109	61	,	,	PUNCT
ijassa-659	109	62	as	as	SCONJ
ijassa-659	109	63	follows	follow	VERB
ijassa-659	109	64	:	:	PUNCT
ijassa-659	109	65	𝑓(𝑆	𝑓(𝑆	X
ijassa-659	109	66	)	)	PUNCT
ijassa-659	110	1	=	=	PUNCT
ijassa-659	110	2	‖	‖	PROPN
ijassa-659	110	3	1	1	NUM
ijassa-659	110	4	𝑙+𝑏	𝑙+𝑏	NUM
ijassa-659	110	5	∑	∑	PUNCT
ijassa-659	110	6	φ(𝑥𝑗	φ(𝑥𝑗	NOUN
ijassa-659	110	7	)	)	PUNCT
ijassa-659	110	8	−	−	NUM
ijassa-659	110	9	1	1	NUM
ijassa-659	110	10	𝑢−𝑏	𝑢−𝑏	NOUN
ijassa-659	110	11	∑	∑	ADP
ijassa-659	110	12	φ(𝑥𝑖)𝑖∈𝐷𝑈∖𝑆𝑗∈𝐷𝐿∪𝑆	φ(𝑥𝑖)𝑖∈𝐷𝑈∖𝑆𝑗∈𝐷𝐿∪𝑆	ADP
ijassa-659	110	13	‖	‖	PROPN
ijassa-659	110	14	ℋ	ℋ	PROPN
ijassa-659	110	15	2	2	NUM
ijassa-659	110	16	(	(	PUNCT
ijassa-659	110	17	5	5	NUM
ijassa-659	110	18	)	)	PUNCT
ijassa-659	110	19	improving	improve	VERB
ijassa-659	110	20	semi	semi	ADJ
ijassa-659	110	21	-	-	ADJ
ijassa-659	110	22	supervised	supervised	ADJ
ijassa-659	110	23	clustering	clustering	ADJ
ijassa-659	110	24	algorithms	algorithm	NOUN
ijassa-659	110	25	with	with	ADP
ijassa-659	110	26	active	active	ADJ
ijassa-659	110	27	query	query	NOUN
ijassa-659	110	28	selection	selection	NOUN
ijassa-659	110	29	29	29	NUM
ijassa-659	110	30	copyright	copyright	NOUN
ijassa-659	110	31	©	©	PROPN
ijassa-659	110	32	2019	2019	NUM
ijassa-659	110	33	assa	assa	PROPN
ijassa-659	110	34	adv	adv	PROPN
ijassa-659	110	35	.	.	PUNCT
ijassa-659	111	1	in	in	ADP
ijassa-659	111	2	systems	system	NOUN
ijassa-659	111	3	science	science	NOUN
ijassa-659	111	4	and	and	CCONJ
ijassa-659	111	5	appl	appl	NOUN
ijassa-659	111	6	.	.	PUNCT
ijassa-659	112	1	(	(	PUNCT
ijassa-659	112	2	2019	2019	NUM
ijassa-659	112	3	)	)	PUNCT
ijassa-659	112	4	since	since	SCONJ
ijassa-659	112	5	we	we	PRON
ijassa-659	112	6	want	want	VERB
ijassa-659	112	7	to	to	PART
ijassa-659	112	8	select	select	VERB
ijassa-659	112	9	a	a	DET
ijassa-659	112	10	set	set	NOUN
ijassa-659	112	11	s	s	VERB
ijassa-659	112	12	from	from	ADP
ijassa-659	112	13	unlabeled	unlabeled	ADJ
ijassa-659	112	14	data	datum	NOUN
ijassa-659	112	15	set	set	VERB
ijassa-659	112	16	du	du	NOUN
ijassa-659	112	17	to	to	PART
ijassa-659	112	18	minimize	minimize	VERB
ijassa-659	112	19	the	the	DET
ijassa-659	112	20	mismatch	mismatch	NOUN
ijassa-659	112	21	between	between	ADP
ijassa-659	112	22	dl	dl	PROPN
ijassa-659	112	23	∪	∪	PROPN
ijassa-659	112	24	s	s	PROPN
ijassa-659	112	25	and	and	CCONJ
ijassa-659	112	26	du	du	PROPN
ijassa-659	112	27	\	\	PROPN
ijassa-659	112	28	s.	s.	PROPN
ijassa-659	112	29	defining	define	VERB
ijassa-659	112	30	a	a	DET
ijassa-659	112	31	binary	binary	ADJ
ijassa-659	112	32	vector	vector	NOUN
ijassa-659	112	33	𝛼	𝛼	NOUN
ijassa-659	112	34	of	of	ADP
ijassa-659	112	35	size	size	NOUN
ijassa-659	112	36	u	u	NOUN
ijassa-659	112	37	where	where	SCONJ
ijassa-659	112	38	each	each	DET
ijassa-659	112	39	entry	entry	NOUN
ijassa-659	112	40	𝛼𝑖	𝛼𝑖	PROPN
ijassa-659	112	41	indicates	indicate	VERB
ijassa-659	112	42	whether	whether	SCONJ
ijassa-659	112	43	the	the	DET
ijassa-659	112	44	data	datum	NOUN
ijassa-659	112	45	xi	xi	X
ijassa-659	112	46	∈	∈	PROPN
ijassa-659	112	47	du	du	NOUN
ijassa-659	112	48	is	be	AUX
ijassa-659	112	49	selected	select	VERB
ijassa-659	112	50	or	or	CCONJ
ijassa-659	112	51	not	not	PART
ijassa-659	112	52	.	.	PUNCT
ijassa-659	113	1	if	if	SCONJ
ijassa-659	113	2	a	a	DET
ijassa-659	113	3	point	point	NOUN
ijassa-659	113	4	is	be	AUX
ijassa-659	113	5	selected	select	VERB
ijassa-659	113	6	,	,	PUNCT
ijassa-659	113	7	the	the	DET
ijassa-659	113	8	corresponding	corresponding	ADJ
ijassa-659	113	9	entry	entry	NOUN
ijassa-659	113	10	𝛼𝑖	𝛼𝑖	PROPN
ijassa-659	113	11	is	be	AUX
ijassa-659	113	12	1	1	NUM
ijassa-659	113	13	else	else	ADV
ijassa-659	113	14	0	0	NUM
ijassa-659	113	15	.	.	PUNCT
ijassa-659	114	1	thus	thus	ADV
ijassa-659	114	2	the	the	DET
ijassa-659	114	3	minimization	minimization	NOUN
ijassa-659	114	4	problem	problem	NOUN
ijassa-659	114	5	reduces	reduce	VERB
ijassa-659	114	6	to	to	ADP
ijassa-659	114	7	finding	find	VERB
ijassa-659	114	8	𝛼	𝛼	PRON
ijassa-659	114	9	that	that	PRON
ijassa-659	114	10	minimizes	minimize	VERB
ijassa-659	114	11	the	the	DET
ijassa-659	114	12	cost	cost	NOUN
ijassa-659	114	13	function	function	NOUN
ijassa-659	114	14	f(s	f(	NOUN
ijassa-659	114	15	):	):	PUNCT
ijassa-659	114	16	min	min	PROPN
ijassa-659	114	17	𝛼:𝛼𝑖∈{0,1},𝛼𝑇1=𝑏	𝛼:𝛼𝑖∈{0,1},𝛼𝑇1=𝑏	NOUN
ijassa-659	114	18	‖	‖	PROPN
ijassa-659	114	19	1	1	NUM
ijassa-659	114	20	𝑙+𝑏	𝑙+𝑏	NUM
ijassa-659	114	21	(	(	PUNCT
ijassa-659	114	22	∑	∑	PART
ijassa-659	114	23	φ(𝑥𝑗)𝑗∈𝐷𝐿	φ(𝑥𝑗)𝑗∈𝐷𝐿	PROPN
ijassa-659	114	24	+	+	CCONJ
ijassa-659	114	25	∑	∑	PROPN
ijassa-659	114	26	αiφ(𝑥𝑖)𝑖∈𝐷𝑈	αiφ(𝑥𝑖)𝑖∈𝐷𝑈	ADJ
ijassa-659	114	27	)	)	PUNCT
ijassa-659	114	28	−	−	PROPN
ijassa-659	114	29	1	1	NUM
ijassa-659	114	30	𝑢−𝑏	𝑢−𝑏	NOUN
ijassa-659	114	31	∑	∑	PUNCT
ijassa-659	114	32	(	(	PUNCT
ijassa-659	114	33	1	1	NUM
ijassa-659	114	34	−	−	NOUN
ijassa-659	114	35	αi)φ(𝑥𝑖)𝑖∈𝐷𝑈	αi)φ(𝑥𝑖)𝑖∈𝐷𝑈	NUM
ijassa-659	114	36	‖	‖	PROPN
ijassa-659	114	37	ℋ	ℋ	PROPN
ijassa-659	114	38	2	2	NUM
ijassa-659	114	39	(	(	PUNCT
ijassa-659	114	40	6	6	NUM
ijassa-659	114	41	)	)	PUNCT
ijassa-659	114	42	where	where	SCONJ
ijassa-659	114	43	1	1	NUM
ijassa-659	114	44	is	be	AUX
ijassa-659	114	45	a	a	DET
ijassa-659	114	46	vector	vector	NOUN
ijassa-659	114	47	of	of	ADP
ijassa-659	114	48	the	the	DET
ijassa-659	114	49	same	same	ADJ
ijassa-659	114	50	dimension	dimension	NOUN
ijassa-659	114	51	as	as	ADP
ijassa-659	114	52	𝛼	𝛼	NOUN
ijassa-659	114	53	with	with	ADP
ijassa-659	114	54	all	all	DET
ijassa-659	114	55	entries	entry	NOUN
ijassa-659	114	56	1	1	NUM
ijassa-659	114	57	and	and	CCONJ
ijassa-659	114	58	symbol	symbol	PROPN
ijassa-659	114	59	t	t	PROPN
ijassa-659	114	60	is	be	AUX
ijassa-659	114	61	used	use	VERB
ijassa-659	114	62	to	to	PART
ijassa-659	114	63	represent	represent	VERB
ijassa-659	114	64	the	the	DET
ijassa-659	114	65	matrix	matrix	NOUN
ijassa-659	114	66	or	or	CCONJ
ijassa-659	114	67	vector	vector	NOUN
ijassa-659	114	68	transpose	transpose	NOUN
ijassa-659	114	69	operation	operation	NOUN
ijassa-659	114	70	.	.	PUNCT
ijassa-659	115	1	evidently	evidently	ADV
ijassa-659	115	2	,	,	PUNCT
ijassa-659	115	3	the	the	DET
ijassa-659	115	4	cost	cost	NOUN
ijassa-659	115	5	function	function	NOUN
ijassa-659	115	6	in	in	ADP
ijassa-659	115	7	equation	equation	NOUN
ijassa-659	115	8	(	(	PUNCT
ijassa-659	115	9	6	6	NUM
ijassa-659	115	10	)	)	PUNCT
ijassa-659	115	11	is	be	AUX
ijassa-659	115	12	an	an	DET
ijassa-659	115	13	alternative	alternative	ADJ
ijassa-659	115	14	(	(	PUNCT
ijassa-659	115	15	equivalent	equivalent	ADJ
ijassa-659	115	16	)	)	PUNCT
ijassa-659	115	17	representation	representation	NOUN
ijassa-659	115	18	of	of	ADP
ijassa-659	115	19	the	the	DET
ijassa-659	115	20	cost	cost	NOUN
ijassa-659	115	21	function	function	NOUN
ijassa-659	115	22	f(s	f(	NOUN
ijassa-659	115	23	)	)	PUNCT
ijassa-659	115	24	in	in	ADP
ijassa-659	115	25	equation	equation	NOUN
ijassa-659	115	26	(	(	PUNCT
ijassa-659	115	27	5	5	NUM
ijassa-659	115	28	)	)	PUNCT
ijassa-659	115	29	.	.	PUNCT
ijassa-659	116	1	the	the	DET
ijassa-659	116	2	first	first	ADJ
ijassa-659	116	3	term	term	NOUN
ijassa-659	116	4	denotes	denote	VERB
ijassa-659	116	5	the	the	DET
ijassa-659	116	6	mean	mean	NOUN
ijassa-659	116	7	of	of	ADP
ijassa-659	116	8	the	the	DET
ijassa-659	116	9	mapped	map	VERB
ijassa-659	116	10	features	feature	NOUN
ijassa-659	116	11	of	of	ADP
ijassa-659	116	12	the	the	DET
ijassa-659	116	13	labeled	label	VERB
ijassa-659	116	14	and	and	CCONJ
ijassa-659	116	15	selected	select	VERB
ijassa-659	116	16	points	point	NOUN
ijassa-659	116	17	.	.	PUNCT
ijassa-659	117	1	note	note	VERB
ijassa-659	117	2	that	that	SCONJ
ijassa-659	117	3	if	if	SCONJ
ijassa-659	117	4	a	a	DET
ijassa-659	117	5	point	point	NOUN
ijassa-659	117	6	xi	xi	INTJ
ijassa-659	117	7	is	be	AUX
ijassa-659	117	8	not	not	PART
ijassa-659	117	9	selected	select	VERB
ijassa-659	117	10	in	in	ADP
ijassa-659	117	11	the	the	DET
ijassa-659	117	12	current	current	ADJ
ijassa-659	117	13	set	set	NOUN
ijassa-659	117	14	then	then	ADV
ijassa-659	117	15	𝛼𝑖	𝛼𝑖	PROPN
ijassa-659	117	16	will	will	AUX
ijassa-659	117	17	be	be	AUX
ijassa-659	117	18	0	0	NUM
ijassa-659	117	19	and	and	CCONJ
ijassa-659	117	20	this	this	DET
ijassa-659	117	21	term	term	NOUN
ijassa-659	117	22	would	would	AUX
ijassa-659	117	23	not	not	PART
ijassa-659	117	24	get	get	AUX
ijassa-659	117	25	added	add	VERB
ijassa-659	117	26	in	in	ADP
ijassa-659	117	27	the	the	DET
ijassa-659	117	28	summation	summation	NOUN
ijassa-659	117	29	.	.	PUNCT
ijassa-659	118	1	the	the	DET
ijassa-659	118	2	second	second	ADJ
ijassa-659	118	3	term	term	NOUN
ijassa-659	118	4	is	be	AUX
ijassa-659	118	5	mean	mean	NOUN
ijassa-659	118	6	of	of	ADP
ijassa-659	118	7	the	the	DET
ijassa-659	118	8	mapped	map	VERB
ijassa-659	118	9	features	feature	NOUN
ijassa-659	118	10	of	of	ADP
ijassa-659	118	11	the	the	DET
ijassa-659	118	12	unlabeled	unlabele	VERB
ijassa-659	118	13	data	datum	NOUN
ijassa-659	118	14	set	set	VERB
ijassa-659	118	15	minus	minus	ADP
ijassa-659	118	16	the	the	DET
ijassa-659	118	17	selected	select	VERB
ijassa-659	118	18	query	query	NOUN
ijassa-659	118	19	set	set	VERB
ijassa-659	118	20	.	.	PUNCT
ijassa-659	119	1	the	the	DET
ijassa-659	119	2	first	first	ADJ
ijassa-659	119	3	constraint	constraint	NOUN
ijassa-659	119	4	ensures	ensure	VERB
ijassa-659	119	5	that	that	SCONJ
ijassa-659	119	6	each	each	DET
ijassa-659	119	7	entry	entry	NOUN
ijassa-659	119	8	in	in	ADP
ijassa-659	119	9	𝛼	𝛼	PROPN
ijassa-659	119	10	is	be	AUX
ijassa-659	119	11	either	either	PRON
ijassa-659	119	12	0	0	NUM
ijassa-659	119	13	or	or	CCONJ
ijassa-659	119	14	1	1	NUM
ijassa-659	119	15	and	and	CCONJ
ijassa-659	119	16	the	the	DET
ijassa-659	119	17	second	second	ADJ
ijassa-659	119	18	constraint	constraint	NOUN
ijassa-659	119	19	ensures	ensure	VERB
ijassa-659	119	20	that	that	SCONJ
ijassa-659	119	21	exactly	exactly	ADV
ijassa-659	119	22	b	b	NUM
ijassa-659	119	23	entries	entry	NOUN
ijassa-659	119	24	of	of	ADP
ijassa-659	119	25	𝛼	𝛼	NOUN
ijassa-659	119	26	are	be	AUX
ijassa-659	119	27	1	1	NUM
ijassa-659	119	28	,	,	PUNCT
ijassa-659	119	29	meaning	mean	VERB
ijassa-659	119	30	exactly	exactly	ADV
ijassa-659	119	31	b	b	NUM
ijassa-659	119	32	instances	instance	NOUN
ijassa-659	119	33	are	be	AUX
ijassa-659	119	34	selected	select	VERB
ijassa-659	119	35	from	from	ADP
ijassa-659	119	36	the	the	DET
ijassa-659	119	37	unlabeled	unlabele	VERB
ijassa-659	119	38	data	datum	NOUN
ijassa-659	119	39	set	set	NOUN
ijassa-659	119	40	,	,	PUNCT
ijassa-659	119	41	where	where	SCONJ
ijassa-659	119	42	b	b	NOUN
ijassa-659	119	43	is	be	AUX
ijassa-659	119	44	specified	specify	VERB
ijassa-659	119	45	a	a	DET
ijassa-659	119	46	priori	priori	NOUN
ijassa-659	119	47	by	by	ADP
ijassa-659	119	48	the	the	DET
ijassa-659	119	49	user	user	NOUN
ijassa-659	119	50	.	.	PUNCT
ijassa-659	120	1	algorithm	algorithm	PROPN
ijassa-659	120	2	1	1	NUM
ijassa-659	120	3	.	.	PUNCT
ijassa-659	121	1	queryinstancesselection(dl	queryinstancesselection(dl	PROPN
ijassa-659	121	2	,	,	PUNCT
ijassa-659	121	3	du	du	X
ijassa-659	121	4	,	,	PUNCT
ijassa-659	121	5	b	b	NOUN
ijassa-659	121	6	)	)	PUNCT
ijassa-659	121	7	;	;	PUNCT
ijassa-659	121	8	input	input	NOUN
ijassa-659	121	9	:	:	PUNCT
ijassa-659	121	10	a	a	DET
ijassa-659	121	11	set	set	NOUN
ijassa-659	121	12	of	of	ADP
ijassa-659	121	13	labeled	label	VERB
ijassa-659	121	14	instances	instance	NOUN
ijassa-659	121	15	dl	dl	NOUN
ijassa-659	121	16	;	;	PUNCT
ijassa-659	121	17	set	set	VERB
ijassa-659	121	18	of	of	ADP
ijassa-659	121	19	unlabeled	unlabeled	ADJ
ijassa-659	121	20	instances	instance	NOUN
ijassa-659	121	21	du	du	X
ijassa-659	121	22	;	;	PUNCT
ijassa-659	121	23	batch	batch	NOUN
ijassa-659	121	24	size	size	NOUN
ijassa-659	121	25	b.	b.	PROPN
ijassa-659	121	26	output	output	PROPN
ijassa-659	121	27	:	:	PUNCT
ijassa-659	121	28	a	a	DET
ijassa-659	121	29	batch	batch	NOUN
ijassa-659	121	30	of	of	ADP
ijassa-659	121	31	query	query	NOUN
ijassa-659	121	32	instances	instance	NOUN
ijassa-659	121	33	s.	s.	PROPN
ijassa-659	121	34	compute	compute	PROPN
ijassa-659	121	35	𝛼	𝛼	PROPN
ijassa-659	121	36	that	that	PRON
ijassa-659	121	37	minimize	minimize	VERB
ijassa-659	121	38	the	the	DET
ijassa-659	121	39	distribution	distribution	NOUN
ijassa-659	121	40	between	between	ADP
ijassa-659	121	41	the	the	DET
ijassa-659	121	42	sets	set	NOUN
ijassa-659	122	1	dl	dl	PROPN
ijassa-659	122	2	∪	∪	PROPN
ijassa-659	122	3	s	s	PROPN
ijassa-659	122	4	and	and	CCONJ
ijassa-659	122	5	du	du	PROPN
ijassa-659	122	6	\	\	PROPN
ijassa-659	122	7	s	s	PART
ijassa-659	122	8	(	(	PUNCT
ijassa-659	122	9	eq	eq	NOUN
ijassa-659	122	10	.	.	PUNCT
ijassa-659	122	11	(	(	PUNCT
ijassa-659	122	12	6	6	NUM
ijassa-659	122	13	)	)	PUNCT
ijassa-659	122	14	)	)	PUNCT
ijassa-659	122	15	sort	sort	ADV
ijassa-659	122	16	du	du	X
ijassa-659	122	17	in	in	ADP
ijassa-659	122	18	descending	descend	VERB
ijassa-659	122	19	order	order	NOUN
ijassa-659	122	20	of	of	ADP
ijassa-659	122	21	𝛼	𝛼	PRON
ijassa-659	122	22	select	select	ADJ
ijassa-659	122	23	top	top	ADJ
ijassa-659	122	24	b	b	NOUN
ijassa-659	122	25	instances	instance	NOUN
ijassa-659	122	26	of	of	ADP
ijassa-659	122	27	du	du	PROPN
ijassa-659	122	28	as	as	ADP
ijassa-659	122	29	s	s	NOUN
ijassa-659	122	30	update	update	NOUN
ijassa-659	122	31	dl	dl	PROPN
ijassa-659	122	32	and	and	CCONJ
ijassa-659	122	33	du	du	PROPN
ijassa-659	122	34	:	:	PUNCT
ijassa-659	122	35	dl←	dl←	VERB
ijassa-659	122	36	dl	dl	PROPN
ijassa-659	122	37	∪	∪	PROPN
ijassa-659	122	38	s	s	PROPN
ijassa-659	122	39	,	,	PUNCT
ijassa-659	122	40	du←	du←	PROPN
ijassa-659	122	41	du	du	PROPN
ijassa-659	122	42	\	\	PROPN
ijassa-659	122	43	s	s	PART
ijassa-659	122	44	return	return	NOUN
ijassa-659	122	45	s	s	VERB
ijassa-659	122	46	3.2	3.2	NUM
ijassa-659	122	47	a	a	DET
ijassa-659	122	48	neighborhood	neighborhood	NOUN
ijassa-659	122	49	-	-	PUNCT
ijassa-659	122	50	based	base	VERB
ijassa-659	122	51	active	active	ADJ
ijassa-659	122	52	learning	learning	NOUN
ijassa-659	122	53	method	method	NOUN
ijassa-659	122	54	once	once	SCONJ
ijassa-659	122	55	the	the	DET
ijassa-659	122	56	batch	batch	NOUN
ijassa-659	122	57	of	of	ADP
ijassa-659	122	58	most	most	ADV
ijassa-659	122	59	informative	informative	ADJ
ijassa-659	122	60	query	query	NOUN
ijassa-659	122	61	instances	instance	NOUN
ijassa-659	122	62	are	be	AUX
ijassa-659	122	63	selected	select	VERB
ijassa-659	122	64	,	,	PUNCT
ijassa-659	122	65	the	the	DET
ijassa-659	122	66	proposed	propose	VERB
ijassa-659	122	67	method	method	NOUN
ijassa-659	122	68	query	query	VERB
ijassa-659	122	69	them	they	PRON
ijassa-659	122	70	against	against	ADP
ijassa-659	122	71	the	the	DET
ijassa-659	122	72	existing	exist	VERB
ijassa-659	122	73	neighborhoods	neighborhood	NOUN
ijassa-659	122	74	to	to	PART
ijassa-659	122	75	determine	determine	VERB
ijassa-659	122	76	which	which	DET
ijassa-659	122	77	neighborhood	neighborhood	NOUN
ijassa-659	122	78	they	they	PRON
ijassa-659	122	79	belong	belong	VERB
ijassa-659	122	80	to	to	ADP
ijassa-659	122	81	.	.	PUNCT
ijassa-659	123	1	in	in	ADP
ijassa-659	123	2	this	this	DET
ijassa-659	123	3	section	section	NOUN
ijassa-659	123	4	,	,	PUNCT
ijassa-659	123	5	introducing	introduce	VERB
ijassa-659	123	6	the	the	DET
ijassa-659	123	7	concept	concept	NOUN
ijassa-659	123	8	of	of	ADP
ijassa-659	123	9	neighborhood	neighborhood	NOUN
ijassa-659	123	10	,	,	PUNCT
ijassa-659	123	11	which	which	PRON
ijassa-659	123	12	is	be	AUX
ijassa-659	123	13	instrumental	instrumental	ADJ
ijassa-659	123	14	in	in	ADP
ijassa-659	123	15	the	the	DET
ijassa-659	123	16	design	design	NOUN
ijassa-659	123	17	of	of	ADP
ijassa-659	123	18	many	many	ADJ
ijassa-659	123	19	existing	exist	VERB
ijassa-659	123	20	methods	method	NOUN
ijassa-659	123	21	for	for	ADP
ijassa-659	123	22	active	active	ADJ
ijassa-659	123	23	learning	learning	NOUN
ijassa-659	123	24	of	of	ADP
ijassa-659	123	25	pairwise	pairwise	NOUN
ijassa-659	123	26	constraints	constraint	NOUN
ijassa-659	123	27	[	[	X
ijassa-659	123	28	1	1	NUM
ijassa-659	123	29	,	,	PUNCT
ijassa-659	123	30	12	12	NUM
ijassa-659	123	31	,	,	PUNCT
ijassa-659	123	32	17	17	NUM
ijassa-659	123	33	]	]	PUNCT
ijassa-659	123	34	.	.	PUNCT
ijassa-659	124	1	then	then	ADV
ijassa-659	124	2	,	,	PUNCT
ijassa-659	124	3	explain	explain	VERB
ijassa-659	124	4	how	how	SCONJ
ijassa-659	124	5	the	the	DET
ijassa-659	124	6	proposed	propose	VERB
ijassa-659	124	7	active	active	ADJ
ijassa-659	124	8	learning	learning	NOUN
ijassa-659	124	9	method	method	NOUN
ijassa-659	124	10	can	can	AUX
ijassa-659	124	11	expand	expand	VERB
ijassa-659	124	12	the	the	DET
ijassa-659	124	13	neighborhoods	neighborhood	NOUN
ijassa-659	124	14	using	use	VERB
ijassa-659	124	15	the	the	DET
ijassa-659	124	16	selected	select	VERB
ijassa-659	124	17	query	query	NOUN
ijassa-659	124	18	instances	instance	NOUN
ijassa-659	124	19	.	.	PUNCT
ijassa-659	125	1	a	a	DET
ijassa-659	125	2	neighborhood	neighborhood	NOUN
ijassa-659	125	3	contains	contain	VERB
ijassa-659	125	4	a	a	DET
ijassa-659	125	5	set	set	NOUN
ijassa-659	125	6	of	of	ADP
ijassa-659	125	7	data	datum	NOUN
ijassa-659	125	8	instances	instance	NOUN
ijassa-659	125	9	that	that	PRON
ijassa-659	125	10	are	be	AUX
ijassa-659	125	11	known	know	VERB
ijassa-659	125	12	to	to	PART
ijassa-659	125	13	belong	belong	VERB
ijassa-659	125	14	to	to	ADP
ijassa-659	125	15	the	the	DET
ijassa-659	125	16	same	same	ADJ
ijassa-659	125	17	cluster	cluster	NOUN
ijassa-659	125	18	(	(	PUNCT
ijassa-659	125	19	i.e.	i.e.	X
ijassa-659	125	20	connected	connect	VERB
ijassa-659	125	21	by	by	ADP
ijassa-659	125	22	must	must	AUX
ijassa-659	125	23	-	-	PUNCT
ijassa-659	125	24	link	link	NOUN
ijassa-659	125	25	constraints	constraint	NOUN
ijassa-659	125	26	)	)	PUNCT
ijassa-659	125	27	.	.	PUNCT
ijassa-659	126	1	different	different	ADJ
ijassa-659	126	2	neighborhoods	neighborhood	NOUN
ijassa-659	126	3	are	be	AUX
ijassa-659	126	4	connected	connect	VERB
ijassa-659	126	5	by	by	ADP
ijassa-659	126	6	cannot	cannot	NOUN
ijassa-659	126	7	-	-	PUNCT
ijassa-659	126	8	link	link	NOUN
ijassa-659	126	9	constraints	constraint	NOUN
ijassa-659	126	10	and	and	CCONJ
ijassa-659	126	11	thus	thus	ADV
ijassa-659	126	12	are	be	AUX
ijassa-659	126	13	known	know	VERB
ijassa-659	126	14	to	to	PART
ijassa-659	126	15	belong	belong	VERB
ijassa-659	126	16	to	to	ADP
ijassa-659	126	17	different	different	ADJ
ijassa-659	126	18	clusters	cluster	NOUN
ijassa-659	126	19	.	.	PUNCT
ijassa-659	127	1	for	for	ADP
ijassa-659	127	2	example	example	NOUN
ijassa-659	127	3	,	,	PUNCT
ijassa-659	127	4	figure	figure	NOUN
ijassa-659	127	5	1	1	NUM
ijassa-659	127	6	shows	show	VERB
ijassa-659	127	7	a	a	DET
ijassa-659	127	8	set	set	NOUN
ijassa-659	127	9	of	of	ADP
ijassa-659	127	10	must	must	AUX
ijassa-659	127	11	-	-	PUNCT
ijassa-659	127	12	link	link	NOUN
ijassa-659	127	13	constraints	constraint	NOUN
ijassa-659	127	14	(	(	PUNCT
ijassa-659	127	15	x1	x1	NOUN
ijassa-659	127	16	;	;	PUNCT
ijassa-659	127	17	x2	x2	PROPN
ijassa-659	127	18	)	)	PUNCT
ijassa-659	127	19	,	,	PUNCT
ijassa-659	127	20	(	(	PUNCT
ijassa-659	127	21	x1	x1	NOUN
ijassa-659	127	22	;	;	PUNCT
ijassa-659	127	23	x3	x3	ADJ
ijassa-659	127	24	)	)	PUNCT
ijassa-659	127	25	,	,	PUNCT
ijassa-659	127	26	(	(	PUNCT
ijassa-659	127	27	x4	x4	PROPN
ijassa-659	127	28	;	;	PUNCT
ijassa-659	127	29	x5	x5	NOUN
ijassa-659	127	30	)	)	PUNCT
ijassa-659	127	31	,	,	PUNCT
ijassa-659	127	32	and	and	CCONJ
ijassa-659	127	33	(	(	PUNCT
ijassa-659	127	34	x4	x4	PROPN
ijassa-659	127	35	;	;	PUNCT
ijassa-659	127	36	x6	x6	PROPN
ijassa-659	127	37	)	)	PUNCT
ijassa-659	127	38	,	,	PUNCT
ijassa-659	127	39	as	as	ADV
ijassa-659	127	40	well	well	ADV
ijassa-659	127	41	as	as	ADP
ijassa-659	127	42	a	a	DET
ijassa-659	127	43	set	set	NOUN
ijassa-659	127	44	of	of	ADP
ijassa-659	127	45	cannot	cannot	NOUN
ijassa-659	127	46	-	-	PUNCT
ijassa-659	127	47	link	link	NOUN
ijassa-659	127	48	constraints	constraint	NOUN
ijassa-659	127	49	(	(	PUNCT
ijassa-659	127	50	x1	x1	PROPN
ijassa-659	127	51	;	;	PUNCT
ijassa-659	127	52	x4	x4	PROPN
ijassa-659	127	53	)	)	PUNCT
ijassa-659	127	54	and	and	CCONJ
ijassa-659	127	55	(	(	PUNCT
ijassa-659	127	56	x1	x1	ADJ
ijassa-659	127	57	;	;	PUNCT
ijassa-659	127	58	x5	x5	PROPN
ijassa-659	127	59	)	)	PUNCT
ijassa-659	127	60	.	.	PUNCT
ijassa-659	128	1	two	two	NUM
ijassa-659	128	2	neighborhoods	neighborhood	NOUN
ijassa-659	128	3	n1	n1	VERB
ijassa-659	128	4	and	and	CCONJ
ijassa-659	128	5	n2	n2	NOUN
ijassa-659	128	6	can	can	AUX
ijassa-659	128	7	be	be	AUX
ijassa-659	128	8	generated	generate	VERB
ijassa-659	128	9	as	as	SCONJ
ijassa-659	128	10	shown	show	VERB
ijassa-659	128	11	in	in	ADP
ijassa-659	128	12	figure	figure	NOUN
ijassa-659	128	13	1	1	NUM
ijassa-659	128	14	.	.	PUNCT
ijassa-659	128	15	neighborhood	neighborhood	NOUN
ijassa-659	128	16	n1	n1	NOUN
ijassa-659	128	17	includes	include	VERB
ijassa-659	128	18	three	three	NUM
ijassa-659	128	19	instances	instance	NOUN
ijassa-659	128	20	x1	x1	NUM
ijassa-659	128	21	,	,	PUNCT
ijassa-659	128	22	x2	x2	PROPN
ijassa-659	128	23	,	,	PUNCT
ijassa-659	128	24	and	and	CCONJ
ijassa-659	128	25	x3	x3	ADJ
ijassa-659	128	26	as	as	SCONJ
ijassa-659	128	27	described	describe	VERB
ijassa-659	128	28	by	by	ADP
ijassa-659	128	29	must	must	AUX
ijassa-659	128	30	-	-	PUNCT
ijassa-659	128	31	link	link	NOUN
ijassa-659	128	32	constraints	constraint	NOUN
ijassa-659	128	33	(	(	PUNCT
ijassa-659	128	34	x1	x1	NOUN
ijassa-659	128	35	;	;	PUNCT
ijassa-659	128	36	x2	x2	NUM
ijassa-659	128	37	)	)	PUNCT
ijassa-659	128	38	and	and	CCONJ
ijassa-659	128	39	(	(	PUNCT
ijassa-659	128	40	x1	x1	PROPN
ijassa-659	128	41	;	;	PUNCT
ijassa-659	128	42	x3	x3	ADJ
ijassa-659	128	43	)	)	PUNCT
ijassa-659	128	44	.	.	PUNCT
ijassa-659	129	1	similarly	similarly	ADV
ijassa-659	129	2	,	,	PUNCT
ijassa-659	129	3	x4	x4	PROPN
ijassa-659	129	4	,	,	PUNCT
ijassa-659	129	5	x5	x5	NOUN
ijassa-659	129	6	,	,	PUNCT
ijassa-659	129	7	and	and	CCONJ
ijassa-659	129	8	x6	x6	PROPN
ijassa-659	129	9	should	should	AUX
ijassa-659	129	10	also	also	ADV
ijassa-659	129	11	be	be	AUX
ijassa-659	129	12	included	include	VERB
ijassa-659	129	13	in	in	ADP
ijassa-659	129	14	the	the	DET
ijassa-659	129	15	same	same	ADJ
ijassa-659	129	16	neighborhood	neighborhood	NOUN
ijassa-659	129	17	.	.	PUNCT
ijassa-659	130	1	indicated	indicate	VERB
ijassa-659	130	2	by	by	ADP
ijassa-659	130	3	cannot	cannot	NOUN
ijassa-659	130	4	-	-	PUNCT
ijassa-659	130	5	link	link	NOUN
ijassa-659	130	6	constraints	constraint	NOUN
ijassa-659	130	7	(	(	PUNCT
ijassa-659	130	8	x1	x1	PROPN
ijassa-659	130	9	;	;	PUNCT
ijassa-659	130	10	x4	x4	PROPN
ijassa-659	130	11	)	)	PUNCT
ijassa-659	130	12	and	and	CCONJ
ijassa-659	130	13	(	(	PUNCT
ijassa-659	130	14	x1	x1	ADJ
ijassa-659	130	15	;	;	PUNCT
ijassa-659	130	16	x5	x5	PROPN
ijassa-659	130	17	)	)	PUNCT
ijassa-659	130	18	,	,	PUNCT
ijassa-659	130	19	x1	x1	PROPN
ijassa-659	130	20	should	should	AUX
ijassa-659	130	21	not	not	PART
ijassa-659	130	22	be	be	AUX
ijassa-659	130	23	in	in	ADP
ijassa-659	130	24	the	the	DET
ijassa-659	130	25	neighborhood	neighborhood	NOUN
ijassa-659	130	26	to	to	PART
ijassa-659	130	27	which	which	PRON
ijassa-659	130	28	x4	x4	PROPN
ijassa-659	130	29	and	and	CCONJ
ijassa-659	130	30	x5	x5	PROPN
ijassa-659	130	31	belong	belong	NOUN
ijassa-659	130	32	.	.	PUNCT
ijassa-659	131	1	therefore	therefore	ADV
ijassa-659	131	2	,	,	PUNCT
ijassa-659	131	3	neighborhood	neighborhood	NOUN
ijassa-659	131	4	n2	n2	NOUN
ijassa-659	131	5	is	be	AUX
ijassa-659	131	6	discovered	discover	VERB
ijassa-659	131	7	and	and	CCONJ
ijassa-659	131	8	it	it	PRON
ijassa-659	131	9	contains	contain	VERB
ijassa-659	131	10	instances	instance	NOUN
ijassa-659	131	11	x4	x4	PROPN
ijassa-659	131	12	,	,	PUNCT
ijassa-659	131	13	x5	x5	NOUN
ijassa-659	131	14	,	,	PUNCT
ijassa-659	131	15	and	and	CCONJ
ijassa-659	131	16	x6	x6	PROPN
ijassa-659	131	17	.	.	PROPN
ijassa-659	132	1	30	30	NUM
ijassa-659	132	2	w.	w.	NOUN
ijassa-659	132	3	atwa	atwa	PROPN
ijassa-659	132	4	,	,	PUNCT
ijassa-659	132	5	m.	m.	NOUN
ijassa-659	132	6	emam	emam	PROPN
ijassa-659	132	7	copyright	copyright	NOUN
ijassa-659	132	8	©	©	PROPN
ijassa-659	132	9	2019	2019	NUM
ijassa-659	132	10	assa	assa	PROPN
ijassa-659	132	11	adv	adv	PROPN
ijassa-659	132	12	.	.	PUNCT
ijassa-659	133	1	in	in	ADP
ijassa-659	133	2	systems	system	NOUN
ijassa-659	133	3	science	science	NOUN
ijassa-659	133	4	and	and	CCONJ
ijassa-659	133	5	appl	appl	NOUN
ijassa-659	133	6	.	.	PUNCT
ijassa-659	134	1	(	(	PUNCT
ijassa-659	134	2	2019	2019	NUM
ijassa-659	134	3	)	)	PUNCT
ijassa-659	134	4	fig	fig	NOUN
ijassa-659	134	5	.	.	PUNCT
ijassa-659	135	1	1	1	NUM
ijassa-659	135	2	:	:	PUNCT
ijassa-659	135	3	an	an	DET
ijassa-659	135	4	example	example	NOUN
ijassa-659	135	5	of	of	ADP
ijassa-659	135	6	neighborhoods	neighborhood	NOUN
ijassa-659	135	7	a	a	DET
ijassa-659	135	8	key	key	ADJ
ijassa-659	135	9	advantage	advantage	NOUN
ijassa-659	135	10	of	of	ADP
ijassa-659	135	11	using	use	VERB
ijassa-659	135	12	the	the	DET
ijassa-659	135	13	neighborhood	neighborhood	NOUN
ijassa-659	135	14	concepts	concept	NOUN
ijassa-659	135	15	is	be	AUX
ijassa-659	135	16	that	that	SCONJ
ijassa-659	135	17	by	by	ADP
ijassa-659	135	18	leveraging	leverage	VERB
ijassa-659	135	19	the	the	DET
ijassa-659	135	20	knowledge	knowledge	NOUN
ijassa-659	135	21	of	of	ADP
ijassa-659	135	22	the	the	DET
ijassa-659	135	23	neighborhoods	neighborhood	NOUN
ijassa-659	135	24	,	,	PUNCT
ijassa-659	135	25	we	we	PRON
ijassa-659	135	26	can	can	AUX
ijassa-659	135	27	acquire	acquire	VERB
ijassa-659	135	28	a	a	DET
ijassa-659	135	29	large	large	ADJ
ijassa-659	135	30	number	number	NOUN
ijassa-659	135	31	of	of	ADP
ijassa-659	135	32	pairwise	pairwise	NOUN
ijassa-659	135	33	constraints	constraint	NOUN
ijassa-659	135	34	via	via	ADP
ijassa-659	135	35	a	a	DET
ijassa-659	135	36	small	small	ADJ
ijassa-659	135	37	number	number	NOUN
ijassa-659	135	38	of	of	ADP
ijassa-659	135	39	queries	query	NOUN
ijassa-659	135	40	.	.	PUNCT
ijassa-659	136	1	in	in	ADP
ijassa-659	136	2	particular	particular	ADJ
ijassa-659	136	3	,	,	PUNCT
ijassa-659	136	4	if	if	SCONJ
ijassa-659	136	5	we	we	PRON
ijassa-659	136	6	can	can	AUX
ijassa-659	136	7	identify	identify	VERB
ijassa-659	136	8	the	the	DET
ijassa-659	136	9	neighborhood	neighborhood	NOUN
ijassa-659	136	10	of	of	ADP
ijassa-659	136	11	an	an	DET
ijassa-659	136	12	instance	instance	NOUN
ijassa-659	136	13	x	x	NOUN
ijassa-659	136	14	,	,	PUNCT
ijassa-659	136	15	we	we	PRON
ijassa-659	136	16	can	can	AUX
ijassa-659	136	17	infer	infer	VERB
ijassa-659	136	18	its	its	PRON
ijassa-659	136	19	pairwise	pairwise	NOUN
ijassa-659	136	20	relationship	relationship	NOUN
ijassa-659	136	21	with	with	ADP
ijassa-659	136	22	all	all	DET
ijassa-659	136	23	other	other	ADJ
ijassa-659	136	24	points	point	NOUN
ijassa-659	136	25	that	that	PRON
ijassa-659	136	26	are	be	AUX
ijassa-659	136	27	currently	currently	ADV
ijassa-659	136	28	confirmed	confirm	VERB
ijassa-659	136	29	to	to	PART
ijassa-659	136	30	belong	belong	VERB
ijassa-659	136	31	to	to	ADP
ijassa-659	136	32	any	any	PRON
ijassa-659	136	33	of	of	ADP
ijassa-659	136	34	the	the	DET
ijassa-659	136	35	existing	exist	VERB
ijassa-659	136	36	neighborhoods	neighborhood	NOUN
ijassa-659	136	37	.	.	PUNCT
ijassa-659	137	1	this	this	PRON
ijassa-659	137	2	naturally	naturally	ADV
ijassa-659	137	3	encourages	encourage	VERB
ijassa-659	137	4	us	we	PRON
ijassa-659	137	5	to	to	PART
ijassa-659	137	6	consider	consider	VERB
ijassa-659	137	7	an	an	DET
ijassa-659	137	8	active	active	ADJ
ijassa-659	137	9	learning	learning	NOUN
ijassa-659	137	10	strategy	strategy	NOUN
ijassa-659	137	11	that	that	PRON
ijassa-659	137	12	incrementally	incrementally	ADV
ijassa-659	137	13	expands	expand	VERB
ijassa-659	137	14	the	the	DET
ijassa-659	137	15	neighborhoods	neighborhood	NOUN
ijassa-659	137	16	by	by	ADP
ijassa-659	137	17	selecting	select	VERB
ijassa-659	137	18	the	the	DET
ijassa-659	137	19	most	most	ADV
ijassa-659	137	20	informative	informative	ADJ
ijassa-659	137	21	instances	instance	NOUN
ijassa-659	137	22	and	and	CCONJ
ijassa-659	137	23	querying	query	VERB
ijassa-659	137	24	them	they	PRON
ijassa-659	137	25	against	against	ADP
ijassa-659	137	26	the	the	DET
ijassa-659	137	27	known	know	VERB
ijassa-659	137	28	neighborhoods	neighborhood	NOUN
ijassa-659	137	29	.	.	PUNCT
ijassa-659	138	1	we	we	PRON
ijassa-659	138	2	summarize	summarize	VERB
ijassa-659	138	3	the	the	DET
ijassa-659	138	4	active	active	ADJ
ijassa-659	138	5	learning	learning	NOUN
ijassa-659	138	6	method	method	NOUN
ijassa-659	138	7	in	in	ADP
ijassa-659	138	8	algorithm	algorithm	NOUN
ijassa-659	138	9	2	2	NUM
ijassa-659	138	10	.	.	PUNCT
ijassa-659	139	1	we	we	PRON
ijassa-659	139	2	begin	begin	VERB
ijassa-659	139	3	by	by	ADP
ijassa-659	139	4	initializing	initialize	VERB
ijassa-659	139	5	the	the	DET
ijassa-659	139	6	neighborhoods	neighborhood	NOUN
ijassa-659	139	7	by	by	ADP
ijassa-659	139	8	selecting	select	VERB
ijassa-659	139	9	a	a	DET
ijassa-659	139	10	random	random	ADJ
ijassa-659	139	11	point	point	NOUN
ijassa-659	139	12	to	to	PART
ijassa-659	139	13	be	be	AUX
ijassa-659	139	14	the	the	DET
ijassa-659	139	15	initial	initial	ADJ
ijassa-659	139	16	neighborhood	neighborhood	NOUN
ijassa-659	139	17	(	(	PUNCT
ijassa-659	139	18	line	line	NOUN
ijassa-659	139	19	1	1	NUM
ijassa-659	139	20	)	)	PUNCT
ijassa-659	139	21	.	.	PUNCT
ijassa-659	140	1	a	a	DET
ijassa-659	140	2	selection	selection	NOUN
ijassa-659	140	3	criterion	criterion	NOUN
ijassa-659	140	4	is	be	AUX
ijassa-659	140	5	then	then	ADV
ijassa-659	140	6	applied	apply	VERB
ijassa-659	140	7	to	to	PART
ijassa-659	140	8	select	select	VERB
ijassa-659	140	9	the	the	DET
ijassa-659	140	10	batch	batch	NOUN
ijassa-659	140	11	query	query	NOUN
ijassa-659	140	12	instances	instance	VERB
ijassa-659	140	13	s	s	PRON
ijassa-659	140	14	as	as	SCONJ
ijassa-659	140	15	explained	explain	VERB
ijassa-659	140	16	in	in	ADP
ijassa-659	140	17	the	the	DET
ijassa-659	140	18	previous	previous	ADJ
ijassa-659	140	19	section	section	NOUN
ijassa-659	140	20	(	(	PUNCT
ijassa-659	140	21	line	line	NOUN
ijassa-659	140	22	2	2	NUM
ijassa-659	140	23	)	)	PUNCT
ijassa-659	140	24	.	.	PUNCT
ijassa-659	141	1	each	each	DET
ijassa-659	141	2	selected	select	VERB
ijassa-659	141	3	instance	instance	NOUN
ijassa-659	141	4	𝑠	𝑠	PROPN
ijassa-659	141	5	∈	∈	PROPN
ijassa-659	141	6	𝑆	𝑆	PROPN
ijassa-659	141	7	is	be	AUX
ijassa-659	141	8	then	then	ADV
ijassa-659	141	9	queried	query	VERB
ijassa-659	141	10	against	against	ADP
ijassa-659	141	11	each	each	DET
ijassa-659	141	12	existing	exist	VERB
ijassa-659	141	13	neighborhood	neighborhood	NOUN
ijassa-659	141	14	ni	ni	NOUN
ijassa-659	141	15	to	to	PART
ijassa-659	141	16	identify	identify	VERB
ijassa-659	141	17	where	where	SCONJ
ijassa-659	141	18	s	s	NOUN
ijassa-659	141	19	belongs	belong	VERB
ijassa-659	141	20	,	,	PUNCT
ijassa-659	141	21	during	during	ADP
ijassa-659	141	22	which	which	PRON
ijassa-659	141	23	the	the	DET
ijassa-659	141	24	constraint	constraint	NOUN
ijassa-659	141	25	set	set	NOUN
ijassa-659	141	26	c	c	PROPN
ijassa-659	141	27	is	be	AUX
ijassa-659	141	28	updated	update	VERB
ijassa-659	141	29	(	(	PUNCT
ijassa-659	141	30	lines	line	NOUN
ijassa-659	141	31	5	5	NUM
ijassa-659	141	32	-	-	SYM
ijassa-659	141	33	13	13	NUM
ijassa-659	141	34	)	)	PUNCT
ijassa-659	141	35	.	.	PUNCT
ijassa-659	142	1	to	to	PART
ijassa-659	142	2	determine	determine	VERB
ijassa-659	142	3	the	the	DET
ijassa-659	142	4	neighborhood	neighborhood	NOUN
ijassa-659	142	5	of	of	ADP
ijassa-659	142	6	s	s	NOUN
ijassa-659	142	7	with	with	ADP
ijassa-659	142	8	the	the	DET
ijassa-659	142	9	smallest	small	ADJ
ijassa-659	142	10	number	number	NOUN
ijassa-659	142	11	of	of	ADP
ijassa-659	142	12	queries	query	NOUN
ijassa-659	142	13	,	,	PUNCT
ijassa-659	142	14	we	we	PRON
ijassa-659	142	15	go	go	VERB
ijassa-659	142	16	through	through	ADP
ijassa-659	142	17	the	the	DET
ijassa-659	142	18	neighborhoods	neighborhood	NOUN
ijassa-659	142	19	in	in	ADP
ijassa-659	142	20	decreasing	decrease	VERB
ijassa-659	142	21	order	order	NOUN
ijassa-659	142	22	based	base	VERB
ijassa-659	142	23	on	on	ADP
ijassa-659	142	24	p(s	p(s	PROPN
ijassa-659	142	25	∈	∈	PROPN
ijassa-659	142	26	ni	ni	PROPN
ijassa-659	142	27	)	)	PUNCT
ijassa-659	142	28	,	,	PUNCT
ijassa-659	142	29	i	i	PRON
ijassa-659	142	30	∈	∈	PROPN
ijassa-659	142	31	{	{	PUNCT
ijassa-659	142	32	1	1	NUM
ijassa-659	142	33	,	,	PUNCT
ijassa-659	142	34	·	·	PUNCT
ijassa-659	142	35	·	·	PUNCT
ijassa-659	142	36	·	·	PUNCT
ijassa-659	142	37	,	,	PUNCT
ijassa-659	142	38	k	k	X
ijassa-659	142	39	}	}	PUNCT
ijassa-659	142	40	,	,	PUNCT
ijassa-659	142	41	i.e.	i.e.	X
ijassa-659	142	42	the	the	DET
ijassa-659	142	43	probability	probability	NOUN
ijassa-659	142	44	of	of	ADP
ijassa-659	142	45	selected	select	VERB
ijassa-659	142	46	instance	instance	NOUN
ijassa-659	142	47	s	s	AUX
ijassa-659	142	48	belonging	belong	VERB
ijassa-659	142	49	to	to	ADP
ijassa-659	142	50	the	the	DET
ijassa-659	142	51	neighborhood	neighborhood	NOUN
ijassa-659	142	52	ni	ni	NOUN
ijassa-659	142	53	,	,	PUNCT
ijassa-659	142	54	which	which	PRON
ijassa-659	142	55	is	be	AUX
ijassa-659	142	56	assumed	assume	VERB
ijassa-659	142	57	to	to	PART
ijassa-659	142	58	be	be	AUX
ijassa-659	142	59	the	the	DET
ijassa-659	142	60	average	average	ADJ
ijassa-659	142	61	similarity	similarity	NOUN
ijassa-659	142	62	between	between	ADP
ijassa-659	142	63	s	s	PRON
ijassa-659	142	64	and	and	CCONJ
ijassa-659	142	65	the	the	DET
ijassa-659	142	66	instances	instance	NOUN
ijassa-659	142	67	in	in	ADP
ijassa-659	142	68	ni	ni	PROPN
ijassa-659	142	69	.	.	PUNCT
ijassa-659	143	1	𝑝(𝑠	𝑝(𝑠	PROPN
ijassa-659	143	2	∈	∈	PROPN
ijassa-659	144	1	𝑁𝑖	𝑁𝑖	PROPN
ijassa-659	144	2	)	)	PUNCT
ijassa-659	144	3	=	=	SYM
ijassa-659	144	4	1	1	NUM
ijassa-659	144	5	|𝑁𝑖|	|𝑁𝑖|	PROPN
ijassa-659	144	6	∑	∑	PUNCT
ijassa-659	144	7	𝑀(𝑠,𝑥𝑗)𝑥𝑗∈𝑁𝑖	𝑀(𝑠,𝑥𝑗)𝑥𝑗∈𝑁𝑖	PRON
ijassa-659	144	8	∑	∑	PROPN
ijassa-659	144	9	1	1	NUM
ijassa-659	144	10	|𝑁𝑝|	|𝑁𝑝|	PROPN
ijassa-659	144	11	∑	∑	PUNCT
ijassa-659	144	12	𝑀(𝑠,𝑥𝑗)𝑥𝑗∈𝑁𝑝	𝑀(𝑠,𝑥𝑗)𝑥𝑗∈𝑁𝑝	PROPN
ijassa-659	144	13	𝑘	𝑘	PRON
ijassa-659	144	14	𝑝=1	𝑝=1	PROPN
ijassa-659	144	15	(	(	PUNCT
ijassa-659	144	16	7	7	NUM
ijassa-659	144	17	)	)	PUNCT
ijassa-659	144	18	where	where	SCONJ
ijassa-659	144	19	m(s	m(s	PROPN
ijassa-659	144	20	,	,	PUNCT
ijassa-659	144	21	xj	xj	NOUN
ijassa-659	144	22	)	)	PUNCT
ijassa-659	144	23	denote	denote	VERB
ijassa-659	144	24	the	the	DET
ijassa-659	144	25	similarity	similarity	NOUN
ijassa-659	144	26	between	between	ADP
ijassa-659	144	27	instance	instance	NOUN
ijassa-659	144	28	s	s	PROPN
ijassa-659	144	29	and	and	CCONJ
ijassa-659	144	30	instance	instance	PROPN
ijassa-659	144	31	xj	xj	PROPN
ijassa-659	144	32	,	,	PUNCT
ijassa-659	144	33	|ni|	|ni|	PROPN
ijassa-659	144	34	indicates	indicate	VERB
ijassa-659	144	35	the	the	DET
ijassa-659	144	36	number	number	NOUN
ijassa-659	144	37	of	of	ADP
ijassa-659	144	38	instances	instance	NOUN
ijassa-659	144	39	in	in	ADP
ijassa-659	144	40	neighborhood	neighborhood	NOUN
ijassa-659	144	41	ni	ni	NOUN
ijassa-659	144	42	,	,	PUNCT
ijassa-659	144	43	and	and	CCONJ
ijassa-659	144	44	k	k	PROPN
ijassa-659	144	45	is	be	AUX
ijassa-659	144	46	the	the	DET
ijassa-659	144	47	total	total	ADJ
ijassa-659	144	48	number	number	NOUN
ijassa-659	144	49	of	of	ADP
ijassa-659	144	50	existing	exist	VERB
ijassa-659	144	51	neighborhoods	neighborhood	NOUN
ijassa-659	144	52	.	.	PUNCT
ijassa-659	145	1	we	we	PRON
ijassa-659	145	2	should	should	AUX
ijassa-659	145	3	always	always	ADV
ijassa-659	145	4	start	start	VERB
ijassa-659	145	5	by	by	ADP
ijassa-659	145	6	querying	query	VERB
ijassa-659	145	7	s	s	PRON
ijassa-659	145	8	against	against	ADP
ijassa-659	145	9	the	the	DET
ijassa-659	145	10	neighborhood	neighborhood	NOUN
ijassa-659	145	11	that	that	PRON
ijassa-659	145	12	has	have	VERB
ijassa-659	145	13	the	the	DET
ijassa-659	145	14	highest	high	ADJ
ijassa-659	145	15	probability	probability	NOUN
ijassa-659	145	16	of	of	ADP
ijassa-659	145	17	containing	contain	VERB
ijassa-659	145	18	s	s	NOUN
ijassa-659	145	19	to	to	PART
ijassa-659	145	20	minimize	minimize	VERB
ijassa-659	145	21	the	the	DET
ijassa-659	145	22	total	total	ADJ
ijassa-659	145	23	number	number	NOUN
ijassa-659	145	24	of	of	ADP
ijassa-659	145	25	required	require	VERB
ijassa-659	145	26	queries	query	NOUN
ijassa-659	145	27	.	.	PUNCT
ijassa-659	146	1	if	if	SCONJ
ijassa-659	146	2	a	a	DET
ijassa-659	146	3	must	must	AUX
ijassa-659	146	4	-	-	PUNCT
ijassa-659	146	5	link	link	NOUN
ijassa-659	146	6	is	be	AUX
ijassa-659	146	7	returned	return	VERB
ijassa-659	146	8	,	,	PUNCT
ijassa-659	146	9	we	we	PRON
ijassa-659	146	10	can	can	AUX
ijassa-659	146	11	stop	stop	VERB
ijassa-659	146	12	with	with	ADP
ijassa-659	146	13	only	only	ADV
ijassa-659	146	14	one	one	NUM
ijassa-659	146	15	query	query	NOUN
ijassa-659	146	16	.	.	PUNCT
ijassa-659	147	1	otherwise	otherwise	ADV
ijassa-659	147	2	,	,	PUNCT
ijassa-659	147	3	one	one	PRON
ijassa-659	147	4	should	should	AUX
ijassa-659	147	5	ask	ask	VERB
ijassa-659	147	6	the	the	DET
ijassa-659	147	7	next	next	ADJ
ijassa-659	147	8	query	query	NOUN
ijassa-659	147	9	against	against	ADP
ijassa-659	147	10	the	the	DET
ijassa-659	147	11	neighborhood	neighborhood	NOUN
ijassa-659	147	12	that	that	PRON
ijassa-659	147	13	has	have	VERB
ijassa-659	147	14	the	the	DET
ijassa-659	147	15	next	next	ADJ
ijassa-659	147	16	highest	high	ADJ
ijassa-659	147	17	probability	probability	NOUN
ijassa-659	147	18	of	of	ADP
ijassa-659	147	19	containing	contain	VERB
ijassa-659	147	20	s.	s.	PROPN
ijassa-659	147	21	this	this	DET
ijassa-659	147	22	process	process	NOUN
ijassa-659	147	23	is	be	AUX
ijassa-659	147	24	repeated	repeat	VERB
ijassa-659	147	25	until	until	SCONJ
ijassa-659	147	26	a	a	DET
ijassa-659	147	27	must	must	AUX
ijassa-659	147	28	-	-	PUNCT
ijassa-659	147	29	link	link	NOUN
ijassa-659	147	30	constraint	constraint	NOUN
ijassa-659	147	31	is	be	AUX
ijassa-659	147	32	returned	return	VERB
ijassa-659	147	33	or	or	CCONJ
ijassa-659	147	34	we	we	PRON
ijassa-659	147	35	have	have	VERB
ijassa-659	147	36	a	a	DET
ijassa-659	147	37	cannot	cannot	ADJ
ijassa-659	147	38	-	-	PUNCT
ijassa-659	147	39	link	link	NOUN
ijassa-659	147	40	constraint	constraint	NOUN
ijassa-659	147	41	against	against	ADP
ijassa-659	147	42	all	all	DET
ijassa-659	147	43	neighborhoods	neighborhood	NOUN
ijassa-659	147	44	.	.	PUNCT
ijassa-659	148	1	if	if	SCONJ
ijassa-659	148	2	no	no	DET
ijassa-659	148	3	must	must	AUX
ijassa-659	148	4	-	-	PUNCT
ijassa-659	148	5	link	link	NOUN
ijassa-659	148	6	is	be	AUX
ijassa-659	148	7	achieved	achieve	VERB
ijassa-659	148	8	,	,	PUNCT
ijassa-659	148	9	a	a	DET
ijassa-659	148	10	new	new	ADJ
ijassa-659	148	11	neighborhood	neighborhood	NOUN
ijassa-659	148	12	will	will	AUX
ijassa-659	148	13	be	be	AUX
ijassa-659	148	14	created	create	VERB
ijassa-659	148	15	using	use	VERB
ijassa-659	148	16	the	the	DET
ijassa-659	148	17	instance	instance	NOUN
ijassa-659	148	18	s	s	PROPN
ijassa-659	148	19	(	(	PUNCT
ijassa-659	148	20	lines	line	NOUN
ijassa-659	148	21	14	14	NUM
ijassa-659	148	22	-	-	SYM
ijassa-659	148	23	16	16	NUM
ijassa-659	148	24	)	)	PUNCT
ijassa-659	148	25	.	.	PUNCT
ijassa-659	149	1	finally	finally	ADV
ijassa-659	149	2	,	,	PUNCT
ijassa-659	149	3	we	we	PRON
ijassa-659	149	4	apply	apply	VERB
ijassa-659	149	5	the	the	DET
ijassa-659	149	6	semi	semi	ADJ
ijassa-659	149	7	-	-	ADJ
ijassa-659	149	8	supervised	supervised	ADJ
ijassa-659	149	9	clustering	clustering	ADJ
ijassa-659	149	10	algorithm	algorithm	NOUN
ijassa-659	149	11	using	use	VERB
ijassa-659	149	12	the	the	DET
ijassa-659	149	13	selected	select	VERB
ijassa-659	149	14	active	active	ADJ
ijassa-659	149	15	pairwise	pairwise	NOUN
ijassa-659	149	16	constraints	constraint	NOUN
ijassa-659	149	17	to	to	PART
ijassa-659	149	18	generate	generate	VERB
ijassa-659	149	19	the	the	DET
ijassa-659	149	20	final	final	ADJ
ijassa-659	149	21	clusters	cluster	NOUN
ijassa-659	149	22	(	(	PUNCT
ijassa-659	149	23	line	line	NOUN
ijassa-659	149	24	18	18	NUM
ijassa-659	149	25	)	)	PUNCT
ijassa-659	149	26	.	.	PUNCT
ijassa-659	150	1	in	in	ADP
ijassa-659	150	2	this	this	DET
ijassa-659	150	3	paper	paper	NOUN
ijassa-659	150	4	,	,	PUNCT
ijassa-659	150	5	we	we	PRON
ijassa-659	150	6	consider	consider	VERB
ijassa-659	150	7	the	the	DET
ijassa-659	150	8	semi	semi	ADJ
ijassa-659	150	9	-	-	ADJ
ijassa-659	150	10	supervised	supervised	ADJ
ijassa-659	150	11	clustering	clustering	ADJ
ijassa-659	150	12	algorithm	algorithm	NOUN
ijassa-659	150	13	as	as	ADP
ijassa-659	150	14	a	a	DET
ijassa-659	150	15	black	black	ADJ
ijassa-659	150	16	-	-	PUNCT
ijassa-659	150	17	box	box	NOUN
ijassa-659	150	18	and	and	CCONJ
ijassa-659	150	19	any	any	DET
ijassa-659	150	20	existing	exist	VERB
ijassa-659	150	21	algorithm	algorithm	NOUN
ijassa-659	150	22	can	can	AUX
ijassa-659	150	23	be	be	AUX
ijassa-659	150	24	used	use	VERB
ijassa-659	150	25	here	here	ADV
ijassa-659	150	26	.	.	PUNCT
ijassa-659	151	1	n1	n1	PROPN
ijassa-659	151	2	n2	n2	NOUN
ijassa-659	152	1	x1	x1	NUM
ijassa-659	152	2	x3	x3	NOUN
ijassa-659	153	1	x2	x2	PROPN
ijassa-659	153	2	x4	x4	PROPN
ijassa-659	153	3	x5	x5	PROPN
ijassa-659	153	4	x6	x6	PROPN
ijassa-659	153	5	improving	improve	VERB
ijassa-659	153	6	semi	semi	ADJ
ijassa-659	153	7	-	-	ADJ
ijassa-659	153	8	supervised	supervised	ADJ
ijassa-659	153	9	clustering	clustering	ADJ
ijassa-659	153	10	algorithms	algorithm	NOUN
ijassa-659	153	11	with	with	ADP
ijassa-659	153	12	active	active	ADJ
ijassa-659	153	13	query	query	NOUN
ijassa-659	153	14	selection	selection	NOUN
ijassa-659	153	15	31	31	NUM
ijassa-659	153	16	copyright	copyright	NOUN
ijassa-659	153	17	©	©	PROPN
ijassa-659	153	18	2019	2019	NUM
ijassa-659	153	19	assa	assa	PROPN
ijassa-659	153	20	adv	adv	PROPN
ijassa-659	153	21	.	.	PUNCT
ijassa-659	154	1	in	in	ADP
ijassa-659	154	2	systems	system	NOUN
ijassa-659	154	3	science	science	NOUN
ijassa-659	154	4	and	and	CCONJ
ijassa-659	154	5	appl	appl	NOUN
ijassa-659	154	6	.	.	PUNCT
ijassa-659	155	1	(	(	PUNCT
ijassa-659	155	2	2019	2019	NUM
ijassa-659	155	3	)	)	PUNCT
ijassa-659	155	4	algorithm	algorithm	NOUN
ijassa-659	155	5	2	2	NUM
ijassa-659	155	6	.	.	PUNCT
ijassa-659	156	1	the	the	DET
ijassa-659	156	2	proposed	propose	VERB
ijassa-659	156	3	active	active	ADJ
ijassa-659	156	4	learning	learning	NOUN
ijassa-659	156	5	method	method	NOUN
ijassa-659	156	6	input	input	NOUN
ijassa-659	156	7	:	:	PUNCT
ijassa-659	156	8	a	a	DET
ijassa-659	156	9	set	set	NOUN
ijassa-659	156	10	of	of	ADP
ijassa-659	156	11	instances	instance	NOUN
ijassa-659	156	12	d	d	X
ijassa-659	156	13	(	(	PUNCT
ijassa-659	156	14	divided	divide	VERB
ijassa-659	156	15	into	into	ADP
ijassa-659	156	16	labeled	label	VERB
ijassa-659	156	17	set	set	VERB
ijassa-659	156	18	dl	dl	PROPN
ijassa-659	156	19	and	and	CCONJ
ijassa-659	156	20	unlabeled	unlabele	VERB
ijassa-659	156	21	set	set	VERB
ijassa-659	156	22	du	du	PROPN
ijassa-659	156	23	)	)	PUNCT
ijassa-659	156	24	;	;	PUNCT
ijassa-659	156	25	total	total	ADJ
ijassa-659	156	26	number	number	NOUN
ijassa-659	156	27	of	of	ADP
ijassa-659	156	28	queries	query	NOUN
ijassa-659	156	29	q	q	NOUN
ijassa-659	156	30	;	;	PUNCT
ijassa-659	156	31	batch	batch	NOUN
ijassa-659	156	32	size	size	NOUN
ijassa-659	156	33	b.	b.	PROPN
ijassa-659	156	34	output	output	PROPN
ijassa-659	156	35	:	:	PUNCT
ijassa-659	156	36	a	a	DET
ijassa-659	156	37	set	set	NOUN
ijassa-659	156	38	of	of	ADP
ijassa-659	156	39	clusters	cluster	NOUN
ijassa-659	156	40	.	.	PUNCT
ijassa-659	157	1	1	1	X
ijassa-659	157	2	.	.	X
ijassa-659	157	3	initialization	initialization	NOUN
ijassa-659	157	4	:	:	PUNCT
ijassa-659	157	5	set	set	VERB
ijassa-659	157	6	n1	n1	NOUN
ijassa-659	157	7	=	=	SYM
ijassa-659	157	8	{	{	PUNCT
ijassa-659	157	9	x	x	NOUN
ijassa-659	157	10	}	}	PUNCT
ijassa-659	157	11	,	,	PUNCT
ijassa-659	157	12	where	where	SCONJ
ijassa-659	157	13	x	x	PRON
ijassa-659	157	14	is	be	AUX
ijassa-659	157	15	a	a	DET
ijassa-659	157	16	random	random	ADJ
ijassa-659	157	17	instance	instance	NOUN
ijassa-659	157	18	;	;	PUNCT
ijassa-659	157	19	c	c	NOUN
ijassa-659	157	20	=	=	SYM
ijassa-659	157	21	∅	∅	NOUN
ijassa-659	157	22	;	;	PUNCT
ijassa-659	157	23	q	q	X
ijassa-659	157	24	=	=	SYM
ijassa-659	157	25	0	0	NUM
ijassa-659	157	26	;	;	PUNCT
ijassa-659	157	27	2	2	NUM
ijassa-659	157	28	.	.	X
ijassa-659	158	1	while	while	SCONJ
ijassa-659	158	2	q	q	PUNCT
ijassa-659	158	3	<	<	X
ijassa-659	158	4	q	q	NOUN
ijassa-659	158	5	3	3	NUM
ijassa-659	158	6	.	.	PUNCT
ijassa-659	158	7	s	s	PART
ijassa-659	158	8	=	=	X
ijassa-659	158	9	queryinstancesselection(dl	queryinstancesselection(dl	PROPN
ijassa-659	158	10	,	,	PUNCT
ijassa-659	158	11	du	du	X
ijassa-659	158	12	,	,	PUNCT
ijassa-659	158	13	b	b	NOUN
ijassa-659	158	14	)	)	PUNCT
ijassa-659	158	15	;	;	PUNCT
ijassa-659	158	16	4	4	X
ijassa-659	158	17	.	.	X
ijassa-659	158	18	for	for	ADP
ijassa-659	158	19	each	each	DET
ijassa-659	158	20	instance	instance	NOUN
ijassa-659	158	21	s	s	PROPN
ijassa-659	158	22	∈	∈	PROPN
ijassa-659	158	23	𝑆	𝑆	PROPN
ijassa-659	158	24	5	5	NUM
ijassa-659	158	25	.	.	PUNCT
ijassa-659	158	26	for	for	ADP
ijassa-659	158	27	each	each	DET
ijassa-659	158	28	neighborhood	neighborhood	NOUN
ijassa-659	158	29	ni	ni	NOUN
ijassa-659	158	30	∈	∈	PROPN
ijassa-659	158	31	𝑁	𝑁	PROPN
ijassa-659	158	32	in	in	ADP
ijassa-659	158	33	decreasing	decrease	VERB
ijassa-659	158	34	order	order	NOUN
ijassa-659	158	35	of	of	ADP
ijassa-659	158	36	p(s∈	p(s∈	NOUN
ijassa-659	158	37	ni	ni	PROPN
ijassa-659	158	38	)	)	PUNCT
ijassa-659	158	39	6	6	NUM
ijassa-659	158	40	.	.	PUNCT
ijassa-659	158	41	query	query	PROPN
ijassa-659	158	42	instance	instance	PROPN
ijassa-659	158	43	s	s	VERB
ijassa-659	158	44	against	against	ADP
ijassa-659	158	45	any	any	DET
ijassa-659	158	46	instance	instance	NOUN
ijassa-659	158	47	xi	xi	PROPN
ijassa-659	158	48	∈	∈	PROPN
ijassa-659	158	49	ni	ni	PROPN
ijassa-659	158	50	;	;	PUNCT
ijassa-659	158	51	7	7	X
ijassa-659	158	52	.	.	NUM
ijassa-659	158	53	q++	q++	VERB
ijassa-659	158	54	;	;	PUNCT
ijassa-659	158	55	8	8	X
ijassa-659	158	56	.	.	X
ijassa-659	158	57	update	update	VERB
ijassa-659	158	58	the	the	DET
ijassa-659	158	59	constraint	constraint	NOUN
ijassa-659	158	60	set	set	NOUN
ijassa-659	158	61	c	c	PROPN
ijassa-659	158	62	based	base	VERB
ijassa-659	158	63	on	on	ADP
ijassa-659	158	64	the	the	DET
ijassa-659	158	65	results	result	NOUN
ijassa-659	158	66	;	;	PUNCT
ijassa-659	158	67	9	9	X
ijassa-659	158	68	.	.	X
ijassa-659	159	1	if	if	SCONJ
ijassa-659	159	2	a	a	DET
ijassa-659	159	3	must	must	AUX
ijassa-659	159	4	-	-	PUNCT
ijassa-659	159	5	link	link	NOUN
ijassa-659	159	6	achieved	achieve	VERB
ijassa-659	159	7	between	between	ADP
ijassa-659	159	8	s	s	PRON
ijassa-659	159	9	and	and	CCONJ
ijassa-659	159	10	xi	xi	X
ijassa-659	159	11	then	then	ADV
ijassa-659	159	12	10	10	NUM
ijassa-659	159	13	.	.	PUNCT
ijassa-659	159	14	add	add	VERB
ijassa-659	159	15	instance	instance	NOUN
ijassa-659	159	16	s	s	VERB
ijassa-659	159	17	to	to	PART
ijassa-659	159	18	neighborhood	neighborhood	VERB
ijassa-659	159	19	ni	ni	NOUN
ijassa-659	159	20	;	;	PUNCT
ijassa-659	159	21	11	11	NUM
ijassa-659	159	22	.	.	X
ijassa-659	160	1	break	break	NOUN
ijassa-659	160	2	;	;	PUNCT
ijassa-659	160	3	12	12	NUM
ijassa-659	160	4	.	.	PUNCT
ijassa-659	161	1	end	end	VERB
ijassa-659	161	2	if	if	SCONJ
ijassa-659	161	3	13	13	NUM
ijassa-659	161	4	.	.	PUNCT
ijassa-659	161	5	end	end	VERB
ijassa-659	161	6	for	for	ADP
ijassa-659	161	7	14	14	NUM
ijassa-659	161	8	.	.	PUNCT
ijassa-659	162	1	if	if	SCONJ
ijassa-659	162	2	no	no	DET
ijassa-659	162	3	must	must	AUX
ijassa-659	162	4	-	-	PUNCT
ijassa-659	162	5	link	link	NOUN
ijassa-659	162	6	is	be	AUX
ijassa-659	162	7	achieved	achieve	VERB
ijassa-659	162	8	then	then	ADV
ijassa-659	162	9	15	15	NUM
ijassa-659	162	10	.	.	PUNCT
ijassa-659	162	11	create	create	VERB
ijassa-659	162	12	new	new	ADJ
ijassa-659	162	13	neighborhood	neighborhood	NOUN
ijassa-659	162	14	with	with	ADP
ijassa-659	162	15	the	the	DET
ijassa-659	162	16	instance	instance	NOUN
ijassa-659	162	17	s	s	NOUN
ijassa-659	162	18	;	;	PUNCT
ijassa-659	162	19	16	16	NUM
ijassa-659	162	20	.	.	PUNCT
ijassa-659	163	1	end	end	VERB
ijassa-659	163	2	if	if	SCONJ
ijassa-659	163	3	17	17	NUM
ijassa-659	163	4	.	.	PUNCT
ijassa-659	164	1	end	end	NOUN
ijassa-659	164	2	for	for	ADP
ijassa-659	164	3	18	18	NUM
ijassa-659	164	4	.	.	PUNCT
ijassa-659	165	1	apply	apply	VERB
ijassa-659	165	2	semi	semi	ADJ
ijassa-659	165	3	-	-	ADJ
ijassa-659	165	4	supervised	supervised	ADJ
ijassa-659	165	5	clustering(d	clustering(d	PROPN
ijassa-659	165	6	,	,	PUNCT
ijassa-659	165	7	c	c	NOUN
ijassa-659	165	8	)	)	PUNCT
ijassa-659	165	9	;	;	PUNCT
ijassa-659	165	10	19	19	NUM
ijassa-659	165	11	.	.	NOUN
ijassa-659	165	12	end	end	NOUN
ijassa-659	165	13	while	while	SCONJ
ijassa-659	165	14	4	4	NUM
ijassa-659	165	15	.	.	PUNCT
ijassa-659	165	16	experiments	experiment	NOUN
ijassa-659	165	17	in	in	ADP
ijassa-659	165	18	this	this	DET
ijassa-659	165	19	section	section	NOUN
ijassa-659	165	20	,	,	PUNCT
ijassa-659	165	21	we	we	PRON
ijassa-659	165	22	evaluated	evaluate	VERB
ijassa-659	165	23	the	the	DET
ijassa-659	165	24	accuracy	accuracy	NOUN
ijassa-659	165	25	and	and	CCONJ
ijassa-659	165	26	efficiency	efficiency	NOUN
ijassa-659	165	27	of	of	ADP
ijassa-659	165	28	the	the	DET
ijassa-659	165	29	proposed	propose	VERB
ijassa-659	165	30	method	method	NOUN
ijassa-659	165	31	on	on	ADP
ijassa-659	165	32	a	a	DET
ijassa-659	165	33	diffrent	diffrent	ADJ
ijassa-659	165	34	realworld	realworld	PROPN
ijassa-659	165	35	datasets	dataset	NOUN
ijassa-659	165	36	.	.	PUNCT
ijassa-659	166	1	the	the	DET
ijassa-659	166	2	results	result	NOUN
ijassa-659	166	3	compared	compare	VERB
ijassa-659	166	4	the	the	DET
ijassa-659	166	5	proposed	propose	VERB
ijassa-659	166	6	method	method	NOUN
ijassa-659	166	7	with	with	ADP
ijassa-659	166	8	other	other	ADJ
ijassa-659	166	9	constraint	constraint	NOUN
ijassa-659	166	10	selection	selection	NOUN
ijassa-659	166	11	heuristics	heuristic	NOUN
ijassa-659	166	12	and	and	CCONJ
ijassa-659	166	13	evaluated	evaluate	VERB
ijassa-659	166	14	in	in	ADP
ijassa-659	166	15	conjunction	conjunction	NOUN
ijassa-659	166	16	with	with	ADP
ijassa-659	166	17	two	two	NUM
ijassa-659	166	18	different	different	ADJ
ijassa-659	166	19	constraint	constraint	NOUN
ijassa-659	166	20	-	-	PUNCT
ijassa-659	166	21	based	base	VERB
ijassa-659	166	22	clustering	clustering	ADJ
ijassa-659	166	23	algorithms	algorithm	NOUN
ijassa-659	166	24	to	to	PART
ijassa-659	166	25	show	show	VERB
ijassa-659	166	26	the	the	DET
ijassa-659	166	27	adaptability	adaptability	NOUN
ijassa-659	166	28	of	of	ADP
ijassa-659	166	29	the	the	DET
ijassa-659	166	30	proposed	propose	VERB
ijassa-659	166	31	method	method	NOUN
ijassa-659	166	32	.	.	PUNCT
ijassa-659	167	1	the	the	DET
ijassa-659	167	2	rest	rest	NOUN
ijassa-659	167	3	of	of	ADP
ijassa-659	167	4	this	this	DET
ijassa-659	167	5	section	section	NOUN
ijassa-659	167	6	is	be	AUX
ijassa-659	167	7	organized	organize	VERB
ijassa-659	167	8	as	as	SCONJ
ijassa-659	167	9	follows	follow	VERB
ijassa-659	167	10	.	.	PUNCT
ijassa-659	168	1	section	section	NOUN
ijassa-659	168	2	4.1	4.1	NUM
ijassa-659	168	3	mentions	mention	VERB
ijassa-659	168	4	the	the	DET
ijassa-659	168	5	datasets	dataset	NOUN
ijassa-659	168	6	used	use	VERB
ijassa-659	168	7	for	for	ADP
ijassa-659	168	8	evaluating	evaluate	VERB
ijassa-659	168	9	the	the	DET
ijassa-659	168	10	proposed	propose	VERB
ijassa-659	168	11	method	method	NOUN
ijassa-659	168	12	.	.	PUNCT
ijassa-659	169	1	section	section	NOUN
ijassa-659	169	2	4.2	4.2	NUM
ijassa-659	169	3	describes	describe	VERB
ijassa-659	169	4	some	some	DET
ijassa-659	169	5	constraint	constraint	NOUN
ijassa-659	169	6	selection	selection	NOUN
ijassa-659	169	7	heuristics	heuristic	NOUN
ijassa-659	169	8	that	that	PRON
ijassa-659	169	9	are	be	AUX
ijassa-659	169	10	compared	compare	VERB
ijassa-659	169	11	with	with	ADP
ijassa-659	169	12	the	the	DET
ijassa-659	169	13	proposed	propose	VERB
ijassa-659	169	14	method	method	NOUN
ijassa-659	169	15	and	and	CCONJ
ijassa-659	169	16	section	section	NOUN
ijassa-659	169	17	4.3	4.3	NUM
ijassa-659	169	18	describes	describe	VERB
ijassa-659	169	19	different	different	ADJ
ijassa-659	169	20	constraint	constraint	NOUN
ijassa-659	169	21	-	-	PUNCT
ijassa-659	169	22	based	base	VERB
ijassa-659	169	23	clustering	clustering	ADJ
ijassa-659	169	24	algorithms	algorithm	NOUN
ijassa-659	169	25	used	use	VERB
ijassa-659	169	26	for	for	ADP
ijassa-659	169	27	constraints	constraint	NOUN
ijassa-659	169	28	evaluation	evaluation	NOUN
ijassa-659	169	29	.	.	PUNCT
ijassa-659	170	1	section	section	NOUN
ijassa-659	170	2	4.4	4.4	NUM
ijassa-659	170	3	explains	explain	VERB
ijassa-659	170	4	the	the	DET
ijassa-659	170	5	evaluation	evaluation	NOUN
ijassa-659	170	6	metrics	metric	NOUN
ijassa-659	170	7	used	use	VERB
ijassa-659	170	8	in	in	ADP
ijassa-659	170	9	this	this	DET
ijassa-659	170	10	paper	paper	NOUN
ijassa-659	170	11	.	.	PUNCT
ijassa-659	171	1	the	the	DET
ijassa-659	171	2	experimental	experimental	ADJ
ijassa-659	171	3	results	result	NOUN
ijassa-659	171	4	are	be	AUX
ijassa-659	171	5	described	describe	VERB
ijassa-659	171	6	in	in	ADP
ijassa-659	171	7	section	section	NOUN
ijassa-659	171	8	4.5	4.5	NUM
ijassa-659	171	9	.	.	PUNCT
ijassa-659	172	1	4.1	4.1	NUM
ijassa-659	172	2	datasets	dataset	NOUN
ijassa-659	172	3	experiments	experiment	NOUN
ijassa-659	172	4	are	be	AUX
ijassa-659	172	5	conducted	conduct	VERB
ijassa-659	172	6	on	on	ADP
ijassa-659	172	7	8	8	NUM
ijassa-659	172	8	datasets	dataset	NOUN
ijassa-659	172	9	from	from	ADP
ijassa-659	172	10	uci	uci	PROPN
ijassa-659	172	11	machine	machine	NOUN
ijassa-659	172	12	learning	learn	VERB
ijassa-659	172	13	repository†	repository†	NOUN
ijassa-659	172	14	(	(	PUNCT
ijassa-659	172	15	each	each	PRON
ijassa-659	172	16	with	with	ADP
ijassa-659	172	17	the	the	DET
ijassa-659	172	18	following	follow	VERB
ijassa-659	172	19	number	number	NOUN
ijassa-659	172	20	of	of	ADP
ijassa-659	172	21	instances	instance	NOUN
ijassa-659	172	22	,	,	PUNCT
ijassa-659	172	23	attributes	attribute	NOUN
ijassa-659	172	24	and	and	CCONJ
ijassa-659	172	25	clusters	cluster	NOUN
ijassa-659	172	26	):	):	PUNCT
ijassa-659	172	27	protein	protein	NOUN
ijassa-659	172	28	(	(	PUNCT
ijassa-659	172	29	116/20/6	116/20/6	NUM
ijassa-659	172	30	)	)	PUNCT
ijassa-659	172	31	[	[	X
ijassa-659	172	32	15	15	NUM
ijassa-659	172	33	]	]	PUNCT
ijassa-659	172	34	,	,	PUNCT
ijassa-659	172	35	heart	heart	NOUN
ijassa-659	172	36	(	(	PUNCT
ijassa-659	172	37	270/13/2	270/13/2	NUM
ijassa-659	172	38	)	)	PUNCT
ijassa-659	173	1	[	[	X
ijassa-659	173	2	17	17	NUM
ijassa-659	173	3	]	]	PUNCT
ijassa-659	173	4	,	,	PUNCT
ijassa-659	173	5	ionosphere	ionosphere	X
ijassa-659	173	6	(	(	PUNCT
ijassa-659	173	7	351/34/2	351/34/2	NOUN
ijassa-659	173	8	)	)	PUNCT
ijassa-659	174	1	[	[	X
ijassa-659	174	2	13	13	NUM
ijassa-659	174	3	,	,	PUNCT
ijassa-659	174	4	14	14	NUM
ijassa-659	174	5	]	]	PUNCT
ijassa-659	174	6	,	,	PUNCT
ijassa-659	174	7	breast	breast	NOUN
ijassa-659	174	8	(	(	PUNCT
ijassa-659	174	9	683/9/2	683/9/2	NUM
ijassa-659	174	10	)	)	PUNCT
ijassa-659	175	1	[	[	X
ijassa-659	175	2	15	15	NUM
ijassa-659	175	3	,	,	PUNCT
ijassa-659	175	4	17	17	NUM
ijassa-659	175	5	]	]	PUNCT
ijassa-659	175	6	,	,	PUNCT
ijassa-659	175	7	yeast	yeast	NOUN
ijassa-659	175	8	(	(	PUNCT
ijassa-659	175	9	1484/8/10	1484/8/10	NUM
ijassa-659	175	10	)	)	PUNCT
ijassa-659	176	1	[	[	X
ijassa-659	176	2	15	15	NUM
ijassa-659	176	3	]	]	PUNCT
ijassa-659	176	4	,	,	PUNCT
ijassa-659	176	5	image	image	NOUN
ijassa-659	176	6	segmentation	segmentation	NOUN
ijassa-659	176	7	(	(	PUNCT
ijassa-659	176	8	2310/19/7	2310/19/7	NUM
ijassa-659	176	9	)	)	PUNCT
ijassa-659	177	1	[	[	X
ijassa-659	177	2	15	15	NUM
ijassa-659	177	3	,	,	PUNCT
ijassa-659	177	4	17	17	NUM
ijassa-659	177	5	]	]	PUNCT
ijassa-659	177	6	,	,	PUNCT
ijassa-659	177	7	digit-389	digit-389	PROPN
ijassa-659	177	8	(	(	PUNCT
ijassa-659	177	9	3165/16/3	3165/16/3	NUM
ijassa-659	177	10	)	)	PUNCT
ijassa-659	178	1	[	[	X
ijassa-659	178	2	17	17	NUM
ijassa-659	178	3	]	]	PUNCT
ijassa-659	178	4	and	and	CCONJ
ijassa-659	178	5	magic	magic	NOUN
ijassa-659	178	6	(	(	PUNCT
ijassa-659	178	7	19020/10/2	19020/10/2	NUM
ijassa-659	178	8	)	)	PUNCT
ijassa-659	179	1	[	[	X
ijassa-659	179	2	10	10	NUM
ijassa-659	179	3	]	]	PUNCT
ijassa-659	179	4	.	.	PUNCT
ijassa-659	180	1	these	these	DET
ijassa-659	180	2	datasets	dataset	NOUN
ijassa-659	180	3	have	have	AUX
ijassa-659	180	4	been	be	AUX
ijassa-659	180	5	chosen	choose	VERB
ijassa-659	180	6	because	because	SCONJ
ijassa-659	180	7	they	they	PRON
ijassa-659	180	8	facilitate	facilitate	VERB
ijassa-659	180	9	the	the	DET
ijassa-659	180	10	reproducibility	reproducibility	NOUN
ijassa-659	180	11	of	of	ADP
ijassa-659	180	12	the	the	DET
ijassa-659	180	13	experiments	experiment	NOUN
ijassa-659	180	14	and	and	CCONJ
ijassa-659	180	15	because	because	SCONJ
ijassa-659	180	16	some	some	PRON
ijassa-659	180	17	of	of	ADP
ijassa-659	180	18	them	they	PRON
ijassa-659	180	19	have	have	AUX
ijassa-659	180	20	already	already	ADV
ijassa-659	180	21	been	be	AUX
ijassa-659	180	22	used	use	VERB
ijassa-659	180	23	in	in	ADP
ijassa-659	180	24	constraint	constraint	NOUN
ijassa-659	180	25	-	-	PUNCT
ijassa-659	180	26	based	base	VERB
ijassa-659	180	27	clustering	clustering	ADJ
ijassa-659	180	28	articles	article	NOUN
ijassa-659	180	29	.	.	PUNCT
ijassa-659	181	1	also	also	ADV
ijassa-659	181	2	,	,	PUNCT
ijassa-659	181	3	provide	provide	VERB
ijassa-659	181	4	†	†	NOUN
ijassa-659	181	5	http://www.ics.uci.edu/~mlearn/mlrepository.html	http://www.ics.uci.edu/~mlearn/mlrepository.html	PROPN
ijassa-659	181	6	http://www.ics.uci.edu/~mlearn/mlrepository.html	http://www.ics.uci.edu/~mlearn/mlrepository.html	PROPN
ijassa-659	181	7	32	32	NUM
ijassa-659	181	8	w.	w.	NOUN
ijassa-659	181	9	atwa	atwa	PROPN
ijassa-659	181	10	,	,	PUNCT
ijassa-659	181	11	m.	m.	NOUN
ijassa-659	181	12	emam	emam	PROPN
ijassa-659	181	13	copyright	copyright	NOUN
ijassa-659	181	14	©	©	PROPN
ijassa-659	181	15	2019	2019	NUM
ijassa-659	181	16	assa	assa	PROPN
ijassa-659	181	17	adv	adv	PROPN
ijassa-659	181	18	.	.	PUNCT
ijassa-659	182	1	in	in	ADP
ijassa-659	182	2	systems	system	NOUN
ijassa-659	182	3	science	science	NOUN
ijassa-659	182	4	and	and	CCONJ
ijassa-659	182	5	appl	appl	NOUN
ijassa-659	182	6	.	.	PUNCT
ijassa-659	183	1	(	(	PUNCT
ijassa-659	183	2	2019	2019	NUM
ijassa-659	183	3	)	)	PUNCT
ijassa-659	183	4	a	a	DET
ijassa-659	183	5	good	good	ADJ
ijassa-659	183	6	representation	representation	NOUN
ijassa-659	183	7	of	of	ADP
ijassa-659	183	8	different	different	ADJ
ijassa-659	183	9	characteristics	characteristic	NOUN
ijassa-659	183	10	:	:	PUNCT
ijassa-659	183	11	number	number	NOUN
ijassa-659	183	12	of	of	ADP
ijassa-659	183	13	instances	instance	NOUN
ijassa-659	183	14	ranges	range	VERB
ijassa-659	183	15	from	from	ADP
ijassa-659	183	16	116	116	NUM
ijassa-659	183	17	to	to	ADP
ijassa-659	183	18	19020	19020	NUM
ijassa-659	183	19	,	,	PUNCT
ijassa-659	183	20	dimensionalities	dimensionality	NOUN
ijassa-659	183	21	from	from	ADP
ijassa-659	183	22	8	8	NUM
ijassa-659	183	23	to	to	ADP
ijassa-659	183	24	34	34	NUM
ijassa-659	183	25	,	,	PUNCT
ijassa-659	183	26	and	and	CCONJ
ijassa-659	183	27	number	number	NOUN
ijassa-659	183	28	of	of	ADP
ijassa-659	183	29	clusters	cluster	NOUN
ijassa-659	183	30	from	from	ADP
ijassa-659	183	31	2	2	NUM
ijassa-659	183	32	to	to	ADP
ijassa-659	183	33	10	10	NUM
ijassa-659	183	34	.	.	PUNCT
ijassa-659	184	1	4.2	4.2	NUM
ijassa-659	184	2	constraint	constraint	NOUN
ijassa-659	184	3	selection	selection	NOUN
ijassa-659	184	4	heuristics	heuristic	NOUN
ijassa-659	184	5	in	in	ADP
ijassa-659	184	6	all	all	DET
ijassa-659	184	7	experiments	experiment	NOUN
ijassa-659	184	8	,	,	PUNCT
ijassa-659	184	9	we	we	PRON
ijassa-659	184	10	consider	consider	VERB
ijassa-659	184	11	the	the	DET
ijassa-659	184	12	following	follow	VERB
ijassa-659	184	13	heuristics	heuristic	NOUN
ijassa-659	184	14	to	to	PART
ijassa-659	184	15	select	select	VERB
ijassa-659	184	16	the	the	DET
ijassa-659	184	17	constraints	constraint	NOUN
ijassa-659	184	18	:	:	PUNCT
ijassa-659	185	1			NOUN
ijassa-659	185	2	random	random	ADJ
ijassa-659	185	3	:	:	PUNCT
ijassa-659	185	4	this	this	DET
ijassa-659	185	5	policy	policy	NOUN
ijassa-659	185	6	corresponds	correspond	VERB
ijassa-659	185	7	to	to	ADP
ijassa-659	185	8	a	a	DET
ijassa-659	185	9	completely	completely	ADV
ijassa-659	185	10	random	random	ADJ
ijassa-659	185	11	selection	selection	NOUN
ijassa-659	185	12	of	of	ADP
ijassa-659	185	13	the	the	DET
ijassa-659	185	14	constraints	constraint	NOUN
ijassa-659	185	15	.	.	PUNCT
ijassa-659	186	1	this	this	DET
ijassa-659	186	2	method	method	NOUN
ijassa-659	186	3	generates	generate	VERB
ijassa-659	186	4	a	a	DET
ijassa-659	186	5	set	set	NOUN
ijassa-659	186	6	of	of	ADP
ijassa-659	186	7	ml	ml	NOUN
ijassa-659	186	8	and	and	CCONJ
ijassa-659	186	9	cl	cl	NOUN
ijassa-659	186	10	constraints	constraint	NOUN
ijassa-659	186	11	based	base	VERB
ijassa-659	186	12	on	on	ADP
ijassa-659	186	13	the	the	DET
ijassa-659	186	14	comparison	comparison	NOUN
ijassa-659	186	15	of	of	ADP
ijassa-659	186	16	the	the	DET
ijassa-659	186	17	labels	label	NOUN
ijassa-659	186	18	of	of	ADP
ijassa-659	186	19	randomly	randomly	ADV
ijassa-659	186	20	chosen	choose	VERB
ijassa-659	186	21	objects	object	NOUN
ijassa-659	186	22	.	.	PUNCT
ijassa-659	187	1	if	if	SCONJ
ijassa-659	187	2	both	both	DET
ijassa-659	187	3	labels	label	NOUN
ijassa-659	187	4	are	be	AUX
ijassa-659	187	5	in	in	ADP
ijassa-659	187	6	the	the	DET
ijassa-659	187	7	same	same	ADJ
ijassa-659	187	8	cluster	cluster	NOUN
ijassa-659	187	9	,	,	PUNCT
ijassa-659	187	10	a	a	DET
ijassa-659	187	11	ml	ml	X
ijassa-659	187	12	constraint	constraint	NOUN
ijassa-659	187	13	is	be	AUX
ijassa-659	187	14	generated	generate	VERB
ijassa-659	187	15	,	,	PUNCT
ijassa-659	187	16	and	and	CCONJ
ijassa-659	187	17	else	else	ADV
ijassa-659	187	18	,	,	PUNCT
ijassa-659	187	19	a	a	DET
ijassa-659	187	20	cl	cl	NOUN
ijassa-659	187	21	constraint	constraint	NOUN
ijassa-659	187	22	is	be	AUX
ijassa-659	187	23	generated	generate	VERB
ijassa-659	187	24	.	.	PUNCT
ijassa-659	188	1			PRON
ijassa-659	189	1	min	min	PROPN
ijassa-659	189	2	-	-	PUNCT
ijassa-659	189	3	max	max	NOUN
ijassa-659	189	4	:	:	PUNCT
ijassa-659	189	5	this	this	DET
ijassa-659	189	6	approach	approach	NOUN
ijassa-659	189	7	is	be	AUX
ijassa-659	189	8	neighborhood	neighborhood	NOUN
ijassa-659	189	9	-	-	PUNCT
ijassa-659	189	10	based	base	VERB
ijassa-659	189	11	approach	approach	NOUN
ijassa-659	189	12	that	that	PRON
ijassa-659	189	13	works	work	VERB
ijassa-659	189	14	in	in	ADP
ijassa-659	189	15	two	two	NUM
ijassa-659	189	16	phases	phase	NOUN
ijassa-659	189	17	[	[	X
ijassa-659	189	18	12	12	NUM
ijassa-659	189	19	]	]	PUNCT
ijassa-659	189	20	.	.	PUNCT
ijassa-659	190	1	in	in	ADP
ijassa-659	190	2	the	the	DET
ijassa-659	190	3	first	first	ADJ
ijassa-659	190	4	phase	phase	NOUN
ijassa-659	190	5	,	,	PUNCT
ijassa-659	190	6	it	it	PRON
ijassa-659	190	7	builds	build	VERB
ijassa-659	190	8	c	c	PROPN
ijassa-659	190	9	disjoint	disjoint	NOUN
ijassa-659	190	10	neighborhoods	neighborhood	NOUN
ijassa-659	190	11	using	use	VERB
ijassa-659	190	12	farthest	farth	ADJ
ijassa-659	190	13	-	-	PUNCT
ijassa-659	190	14	first	first	ADJ
ijassa-659	190	15	traversal	traversal	NOUN
ijassa-659	190	16	,	,	PUNCT
ijassa-659	190	17	where	where	SCONJ
ijassa-659	190	18	c	c	PROPN
ijassa-659	190	19	is	be	AUX
ijassa-659	190	20	the	the	DET
ijassa-659	190	21	total	total	ADJ
ijassa-659	190	22	number	number	NOUN
ijassa-659	190	23	of	of	ADP
ijassa-659	190	24	clusters	cluster	NOUN
ijassa-659	190	25	.	.	PUNCT
ijassa-659	191	1	in	in	ADP
ijassa-659	191	2	the	the	DET
ijassa-659	191	3	second	second	ADJ
ijassa-659	191	4	phase	phase	NOUN
ijassa-659	191	5	,	,	PUNCT
ijassa-659	191	6	it	it	PRON
ijassa-659	191	7	incrementally	incrementally	ADV
ijassa-659	191	8	expands	expand	VERB
ijassa-659	191	9	the	the	DET
ijassa-659	191	10	neighborhoods	neighborhood	NOUN
ijassa-659	191	11	by	by	ADP
ijassa-659	191	12	selecting	select	VERB
ijassa-659	191	13	a	a	DET
ijassa-659	191	14	point	point	NOUN
ijassa-659	191	15	to	to	ADP
ijassa-659	191	16	query	query	NOUN
ijassa-659	191	17	using	use	VERB
ijassa-659	191	18	a	a	DET
ijassa-659	191	19	distance	distance	NOUN
ijassa-659	191	20	-	-	PUNCT
ijassa-659	191	21	based	base	VERB
ijassa-659	191	22	min	min	PROPN
ijassa-659	191	23	-	-	ADJ
ijassa-659	191	24	max	max	PROPN
ijassa-659	191	25	criterion	criterion	NOUN
ijassa-659	191	26	.	.	PUNCT
ijassa-659	192	1			PROPN
ijassa-659	192	2	asc	asc	PROPN
ijassa-659	192	3	:	:	PUNCT
ijassa-659	192	4	an	an	DET
ijassa-659	192	5	active	active	ADJ
ijassa-659	192	6	learning	learning	NOUN
ijassa-659	192	7	algorithm	algorithm	NOUN
ijassa-659	192	8	that	that	PRON
ijassa-659	192	9	relies	rely	VERB
ijassa-659	192	10	on	on	ADP
ijassa-659	192	11	a	a	DET
ijassa-659	192	12	k	k	NOUN
ijassa-659	192	13	-	-	PUNCT
ijassa-659	192	14	nearest	near	ADJ
ijassa-659	192	15	neighbors	neighbor	NOUN
ijassa-659	192	16	graph	graph	VERB
ijassa-659	192	17	and	and	CCONJ
ijassa-659	192	18	a	a	DET
ijassa-659	192	19	new	new	ADJ
ijassa-659	192	20	constraint	constraint	NOUN
ijassa-659	192	21	utility	utility	NOUN
ijassa-659	192	22	function	function	NOUN
ijassa-659	192	23	to	to	PART
ijassa-659	192	24	generate	generate	VERB
ijassa-659	192	25	queries	query	NOUN
ijassa-659	192	26	to	to	ADP
ijassa-659	192	27	the	the	DET
ijassa-659	192	28	human	human	ADJ
ijassa-659	192	29	expert	expert	NOUN
ijassa-659	192	30	.	.	PUNCT
ijassa-659	193	1	asc	asc	PROPN
ijassa-659	193	2	is	be	AUX
ijassa-659	193	3	based	base	VERB
ijassa-659	193	4	on	on	ADP
ijassa-659	193	5	two	two	NUM
ijassa-659	193	6	parameters	parameter	NOUN
ijassa-659	193	7	(	(	PUNCT
ijassa-659	193	8	i.e.	i.e.	X
ijassa-659	193	9	the	the	DET
ijassa-659	193	10	number	number	NOUN
ijassa-659	193	11	of	of	ADP
ijassa-659	193	12	nearest	near	ADJ
ijassa-659	193	13	neighbors	neighbor	NOUN
ijassa-659	193	14	k	k	PROPN
ijassa-659	193	15	and	and	CCONJ
ijassa-659	193	16	the	the	DET
ijassa-659	193	17	threshold	threshold	NOUN
ijassa-659	193	18	θ	θ	PROPN
ijassa-659	193	19	)	)	PUNCT
ijassa-659	193	20	.	.	PUNCT
ijassa-659	194	1	these	these	DET
ijassa-659	194	2	parameters	parameter	NOUN
ijassa-659	194	3	k	k	PROPN
ijassa-659	194	4	and	and	CCONJ
ijassa-659	194	5	θ	θ	PROPN
ijassa-659	194	6	are	be	AUX
ijassa-659	194	7	set	set	VERB
ijassa-659	194	8	to	to	ADP
ijassa-659	194	9	6	6	NUM
ijassa-659	194	10	and	and	CCONJ
ijassa-659	194	11	⌊(𝑘/2	⌊(𝑘/2	NUM
ijassa-659	194	12	)	)	PUNCT
ijassa-659	195	1	+	+	NOUN
ijassa-659	195	2	1⌋	1⌋	NUM
ijassa-659	195	3	respectively	respectively	ADV
ijassa-659	195	4	as	as	SCONJ
ijassa-659	195	5	recommended	recommend	VERB
ijassa-659	195	6	in	in	ADP
ijassa-659	195	7	their	their	PRON
ijassa-659	195	8	method	method	NOUN
ijassa-659	195	9	[	[	X
ijassa-659	195	10	15	15	NUM
ijassa-659	195	11	]	]	PUNCT
ijassa-659	195	12	.	.	PUNCT
ijassa-659	196	1			PROPN
ijassa-659	197	1	npu	npu	ADJ
ijassa-659	197	2	:	:	PUNCT
ijassa-659	197	3	an	an	DET
ijassa-659	197	4	active	active	ADJ
ijassa-659	197	5	learning	learning	NOUN
ijassa-659	197	6	method	method	NOUN
ijassa-659	197	7	based	base	VERB
ijassa-659	197	8	on	on	ADP
ijassa-659	197	9	the	the	DET
ijassa-659	197	10	classic	classic	ADJ
ijassa-659	197	11	uncertainty	uncertainty	NOUN
ijassa-659	197	12	-	-	PUNCT
ijassa-659	197	13	based	base	VERB
ijassa-659	197	14	principle	principle	NOUN
ijassa-659	197	15	that	that	PRON
ijassa-659	197	16	takes	take	VERB
ijassa-659	197	17	a	a	DET
ijassa-659	197	18	neighborhood	neighborhood	NOUN
ijassa-659	197	19	based	base	VERB
ijassa-659	197	20	approach	approach	NOUN
ijassa-659	197	21	,	,	PUNCT
ijassa-659	197	22	and	and	CCONJ
ijassa-659	197	23	incrementally	incrementally	ADV
ijassa-659	197	24	expands	expand	VERB
ijassa-659	197	25	the	the	DET
ijassa-659	197	26	neighborhoods	neighborhood	NOUN
ijassa-659	197	27	by	by	ADP
ijassa-659	197	28	selecting	select	VERB
ijassa-659	197	29	a	a	DET
ijassa-659	197	30	single	single	ADJ
ijassa-659	197	31	instance	instance	NOUN
ijassa-659	197	32	to	to	PART
ijassa-659	197	33	query	query	VERB
ijassa-659	197	34	each	each	DET
ijassa-659	197	35	time	time	NOUN
ijassa-659	197	36	[	[	X
ijassa-659	197	37	17	17	NUM
ijassa-659	197	38	]	]	PUNCT
ijassa-659	197	39	.	.	PUNCT
ijassa-659	198	1	4.3	4.3	NUM
ijassa-659	198	2	constraint	constraint	NOUN
ijassa-659	198	3	-	-	PUNCT
ijassa-659	198	4	based	base	VERB
ijassa-659	198	5	clustering	clustering	ADJ
ijassa-659	198	6	algorithms	algorithm	NOUN
ijassa-659	198	7	in	in	ADP
ijassa-659	198	8	all	all	DET
ijassa-659	198	9	experiments	experiment	NOUN
ijassa-659	198	10	,	,	PUNCT
ijassa-659	198	11	we	we	PRON
ijassa-659	198	12	report	report	VERB
ijassa-659	198	13	the	the	DET
ijassa-659	198	14	obtained	obtain	VERB
ijassa-659	198	15	results	result	NOUN
ijassa-659	198	16	in	in	ADP
ijassa-659	198	17	conjunction	conjunction	NOUN
ijassa-659	198	18	with	with	ADP
ijassa-659	198	19	two	two	NUM
ijassa-659	198	20	different	different	ADJ
ijassa-659	198	21	constraintbased	constraintbased	ADJ
ijassa-659	198	22	clustering	cluster	VERB
ijassa-659	198	23	algorithms	algorithm	NOUN
ijassa-659	198	24	:	:	PUNCT
ijassa-659	198	25	the	the	DET
ijassa-659	198	26	constrained	constrained	ADJ
ijassa-659	198	27	k	k	NOUN
ijassa-659	198	28	-	-	PUNCT
ijassa-659	198	29	means	means	NOUN
ijassa-659	198	30	(	(	PUNCT
ijassa-659	198	31	mpckmeans	mpckmean	NOUN
ijassa-659	198	32	)	)	PUNCT
ijassa-659	199	1	[	[	X
ijassa-659	199	2	19	19	NUM
ijassa-659	199	3	]	]	PUNCT
ijassa-659	199	4	and	and	CCONJ
ijassa-659	199	5	the	the	DET
ijassa-659	199	6	agglomerative	agglomerative	ADJ
ijassa-659	199	7	hierarchical	hierarchical	ADJ
ijassa-659	199	8	clustering	clustering	NOUN
ijassa-659	199	9	with	with	ADP
ijassa-659	199	10	constraints	constraint	NOUN
ijassa-659	199	11	(	(	PUNCT
ijassa-659	199	12	ahcc	ahcc	NOUN
ijassa-659	199	13	)	)	PUNCT
ijassa-659	200	1	[	[	X
ijassa-659	200	2	20	20	NUM
ijassa-659	200	3	]	]	PUNCT
ijassa-659	200	4	.	.	PUNCT
ijassa-659	201	1	the	the	DET
ijassa-659	201	2	choices	choice	NOUN
ijassa-659	201	3	of	of	ADP
ijassa-659	201	4	these	these	DET
ijassa-659	201	5	algorithms	algorithm	NOUN
ijassa-659	201	6	are	be	AUX
ijassa-659	201	7	not	not	PART
ijassa-659	201	8	critical	critical	ADJ
ijassa-659	201	9	and	and	CCONJ
ijassa-659	201	10	the	the	DET
ijassa-659	201	11	proposed	propose	VERB
ijassa-659	201	12	method	method	NOUN
ijassa-659	201	13	can	can	AUX
ijassa-659	201	14	be	be	AUX
ijassa-659	201	15	used	use	VERB
ijassa-659	201	16	with	with	ADP
ijassa-659	201	17	any	any	DET
ijassa-659	201	18	constraint	constraint	NOUN
ijassa-659	201	19	-	-	PUNCT
ijassa-659	201	20	based	base	VERB
ijassa-659	201	21	clustering	clustering	ADJ
ijassa-659	201	22	algorithm	algorithm	NOUN
ijassa-659	201	23	.	.	PUNCT
ijassa-659	202	1	they	they	PRON
ijassa-659	202	2	have	have	AUX
ijassa-659	202	3	been	be	AUX
ijassa-659	202	4	chosen	choose	VERB
ijassa-659	202	5	because	because	SCONJ
ijassa-659	202	6	they	they	PRON
ijassa-659	202	7	are	be	AUX
ijassa-659	202	8	representative	representative	ADJ
ijassa-659	202	9	of	of	ADP
ijassa-659	202	10	the	the	DET
ijassa-659	202	11	most	most	ADV
ijassa-659	202	12	popular	popular	ADJ
ijassa-659	202	13	clustering	clustering	ADJ
ijassa-659	202	14	algorithms	algorithm	NOUN
ijassa-659	202	15	.	.	PUNCT
ijassa-659	203	1	when	when	SCONJ
ijassa-659	203	2	evaluating	evaluate	VERB
ijassa-659	203	3	the	the	DET
ijassa-659	203	4	performance	performance	NOUN
ijassa-659	203	5	of	of	ADP
ijassa-659	203	6	particular	particular	ADJ
ijassa-659	203	7	methods	method	NOUN
ijassa-659	203	8	on	on	ADP
ijassa-659	203	9	a	a	DET
ijassa-659	203	10	given	give	VERB
ijassa-659	203	11	dataset	dataset	VERB
ijassa-659	203	12	d	d	NOUN
ijassa-659	203	13	,	,	PUNCT
ijassa-659	203	14	we	we	PRON
ijassa-659	203	15	apply	apply	VERB
ijassa-659	203	16	it	it	PRON
ijassa-659	203	17	to	to	PART
ijassa-659	203	18	select	select	VERB
ijassa-659	203	19	up	up	ADP
ijassa-659	203	20	to	to	PART
ijassa-659	203	21	150	150	NUM
ijassa-659	203	22	pairwise	pairwise	NOUN
ijassa-659	203	23	queries	query	NOUN
ijassa-659	203	24	,	,	PUNCT
ijassa-659	203	25	starting	start	VERB
ijassa-659	203	26	from	from	ADP
ijassa-659	203	27	no	no	DET
ijassa-659	203	28	query	query	NOUN
ijassa-659	203	29	at	at	ADV
ijassa-659	203	30	all	all	ADV
ijassa-659	203	31	.	.	PUNCT
ijassa-659	204	1	the	the	DET
ijassa-659	204	2	queries	query	NOUN
ijassa-659	204	3	are	be	AUX
ijassa-659	204	4	answered	answer	VERB
ijassa-659	204	5	based	base	VERB
ijassa-659	204	6	on	on	ADP
ijassa-659	204	7	the	the	DET
ijassa-659	204	8	ground	ground	NOUN
ijassa-659	204	9	-	-	PUNCT
ijassa-659	204	10	truth	truth	NOUN
ijassa-659	204	11	class	class	NOUN
ijassa-659	204	12	label	label	NOUN
ijassa-659	204	13	for	for	ADP
ijassa-659	204	14	the	the	DET
ijassa-659	204	15	dataset	dataset	NOUN
ijassa-659	204	16	.	.	PUNCT
ijassa-659	205	1	the	the	DET
ijassa-659	205	2	constraint	constraint	NOUN
ijassa-659	205	3	-	-	PUNCT
ijassa-659	205	4	based	base	VERB
ijassa-659	205	5	clustering	clustering	ADJ
ijassa-659	205	6	algorithms	algorithm	NOUN
ijassa-659	205	7	(	(	PUNCT
ijassa-659	205	8	mpckmeans	mpckmean	NOUN
ijassa-659	205	9	and	and	CCONJ
ijassa-659	205	10	ahcc	ahcc	NOUN
ijassa-659	205	11	)	)	PUNCT
ijassa-659	205	12	are	be	AUX
ijassa-659	205	13	then	then	ADV
ijassa-659	205	14	applied	apply	VERB
ijassa-659	205	15	to	to	ADP
ijassa-659	205	16	the	the	DET
ijassa-659	205	17	data	datum	NOUN
ijassa-659	205	18	with	with	ADP
ijassa-659	205	19	the	the	DET
ijassa-659	205	20	selecting	selecting	NOUN
ijassa-659	205	21	constraints	constraint	NOUN
ijassa-659	205	22	.	.	PUNCT
ijassa-659	206	1	4.4	4.4	NUM
ijassa-659	206	2	evaluation	evaluation	NOUN
ijassa-659	206	3	metricsto	metricsto	NOUN
ijassa-659	206	4	evaluate	evaluate	VERB
ijassa-659	206	5	the	the	DET
ijassa-659	206	6	performance	performance	NOUN
ijassa-659	206	7	of	of	ADP
ijassa-659	206	8	the	the	DET
ijassa-659	206	9	methods	method	NOUN
ijassa-659	206	10	,	,	PUNCT
ijassa-659	206	11	we	we	PRON
ijassa-659	206	12	used	use	VERB
ijassa-659	206	13	normalized	normalize	VERB
ijassa-659	206	14	mutual	mutual	ADJ
ijassa-659	206	15	information	information	NOUN
ijassa-659	206	16	(	(	PUNCT
ijassa-659	206	17	nmi	nmi	PROPN
ijassa-659	206	18	)	)	PUNCT
ijassa-659	206	19	and	and	CCONJ
ijassa-659	206	20	pairwise	pairwise	PROPN
ijassa-659	206	21	f	f	NOUN
ijassa-659	206	22	-	-	PUNCT
ijassa-659	206	23	measure	measure	NOUN
ijassa-659	206	24	as	as	ADP
ijassa-659	206	25	the	the	DET
ijassa-659	206	26	clustering	clustering	ADJ
ijassa-659	206	27	validation	validation	NOUN
ijassa-659	206	28	metrics	metric	NOUN
ijassa-659	206	29	.	.	PUNCT
ijassa-659	207	1	nmi	nmi	PROPN
ijassa-659	207	2	is	be	AUX
ijassa-659	207	3	an	an	DET
ijassa-659	207	4	external	external	ADJ
ijassa-659	207	5	validation	validation	NOUN
ijassa-659	207	6	metric	metric	NOUN
ijassa-659	207	7	,	,	PUNCT
ijassa-659	207	8	which	which	PRON
ijassa-659	207	9	is	be	AUX
ijassa-659	207	10	used	use	VERB
ijassa-659	207	11	to	to	PART
ijassa-659	207	12	estimate	estimate	VERB
ijassa-659	207	13	the	the	DET
ijassa-659	207	14	quality	quality	NOUN
ijassa-659	207	15	of	of	ADP
ijassa-659	207	16	clustering	cluster	VERB
ijassa-659	207	17	with	with	ADP
ijassa-659	207	18	respect	respect	NOUN
ijassa-659	207	19	to	to	ADP
ijassa-659	207	20	the	the	DET
ijassa-659	207	21	given	give	VERB
ijassa-659	207	22	true	true	ADJ
ijassa-659	207	23	labels	label	NOUN
ijassa-659	207	24	of	of	ADP
ijassa-659	207	25	the	the	DET
ijassa-659	207	26	datasets	dataset	NOUN
ijassa-659	207	27	.	.	PUNCT
ijassa-659	208	1	nmi	nmi	NOUN
ijassa-659	208	2	measures	measure	VERB
ijassa-659	208	3	how	how	SCONJ
ijassa-659	208	4	closely	closely	ADV
ijassa-659	208	5	the	the	DET
ijassa-659	208	6	clustering	cluster	VERB
ijassa-659	208	7	algorithm	algorithm	NOUN
ijassa-659	208	8	could	could	AUX
ijassa-659	208	9	reconstruct	reconstruct	VERB
ijassa-659	208	10	the	the	DET
ijassa-659	208	11	underlying	underlie	VERB
ijassa-659	208	12	label	label	NOUN
ijassa-659	208	13	distribution	distribution	NOUN
ijassa-659	208	14	in	in	ADP
ijassa-659	208	15	the	the	DET
ijassa-659	208	16	data	datum	NOUN
ijassa-659	208	17	.	.	PUNCT
ijassa-659	209	1	if	if	SCONJ
ijassa-659	209	2	x	x	PRON
ijassa-659	209	3	is	be	AUX
ijassa-659	209	4	the	the	DET
ijassa-659	209	5	random	random	ADJ
ijassa-659	209	6	variable	variable	NOUN
ijassa-659	209	7	representing	represent	VERB
ijassa-659	209	8	the	the	DET
ijassa-659	209	9	cluster	cluster	NOUN
ijassa-659	209	10	assignments	assignment	NOUN
ijassa-659	209	11	of	of	ADP
ijassa-659	209	12	the	the	DET
ijassa-659	209	13	instances	instance	NOUN
ijassa-659	209	14	and	and	CCONJ
ijassa-659	209	15	y	y	PROPN
ijassa-659	209	16	is	be	AUX
ijassa-659	209	17	the	the	DET
ijassa-659	209	18	random	random	ADJ
ijassa-659	209	19	variable	variable	NOUN
ijassa-659	209	20	representing	represent	VERB
ijassa-659	209	21	the	the	DET
ijassa-659	209	22	class	class	NOUN
ijassa-659	209	23	labels	label	NOUN
ijassa-659	209	24	of	of	ADP
ijassa-659	209	25	the	the	DET
ijassa-659	209	26	instances	instance	NOUN
ijassa-659	209	27	,	,	PUNCT
ijassa-659	209	28	then	then	ADV
ijassa-659	209	29	nmi	nmi	PROPN
ijassa-659	209	30	is	be	AUX
ijassa-659	209	31	defined	define	VERB
ijassa-659	209	32	as	as	SCONJ
ijassa-659	209	33	follows	follow	VERB
ijassa-659	209	34	:	:	PUNCT
ijassa-659	209	35	𝑁𝑀𝐼	𝑁𝑀𝐼	PROPN
ijassa-659	209	36	=	=	SYM
ijassa-659	209	37	𝐼(𝑋;𝑌	𝐼(𝑋;𝑌	NUM
ijassa-659	209	38	)	)	PUNCT
ijassa-659	209	39	(	(	PUNCT
ijassa-659	209	40	𝐻(𝑋)+𝐻(𝑌))/2	𝐻(𝑋)+𝐻(𝑌))/2	PROPN
ijassa-659	209	41	(	(	PUNCT
ijassa-659	209	42	8)	8)	NUM
ijassa-659	209	43	where	where	SCONJ
ijassa-659	209	44	i	i	PRON
ijassa-659	209	45	(	(	PUNCT
ijassa-659	209	46	x	x	X
ijassa-659	209	47	;	;	PUNCT
ijassa-659	209	48	y	y	X
ijassa-659	209	49	)	)	PUNCT
ijassa-659	209	50	=	=	SYM
ijassa-659	210	1	h(y	h(y	X
ijassa-659	210	2	)	)	PUNCT
ijassa-659	210	3	−	−	ADP
ijassa-659	210	4	h(y|x	h(y|x	NUM
ijassa-659	210	5	)	)	PUNCT
ijassa-659	210	6	is	be	AUX
ijassa-659	210	7	the	the	DET
ijassa-659	210	8	mutual	mutual	ADJ
ijassa-659	210	9	information	information	NOUN
ijassa-659	210	10	between	between	ADP
ijassa-659	210	11	the	the	DET
ijassa-659	210	12	random	random	ADJ
ijassa-659	210	13	variables	variable	NOUN
ijassa-659	210	14	x	x	PUNCT
ijassa-659	210	15	and	and	CCONJ
ijassa-659	210	16	y	y	PROPN
ijassa-659	210	17	,	,	PUNCT
ijassa-659	210	18	h(y	h(y	ADV
ijassa-659	210	19	)	)	PUNCT
ijassa-659	210	20	is	be	AUX
ijassa-659	210	21	the	the	DET
ijassa-659	210	22	shannon	shannon	PROPN
ijassa-659	210	23	entropy	entropy	PROPN
ijassa-659	210	24	of	of	ADP
ijassa-659	210	25	y	y	PROPN
ijassa-659	210	26	,	,	PUNCT
ijassa-659	210	27	and	and	CCONJ
ijassa-659	210	28	h(y|x	h(y|x	NUM
ijassa-659	210	29	)	)	PUNCT
ijassa-659	210	30	is	be	AUX
ijassa-659	210	31	the	the	DET
ijassa-659	210	32	conditional	conditional	ADJ
ijassa-659	210	33	entropy	entropy	NOUN
ijassa-659	210	34	of	of	ADP
ijassa-659	210	35	y	y	PROPN
ijassa-659	210	36	given	give	VERB
ijassa-659	210	37	x	x	PUNCT
ijassa-659	211	1	[	[	X
ijassa-659	211	2	21	21	NUM
ijassa-659	211	3	]	]	PUNCT
ijassa-659	211	4	.	.	PUNCT
ijassa-659	212	1	the	the	DET
ijassa-659	212	2	range	range	NOUN
ijassa-659	212	3	of	of	ADP
ijassa-659	212	4	nmi	nmi	PROPN
ijassa-659	212	5	values	value	NOUN
ijassa-659	212	6	is	be	AUX
ijassa-659	212	7	0–1	0–1	NOUN
ijassa-659	212	8	.	.	PUNCT
ijassa-659	213	1	in	in	ADP
ijassa-659	213	2	general	general	ADJ
ijassa-659	213	3	,	,	PUNCT
ijassa-659	213	4	the	the	PRON
ijassa-659	213	5	larger	large	ADJ
ijassa-659	213	6	the	the	DET
ijassa-659	213	7	nmi	nmi	PROPN
ijassa-659	213	8	value	value	NOUN
ijassa-659	213	9	,	,	PUNCT
ijassa-659	213	10	the	the	PRON
ijassa-659	213	11	better	well	ADJ
ijassa-659	213	12	the	the	DET
ijassa-659	213	13	clustering	clustering	ADJ
ijassa-659	213	14	quality	quality	NOUN
ijassa-659	213	15	.	.	PUNCT
ijassa-659	214	1	improving	improve	VERB
ijassa-659	214	2	semi	semi	ADJ
ijassa-659	214	3	-	-	ADJ
ijassa-659	214	4	supervised	supervised	ADJ
ijassa-659	214	5	clustering	clustering	ADJ
ijassa-659	214	6	algorithms	algorithm	NOUN
ijassa-659	214	7	with	with	ADP
ijassa-659	214	8	active	active	ADJ
ijassa-659	214	9	query	query	NOUN
ijassa-659	214	10	selection	selection	NOUN
ijassa-659	214	11	33	33	NUM
ijassa-659	214	12	copyright	copyright	NOUN
ijassa-659	214	13	©	©	PROPN
ijassa-659	214	14	2019	2019	NUM
ijassa-659	214	15	assa	assa	PROPN
ijassa-659	214	16	adv	adv	PROPN
ijassa-659	214	17	.	.	PUNCT
ijassa-659	215	1	in	in	ADP
ijassa-659	215	2	systems	system	NOUN
ijassa-659	215	3	science	science	NOUN
ijassa-659	215	4	and	and	CCONJ
ijassa-659	215	5	appl	appl	NOUN
ijassa-659	215	6	.	.	PUNCT
ijassa-659	216	1	(	(	PUNCT
ijassa-659	216	2	2019	2019	NUM
ijassa-659	216	3	)	)	PUNCT
ijassa-659	216	4	pairwise	pairwise	NOUN
ijassa-659	216	5	f	f	NOUN
ijassa-659	216	6	-	-	PUNCT
ijassa-659	216	7	measure	measure	NOUN
ijassa-659	216	8	is	be	AUX
ijassa-659	216	9	another	another	DET
ijassa-659	216	10	evaluation	evaluation	NOUN
ijassa-659	216	11	metric	metric	ADJ
ijassa-659	216	12	to	to	PART
ijassa-659	216	13	evaluate	evaluate	VERB
ijassa-659	216	14	how	how	SCONJ
ijassa-659	216	15	well	well	ADV
ijassa-659	216	16	we	we	PRON
ijassa-659	216	17	can	can	AUX
ijassa-659	216	18	predict	predict	VERB
ijassa-659	216	19	the	the	DET
ijassa-659	216	20	pairwise	pairwise	NOUN
ijassa-659	216	21	relationship	relationship	NOUN
ijassa-659	216	22	between	between	ADP
ijassa-659	216	23	each	each	DET
ijassa-659	216	24	pair	pair	NOUN
ijassa-659	216	25	of	of	ADP
ijassa-659	216	26	instances	instance	NOUN
ijassa-659	216	27	in	in	ADP
ijassa-659	216	28	comparison	comparison	NOUN
ijassa-659	216	29	to	to	ADP
ijassa-659	216	30	the	the	DET
ijassa-659	216	31	relationship	relationship	NOUN
ijassa-659	216	32	defined	define	VERB
ijassa-659	216	33	by	by	ADP
ijassa-659	216	34	the	the	DET
ijassa-659	216	35	ground	ground	NOUN
ijassa-659	216	36	truth	truth	NOUN
ijassa-659	216	37	class	class	NOUN
ijassa-659	216	38	labels	label	NOUN
ijassa-659	216	39	.	.	PUNCT
ijassa-659	217	1	pairwise	pairwise	NOUN
ijassa-659	217	2	f	f	NOUN
ijassa-659	217	3	-	-	PUNCT
ijassa-659	217	4	measure	measure	NOUN
ijassa-659	217	5	is	be	AUX
ijassa-659	217	6	defined	define	VERB
ijassa-659	217	7	as	as	ADP
ijassa-659	217	8	the	the	DET
ijassa-659	217	9	harmonic	harmonic	ADJ
ijassa-659	217	10	mean	mean	NOUN
ijassa-659	217	11	of	of	ADP
ijassa-659	217	12	pairwise	pairwise	NOUN
ijassa-659	217	13	precision	precision	NOUN
ijassa-659	217	14	and	and	CCONJ
ijassa-659	217	15	recall	recall	NOUN
ijassa-659	217	16	,	,	PUNCT
ijassa-659	217	17	where	where	SCONJ
ijassa-659	217	18	the	the	DET
ijassa-659	217	19	traditional	traditional	ADJ
ijassa-659	217	20	information	information	NOUN
ijassa-659	217	21	retrieval	retrieval	NOUN
ijassa-659	217	22	measures	measure	NOUN
ijassa-659	217	23	are	be	AUX
ijassa-659	217	24	adapted	adapt	VERB
ijassa-659	217	25	for	for	ADP
ijassa-659	217	26	evaluating	evaluate	VERB
ijassa-659	217	27	clustering	cluster	VERB
ijassa-659	217	28	by	by	ADP
ijassa-659	217	29	considering	consider	VERB
ijassa-659	217	30	pairs	pair	NOUN
ijassa-659	217	31	of	of	ADP
ijassa-659	217	32	points	point	NOUN
ijassa-659	217	33	.	.	PUNCT
ijassa-659	218	1	for	for	ADP
ijassa-659	218	2	any	any	DET
ijassa-659	218	3	pair	pair	NOUN
ijassa-659	218	4	of	of	ADP
ijassa-659	218	5	points	point	NOUN
ijassa-659	218	6	,	,	PUNCT
ijassa-659	218	7	the	the	DET
ijassa-659	218	8	decision	decision	NOUN
ijassa-659	218	9	to	to	PART
ijassa-659	218	10	cluster	cluster	VERB
ijassa-659	218	11	this	this	DET
ijassa-659	218	12	pair	pair	NOUN
ijassa-659	218	13	into	into	ADP
ijassa-659	218	14	same	same	ADJ
ijassa-659	218	15	or	or	CCONJ
ijassa-659	218	16	different	different	ADJ
ijassa-659	218	17	clusters	cluster	NOUN
ijassa-659	218	18	is	be	AUX
ijassa-659	218	19	considered	consider	VERB
ijassa-659	218	20	to	to	PART
ijassa-659	218	21	be	be	AUX
ijassa-659	218	22	correct	correct	ADJ
ijassa-659	218	23	if	if	SCONJ
ijassa-659	218	24	it	it	PRON
ijassa-659	218	25	matches	match	VERB
ijassa-659	218	26	with	with	ADP
ijassa-659	218	27	the	the	DET
ijassa-659	218	28	underlying	underlie	VERB
ijassa-659	218	29	class	class	NOUN
ijassa-659	218	30	labeling	labeling	NOUN
ijassa-659	218	31	available	available	ADJ
ijassa-659	218	32	for	for	ADP
ijassa-659	218	33	the	the	DET
ijassa-659	218	34	points	point	NOUN
ijassa-659	218	35	.	.	PUNCT
ijassa-659	219	1	pairwise	pairwise	NOUN
ijassa-659	219	2	f	f	NOUN
ijassa-659	219	3	-	-	PUNCT
ijassa-659	219	4	measure	measure	NOUN
ijassa-659	219	5	is	be	AUX
ijassa-659	219	6	defined	define	VERB
ijassa-659	219	7	as	as	SCONJ
ijassa-659	219	8	follows	follow	VERB
ijassa-659	219	9	:	:	PUNCT
ijassa-659	219	10	𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛	𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛	PROPN
ijassa-659	219	11	=	=	PUNCT
ijassa-659	219	12	𝑛𝑐	𝑛𝑐	PART
ijassa-659	219	13	𝑛𝑠	𝑛𝑠	VERB
ijassa-659	219	14	𝑅𝑒𝑐𝑎𝑙𝑙	𝑅𝑒𝑐𝑎𝑙𝑙	PROPN
ijassa-659	219	15	=	=	PUNCT
ijassa-659	220	1	𝑛𝑐	𝑛𝑐	VERB
ijassa-659	220	2	𝑛𝑓	𝑛𝑓	NOUN
ijassa-659	220	3	𝐹	𝐹	PROPN
ijassa-659	220	4	−	−	NOUN
ijassa-659	220	5	𝑚𝑒𝑎𝑠𝑢𝑟𝑒	𝑚𝑒𝑎𝑠𝑢𝑟𝑒	ADJ
ijassa-659	220	6	=	=	SYM
ijassa-659	220	7	2×𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛×𝑅𝑒𝑐𝑎𝑙𝑙	2×𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛×𝑅𝑒𝑐𝑎𝑙𝑙	NUM
ijassa-659	220	8	𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛+𝑅𝑒𝑐𝑎𝑙𝑙	𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛+𝑅𝑒𝑐𝑎𝑙𝑙	X
ijassa-659	220	9	(	(	PUNCT
ijassa-659	220	10	9	9	NUM
ijassa-659	220	11	)	)	PUNCT
ijassa-659	220	12	where	where	SCONJ
ijassa-659	220	13	nc	nc	PROPN
ijassa-659	220	14	is	be	AUX
ijassa-659	220	15	the	the	DET
ijassa-659	220	16	number	number	NOUN
ijassa-659	220	17	of	of	ADP
ijassa-659	220	18	point	point	NOUN
ijassa-659	220	19	pairs	pair	NOUN
ijassa-659	220	20	that	that	PRON
ijassa-659	220	21	are	be	AUX
ijassa-659	220	22	correctly	correctly	ADV
ijassa-659	220	23	predicted	predict	VERB
ijassa-659	220	24	as	as	ADP
ijassa-659	220	25	in	in	ADP
ijassa-659	220	26	the	the	DET
ijassa-659	220	27	same	same	ADJ
ijassa-659	220	28	cluster	cluster	NOUN
ijassa-659	220	29	;	;	PUNCT
ijassa-659	220	30	ns	ns	NUM
ijassa-659	220	31	is	be	AUX
ijassa-659	220	32	the	the	DET
ijassa-659	220	33	number	number	NOUN
ijassa-659	220	34	of	of	ADP
ijassa-659	220	35	point	point	NOUN
ijassa-659	220	36	pairs	pair	NOUN
ijassa-659	220	37	that	that	PRON
ijassa-659	220	38	are	be	AUX
ijassa-659	220	39	predicted	predict	VERB
ijassa-659	220	40	as	as	ADP
ijassa-659	220	41	in	in	ADP
ijassa-659	220	42	the	the	DET
ijassa-659	220	43	same	same	ADJ
ijassa-659	220	44	cluster	cluster	NOUN
ijassa-659	220	45	;	;	PUNCT
ijassa-659	220	46	nf	nf	NUM
ijassa-659	220	47	is	be	AUX
ijassa-659	220	48	the	the	DET
ijassa-659	220	49	number	number	NOUN
ijassa-659	220	50	of	of	ADP
ijassa-659	220	51	point	point	NOUN
ijassa-659	220	52	pairs	pair	NOUN
ijassa-659	220	53	that	that	PRON
ijassa-659	220	54	are	be	AUX
ijassa-659	220	55	actually	actually	ADV
ijassa-659	220	56	in	in	ADP
ijassa-659	220	57	the	the	DET
ijassa-659	220	58	same	same	ADJ
ijassa-659	220	59	cluster	cluster	NOUN
ijassa-659	220	60	.	.	PUNCT
ijassa-659	221	1	4.5	4.5	NUM
ijassa-659	221	2	evaluation	evaluation	NOUN
ijassa-659	221	3	results	result	VERB
ijassa-659	221	4	4.5.1	4.5.1	NUM
ijassa-659	221	5	effectiveness	effectiveness	NOUN
ijassa-659	221	6	and	and	CCONJ
ijassa-659	221	7	performance	performance	NOUN
ijassa-659	221	8	analysis	analysis	NOUN
ijassa-659	221	9	in	in	ADP
ijassa-659	221	10	this	this	DET
ijassa-659	221	11	section	section	NOUN
ijassa-659	221	12	we	we	PRON
ijassa-659	221	13	present	present	VERB
ijassa-659	221	14	the	the	DET
ijassa-659	221	15	evaluation	evaluation	NOUN
ijassa-659	221	16	results	result	NOUN
ijassa-659	221	17	of	of	ADP
ijassa-659	221	18	the	the	DET
ijassa-659	221	19	proposed	propose	VERB
ijassa-659	221	20	method	method	NOUN
ijassa-659	221	21	compared	compare	VERB
ijassa-659	221	22	to	to	ADP
ijassa-659	221	23	min	min	PROPN
ijassa-659	221	24	–	–	PUNCT
ijassa-659	221	25	max	max	PROPN
ijassa-659	222	1	[	[	X
ijassa-659	222	2	12	12	NUM
ijassa-659	222	3	]	]	PUNCT
ijassa-659	222	4	,	,	PUNCT
ijassa-659	222	5	asc	asc	PROPN
ijassa-659	223	1	[	[	X
ijassa-659	223	2	15	15	NUM
ijassa-659	223	3	]	]	PUNCT
ijassa-659	223	4	,	,	PUNCT
ijassa-659	223	5	npu	npu	ADJ
ijassa-659	223	6	[	[	X
ijassa-659	223	7	17	17	NUM
ijassa-659	223	8	]	]	PUNCT
ijassa-659	223	9	heuristics	heuristic	NOUN
ijassa-659	223	10	and	and	CCONJ
ijassa-659	223	11	the	the	DET
ijassa-659	223	12	random	random	ADJ
ijassa-659	223	13	selection	selection	NOUN
ijassa-659	223	14	of	of	ADP
ijassa-659	223	15	the	the	DET
ijassa-659	223	16	constraints	constraint	NOUN
ijassa-659	223	17	.	.	PUNCT
ijassa-659	224	1	to	to	PART
ijassa-659	224	2	evaluate	evaluate	VERB
ijassa-659	224	3	the	the	DET
ijassa-659	224	4	performance	performance	NOUN
ijassa-659	224	5	of	of	ADP
ijassa-659	224	6	the	the	DET
ijassa-659	224	7	proposed	propose	VERB
ijassa-659	224	8	method	method	NOUN
ijassa-659	224	9	to	to	PART
ijassa-659	224	10	adapt	adapt	VERB
ijassa-659	224	11	to	to	ADP
ijassa-659	224	12	distinct	distinct	ADJ
ijassa-659	224	13	clustering	clustering	ADJ
ijassa-659	224	14	algorithms	algorithm	NOUN
ijassa-659	224	15	,	,	PUNCT
ijassa-659	224	16	the	the	DET
ijassa-659	224	17	heuristics	heuristic	NOUN
ijassa-659	224	18	are	be	AUX
ijassa-659	224	19	compared	compare	VERB
ijassa-659	224	20	in	in	ADP
ijassa-659	224	21	conjunction	conjunction	NOUN
ijassa-659	224	22	with	with	ADP
ijassa-659	224	23	the	the	DET
ijassa-659	224	24	clustering	cluster	VERB
ijassa-659	224	25	algorithms	algorithms	NOUN
ijassa-659	224	26	(	(	PUNCT
ijassa-659	224	27	mpckmeans	mpckmean	NOUN
ijassa-659	224	28	and	and	CCONJ
ijassa-659	224	29	ahcc	ahcc	NOUN
ijassa-659	224	30	)	)	PUNCT
ijassa-659	224	31	.	.	PUNCT
ijassa-659	225	1	figures	figure	NOUN
ijassa-659	225	2	2	2	NUM
ijassa-659	225	3	and	and	CCONJ
ijassa-659	225	4	3	3	NUM
ijassa-659	225	5	show	show	VERB
ijassa-659	225	6	the	the	DET
ijassa-659	225	7	clustering	cluster	VERB
ijassa-659	225	8	performance	performance	NOUN
ijassa-659	225	9	on	on	ADP
ijassa-659	225	10	mpckmeans	mpckmean	NOUN
ijassa-659	225	11	and	and	CCONJ
ijassa-659	225	12	ahcc	ahcc	VERB
ijassa-659	225	13	respectively	respectively	ADV
ijassa-659	225	14	.	.	PUNCT
ijassa-659	226	1	to	to	PART
ijassa-659	226	2	evaluate	evaluate	VERB
ijassa-659	226	3	the	the	DET
ijassa-659	226	4	performance	performance	NOUN
ijassa-659	226	5	in	in	ADP
ijassa-659	226	6	both	both	DET
ijassa-659	226	7	mpckmeans	mpckmean	NOUN
ijassa-659	226	8	and	and	CCONJ
ijassa-659	226	9	ahcc	ahcc	NOUN
ijassa-659	226	10	,	,	PUNCT
ijassa-659	226	11	we	we	PRON
ijassa-659	226	12	repeat	repeat	VERB
ijassa-659	226	13	this	this	DET
ijassa-659	226	14	process	process	NOUN
ijassa-659	226	15	for	for	ADP
ijassa-659	226	16	50	50	NUM
ijassa-659	226	17	independent	independent	ADJ
ijassa-659	226	18	runs	run	NOUN
ijassa-659	226	19	and	and	CCONJ
ijassa-659	226	20	report	report	VERB
ijassa-659	226	21	the	the	DET
ijassa-659	226	22	average	average	ADJ
ijassa-659	226	23	performance	performance	NOUN
ijassa-659	226	24	using	use	VERB
ijassa-659	226	25	evaluation	evaluation	NOUN
ijassa-659	226	26	criteria	criterion	NOUN
ijassa-659	226	27	described	describe	VERB
ijassa-659	226	28	above	above	ADV
ijassa-659	226	29	.	.	PUNCT
ijassa-659	227	1	it	it	PRON
ijassa-659	227	2	can	can	AUX
ijassa-659	227	3	be	be	AUX
ijassa-659	227	4	observed	observe	VERB
ijassa-659	227	5	from	from	ADP
ijassa-659	227	6	figures	figure	NOUN
ijassa-659	227	7	2	2	NUM
ijassa-659	227	8	and	and	CCONJ
ijassa-659	227	9	3	3	NUM
ijassa-659	227	10	,	,	PUNCT
ijassa-659	227	11	that	that	SCONJ
ijassa-659	227	12	the	the	DET
ijassa-659	227	13	proposed	propose	VERB
ijassa-659	227	14	method	method	NOUN
ijassa-659	227	15	generally	generally	ADV
ijassa-659	227	16	outperforms	outperform	VERB
ijassa-659	227	17	other	other	ADJ
ijassa-659	227	18	constraint	constraint	NOUN
ijassa-659	227	19	selection	selection	NOUN
ijassa-659	227	20	heuristics	heuristic	NOUN
ijassa-659	227	21	in	in	ADP
ijassa-659	227	22	conjunction	conjunction	NOUN
ijassa-659	227	23	with	with	ADP
ijassa-659	227	24	the	the	DET
ijassa-659	227	25	clustering	cluster	VERB
ijassa-659	227	26	algorithms	algorithms	NOUN
ijassa-659	227	27	mpckmeans	mpckmean	NOUN
ijassa-659	227	28	and	and	CCONJ
ijassa-659	227	29	ahcc	ahcc	NOUN
ijassa-659	227	30	.	.	PUNCT
ijassa-659	228	1	this	this	PRON
ijassa-659	228	2	implies	imply	VERB
ijassa-659	228	3	that	that	SCONJ
ijassa-659	228	4	the	the	DET
ijassa-659	228	5	usefulness	usefulness	NOUN
ijassa-659	228	6	of	of	ADP
ijassa-659	228	7	constraints	constraint	NOUN
ijassa-659	228	8	depends	depend	VERB
ijassa-659	228	9	on	on	ADP
ijassa-659	228	10	how	how	SCONJ
ijassa-659	228	11	they	they	PRON
ijassa-659	228	12	are	be	AUX
ijassa-659	228	13	utilized	utilize	VERB
ijassa-659	228	14	by	by	ADP
ijassa-659	228	15	a	a	DET
ijassa-659	228	16	clustering	clustering	ADJ
ijassa-659	228	17	algorithm	algorithm	NOUN
ijassa-659	228	18	.	.	PUNCT
ijassa-659	229	1	furthermore	furthermore	ADV
ijassa-659	229	2	,	,	PUNCT
ijassa-659	229	3	the	the	DET
ijassa-659	229	4	proposed	propose	VERB
ijassa-659	229	5	method	method	NOUN
ijassa-659	229	6	keeps	keep	VERB
ijassa-659	229	7	a	a	DET
ijassa-659	229	8	smooth	smooth	ADJ
ijassa-659	229	9	increase	increase	NOUN
ijassa-659	229	10	in	in	ADP
ijassa-659	229	11	the	the	DET
ijassa-659	229	12	clustering	clustering	ADJ
ijassa-659	229	13	performance	performance	NOUN
ijassa-659	229	14	while	while	SCONJ
ijassa-659	229	15	the	the	DET
ijassa-659	229	16	other	other	ADJ
ijassa-659	229	17	constraint	constraint	NOUN
ijassa-659	229	18	selection	selection	NOUN
ijassa-659	229	19	heuristics	heuristic	NOUN
ijassa-659	229	20	drop	drop	VERB
ijassa-659	229	21	in	in	ADP
ijassa-659	229	22	performance	performance	NOUN
ijassa-659	229	23	when	when	SCONJ
ijassa-659	229	24	the	the	DET
ijassa-659	229	25	number	number	NOUN
ijassa-659	229	26	of	of	ADP
ijassa-659	229	27	queries	query	NOUN
ijassa-659	229	28	increases	increase	NOUN
ijassa-659	229	29	.	.	PUNCT
ijassa-659	230	1	it	it	PRON
ijassa-659	230	2	is	be	AUX
ijassa-659	230	3	interesting	interesting	ADJ
ijassa-659	230	4	to	to	PART
ijassa-659	230	5	note	note	VERB
ijassa-659	230	6	that	that	SCONJ
ijassa-659	230	7	the	the	DET
ijassa-659	230	8	random	random	ADJ
ijassa-659	230	9	selection	selection	NOUN
ijassa-659	230	10	of	of	ADP
ijassa-659	230	11	constraints	constraint	NOUN
ijassa-659	230	12	degrade	degrade	VERB
ijassa-659	230	13	the	the	DET
ijassa-659	230	14	clustering	clustering	ADJ
ijassa-659	230	15	performance	performance	NOUN
ijassa-659	230	16	in	in	ADP
ijassa-659	230	17	some	some	DET
ijassa-659	230	18	datasets	dataset	NOUN
ijassa-659	230	19	when	when	SCONJ
ijassa-659	230	20	the	the	DET
ijassa-659	230	21	number	number	NOUN
ijassa-659	230	22	of	of	ADP
ijassa-659	230	23	queries	query	NOUN
ijassa-659	230	24	increases	increase	NOUN
ijassa-659	230	25	(	(	PUNCT
ijassa-659	230	26	e.g.	e.g.	ADV
ijassa-659	230	27	heart	heart	NOUN
ijassa-659	230	28	,	,	PUNCT
ijassa-659	230	29	breast	breast	NOUN
ijassa-659	230	30	,	,	PUNCT
ijassa-659	230	31	segment	segment	NOUN
ijassa-659	230	32	,	,	PUNCT
ijassa-659	230	33	and	and	CCONJ
ijassa-659	230	34	digit-389	digit-389	ADJ
ijassa-659	230	35	)	)	PUNCT
ijassa-659	230	36	,	,	PUNCT
ijassa-659	230	37	while	while	SCONJ
ijassa-659	230	38	it	it	PRON
ijassa-659	230	39	is	be	AUX
ijassa-659	230	40	expected	expect	VERB
ijassa-659	230	41	that	that	SCONJ
ijassa-659	230	42	the	the	DET
ijassa-659	230	43	performance	performance	NOUN
ijassa-659	230	44	increases	increase	VERB
ijassa-659	230	45	monotonically	monotonically	ADV
ijassa-659	230	46	with	with	ADP
ijassa-659	230	47	the	the	DET
ijassa-659	230	48	number	number	NOUN
ijassa-659	230	49	of	of	ADP
ijassa-659	230	50	queries	query	NOUN
ijassa-659	230	51	.	.	PUNCT
ijassa-659	231	1	this	this	DET
ijassa-659	231	2	problem	problem	NOUN
ijassa-659	231	3	is	be	AUX
ijassa-659	231	4	a	a	DET
ijassa-659	231	5	well	well	ADV
ijassa-659	231	6	-	-	PUNCT
ijassa-659	231	7	known	know	VERB
ijassa-659	231	8	issue	issue	NOUN
ijassa-659	231	9	in	in	ADP
ijassa-659	231	10	constraint	constraint	NOUN
ijassa-659	231	11	-	-	PUNCT
ijassa-659	231	12	based	base	VERB
ijassa-659	231	13	clustering	clustering	NOUN
ijassa-659	231	14	that	that	PRON
ijassa-659	231	15	has	have	AUX
ijassa-659	231	16	been	be	AUX
ijassa-659	231	17	addressed	address	VERB
ijassa-659	231	18	in	in	ADP
ijassa-659	231	19	[	[	X
ijassa-659	231	20	22	22	NUM
ijassa-659	231	21	,	,	PUNCT
ijassa-659	231	22	23	23	NUM
ijassa-659	231	23	]	]	PUNCT
ijassa-659	231	24	and	and	CCONJ
ijassa-659	231	25	may	may	AUX
ijassa-659	231	26	be	be	AUX
ijassa-659	231	27	due	due	ADJ
ijassa-659	231	28	to	to	ADP
ijassa-659	231	29	either	either	CCONJ
ijassa-659	231	30	the	the	DET
ijassa-659	231	31	variability	variability	NOUN
ijassa-659	231	32	of	of	ADP
ijassa-659	231	33	the	the	DET
ijassa-659	231	34	random	random	ADJ
ijassa-659	231	35	approaches	approach	NOUN
ijassa-659	231	36	in	in	ADP
ijassa-659	231	37	some	some	DET
ijassa-659	231	38	cases	case	NOUN
ijassa-659	231	39	or	or	CCONJ
ijassa-659	231	40	due	due	ADP
ijassa-659	231	41	to	to	ADP
ijassa-659	231	42	a	a	DET
ijassa-659	231	43	bad	bad	ADJ
ijassa-659	231	44	selection	selection	NOUN
ijassa-659	231	45	of	of	ADP
ijassa-659	231	46	some	some	DET
ijassa-659	231	47	constraints	constraint	NOUN
ijassa-659	231	48	that	that	PRON
ijassa-659	231	49	leads	lead	VERB
ijassa-659	231	50	constraint	constraint	NOUN
ijassa-659	231	51	-	-	PUNCT
ijassa-659	231	52	based	base	VERB
ijassa-659	231	53	clustering	clustering	ADJ
ijassa-659	231	54	algorithm	algorithm	NOUN
ijassa-659	231	55	to	to	ADP
ijassa-659	231	56	poorer	poor	ADJ
ijassa-659	231	57	clustering	clustering	ADJ
ijassa-659	231	58	results	result	NOUN
ijassa-659	231	59	.	.	PUNCT
ijassa-659	232	1	this	this	DET
ijassa-659	232	2	further	far	ADV
ijassa-659	232	3	demonstrates	demonstrate	VERB
ijassa-659	232	4	the	the	DET
ijassa-659	232	5	importance	importance	NOUN
ijassa-659	232	6	of	of	ADP
ijassa-659	232	7	selecting	select	VERB
ijassa-659	232	8	the	the	DET
ijassa-659	232	9	right	right	ADJ
ijassa-659	232	10	set	set	NOUN
ijassa-659	232	11	of	of	ADP
ijassa-659	232	12	constraints	constraint	NOUN
ijassa-659	232	13	.	.	PUNCT
ijassa-659	233	1	in	in	ADP
ijassa-659	233	2	comparison	comparison	NOUN
ijassa-659	233	3	,	,	PUNCT
ijassa-659	233	4	min	min	ADJ
ijassa-659	233	5	-	-	ADJ
ijassa-659	233	6	max	max	PROPN
ijassa-659	233	7	method	method	NOUN
ijassa-659	233	8	obtains	obtain	VERB
ijassa-659	233	9	better	well	ADJ
ijassa-659	233	10	results	result	NOUN
ijassa-659	233	11	with	with	ADP
ijassa-659	233	12	mpckmeans	mpckmean	NOUN
ijassa-659	233	13	only	only	ADV
ijassa-659	233	14	for	for	SCONJ
ijassa-659	233	15	the	the	DET
ijassa-659	233	16	heart	heart	NOUN
ijassa-659	233	17	dataset	dataset	VERB
ijassa-659	233	18	.	.	PUNCT
ijassa-659	234	1	this	this	PRON
ijassa-659	234	2	can	can	AUX
ijassa-659	234	3	be	be	AUX
ijassa-659	234	4	explained	explain	VERB
ijassa-659	234	5	by	by	ADP
ijassa-659	234	6	the	the	DET
ijassa-659	234	7	fact	fact	NOUN
ijassa-659	234	8	that	that	SCONJ
ijassa-659	234	9	for	for	ADP
ijassa-659	234	10	simpler	simple	ADJ
ijassa-659	234	11	datasets	dataset	NOUN
ijassa-659	234	12	with	with	ADP
ijassa-659	234	13	a	a	DET
ijassa-659	234	14	small	small	ADJ
ijassa-659	234	15	number	number	NOUN
ijassa-659	234	16	of	of	ADP
ijassa-659	234	17	clusters	cluster	NOUN
ijassa-659	234	18	like	like	ADP
ijassa-659	234	19	heart	heart	NOUN
ijassa-659	234	20	dataset	dataset	NOUN
ijassa-659	234	21	,	,	PUNCT
ijassa-659	234	22	the	the	DET
ijassa-659	234	23	min	min	PROPN
ijassa-659	234	24	-	-	ADJ
ijassa-659	234	25	max	max	PROPN
ijassa-659	234	26	method	method	NOUN
ijassa-659	234	27	generally	generally	ADV
ijassa-659	234	28	better	well	ADV
ijassa-659	234	29	manages	manage	VERB
ijassa-659	234	30	to	to	PART
ijassa-659	234	31	define	define	VERB
ijassa-659	234	32	the	the	DET
ijassa-659	234	33	skeleton	skeleton	NOUN
ijassa-659	234	34	of	of	ADP
ijassa-659	234	35	cl	cl	NOUN
ijassa-659	234	36	constraints	constraint	NOUN
ijassa-659	234	37	during	during	ADP
ijassa-659	234	38	its	its	PRON
ijassa-659	234	39	exploration	exploration	NOUN
ijassa-659	234	40	step	step	NOUN
ijassa-659	234	41	,	,	PUNCT
ijassa-659	234	42	which	which	PRON
ijassa-659	234	43	in	in	ADP
ijassa-659	234	44	turn	turn	NOUN
ijassa-659	234	45	promotes	promote	VERB
ijassa-659	234	46	the	the	DET
ijassa-659	234	47	initialization	initialization	NOUN
ijassa-659	234	48	of	of	ADP
ijassa-659	234	49	the	the	DET
ijassa-659	234	50	centers	center	NOUN
ijassa-659	234	51	in	in	ADP
ijassa-659	234	52	34	34	NUM
ijassa-659	234	53	w.	w.	PROPN
ijassa-659	234	54	atwa	atwa	PROPN
ijassa-659	234	55	,	,	PUNCT
ijassa-659	234	56	m.	m.	NOUN
ijassa-659	234	57	emam	emam	PROPN
ijassa-659	234	58	copyright	copyright	NOUN
ijassa-659	234	59	©	©	PROPN
ijassa-659	234	60	2019	2019	NUM
ijassa-659	234	61	assa	assa	PROPN
ijassa-659	234	62	adv	adv	PROPN
ijassa-659	234	63	.	.	PUNCT
ijassa-659	235	1	in	in	ADP
ijassa-659	235	2	systems	system	NOUN
ijassa-659	235	3	science	science	NOUN
ijassa-659	235	4	and	and	CCONJ
ijassa-659	235	5	appl	appl	NOUN
ijassa-659	235	6	.	.	PUNCT
ijassa-659	236	1	(	(	PUNCT
ijassa-659	236	2	2019	2019	NUM
ijassa-659	236	3	)	)	PUNCT
ijassa-659	236	4	mpckmeans	mpckmean	NOUN
ijassa-659	236	5	.	.	PUNCT
ijassa-659	237	1	in	in	ADP
ijassa-659	237	2	the	the	DET
ijassa-659	237	3	case	case	NOUN
ijassa-659	237	4	of	of	ADP
ijassa-659	237	5	more	more	ADJ
ijassa-659	237	6	complex	complex	ADJ
ijassa-659	237	7	datasets	dataset	NOUN
ijassa-659	237	8	like	like	ADP
ijassa-659	237	9	segment	segment	NOUN
ijassa-659	237	10	and	and	CCONJ
ijassa-659	237	11	magic	magic	ADJ
ijassa-659	237	12	datasets	dataset	NOUN
ijassa-659	237	13	,	,	PUNCT
ijassa-659	237	14	the	the	DET
ijassa-659	237	15	minmax	minmax	PROPN
ijassa-659	237	16	method	method	NOUN
ijassa-659	237	17	needs	need	VERB
ijassa-659	237	18	more	more	ADJ
ijassa-659	237	19	constraints	constraint	NOUN
ijassa-659	237	20	to	to	PART
ijassa-659	237	21	improve	improve	VERB
ijassa-659	237	22	the	the	DET
ijassa-659	237	23	clustering	clustering	ADJ
ijassa-659	237	24	performance	performance	NOUN
ijassa-659	237	25	.	.	PUNCT
ijassa-659	238	1	more	more	ADV
ijassa-659	238	2	generally	generally	ADV
ijassa-659	238	3	,	,	PUNCT
ijassa-659	238	4	it	it	PRON
ijassa-659	238	5	can	can	AUX
ijassa-659	238	6	be	be	AUX
ijassa-659	238	7	seen	see	VERB
ijassa-659	238	8	from	from	ADP
ijassa-659	238	9	figures	figure	NOUN
ijassa-659	238	10	2	2	NUM
ijassa-659	238	11	and	and	CCONJ
ijassa-659	238	12	3	3	NUM
ijassa-659	238	13	,	,	PUNCT
ijassa-659	238	14	that	that	SCONJ
ijassa-659	238	15	the	the	DET
ijassa-659	238	16	asc	asc	PROPN
ijassa-659	238	17	method	method	NOUN
ijassa-659	238	18	obtains	obtain	VERB
ijassa-659	238	19	better	well	ADJ
ijassa-659	238	20	results	result	NOUN
ijassa-659	238	21	than	than	ADP
ijassa-659	238	22	the	the	DET
ijassa-659	238	23	proposed	propose	VERB
ijassa-659	238	24	active	active	ADJ
ijassa-659	238	25	learning	learning	NOUN
ijassa-659	238	26	method	method	NOUN
ijassa-659	238	27	with	with	ADP
ijassa-659	238	28	small	small	ADJ
ijassa-659	238	29	number	number	NOUN
ijassa-659	238	30	of	of	ADP
ijassa-659	238	31	queries	query	NOUN
ijassa-659	238	32	(	(	PUNCT
ijassa-659	238	33	e.g.	e.g.	ADV
ijassa-659	238	34	yeast	yeast	NOUN
ijassa-659	238	35	,	,	PUNCT
ijassa-659	238	36	segment	segment	NOUN
ijassa-659	238	37	and	and	CCONJ
ijassa-659	238	38	digit-389	digit-389	ADJ
ijassa-659	238	39	datasets	dataset	NOUN
ijassa-659	238	40	)	)	PUNCT
ijassa-659	238	41	.	.	PUNCT
ijassa-659	239	1	however	however	ADV
ijassa-659	239	2	,	,	PUNCT
ijassa-659	239	3	drop	drop	VERB
ijassa-659	239	4	inefficiency	inefficiency	NOUN
ijassa-659	239	5	with	with	ADP
ijassa-659	239	6	increasing	increase	VERB
ijassa-659	239	7	the	the	DET
ijassa-659	239	8	number	number	NOUN
ijassa-659	239	9	of	of	ADP
ijassa-659	239	10	queries	query	NOUN
ijassa-659	239	11	makes	make	VERB
ijassa-659	239	12	the	the	DET
ijassa-659	239	13	proposed	propose	VERB
ijassa-659	239	14	method	method	NOUN
ijassa-659	239	15	superior	superior	ADJ
ijassa-659	239	16	to	to	ADP
ijassa-659	239	17	asc	asc	PROPN
ijassa-659	239	18	.	.	PUNCT
ijassa-659	240	1	the	the	DET
ijassa-659	240	2	superiority	superiority	NOUN
ijassa-659	240	3	of	of	ADP
ijassa-659	240	4	asc	asc	PROPN
ijassa-659	240	5	in	in	ADP
ijassa-659	240	6	a	a	DET
ijassa-659	240	7	small	small	ADJ
ijassa-659	240	8	number	number	NOUN
ijassa-659	240	9	of	of	ADP
ijassa-659	240	10	queries	query	NOUN
ijassa-659	240	11	comes	come	VERB
ijassa-659	240	12	from	from	ADP
ijassa-659	240	13	its	its	PRON
ijassa-659	240	14	propagation	propagation	NOUN
ijassa-659	240	15	procedure	procedure	NOUN
ijassa-659	240	16	that	that	PRON
ijassa-659	240	17	discovers	discover	VERB
ijassa-659	240	18	new	new	ADJ
ijassa-659	240	19	constraints	constraint	NOUN
ijassa-659	240	20	from	from	ADP
ijassa-659	240	21	the	the	DET
ijassa-659	240	22	information	information	NOUN
ijassa-659	240	23	stored	store	VERB
ijassa-659	240	24	in	in	ADP
ijassa-659	240	25	previously	previously	ADV
ijassa-659	240	26	chosen	choose	VERB
ijassa-659	240	27	constraints	constraint	NOUN
ijassa-659	240	28	and	and	CCONJ
ijassa-659	240	29	gives	give	VERB
ijassa-659	240	30	asc	asc	PROPN
ijassa-659	240	31	the	the	DET
ijassa-659	240	32	capability	capability	NOUN
ijassa-659	240	33	to	to	PART
ijassa-659	240	34	select	select	VERB
ijassa-659	240	35	a	a	DET
ijassa-659	240	36	well	well	ADV
ijassa-659	240	37	propagated	propagate	VERB
ijassa-659	240	38	set	set	NOUN
ijassa-659	240	39	of	of	ADP
ijassa-659	240	40	constraints	constraint	NOUN
ijassa-659	240	41	in	in	ADP
ijassa-659	240	42	a	a	DET
ijassa-659	240	43	small	small	ADJ
ijassa-659	240	44	number	number	NOUN
ijassa-659	240	45	of	of	ADP
ijassa-659	240	46	queries	query	NOUN
ijassa-659	240	47	.	.	PUNCT
ijassa-659	241	1	on	on	ADP
ijassa-659	241	2	the	the	DET
ijassa-659	241	3	other	other	ADJ
ijassa-659	241	4	hand	hand	NOUN
ijassa-659	241	5	,	,	PUNCT
ijassa-659	241	6	the	the	DET
ijassa-659	241	7	efficiency	efficiency	NOUN
ijassa-659	241	8	of	of	ADP
ijassa-659	241	9	asc	asc	PROPN
ijassa-659	241	10	is	be	AUX
ijassa-659	241	11	highly	highly	ADV
ijassa-659	241	12	depending	depend	VERB
ijassa-659	241	13	on	on	ADP
ijassa-659	241	14	its	its	PRON
ijassa-659	241	15	parameters	parameter	NOUN
ijassa-659	241	16	k	k	PROPN
ijassa-659	241	17	and	and	CCONJ
ijassa-659	241	18	θ	θ	PROPN
ijassa-659	241	19	.	.	PUNCT
ijassa-659	242	1	any	any	DET
ijassa-659	242	2	improper	improper	ADJ
ijassa-659	242	3	assignments	assignment	NOUN
ijassa-659	242	4	of	of	ADP
ijassa-659	242	5	k	k	PROPN
ijassa-659	242	6	and	and	CCONJ
ijassa-659	242	7	θ	θ	PROPN
ijassa-659	242	8	may	may	AUX
ijassa-659	242	9	result	result	VERB
ijassa-659	242	10	in	in	ADP
ijassa-659	242	11	an	an	DET
ijassa-659	242	12	inefficient	inefficient	ADJ
ijassa-659	242	13	set	set	NOUN
ijassa-659	242	14	of	of	ADP
ijassa-659	242	15	constraints	constraint	NOUN
ijassa-659	242	16	.	.	PUNCT
ijassa-659	243	1	in	in	ADP
ijassa-659	243	2	addition	addition	NOUN
ijassa-659	243	3	,	,	PUNCT
ijassa-659	243	4	some	some	DET
ijassa-659	243	5	values	value	NOUN
ijassa-659	243	6	of	of	ADP
ijassa-659	243	7	k	k	PROPN
ijassa-659	243	8	and	and	CCONJ
ijassa-659	243	9	θ	θ	PROPN
ijassa-659	243	10	may	may	AUX
ijassa-659	243	11	result	result	VERB
ijassa-659	243	12	in	in	ADP
ijassa-659	243	13	incorrectly	incorrectly	ADV
ijassa-659	243	14	propagated	propagate	VERB
ijassa-659	243	15	constraints	constraint	NOUN
ijassa-659	243	16	in	in	ADP
ijassa-659	243	17	the	the	DET
ijassa-659	243	18	propagation	propagation	NOUN
ijassa-659	243	19	step	step	NOUN
ijassa-659	243	20	.	.	PUNCT
ijassa-659	244	1	it	it	PRON
ijassa-659	244	2	means	mean	VERB
ijassa-659	244	3	that	that	SCONJ
ijassa-659	244	4	asc	asc	PROPN
ijassa-659	244	5	may	may	AUX
ijassa-659	244	6	provide	provide	VERB
ijassa-659	244	7	incorrect	incorrect	ADJ
ijassa-659	244	8	-	-	PUNCT
ijassa-659	244	9	labeled	label	VERB
ijassa-659	244	10	constraints	constraint	NOUN
ijassa-659	244	11	and	and	CCONJ
ijassa-659	244	12	mislead	mislead	VERB
ijassa-659	244	13	the	the	DET
ijassa-659	244	14	clustering	clustering	ADJ
ijassa-659	244	15	algorithms	algorithm	NOUN
ijassa-659	244	16	.	.	PUNCT
ijassa-659	245	1	in	in	ADP
ijassa-659	245	2	comparison	comparison	NOUN
ijassa-659	245	3	with	with	ADP
ijassa-659	245	4	npu	npu	ADJ
ijassa-659	245	5	method	method	NOUN
ijassa-659	245	6	that	that	PRON
ijassa-659	245	7	generally	generally	ADV
ijassa-659	245	8	outperforms	outperform	VERB
ijassa-659	245	9	random	random	ADJ
ijassa-659	245	10	,	,	PUNCT
ijassa-659	245	11	min	min	NOUN
ijassa-659	245	12	-	-	PUNCT
ijassa-659	245	13	max	max	NOUN
ijassa-659	245	14	,	,	PUNCT
ijassa-659	245	15	and	and	CCONJ
ijassa-659	245	16	asc	asc	PROPN
ijassa-659	245	17	method	method	PROPN
ijassa-659	245	18	.	.	PUNCT
ijassa-659	246	1	npu	npu	ADJ
ijassa-659	246	2	is	be	AUX
ijassa-659	246	3	generally	generally	ADV
ijassa-659	246	4	able	able	ADJ
ijassa-659	246	5	to	to	PART
ijassa-659	246	6	improve	improve	VERB
ijassa-659	246	7	the	the	DET
ijassa-659	246	8	clustering	clustering	ADJ
ijassa-659	246	9	performance	performance	NOUN
ijassa-659	246	10	consistently	consistently	ADV
ijassa-659	246	11	as	as	ADP
ijassa-659	246	12	increasing	increase	VERB
ijassa-659	246	13	the	the	DET
ijassa-659	246	14	number	number	NOUN
ijassa-659	246	15	of	of	ADP
ijassa-659	246	16	queries	query	NOUN
ijassa-659	246	17	.	.	PUNCT
ijassa-659	247	1	however	however	ADV
ijassa-659	247	2	,	,	PUNCT
ijassa-659	247	3	its	its	PRON
ijassa-659	247	4	performance	performance	NOUN
ijassa-659	247	5	is	be	AUX
ijassa-659	247	6	dominated	dominate	VERB
ijassa-659	247	7	by	by	ADP
ijassa-659	247	8	the	the	DET
ijassa-659	247	9	proposed	propose	VERB
ijassa-659	247	10	method	method	NOUN
ijassa-659	247	11	in	in	ADP
ijassa-659	247	12	most	most	ADJ
ijassa-659	247	13	cases	case	NOUN
ijassa-659	247	14	.	.	PUNCT
ijassa-659	248	1	improving	improve	VERB
ijassa-659	248	2	semi	semi	ADJ
ijassa-659	248	3	-	-	ADJ
ijassa-659	248	4	supervised	supervised	ADJ
ijassa-659	248	5	clustering	clustering	ADJ
ijassa-659	248	6	algorithms	algorithm	NOUN
ijassa-659	248	7	with	with	ADP
ijassa-659	248	8	active	active	ADJ
ijassa-659	248	9	query	query	NOUN
ijassa-659	248	10	selection	selection	NOUN
ijassa-659	248	11	35	35	NUM
ijassa-659	248	12	copyright	copyright	NOUN
ijassa-659	248	13	©	©	PROPN
ijassa-659	248	14	2019	2019	NUM
ijassa-659	248	15	assa	assa	PROPN
ijassa-659	248	16	adv	adv	PROPN
ijassa-659	248	17	.	.	PUNCT
ijassa-659	249	1	in	in	ADP
ijassa-659	249	2	systems	system	NOUN
ijassa-659	249	3	science	science	NOUN
ijassa-659	249	4	and	and	CCONJ
ijassa-659	249	5	appl	appl	NOUN
ijassa-659	249	6	.	.	PUNCT
ijassa-659	250	1	(	(	PUNCT
ijassa-659	250	2	2019	2019	NUM
ijassa-659	250	3	)	)	PUNCT
ijassa-659	250	4	protein	protein	NOUN
ijassa-659	250	5	heart	heart	NOUN
ijassa-659	250	6	ionosphere	ionosphere	NOUN
ijassa-659	250	7	breast	breast	NOUN
ijassa-659	250	8	yeast	yeast	NOUN
ijassa-659	250	9	segment	segment	NOUN
ijassa-659	250	10	digit-389	digit-389	NOUN
ijassa-659	250	11	magic	magic	ADJ
ijassa-659	250	12	figure	figure	NOUN
ijassa-659	250	13	2	2	NUM
ijassa-659	250	14	:	:	PUNCT
ijassa-659	250	15	comparison	comparison	NOUN
ijassa-659	250	16	of	of	ADP
ijassa-659	250	17	the	the	DET
ijassa-659	250	18	constraint	constraint	NOUN
ijassa-659	250	19	selection	selection	NOUN
ijassa-659	250	20	heuristics	heuristic	NOUN
ijassa-659	250	21	with	with	ADP
ijassa-659	250	22	mpckmeans	mpckmean	NOUN
ijassa-659	250	23	algorithm	algorithm	NOUN
ijassa-659	250	24	.	.	PUNCT
ijassa-659	251	1	0	0	NUM
ijassa-659	251	2	0,1	0,1	NUM
ijassa-659	251	3	0,2	0,2	NUM
ijassa-659	251	4	0,3	0,3	NUM
ijassa-659	251	5	0,4	0,4	NUM
ijassa-659	251	6	0,5	0,5	NUM
ijassa-659	251	7	0,6	0,6	NUM
ijassa-659	251	8	0,7	0,7	NUM
ijassa-659	251	9	0	0	NUM
ijassa-659	251	10	25	25	NUM
ijassa-659	251	11	50	50	NUM
ijassa-659	251	12	75	75	NUM
ijassa-659	251	13	100	100	NUM
ijassa-659	251	14	125	125	NUM
ijassa-659	251	15	150	150	NUM
ijassa-659	251	16	n	n	NOUN
ijassa-659	251	17	m	m	VERB
ijassa-659	251	18	i	i	PRON
ijassa-659	251	19	v	v	NUM
ijassa-659	251	20	al	al	PROPN
ijassa-659	251	21	u	u	NOUN
ijassa-659	251	22	e	e	NOUN
ijassa-659	251	23	number	number	NOUN
ijassa-659	251	24	of	of	ADP
ijassa-659	251	25	queries	query	NOUN
ijassa-659	251	26	our	our	PRON
ijassa-659	251	27	method	method	PROPN
ijassa-659	251	28	npu	npu	PROPN
ijassa-659	251	29	asc	asc	PROPN
ijassa-659	251	30	min	min	PROPN
ijassa-659	251	31	-	-	PROPN
ijassa-659	251	32	max	max	ADJ
ijassa-659	251	33	random	random	ADJ
ijassa-659	251	34	0	0	NUM
ijassa-659	251	35	0,1	0,1	NUM
ijassa-659	251	36	0,2	0,2	NUM
ijassa-659	251	37	0,3	0,3	NUM
ijassa-659	251	38	0,4	0,4	NUM
ijassa-659	251	39	0,5	0,5	NUM
ijassa-659	251	40	0,6	0,6	NUM
ijassa-659	251	41	0,7	0,7	NUM
ijassa-659	252	1	0,8	0,8	NUM
ijassa-659	252	2	0	0	NUM
ijassa-659	252	3	25	25	NUM
ijassa-659	252	4	50	50	NUM
ijassa-659	252	5	75	75	NUM
ijassa-659	252	6	100	100	NUM
ijassa-659	252	7	125	125	NUM
ijassa-659	252	8	150	150	NUM
ijassa-659	253	1	n	n	NOUN
ijassa-659	253	2	m	m	VERB
ijassa-659	253	3	i	i	PRON
ijassa-659	253	4	v	v	NUM
ijassa-659	253	5	al	al	PROPN
ijassa-659	253	6	u	u	NOUN
ijassa-659	253	7	e	e	NOUN
ijassa-659	253	8	number	number	NOUN
ijassa-659	253	9	of	of	ADP
ijassa-659	253	10	queries	query	NOUN
ijassa-659	253	11	our	our	PRON
ijassa-659	253	12	method	method	PROPN
ijassa-659	253	13	npu	npu	PROPN
ijassa-659	253	14	asc	asc	PROPN
ijassa-659	253	15	min	min	PROPN
ijassa-659	253	16	-	-	PROPN
ijassa-659	253	17	max	max	ADJ
ijassa-659	253	18	random	random	ADJ
ijassa-659	253	19	0,4	0,4	NUM
ijassa-659	253	20	0,5	0,5	NUM
ijassa-659	253	21	0,6	0,6	NUM
ijassa-659	253	22	0,7	0,7	NUM
ijassa-659	253	23	0,8	0,8	NUM
ijassa-659	253	24	0,9	0,9	NUM
ijassa-659	253	25	0	0	NUM
ijassa-659	253	26	25	25	NUM
ijassa-659	253	27	50	50	NUM
ijassa-659	253	28	75	75	NUM
ijassa-659	253	29	100	100	NUM
ijassa-659	253	30	125	125	NUM
ijassa-659	253	31	150	150	NUM
ijassa-659	253	32	n	n	NOUN
ijassa-659	253	33	m	m	VERB
ijassa-659	253	34	i	i	PRON
ijassa-659	253	35	v	v	NUM
ijassa-659	253	36	al	al	PROPN
ijassa-659	253	37	u	u	NOUN
ijassa-659	253	38	e	e	NOUN
ijassa-659	253	39	number	number	NOUN
ijassa-659	253	40	of	of	ADP
ijassa-659	253	41	queries	query	NOUN
ijassa-659	253	42	our	our	PRON
ijassa-659	253	43	method	method	PROPN
ijassa-659	253	44	npu	npu	PROPN
ijassa-659	253	45	asc	asc	PROPN
ijassa-659	253	46	min	min	PROPN
ijassa-659	253	47	-	-	PROPN
ijassa-659	253	48	max	max	PROPN
ijassa-659	253	49	random	random	NOUN
ijassa-659	253	50	0,65	0,65	NOUN
ijassa-659	253	51	0,75	0,75	NOUN
ijassa-659	253	52	0,85	0,85	ADP
ijassa-659	254	1	0,95	0,95	SYM
ijassa-659	254	2	1,05	1,05	NUM
ijassa-659	254	3	0	0	NUM
ijassa-659	254	4	25	25	NUM
ijassa-659	254	5	50	50	NUM
ijassa-659	254	6	75	75	NUM
ijassa-659	254	7	100	100	NUM
ijassa-659	254	8	125	125	NUM
ijassa-659	254	9	150	150	NUM
ijassa-659	254	10	n	n	NOUN
ijassa-659	254	11	m	m	VERB
ijassa-659	254	12	i	i	PRON
ijassa-659	254	13	v	v	NUM
ijassa-659	254	14	al	al	PROPN
ijassa-659	254	15	u	u	NOUN
ijassa-659	254	16	e	e	NOUN
ijassa-659	254	17	number	number	NOUN
ijassa-659	254	18	of	of	ADP
ijassa-659	254	19	queries	query	NOUN
ijassa-659	254	20	our	our	PRON
ijassa-659	254	21	method	method	PROPN
ijassa-659	254	22	npu	npu	PROPN
ijassa-659	254	23	asc	asc	PROPN
ijassa-659	254	24	min	min	PROPN
ijassa-659	254	25	-	-	PROPN
ijassa-659	254	26	max	max	ADJ
ijassa-659	254	27	random	random	ADJ
ijassa-659	254	28	0,6	0,6	NUM
ijassa-659	254	29	0,7	0,7	NUM
ijassa-659	254	30	0,8	0,8	NUM
ijassa-659	254	31	0,9	0,9	NUM
ijassa-659	254	32	1	1	NUM
ijassa-659	254	33	0	0	NUM
ijassa-659	254	34	25	25	NUM
ijassa-659	254	35	50	50	NUM
ijassa-659	254	36	75	75	NUM
ijassa-659	254	37	100	100	NUM
ijassa-659	254	38	125	125	NUM
ijassa-659	254	39	150	150	NUM
ijassa-659	255	1	n	n	NOUN
ijassa-659	255	2	m	m	VERB
ijassa-659	255	3	i	i	PRON
ijassa-659	255	4	v	v	NUM
ijassa-659	255	5	al	al	PROPN
ijassa-659	255	6	u	u	NOUN
ijassa-659	255	7	e	e	NOUN
ijassa-659	255	8	number	number	NOUN
ijassa-659	255	9	of	of	ADP
ijassa-659	255	10	queries	query	NOUN
ijassa-659	255	11	our	our	PRON
ijassa-659	255	12	method	method	PROPN
ijassa-659	255	13	npu	npu	PROPN
ijassa-659	255	14	asc	asc	PROPN
ijassa-659	255	15	min	min	PROPN
ijassa-659	255	16	-	-	PROPN
ijassa-659	255	17	max	max	PROPN
ijassa-659	255	18	random	random	PROPN
ijassa-659	255	19	0,5	0,5	X
ijassa-659	255	20	0,55	0,55	NOUN
ijassa-659	255	21	0,6	0,6	NUM
ijassa-659	255	22	0,65	0,65	NUM
ijassa-659	256	1	0,7	0,7	NUM
ijassa-659	256	2	0,75	0,75	NUM
ijassa-659	256	3	0,8	0,8	NOUN
ijassa-659	256	4	0	0	NUM
ijassa-659	256	5	25	25	NUM
ijassa-659	256	6	50	50	NUM
ijassa-659	256	7	75	75	NUM
ijassa-659	256	8	100	100	NUM
ijassa-659	256	9	125	125	NUM
ijassa-659	256	10	150	150	NUM
ijassa-659	256	11	n	n	NOUN
ijassa-659	256	12	m	m	VERB
ijassa-659	256	13	i	i	PRON
ijassa-659	256	14	v	v	NUM
ijassa-659	256	15	al	al	PROPN
ijassa-659	256	16	u	u	NOUN
ijassa-659	256	17	e	e	NOUN
ijassa-659	256	18	number	number	NOUN
ijassa-659	256	19	of	of	ADP
ijassa-659	256	20	queries	query	NOUN
ijassa-659	256	21	our	our	PRON
ijassa-659	256	22	method	method	PROPN
ijassa-659	256	23	npu	npu	PROPN
ijassa-659	256	24	asc	asc	PROPN
ijassa-659	256	25	min	min	PROPN
ijassa-659	256	26	-	-	PROPN
ijassa-659	256	27	max	max	ADJ
ijassa-659	256	28	random	random	ADJ
ijassa-659	256	29	0,55	0,55	PUNCT
ijassa-659	257	1	0,6	0,6	NUM
ijassa-659	257	2	0,65	0,65	NUM
ijassa-659	257	3	0,7	0,7	NUM
ijassa-659	257	4	0,75	0,75	NUM
ijassa-659	257	5	0,8	0,8	NUM
ijassa-659	257	6	0,85	0,85	SYM
ijassa-659	257	7	0	0	NUM
ijassa-659	257	8	25	25	NUM
ijassa-659	257	9	50	50	NUM
ijassa-659	257	10	75	75	NUM
ijassa-659	257	11	100	100	NUM
ijassa-659	257	12	125	125	NUM
ijassa-659	257	13	150	150	NUM
ijassa-659	257	14	n	n	NOUN
ijassa-659	257	15	m	m	VERB
ijassa-659	257	16	i	i	PRON
ijassa-659	257	17	v	v	NUM
ijassa-659	257	18	al	al	PROPN
ijassa-659	257	19	u	u	NOUN
ijassa-659	257	20	e	e	NOUN
ijassa-659	257	21	number	number	NOUN
ijassa-659	257	22	of	of	ADP
ijassa-659	257	23	queries	query	NOUN
ijassa-659	257	24	our	our	PRON
ijassa-659	257	25	method	method	PROPN
ijassa-659	257	26	npu	npu	PROPN
ijassa-659	257	27	asc	asc	PROPN
ijassa-659	257	28	min	min	PROPN
ijassa-659	257	29	-	-	PROPN
ijassa-659	257	30	max	max	PROPN
ijassa-659	257	31	random	random	ADJ
ijassa-659	257	32	0,5	0,5	NUM
ijassa-659	257	33	0,6	0,6	NUM
ijassa-659	257	34	0,7	0,7	NUM
ijassa-659	257	35	0,8	0,8	NUM
ijassa-659	257	36	0,9	0,9	NUM
ijassa-659	257	37	0	0	NUM
ijassa-659	257	38	25	25	NUM
ijassa-659	257	39	50	50	NUM
ijassa-659	257	40	75	75	NUM
ijassa-659	257	41	100	100	NUM
ijassa-659	257	42	125	125	NUM
ijassa-659	257	43	150	150	NUM
ijassa-659	257	44	n	n	NOUN
ijassa-659	257	45	m	m	VERB
ijassa-659	257	46	i	i	PRON
ijassa-659	257	47	v	v	NUM
ijassa-659	257	48	al	al	PROPN
ijassa-659	257	49	u	u	NOUN
ijassa-659	257	50	e	e	NOUN
ijassa-659	257	51	number	number	NOUN
ijassa-659	257	52	of	of	ADP
ijassa-659	257	53	queries	query	NOUN
ijassa-659	257	54	our	our	PRON
ijassa-659	257	55	method	method	PROPN
ijassa-659	257	56	npu	npu	PROPN
ijassa-659	257	57	asc	asc	PROPN
ijassa-659	257	58	min	min	PROPN
ijassa-659	257	59	-	-	PROPN
ijassa-659	257	60	max	max	PROPN
ijassa-659	257	61	random	random	ADJ
ijassa-659	257	62	36	36	NUM
ijassa-659	257	63	w.	w.	NOUN
ijassa-659	257	64	atwa	atwa	PROPN
ijassa-659	257	65	,	,	PUNCT
ijassa-659	257	66	m.	m.	NOUN
ijassa-659	257	67	emam	emam	PROPN
ijassa-659	257	68	copyright	copyright	NOUN
ijassa-659	257	69	©	©	PROPN
ijassa-659	257	70	2019	2019	NUM
ijassa-659	257	71	assa	assa	PROPN
ijassa-659	257	72	adv	adv	PROPN
ijassa-659	257	73	.	.	PUNCT
ijassa-659	258	1	in	in	ADP
ijassa-659	258	2	systems	system	NOUN
ijassa-659	258	3	science	science	NOUN
ijassa-659	258	4	and	and	CCONJ
ijassa-659	258	5	appl	appl	NOUN
ijassa-659	258	6	.	.	PUNCT
ijassa-659	259	1	(	(	PUNCT
ijassa-659	259	2	2019	2019	NUM
ijassa-659	259	3	)	)	PUNCT
ijassa-659	259	4	protein	protein	NOUN
ijassa-659	259	5	heart	heart	NOUN
ijassa-659	259	6	ionosphere	ionosphere	NOUN
ijassa-659	259	7	breast	breast	NOUN
ijassa-659	259	8	yeast	yeast	NOUN
ijassa-659	259	9	segment	segment	NOUN
ijassa-659	259	10	digit-389	digit-389	PROPN
ijassa-659	259	11	magic	magic	PROPN
ijassa-659	259	12	fig	fig	NOUN
ijassa-659	259	13	.	.	PUNCT
ijassa-659	260	1	3	3	NUM
ijassa-659	260	2	:	:	PUNCT
ijassa-659	260	3	comparison	comparison	NOUN
ijassa-659	260	4	of	of	ADP
ijassa-659	260	5	the	the	DET
ijassa-659	260	6	constraint	constraint	NOUN
ijassa-659	260	7	selection	selection	NOUN
ijassa-659	260	8	heuristics	heuristic	NOUN
ijassa-659	260	9	with	with	ADP
ijassa-659	260	10	ahcc	ahcc	NOUN
ijassa-659	260	11	algorithm	algorithm	NOUN
ijassa-659	260	12	.	.	PUNCT
ijassa-659	261	1	0	0	NUM
ijassa-659	261	2	0,1	0,1	NUM
ijassa-659	261	3	0,2	0,2	NUM
ijassa-659	261	4	0,3	0,3	NUM
ijassa-659	261	5	0,4	0,4	NUM
ijassa-659	261	6	0,5	0,5	NUM
ijassa-659	261	7	0,6	0,6	NUM
ijassa-659	261	8	0,7	0,7	NUM
ijassa-659	261	9	0,8	0,8	NUM
ijassa-659	261	10	0	0	NUM
ijassa-659	261	11	25	25	NUM
ijassa-659	261	12	50	50	NUM
ijassa-659	261	13	75	75	NUM
ijassa-659	261	14	100	100	NUM
ijassa-659	261	15	125	125	NUM
ijassa-659	261	16	150	150	NUM
ijassa-659	262	1	n	n	NOUN
ijassa-659	263	1	m	m	VERB
ijassa-659	264	1	i	i	PRON
ijassa-659	264	2	v	v	NUM
ijassa-659	264	3	al	al	PROPN
ijassa-659	264	4	u	u	NOUN
ijassa-659	264	5	e	e	NOUN
ijassa-659	264	6	number	number	NOUN
ijassa-659	264	7	of	of	ADP
ijassa-659	264	8	queries	query	NOUN
ijassa-659	264	9	our	our	PRON
ijassa-659	264	10	method	method	PROPN
ijassa-659	264	11	npu	npu	PROPN
ijassa-659	264	12	asc	asc	PROPN
ijassa-659	264	13	min	min	PROPN
ijassa-659	264	14	-	-	PROPN
ijassa-659	264	15	max	max	ADJ
ijassa-659	264	16	random	random	ADJ
ijassa-659	264	17	0	0	NUM
ijassa-659	264	18	0,1	0,1	NUM
ijassa-659	264	19	0,2	0,2	NUM
ijassa-659	264	20	0,3	0,3	NUM
ijassa-659	264	21	0,4	0,4	NUM
ijassa-659	264	22	0,5	0,5	NUM
ijassa-659	264	23	0,6	0,6	NUM
ijassa-659	264	24	0,7	0,7	NUM
ijassa-659	264	25	0,8	0,8	NUM
ijassa-659	265	1	0,9	0,9	NUM
ijassa-659	265	2	0	0	NUM
ijassa-659	265	3	25	25	NUM
ijassa-659	265	4	50	50	NUM
ijassa-659	265	5	75	75	NUM
ijassa-659	265	6	100	100	NUM
ijassa-659	265	7	125	125	NUM
ijassa-659	265	8	150	150	NUM
ijassa-659	265	9	n	n	NOUN
ijassa-659	265	10	m	m	VERB
ijassa-659	265	11	i	i	PRON
ijassa-659	265	12	v	v	NUM
ijassa-659	265	13	al	al	PROPN
ijassa-659	265	14	u	u	NOUN
ijassa-659	265	15	e	e	NOUN
ijassa-659	265	16	number	number	NOUN
ijassa-659	265	17	of	of	ADP
ijassa-659	265	18	queries	query	NOUN
ijassa-659	265	19	our	our	PRON
ijassa-659	265	20	method	method	PROPN
ijassa-659	265	21	npu	npu	PROPN
ijassa-659	265	22	asc	asc	PROPN
ijassa-659	265	23	min	min	PROPN
ijassa-659	265	24	-	-	PROPN
ijassa-659	265	25	max	max	ADJ
ijassa-659	265	26	random	random	ADJ
ijassa-659	265	27	0,4	0,4	NUM
ijassa-659	265	28	0,5	0,5	NUM
ijassa-659	265	29	0,6	0,6	NUM
ijassa-659	265	30	0,7	0,7	NUM
ijassa-659	265	31	0,8	0,8	NUM
ijassa-659	265	32	0,9	0,9	NUM
ijassa-659	265	33	0	0	NUM
ijassa-659	265	34	25	25	NUM
ijassa-659	265	35	50	50	NUM
ijassa-659	265	36	75	75	NUM
ijassa-659	265	37	100	100	NUM
ijassa-659	265	38	125	125	NUM
ijassa-659	265	39	150	150	NUM
ijassa-659	265	40	n	n	NOUN
ijassa-659	265	41	m	m	VERB
ijassa-659	265	42	i	i	PRON
ijassa-659	265	43	v	v	NUM
ijassa-659	265	44	al	al	PROPN
ijassa-659	265	45	u	u	NOUN
ijassa-659	265	46	e	e	NOUN
ijassa-659	265	47	number	number	NOUN
ijassa-659	265	48	of	of	ADP
ijassa-659	265	49	queries	query	NOUN
ijassa-659	265	50	our	our	PRON
ijassa-659	265	51	method	method	PROPN
ijassa-659	265	52	npu	npu	PROPN
ijassa-659	265	53	asc	asc	PROPN
ijassa-659	265	54	min	min	PROPN
ijassa-659	265	55	-	-	PROPN
ijassa-659	265	56	max	max	ADJ
ijassa-659	265	57	random	random	ADJ
ijassa-659	265	58	0,4	0,4	NUM
ijassa-659	265	59	0,5	0,5	NUM
ijassa-659	265	60	0,6	0,6	NUM
ijassa-659	265	61	0,7	0,7	NUM
ijassa-659	265	62	0,8	0,8	NUM
ijassa-659	265	63	0,9	0,9	NUM
ijassa-659	265	64	1	1	NUM
ijassa-659	265	65	0	0	NUM
ijassa-659	265	66	25	25	NUM
ijassa-659	265	67	50	50	NUM
ijassa-659	265	68	75	75	NUM
ijassa-659	265	69	100	100	NUM
ijassa-659	265	70	125	125	NUM
ijassa-659	265	71	150	150	NUM
ijassa-659	266	1	n	n	NOUN
ijassa-659	266	2	m	m	VERB
ijassa-659	266	3	i	i	PRON
ijassa-659	266	4	v	v	NUM
ijassa-659	266	5	al	al	PROPN
ijassa-659	266	6	u	u	NOUN
ijassa-659	266	7	e	e	NOUN
ijassa-659	266	8	number	number	NOUN
ijassa-659	266	9	of	of	ADP
ijassa-659	266	10	queries	query	NOUN
ijassa-659	266	11	our	our	PRON
ijassa-659	266	12	method	method	PROPN
ijassa-659	266	13	npu	npu	PROPN
ijassa-659	266	14	asc	asc	PROPN
ijassa-659	266	15	min	min	PROPN
ijassa-659	266	16	-	-	PROPN
ijassa-659	266	17	max	max	ADJ
ijassa-659	266	18	random	random	ADJ
ijassa-659	266	19	0,6	0,6	NUM
ijassa-659	266	20	0,7	0,7	NUM
ijassa-659	266	21	0,8	0,8	NUM
ijassa-659	266	22	0,9	0,9	NUM
ijassa-659	266	23	1	1	NUM
ijassa-659	266	24	0	0	NUM
ijassa-659	266	25	25	25	NUM
ijassa-659	266	26	50	50	NUM
ijassa-659	266	27	75	75	NUM
ijassa-659	266	28	100	100	NUM
ijassa-659	266	29	125	125	NUM
ijassa-659	266	30	150	150	NUM
ijassa-659	267	1	n	n	NOUN
ijassa-659	267	2	m	m	VERB
ijassa-659	267	3	i	i	PRON
ijassa-659	267	4	v	v	NUM
ijassa-659	267	5	al	al	PROPN
ijassa-659	267	6	u	u	NOUN
ijassa-659	267	7	e	e	NOUN
ijassa-659	267	8	number	number	NOUN
ijassa-659	267	9	of	of	ADP
ijassa-659	267	10	queries	query	NOUN
ijassa-659	267	11	our	our	PRON
ijassa-659	267	12	method	method	PROPN
ijassa-659	267	13	npu	npu	PROPN
ijassa-659	267	14	asc	asc	PROPN
ijassa-659	267	15	min	min	PROPN
ijassa-659	267	16	-	-	PROPN
ijassa-659	267	17	max	max	ADJ
ijassa-659	267	18	random	random	ADJ
ijassa-659	267	19	0,6	0,6	NUM
ijassa-659	267	20	0,65	0,65	NUM
ijassa-659	268	1	0,7	0,7	NUM
ijassa-659	268	2	0,75	0,75	NUM
ijassa-659	268	3	0,8	0,8	NOUN
ijassa-659	268	4	0	0	NUM
ijassa-659	268	5	25	25	NUM
ijassa-659	268	6	50	50	NUM
ijassa-659	268	7	75	75	NUM
ijassa-659	268	8	100	100	NUM
ijassa-659	268	9	125	125	NUM
ijassa-659	268	10	150	150	NUM
ijassa-659	268	11	n	n	NOUN
ijassa-659	268	12	m	m	VERB
ijassa-659	268	13	i	i	PRON
ijassa-659	268	14	v	v	NUM
ijassa-659	268	15	al	al	PROPN
ijassa-659	268	16	u	u	NOUN
ijassa-659	268	17	e	e	NOUN
ijassa-659	268	18	number	number	NOUN
ijassa-659	268	19	of	of	ADP
ijassa-659	268	20	queries	query	NOUN
ijassa-659	268	21	our	our	PRON
ijassa-659	268	22	method	method	PROPN
ijassa-659	268	23	npu	npu	PROPN
ijassa-659	268	24	asc	asc	PROPN
ijassa-659	268	25	min	min	PROPN
ijassa-659	268	26	-	-	PROPN
ijassa-659	268	27	max	max	ADJ
ijassa-659	268	28	random	random	ADJ
ijassa-659	268	29	0,55	0,55	PUNCT
ijassa-659	269	1	0,6	0,6	NUM
ijassa-659	269	2	0,65	0,65	NUM
ijassa-659	269	3	0,7	0,7	NUM
ijassa-659	269	4	0,75	0,75	NUM
ijassa-659	269	5	0,8	0,8	NUM
ijassa-659	269	6	0,85	0,85	PUNCT
ijassa-659	270	1	0,9	0,9	NUM
ijassa-659	270	2	0	0	NUM
ijassa-659	270	3	25	25	NUM
ijassa-659	270	4	50	50	NUM
ijassa-659	270	5	75	75	NUM
ijassa-659	270	6	100	100	NUM
ijassa-659	270	7	125	125	NUM
ijassa-659	270	8	150	150	NUM
ijassa-659	270	9	n	n	NOUN
ijassa-659	270	10	m	m	VERB
ijassa-659	270	11	i	i	PRON
ijassa-659	270	12	v	v	NUM
ijassa-659	270	13	al	al	PROPN
ijassa-659	270	14	u	u	NOUN
ijassa-659	270	15	e	e	NOUN
ijassa-659	270	16	number	number	NOUN
ijassa-659	270	17	of	of	ADP
ijassa-659	270	18	queries	query	NOUN
ijassa-659	270	19	our	our	PRON
ijassa-659	270	20	method	method	PROPN
ijassa-659	270	21	npu	npu	PROPN
ijassa-659	270	22	asc	asc	PROPN
ijassa-659	270	23	min	min	PROPN
ijassa-659	270	24	-	-	PROPN
ijassa-659	270	25	max	max	PROPN
ijassa-659	270	26	random	random	ADJ
ijassa-659	270	27	0,5	0,5	NUM
ijassa-659	270	28	0,6	0,6	NUM
ijassa-659	270	29	0,7	0,7	NUM
ijassa-659	270	30	0,8	0,8	NUM
ijassa-659	270	31	0,9	0,9	NUM
ijassa-659	270	32	0	0	NUM
ijassa-659	270	33	25	25	NUM
ijassa-659	270	34	50	50	NUM
ijassa-659	270	35	75	75	NUM
ijassa-659	270	36	100	100	NUM
ijassa-659	270	37	125	125	NUM
ijassa-659	270	38	150	150	NUM
ijassa-659	270	39	n	n	NOUN
ijassa-659	270	40	m	m	VERB
ijassa-659	270	41	i	i	PRON
ijassa-659	270	42	v	v	NUM
ijassa-659	270	43	al	al	PROPN
ijassa-659	270	44	u	u	NOUN
ijassa-659	270	45	e	e	NOUN
ijassa-659	270	46	number	number	NOUN
ijassa-659	270	47	of	of	ADP
ijassa-659	270	48	queries	query	NOUN
ijassa-659	270	49	our	our	PRON
ijassa-659	270	50	method	method	PROPN
ijassa-659	270	51	npu	npu	PROPN
ijassa-659	270	52	asc	asc	PROPN
ijassa-659	270	53	min	min	PROPN
ijassa-659	270	54	-	-	PROPN
ijassa-659	270	55	max	max	PROPN
ijassa-659	270	56	random	random	ADJ
ijassa-659	270	57	improving	improve	VERB
ijassa-659	270	58	semi	semi	ADJ
ijassa-659	270	59	-	-	ADJ
ijassa-659	270	60	supervised	supervised	ADJ
ijassa-659	270	61	clustering	clustering	ADJ
ijassa-659	270	62	algorithms	algorithm	NOUN
ijassa-659	270	63	with	with	ADP
ijassa-659	270	64	active	active	ADJ
ijassa-659	270	65	query	query	NOUN
ijassa-659	270	66	selection	selection	NOUN
ijassa-659	270	67	37	37	NUM
ijassa-659	270	68	copyright	copyright	NOUN
ijassa-659	270	69	©	©	PROPN
ijassa-659	270	70	2019	2019	NUM
ijassa-659	270	71	assa	assa	PROPN
ijassa-659	270	72	adv	adv	PROPN
ijassa-659	270	73	.	.	PUNCT
ijassa-659	271	1	in	in	ADP
ijassa-659	271	2	systems	system	NOUN
ijassa-659	271	3	science	science	NOUN
ijassa-659	271	4	and	and	CCONJ
ijassa-659	271	5	appl	appl	NOUN
ijassa-659	271	6	.	.	PUNCT
ijassa-659	272	1	(	(	PUNCT
ijassa-659	272	2	2019	2019	NUM
ijassa-659	272	3	)	)	PUNCT
ijassa-659	272	4	we	we	PRON
ijassa-659	272	5	also	also	ADV
ijassa-659	272	6	use	use	VERB
ijassa-659	272	7	pairwise	pairwise	NOUN
ijassa-659	272	8	f	f	NOUN
ijassa-659	272	9	-	-	PUNCT
ijassa-659	272	10	measure	measure	NOUN
ijassa-659	272	11	to	to	PART
ijassa-659	272	12	evaluate	evaluate	VERB
ijassa-659	272	13	the	the	DET
ijassa-659	272	14	clustering	clustering	ADJ
ijassa-659	272	15	quality	quality	NOUN
ijassa-659	272	16	;	;	PUNCT
ijassa-659	272	17	as	as	SCONJ
ijassa-659	272	18	it	it	PRON
ijassa-659	272	19	focuses	focus	VERB
ijassa-659	272	20	on	on	ADP
ijassa-659	272	21	how	how	SCONJ
ijassa-659	272	22	accurately	accurately	ADV
ijassa-659	272	23	we	we	PRON
ijassa-659	272	24	can	can	AUX
ijassa-659	272	25	predict	predict	VERB
ijassa-659	272	26	the	the	DET
ijassa-659	272	27	pairwise	pairwise	NOUN
ijassa-659	272	28	relationship	relationship	NOUN
ijassa-659	272	29	between	between	ADP
ijassa-659	272	30	any	any	DET
ijassa-659	272	31	pair	pair	NOUN
ijassa-659	272	32	of	of	ADP
ijassa-659	272	33	instances	instance	NOUN
ijassa-659	272	34	.	.	PUNCT
ijassa-659	273	1	tables	table	NOUN
ijassa-659	273	2	1	1	NUM
ijassa-659	273	3	2	2	NUM
ijassa-659	273	4	show	show	VERB
ijassa-659	273	5	the	the	DET
ijassa-659	273	6	pairwise	pairwise	NOUN
ijassa-659	273	7	f	f	NOUN
ijassa-659	273	8	-	-	PUNCT
ijassa-659	273	9	measure	measure	NOUN
ijassa-659	273	10	results	result	NOUN
ijassa-659	273	11	with	with	ADP
ijassa-659	273	12	different	different	ADJ
ijassa-659	273	13	query	query	NOUN
ijassa-659	273	14	size	size	NOUN
ijassa-659	273	15	on	on	ADP
ijassa-659	273	16	the	the	DET
ijassa-659	273	17	datasets	dataset	NOUN
ijassa-659	273	18	with	with	ADP
ijassa-659	273	19	mpckmeans	mpckmean	NOUN
ijassa-659	273	20	and	and	CCONJ
ijassa-659	273	21	ahcc	ahcc	NOUN
ijassa-659	273	22	algorithm	algorithm	NOUN
ijassa-659	273	23	respectively	respectively	ADV
ijassa-659	273	24	.	.	PUNCT
ijassa-659	274	1	the	the	DET
ijassa-659	274	2	best	well	ADV
ijassa-659	274	3	performing	performing	NOUN
ijassa-659	274	4	method	method	NOUN
ijassa-659	274	5	is	be	AUX
ijassa-659	274	6	then	then	ADV
ijassa-659	274	7	highlighted	highlight	VERB
ijassa-659	274	8	in	in	ADP
ijassa-659	274	9	boldface	boldface	NOUN
ijassa-659	274	10	.	.	PUNCT
ijassa-659	275	1	once	once	ADV
ijassa-659	275	2	again	again	ADV
ijassa-659	275	3	,	,	PUNCT
ijassa-659	275	4	the	the	DET
ijassa-659	275	5	proposed	propose	VERB
ijassa-659	275	6	active	active	ADJ
ijassa-659	275	7	methods	method	NOUN
ijassa-659	275	8	show	show	VERB
ijassa-659	275	9	a	a	DET
ijassa-659	275	10	clear	clear	ADJ
ijassa-659	275	11	advantage	advantage	NOUN
ijassa-659	275	12	over	over	ADP
ijassa-659	275	13	the	the	DET
ijassa-659	275	14	baseline	baseline	ADJ
ijassa-659	275	15	methods	method	NOUN
ijassa-659	275	16	.	.	PUNCT
ijassa-659	276	1	when	when	SCONJ
ijassa-659	276	2	using	use	VERB
ijassa-659	276	3	small	small	ADJ
ijassa-659	276	4	number	number	NOUN
ijassa-659	276	5	of	of	ADP
ijassa-659	276	6	queries	query	NOUN
ijassa-659	276	7	,	,	PUNCT
ijassa-659	276	8	the	the	DET
ijassa-659	276	9	performance	performance	NOUN
ijassa-659	276	10	of	of	ADP
ijassa-659	276	11	the	the	DET
ijassa-659	276	12	methods	method	NOUN
ijassa-659	276	13	is	be	AUX
ijassa-659	276	14	fairly	fairly	ADV
ijassa-659	276	15	close	close	ADJ
ijassa-659	276	16	.	.	PUNCT
ijassa-659	277	1	however	however	ADV
ijassa-659	277	2	,	,	PUNCT
ijassa-659	277	3	as	as	SCONJ
ijassa-659	277	4	we	we	PRON
ijassa-659	277	5	increase	increase	VERB
ijassa-659	277	6	the	the	DET
ijassa-659	277	7	number	number	NOUN
ijassa-659	277	8	of	of	ADP
ijassa-659	277	9	queries	query	NOUN
ijassa-659	277	10	,	,	PUNCT
ijassa-659	277	11	the	the	DET
ijassa-659	277	12	proposed	propose	VERB
ijassa-659	277	13	method	method	NOUN
ijassa-659	277	14	becomes	become	VERB
ijassa-659	277	15	better	well	ADJ
ijassa-659	277	16	than	than	ADP
ijassa-659	277	17	all	all	DET
ijassa-659	277	18	other	other	ADJ
ijassa-659	277	19	methods	method	NOUN
ijassa-659	277	20	.	.	PUNCT
ijassa-659	278	1	the	the	DET
ijassa-659	278	2	experimental	experimental	ADJ
ijassa-659	278	3	results	result	NOUN
ijassa-659	278	4	demonstrate	demonstrate	VERB
ijassa-659	278	5	that	that	SCONJ
ijassa-659	278	6	the	the	DET
ijassa-659	278	7	constraints	constraint	NOUN
ijassa-659	278	8	selected	select	VERB
ijassa-659	278	9	by	by	ADP
ijassa-659	278	10	the	the	DET
ijassa-659	278	11	proposed	propose	VERB
ijassa-659	278	12	active	active	ADJ
ijassa-659	278	13	learning	learning	NOUN
ijassa-659	278	14	process	process	NOUN
ijassa-659	278	15	are	be	AUX
ijassa-659	278	16	generally	generally	ADV
ijassa-659	278	17	more	more	ADV
ijassa-659	278	18	beneficial	beneficial	ADJ
ijassa-659	278	19	for	for	ADP
ijassa-659	278	20	constraint	constraint	NOUN
ijassa-659	278	21	-	-	PUNCT
ijassa-659	278	22	based	base	VERB
ijassa-659	278	23	clustering	clustering	ADJ
ijassa-659	278	24	algorithms	algorithm	NOUN
ijassa-659	278	25	than	than	ADP
ijassa-659	278	26	other	other	ADJ
ijassa-659	278	27	baseline	baseline	NOUN
ijassa-659	278	28	methods	method	NOUN
ijassa-659	278	29	.	.	PUNCT
ijassa-659	279	1	table	table	NOUN
ijassa-659	279	2	1	1	NUM
ijassa-659	279	3	:	:	PUNCT
ijassa-659	279	4	comparison	comparison	NOUN
ijassa-659	279	5	on	on	ADP
ijassa-659	279	6	pairwise	pairwise	NOUN
ijassa-659	279	7	f	f	NOUN
ijassa-659	279	8	-	-	PUNCT
ijassa-659	279	9	measure	measure	NOUN
ijassa-659	279	10	(	(	PUNCT
ijassa-659	279	11	mean	mean	NOUN
ijassa-659	279	12	±	±	NUM
ijassa-659	279	13	std	std	NOUN
ijassa-659	279	14	)	)	PUNCT
ijassa-659	279	15	with	with	ADP
ijassa-659	279	16	mpckmeans	mpckmean	NOUN
ijassa-659	279	17	algorithm	algorithm	PROPN
ijassa-659	279	18	.	.	PUNCT
ijassa-659	280	1	dataset	dataset	ADJ
ijassa-659	280	2	constraint	constraint	NOUN
ijassa-659	280	3	selection	selection	NOUN
ijassa-659	280	4	heuristics	heuristic	NOUN
ijassa-659	280	5	number	number	NOUN
ijassa-659	280	6	of	of	ADP
ijassa-659	280	7	queries	query	NOUN
ijassa-659	280	8	25	25	NUM
ijassa-659	280	9	50	50	NUM
ijassa-659	280	10	75	75	NUM
ijassa-659	280	11	100	100	NUM
ijassa-659	280	12	125	125	NUM
ijassa-659	280	13	150	150	NUM
ijassa-659	280	14	protein	protein	NOUN
ijassa-659	280	15	proposed	propose	VERB
ijassa-659	280	16	method	method	PROPN
ijassa-659	281	1	npu	npu	PROPN
ijassa-659	281	2	asc	asc	PROPN
ijassa-659	281	3	min	min	PROPN
ijassa-659	281	4	-	-	PROPN
ijassa-659	281	5	max	max	ADJ
ijassa-659	281	6	random	random	ADJ
ijassa-659	282	1	0.63±0.045	0.63±0.045	PUNCT
ijassa-659	283	1	0.59±0.047	0.59±0.047	NOUN
ijassa-659	283	2	0.56±0.019	0.56±0.019	PUNCT
ijassa-659	284	1	0.49±0.001	0.49±0.001	NOUN
ijassa-659	284	2	0.44±0.035	0.44±0.035	PUNCT
ijassa-659	285	1	0.65±0.054	0.65±0.054	PROPN
ijassa-659	286	1	0.60±0.042	0.60±0.042	PUNCT
ijassa-659	286	2	0.58±0.022	0.58±0.022	NOUN
ijassa-659	287	1	0.52±0.004	0.52±0.004	PUNCT
ijassa-659	287	2	0.45±0.035	0.45±0.035	PUNCT
ijassa-659	288	1	0.66±0.055	0.66±0.055	NOUN
ijassa-659	289	1	0.62±0.046	0.62±0.046	PUNCT
ijassa-659	289	2	0.59±0.048	0.59±0.048	PROPN
ijassa-659	290	1	0.54±0.022	0.54±0.022	PROPN
ijassa-659	291	1	0.48±0.038	0.48±0.038	PUNCT
ijassa-659	292	1	0.68±0.066	0.68±0.066	PUNCT
ijassa-659	292	2	0.64±0.048	0.64±0.048	PUNCT
ijassa-659	293	1	0.63±0.041	0.63±0.041	PROPN
ijassa-659	293	2	0.55±0.005	0.55±0.005	NOUN
ijassa-659	294	1	0.52±0.038	0.52±0.038	PROPN
ijassa-659	294	2	0.72±0.066	0.72±0.066	PROPN
ijassa-659	294	3	0.66±0.052	0.66±0.052	PUNCT
ijassa-659	295	1	0.65±0.022	0.65±0.022	NOUN
ijassa-659	296	1	0.55±0.005	0.55±0.005	NUM
ijassa-659	296	2	0.55±0.038	0.55±0.038	PROPN
ijassa-659	297	1	0.74±0.047	0.74±0.047	NUM
ijassa-659	298	1	0.69±0.052	0.69±0.052	PUNCT
ijassa-659	298	2	0.66±0.045	0.66±0.045	NUM
ijassa-659	298	3	0.57±0.015	0.57±0.015	PUNCT
ijassa-659	298	4	0.56±0.038	0.56±0.038	PROPN
ijassa-659	298	5	heart	heart	NOUN
ijassa-659	298	6	proposed	propose	VERB
ijassa-659	298	7	method	method	PROPN
ijassa-659	298	8	npu	npu	PROPN
ijassa-659	298	9	asc	asc	PROPN
ijassa-659	298	10	min	min	PROPN
ijassa-659	298	11	-	-	PROPN
ijassa-659	298	12	max	max	ADJ
ijassa-659	298	13	random	random	ADJ
ijassa-659	298	14	0.68±0.002	0.68±0.002	NOUN
ijassa-659	299	1	0.65±0.011	0.65±0.011	PUNCT
ijassa-659	299	2	0.57±0.024	0.57±0.024	NUM
ijassa-659	299	3	0.55±0.004	0.55±0.004	PUNCT
ijassa-659	299	4	0.48±0.025	0.48±0.025	PUNCT
ijassa-659	299	5	0.71±0.006	0.71±0.006	X
ijassa-659	300	1	0.66±0.044	0.66±0.044	PUNCT
ijassa-659	300	2	0.58±0.055	0.58±0.055	PUNCT
ijassa-659	301	1	0.57±0.012	0.57±0.012	PUNCT
ijassa-659	301	2	0.49±0.030	0.49±0.030	PROPN
ijassa-659	301	3	0.73±0.005	0.73±0.005	VERB
ijassa-659	301	4	0.67±00.72	0.67±00.72	NOUN
ijassa-659	301	5	0.59±0.044	0.59±0.044	PUNCT
ijassa-659	302	1	0.57±0.012	0.57±0.012	PUNCT
ijassa-659	303	1	0.52±0.030	0.52±0.030	NOUN
ijassa-659	304	1	0.75±0.005	0.75±0.005	INTJ
ijassa-659	304	2	0.68±0.007	0.68±0.007	PUNCT
ijassa-659	305	1	0.62±0.006	0.62±0.006	NUM
ijassa-659	306	1	0.57±0.014	0.57±0.014	PROPN
ijassa-659	306	2	0.55±0.033	0.55±0.033	PUNCT
ijassa-659	307	1	0.75±0.008	0.75±0.008	PUNCT
ijassa-659	307	2	0.70±0.036	0.70±0.036	NOUN
ijassa-659	307	3	0.63±0.008	0.63±0.008	NOUN
ijassa-659	307	4	0.58±0.015	0.58±0.015	PUNCT
ijassa-659	307	5	0.55±0.036	0.55±0.036	NOUN
ijassa-659	307	6	0.76±0.009	0.76±0.009	X
ijassa-659	307	7	0.71±0.074	0.71±0.074	PROPN
ijassa-659	307	8	0.64±0.009	0.64±0.009	X
ijassa-659	307	9	0.59±0.015	0.59±0.015	PUNCT
ijassa-659	307	10	0.56±0.036	0.56±0.036	X
ijassa-659	307	11	ionosphere	ionosphere	NOUN
ijassa-659	307	12	proposed	propose	VERB
ijassa-659	307	13	method	method	PROPN
ijassa-659	307	14	npu	npu	PROPN
ijassa-659	307	15	asc	asc	PROPN
ijassa-659	307	16	min	min	PROPN
ijassa-659	307	17	-	-	PROPN
ijassa-659	307	18	max	max	ADJ
ijassa-659	307	19	random	random	NOUN
ijassa-659	308	1	0.71±0.054	0.71±0.054	PUNCT
ijassa-659	309	1	0.65±0.001	0.65±0.001	PUNCT
ijassa-659	310	1	0.67±0.024	0.67±0.024	NOUN
ijassa-659	310	2	0.64±0.033	0.64±0.033	NOUN
ijassa-659	311	1	0.60±0.053	0.60±0.053	NOUN
ijassa-659	311	2	0.77±0.027	0.77±0.027	PROPN
ijassa-659	312	1	0.70±0.028	0.70±0.028	PROPN
ijassa-659	312	2	0.69±0.056	0.69±0.056	PROPN
ijassa-659	313	1	0.68±0.035	0.68±0.035	PROPN
ijassa-659	314	1	0.65±0.005	0.65±0.005	PROPN
ijassa-659	315	1	0.79±0.077	0.79±0.077	PROPN
ijassa-659	316	1	0.75±0.024	0.75±0.024	INTJ
ijassa-659	316	2	0.74±0.032	0.74±0.032	PROPN
ijassa-659	317	1	0.72±0.024	0.72±0.024	VERB
ijassa-659	317	2	0.66±0.006	0.66±0.006	NOUN
ijassa-659	317	3	0.84±0.030	0.84±0.030	NOUN
ijassa-659	318	1	0.83±0.043	0.83±0.043	NOUN
ijassa-659	319	1	0.81±0.064	0.81±0.064	PUNCT
ijassa-659	319	2	0.75±0.034	0.75±0.034	PUNCT
ijassa-659	320	1	0.61±0.009	0.61±0.009	PUNCT
ijassa-659	320	2	0.86±0.034	0.86±0.034	PUNCT
ijassa-659	321	1	0.86±0.025	0.86±0.025	PUNCT
ijassa-659	321	2	0.83±0.057	0.83±0.057	X
ijassa-659	321	3	0.77±0.084	0.77±0.084	PUNCT
ijassa-659	322	1	0.57±0.005	0.57±0.005	NUM
ijassa-659	323	1	0.87±0.077	0.87±0.077	PUNCT
ijassa-659	324	1	0.86±0.035	0.86±0.035	PUNCT
ijassa-659	325	1	0.84±0.065	0.84±0.065	NOUN
ijassa-659	326	1	0.80±0.075	0.80±0.075	PUNCT
ijassa-659	326	2	0.55±0.012	0.55±0.012	NUM
ijassa-659	326	3	breast	breast	NOUN
ijassa-659	326	4	proposed	propose	VERB
ijassa-659	326	5	method	method	PROPN
ijassa-659	326	6	npu	npu	PROPN
ijassa-659	326	7	asc	asc	PROPN
ijassa-659	326	8	min	min	PROPN
ijassa-659	326	9	-	-	PROPN
ijassa-659	326	10	max	max	PROPN
ijassa-659	326	11	random	random	NOUN
ijassa-659	327	1	0.74±0.015	0.74±0.015	PROPN
ijassa-659	328	1	0.67±0.012	0.67±0.012	PROPN
ijassa-659	328	2	0.66±0.018	0.66±0.018	NOUN
ijassa-659	329	1	0.64±0.044	0.64±0.044	PUNCT
ijassa-659	330	1	0.64±0.025	0.64±0.025	X
ijassa-659	330	2	0.83±0.015	0.83±0.015	X
ijassa-659	330	3	0.79±0.012	0.79±0.012	PROPN
ijassa-659	331	1	0.77±0.022	0.77±0.022	X
ijassa-659	332	1	0.77±0.044	0.77±0.044	PUNCT
ijassa-659	333	1	0.62±0.025	0.62±0.025	PUNCT
ijassa-659	333	2	0.88±0.015	0.88±0.015	PUNCT
ijassa-659	334	1	0.85±0.026	0.85±0.026	PUNCT
ijassa-659	334	2	0.84±0.018	0.84±0.018	PUNCT
ijassa-659	335	1	0.83±0.045	0.83±0.045	PUNCT
ijassa-659	335	2	0.63±0.028	0.63±0.028	PUNCT
ijassa-659	335	3	0.88±0.014	0.88±0.014	PUNCT
ijassa-659	336	1	0.87±0.022	0.87±0.022	NOUN
ijassa-659	337	1	0.84±0.041	0.84±0.041	NUM
ijassa-659	338	1	0.85±0.045	0.85±0.045	PUNCT
ijassa-659	338	2	0.61±0.028	0.61±0.028	PUNCT
ijassa-659	339	1	0.91±0.016	0.91±0.016	PROPN
ijassa-659	339	2	0.90±0.022	0.90±0.022	PUNCT
ijassa-659	339	3	0.85±0.034	0.85±0.034	PUNCT
ijassa-659	339	4	0.85±0.047	0.85±0.047	PUNCT
ijassa-659	340	1	0.62±0.028	0.62±0.028	ADJ
ijassa-659	341	1	0.91±0.016	0.91±0.016	PROPN
ijassa-659	341	2	0.90±0.022	0.90±0.022	PUNCT
ijassa-659	342	1	0.86±0.015	0.86±0.015	PUNCT
ijassa-659	342	2	0.85±0.048	0.85±0.048	PUNCT
ijassa-659	342	3	0.62±0.028	0.62±0.028	ADJ
ijassa-659	342	4	yeast	yeast	NOUN
ijassa-659	342	5	proposed	propose	VERB
ijassa-659	342	6	method	method	PROPN
ijassa-659	342	7	npu	npu	PROPN
ijassa-659	342	8	asc	asc	PROPN
ijassa-659	342	9	min	min	PROPN
ijassa-659	342	10	-	-	PROPN
ijassa-659	342	11	max	max	ADJ
ijassa-659	342	12	random	random	NOUN
ijassa-659	343	1	0.83±0.023	0.83±0.023	INTJ
ijassa-659	344	1	0.81±0.021	0.81±0.021	NOUN
ijassa-659	344	2	0.81±0.066	0.81±0.066	PROPN
ijassa-659	345	1	0.78±0.064	0.78±0.064	NOUN
ijassa-659	345	2	0.66±0.012	0.66±0.012	PUNCT
ijassa-659	346	1	0.85±0.035	0.85±0.035	PUNCT
ijassa-659	346	2	0.83±0.022	0.83±0.022	X
ijassa-659	347	1	0.82±0.066	0.82±0.066	PUNCT
ijassa-659	348	1	0.79±0.064	0.79±0.064	PUNCT
ijassa-659	348	2	0.70±0.012	0.70±0.012	NUM
ijassa-659	349	1	0.88±0.039	0.88±0.039	NUM
ijassa-659	350	1	0.84±00.22	0.84±00.22	X
ijassa-659	350	2	0.84±0.064	0.84±0.064	X
ijassa-659	351	1	0.81±0.068	0.81±0.068	PUNCT
ijassa-659	352	1	0.74±0.013	0.74±0.013	NOUN
ijassa-659	352	2	0.91±0.045	0.91±0.045	NOUN
ijassa-659	352	3	0.86±0.023	0.86±0.023	PUNCT
ijassa-659	353	1	0.85±0.067	0.85±0.067	PUNCT
ijassa-659	353	2	0.82±0.067	0.82±0.067	PUNCT
ijassa-659	354	1	0.77±0.014	0.77±0.014	X
ijassa-659	354	2	0.93±0.067	0.93±0.067	NUM
ijassa-659	354	3	0.89±0.024	0.89±0.024	NUM
ijassa-659	354	4	0.87±0.067	0.87±0.067	PUNCT
ijassa-659	355	1	0.84±0.067	0.84±0.067	PROPN
ijassa-659	355	2	0.73±0.013	0.73±0.013	PUNCT
ijassa-659	356	1	0.94±0.078	0.94±0.078	PROPN
ijassa-659	356	2	0.91±0.024	0.91±0.024	NOUN
ijassa-659	357	1	0.88±0.068	0.88±0.068	PUNCT
ijassa-659	357	2	0.86±0.069	0.86±0.069	NOUN
ijassa-659	357	3	0.78±0.014	0.78±0.014	NUM
ijassa-659	357	4	segment	segment	NOUN
ijassa-659	357	5	proposed	propose	VERB
ijassa-659	357	6	method	method	PROPN
ijassa-659	357	7	npu	npu	PROPN
ijassa-659	357	8	asc	asc	PROPN
ijassa-659	357	9	min	min	PROPN
ijassa-659	357	10	-	-	PROPN
ijassa-659	357	11	max	max	ADJ
ijassa-659	357	12	random	random	ADJ
ijassa-659	358	1	0.61±0.024	0.61±0.024	PROPN
ijassa-659	359	1	0.57±0.031	0.57±0.031	PROPN
ijassa-659	359	2	0.64±0.081	0.64±0.081	PROPN
ijassa-659	359	3	0.55±0.014	0.55±0.014	PUNCT
ijassa-659	360	1	0.52±0.024	0.52±0.024	PUNCT
ijassa-659	360	2	0.65±0.024	0.65±0.024	PUNCT
ijassa-659	361	1	0.59±0.032	0.59±0.032	NOUN
ijassa-659	361	2	0.66±0.004	0.66±0.004	PUNCT
ijassa-659	362	1	0.55±0.015	0.55±0.015	PUNCT
ijassa-659	362	2	0.54±0.025	0.54±0.025	PUNCT
ijassa-659	363	1	0.68±0.025	0.68±0.025	NUM
ijassa-659	363	2	0.61±0.035	0.61±0.035	PUNCT
ijassa-659	364	1	0.66±0.022	0.66±0.022	PUNCT
ijassa-659	365	1	0.54±0.017	0.54±0.017	NUM
ijassa-659	365	2	0.54±0.025	0.54±0.025	NUM
ijassa-659	365	3	0.69±0.026	0.69±0.026	PUNCT
ijassa-659	366	1	0.62±0.035	0.62±0.035	X
ijassa-659	367	1	0.67±0.024	0.67±0.024	NOUN
ijassa-659	367	2	0.54±0.017	0.54±0.017	NUM
ijassa-659	368	1	0.52±0.024	0.52±0.024	NUM
ijassa-659	368	2	0.70±0.027	0.70±0.027	PROPN
ijassa-659	368	3	0.63±0.038	0.63±0.038	PUNCT
ijassa-659	369	1	0.68±0.037	0.68±0.037	PROPN
ijassa-659	369	2	0.53±0.020	0.53±0.020	PROPN
ijassa-659	370	1	0.51±0.024	0.51±0.024	NOUN
ijassa-659	370	2	0.71±0.027	0.71±0.027	NUM
ijassa-659	370	3	0.64±0.041	0.64±0.041	NOUN
ijassa-659	370	4	0.68±0.038	0.68±0.038	PUNCT
ijassa-659	371	1	0.53±0.022	0.53±0.022	PROPN
ijassa-659	371	2	0.51±0.027	0.51±0.027	PROPN
ijassa-659	371	3	digit-389	digit-389	NOUN
ijassa-659	371	4	proposed	propose	VERB
ijassa-659	371	5	method	method	PROPN
ijassa-659	371	6	npu	npu	PROPN
ijassa-659	371	7	asc	asc	PROPN
ijassa-659	371	8	min	min	PROPN
ijassa-659	371	9	-	-	PROPN
ijassa-659	371	10	max	max	ADJ
ijassa-659	371	11	random	random	ADJ
ijassa-659	371	12	0.73±0.025	0.73±0.025	PROPN
ijassa-659	372	1	0.68±0.018	0.68±0.018	PUNCT
ijassa-659	373	1	0.66±0.022	0.66±0.022	X
ijassa-659	373	2	0.65±0.014	0.65±0.014	PUNCT
ijassa-659	373	3	0.70±0.055	0.70±0.055	NOUN
ijassa-659	374	1	0.75±0.025	0.75±0.025	PRON
ijassa-659	374	2	0.70±0.018	0.70±0.018	PUNCT
ijassa-659	374	3	0.69±0.022	0.69±0.022	NOUN
ijassa-659	374	4	0.68±0.014	0.68±0.014	X
ijassa-659	374	5	0.73±0.055	0.73±0.055	PUNCT
ijassa-659	374	6	0.77±0.025	0.77±0.025	NUM
ijassa-659	375	1	0.75±0.019	0.75±0.019	PUNCT
ijassa-659	375	2	0.71±0.024	0.71±0.024	PUNCT
ijassa-659	376	1	0.70±0.015	0.70±0.015	PUNCT
ijassa-659	377	1	0.75±0.058	0.75±0.058	NOUN
ijassa-659	378	1	0.84±0.025	0.84±0.025	PUNCT
ijassa-659	379	1	0.81±0.021	0.81±0.021	NUM
ijassa-659	379	2	0.77±0.031	0.77±0.031	NOUN
ijassa-659	380	1	0.73±0.015	0.73±0.015	NUM
ijassa-659	380	2	0.70±0.058	0.70±0.058	NUM
ijassa-659	381	1	0.86±0.026	0.86±0.026	PUNCT
ijassa-659	382	1	0.83±0.022	0.83±0.022	PUNCT
ijassa-659	383	1	0.79±0.035	0.79±0.035	PUNCT
ijassa-659	384	1	0.74±0.017	0.74±0.017	PROPN
ijassa-659	384	2	0.72	0.72	NUM
ijassa-659	384	3	±0.045	±0.045	NOUN
ijassa-659	385	1	0.88±0.026	0.88±0.026	PROPN
ijassa-659	386	1	0.84±0.023	0.84±0.023	NOUN
ijassa-659	387	1	0.81±0.035	0.81±0.035	NOUN
ijassa-659	387	2	0.75±0.018	0.75±0.018	PUNCT
ijassa-659	388	1	0.73±0.046	0.73±0.046	ADJ
ijassa-659	388	2	magic	magic	NOUN
ijassa-659	388	3	proposed	propose	VERB
ijassa-659	388	4	method	method	PROPN
ijassa-659	388	5	npu	npu	PROPN
ijassa-659	388	6	asc	asc	PROPN
ijassa-659	388	7	min	min	PROPN
ijassa-659	388	8	-	-	PROPN
ijassa-659	388	9	max	max	PROPN
ijassa-659	388	10	random	random	NOUN
ijassa-659	389	1	0.69±0.045	0.69±0.045	X
ijassa-659	389	2	0.69±0.062	0.69±0.062	PUNCT
ijassa-659	390	1	0.65±0.011	0.65±0.011	PUNCT
ijassa-659	391	1	0.61±0.057	0.61±0.057	PUNCT
ijassa-659	391	2	0.60±0.074	0.60±0.074	PROPN
ijassa-659	391	3	0.74±0.045	0.74±0.045	VERB
ijassa-659	392	1	0.68±0.062	0.68±0.062	PRON
ijassa-659	392	2	0.68±0.015	0.68±0.015	PUNCT
ijassa-659	393	1	0.61±0.057	0.61±0.057	PUNCT
ijassa-659	394	1	0.60±0.074	0.60±0.074	X
ijassa-659	394	2	0.75±0.046	0.75±0.046	NOUN
ijassa-659	395	1	0.69±0.065	0.69±0.065	NOUN
ijassa-659	396	1	0.71±0.017	0.71±0.017	ADP
ijassa-659	396	2	0.62±0.057	0.62±0.057	NUM
ijassa-659	396	3	0.61±0.074	0.61±0.074	PROPN
ijassa-659	396	4	0.75±0.041	0.75±0.041	NOUN
ijassa-659	396	5	0.70±0.065	0.70±0.065	NUM
ijassa-659	397	1	0.72±0.025	0.72±0.025	PUNCT
ijassa-659	397	2	0.61±0.057	0.61±0.057	PUNCT
ijassa-659	398	1	0.61±0.078	0.61±0.078	NOUN
ijassa-659	399	1	0.76±0.041	0.76±0.041	NOUN
ijassa-659	399	2	0.71±0.066	0.71±0.066	PUNCT
ijassa-659	400	1	0.71±0.018	0.71±0.018	NOUN
ijassa-659	401	1	0.61±0.059	0.61±0.059	PUNCT
ijassa-659	402	1	0.60±0.078	0.60±0.078	NOUN
ijassa-659	402	2	0.76±0.039	0.76±0.039	NOUN
ijassa-659	402	3	0.72±0.067	0.72±0.067	NUM
ijassa-659	402	4	0.71±0.018	0.71±0.018	NOUN
ijassa-659	402	5	0.61±0.062	0.61±0.062	X
ijassa-659	402	6	0.60±0.079	0.60±0.079	PROPN
ijassa-659	402	7	38	38	NUM
ijassa-659	402	8	w.	w.	NOUN
ijassa-659	402	9	atwa	atwa	PROPN
ijassa-659	402	10	,	,	PUNCT
ijassa-659	402	11	m.	m.	NOUN
ijassa-659	402	12	emam	emam	PROPN
ijassa-659	402	13	copyright	copyright	NOUN
ijassa-659	402	14	©	©	PROPN
ijassa-659	402	15	2019	2019	NUM
ijassa-659	402	16	assa	assa	PROPN
ijassa-659	402	17	adv	adv	PROPN
ijassa-659	402	18	.	.	PUNCT
ijassa-659	403	1	in	in	ADP
ijassa-659	403	2	systems	system	NOUN
ijassa-659	403	3	science	science	NOUN
ijassa-659	403	4	and	and	CCONJ
ijassa-659	403	5	appl	appl	NOUN
ijassa-659	403	6	.	.	PUNCT
ijassa-659	404	1	(	(	PUNCT
ijassa-659	404	2	2019	2019	NUM
ijassa-659	404	3	)	)	PUNCT
ijassa-659	404	4	table	table	NOUN
ijassa-659	404	5	2	2	NUM
ijassa-659	404	6	:	:	PUNCT
ijassa-659	404	7	comparison	comparison	NOUN
ijassa-659	404	8	on	on	ADP
ijassa-659	404	9	pairwise	pairwise	NOUN
ijassa-659	404	10	f	f	NOUN
ijassa-659	404	11	-	-	PUNCT
ijassa-659	404	12	measure	measure	NOUN
ijassa-659	404	13	(	(	PUNCT
ijassa-659	404	14	mean	mean	NOUN
ijassa-659	404	15	±	±	NUM
ijassa-659	404	16	std	std	NOUN
ijassa-659	404	17	)	)	PUNCT
ijassa-659	404	18	with	with	ADP
ijassa-659	404	19	ahcc	ahcc	PROPN
ijassa-659	404	20	algorithm	algorithm	NOUN
ijassa-659	404	21	.	.	PUNCT
ijassa-659	405	1	dataset	dataset	ADJ
ijassa-659	405	2	constraint	constraint	NOUN
ijassa-659	405	3	selection	selection	NOUN
ijassa-659	405	4	heuristics	heuristic	NOUN
ijassa-659	405	5	number	number	NOUN
ijassa-659	405	6	of	of	ADP
ijassa-659	405	7	queries	query	NOUN
ijassa-659	405	8	25	25	NUM
ijassa-659	405	9	50	50	NUM
ijassa-659	405	10	75	75	NUM
ijassa-659	405	11	100	100	NUM
ijassa-659	405	12	125	125	NUM
ijassa-659	405	13	150	150	NUM
ijassa-659	405	14	protein	protein	NOUN
ijassa-659	405	15	proposed	propose	VERB
ijassa-659	405	16	method	method	PROPN
ijassa-659	405	17	npu	npu	PROPN
ijassa-659	405	18	asc	asc	PROPN
ijassa-659	405	19	min	min	PROPN
ijassa-659	405	20	-	-	PROPN
ijassa-659	405	21	max	max	PROPN
ijassa-659	405	22	random	random	PROPN
ijassa-659	405	23	0.64±0.034	0.64±0.034	PUNCT
ijassa-659	406	1	0.64±0.061	0.64±0.061	PROPN
ijassa-659	407	1	0.63±0.025	0.63±0.025	PROPN
ijassa-659	407	2	0.64±0.014	0.64±0.014	PROPN
ijassa-659	407	3	0.67±0.002	0.67±0.002	PUNCT
ijassa-659	408	1	0.68±0.034	0.68±0.034	PUNCT
ijassa-659	409	1	0.66±0.063	0.66±0.063	NOUN
ijassa-659	409	2	0.64±0.022	0.64±0.022	PUNCT
ijassa-659	410	1	0.66±0.014	0.66±0.014	PUNCT
ijassa-659	411	1	0.66±0.008	0.66±0.008	NOUN
ijassa-659	411	2	0.71±0.033	0.71±0.033	PROPN
ijassa-659	412	1	0.69±0.063	0.69±0.063	PROPN
ijassa-659	412	2	0.65±0.027	0.65±0.027	PROPN
ijassa-659	413	1	0.68±0.015	0.68±0.015	PUNCT
ijassa-659	413	2	0.68±0.008	0.68±0.008	PUNCT
ijassa-659	413	3	0.73±0.035	0.73±0.035	PROPN
ijassa-659	414	1	0.71±0.066	0.71±0.066	PRON
ijassa-659	415	1	0.65±0.025	0.65±0.025	PUNCT
ijassa-659	415	2	0.70±0.013	0.70±0.013	PUNCT
ijassa-659	416	1	0.64±0.008	0.64±0.008	PUNCT
ijassa-659	416	2	0.74±0.035	0.74±0.035	PUNCT
ijassa-659	417	1	0.73±0.064	0.73±0.064	PUNCT
ijassa-659	417	2	0.66±0.022	0.66±0.022	PUNCT
ijassa-659	417	3	0.71±0.014	0.71±0.014	PUNCT
ijassa-659	418	1	0.66±0.011	0.66±0.011	PUNCT
ijassa-659	418	2	0.77±0.036	0.77±0.036	NOUN
ijassa-659	419	1	0.74±0.063	0.74±0.063	ADJ
ijassa-659	419	2	0.67±0.022	0.67±0.022	X
ijassa-659	419	3	0.73±0.014	0.73±0.014	PROPN
ijassa-659	419	4	0.65±0.012	0.65±0.012	PROPN
ijassa-659	419	5	heart	heart	NOUN
ijassa-659	419	6	proposed	propose	VERB
ijassa-659	419	7	method	method	PROPN
ijassa-659	419	8	npu	npu	PROPN
ijassa-659	419	9	asc	asc	PROPN
ijassa-659	419	10	min	min	PROPN
ijassa-659	419	11	-	-	PROPN
ijassa-659	419	12	max	max	PROPN
ijassa-659	419	13	random	random	ADJ
ijassa-659	419	14	0.70±0.043	0.70±0.043	X
ijassa-659	420	1	0.67±0.012	0.67±0.012	PROPN
ijassa-659	420	2	0.68±0.026	0.68±0.026	PUNCT
ijassa-659	421	1	0.67±0.035	0.67±0.035	PUNCT
ijassa-659	422	1	0.54±0.072	0.54±0.072	X
ijassa-659	422	2	0.74±0.044	0.74±0.044	PUNCT
ijassa-659	422	3	0.72±0.012	0.72±0.012	PROPN
ijassa-659	422	4	0.71±0.022	0.71±0.022	PUNCT
ijassa-659	423	1	0.73±0.045	0.73±0.045	VERB
ijassa-659	423	2	0.51±0.062	0.51±0.062	X
ijassa-659	424	1	0.77±0.042	0.77±0.042	NOUN
ijassa-659	424	2	0.75±0.014	0.75±0.014	NOUN
ijassa-659	424	3	0.74±0.033	0.74±0.033	NOUN
ijassa-659	424	4	0.75±0.054	0.75±0.054	PUNCT
ijassa-659	425	1	0.55±0.015	0.55±0.015	PUNCT
ijassa-659	425	2	0.79±0.043	0.79±0.043	PUNCT
ijassa-659	426	1	0.78±0.013	0.78±0.013	PUNCT
ijassa-659	426	2	0.78±0.064	0.78±0.064	PUNCT
ijassa-659	427	1	0.77±0.003	0.77±0.003	NOUN
ijassa-659	427	2	0.53±0.014	0.53±0.014	PUNCT
ijassa-659	428	1	0.83±0.043	0.83±0.043	NOUN
ijassa-659	428	2	0.83±0.013	0.83±0.013	PUNCT
ijassa-659	429	1	0.82±0.021	0.82±0.021	PRON
ijassa-659	430	1	0.82±0.042	0.82±0.042	NUM
ijassa-659	430	2	0.55±0.016	0.55±0.016	PUNCT
ijassa-659	431	1	0.86±0.043	0.86±0.043	INTJ
ijassa-659	432	1	0.85±0.012	0.85±0.012	INTJ
ijassa-659	433	1	0.84±0.032	0.84±0.032	NUM
ijassa-659	434	1	0.83±0.008	0.83±0.008	PROPN
ijassa-659	435	1	0.52±0.035	0.52±0.035	PROPN
ijassa-659	435	2	ionosphere	ionosphere	NOUN
ijassa-659	435	3	proposed	propose	VERB
ijassa-659	435	4	method	method	PROPN
ijassa-659	435	5	npu	npu	PROPN
ijassa-659	435	6	asc	asc	PROPN
ijassa-659	435	7	min	min	PROPN
ijassa-659	435	8	-	-	PROPN
ijassa-659	435	9	max	max	ADJ
ijassa-659	435	10	random	random	ADJ
ijassa-659	435	11	0.75±0.023	0.75±0.023	PROPN
ijassa-659	435	12	0.74±0.031	0.74±0.031	PUNCT
ijassa-659	436	1	0.72±0.024	0.72±0.024	NOUN
ijassa-659	436	2	0.68±0.031	0.68±0.031	X
ijassa-659	436	3	0.65±0.003	0.65±0.003	PUNCT
ijassa-659	437	1	0.78±0.023	0.78±0.023	PROPN
ijassa-659	437	2	0.77±0.038	0.77±0.038	NOUN
ijassa-659	438	1	0.75±0.026	0.75±0.026	PUNCT
ijassa-659	439	1	0.69±0.031	0.69±0.031	PUNCT
ijassa-659	440	1	0.65±0.005	0.65±0.005	PUNCT
ijassa-659	441	1	0.81±0.024	0.81±0.024	NOUN
ijassa-659	441	2	0.80±0.034	0.80±0.034	X
ijassa-659	442	1	0.77±0.022	0.77±0.022	X
ijassa-659	443	1	0.74±0.032	0.74±0.032	PUNCT
ijassa-659	443	2	0.68±0.006	0.68±0.006	PUNCT
ijassa-659	444	1	0.84±0.024	0.84±0.024	NOUN
ijassa-659	444	2	0.83±0.033	0.83±0.033	NOUN
ijassa-659	445	1	0.81±0.024	0.81±0.024	NOUN
ijassa-659	445	2	0.75±0.034	0.75±0.034	PUNCT
ijassa-659	446	1	0.64±0.007	0.64±0.007	VERB
ijassa-659	447	1	0.87±0.025	0.87±0.025	PROPN
ijassa-659	447	2	0.87±0.035	0.87±0.035	PUNCT
ijassa-659	448	1	0.84±0.027	0.84±0.027	PUNCT
ijassa-659	448	2	0.77±0.034	0.77±0.034	PUNCT
ijassa-659	449	1	0.65±0.009	0.65±0.009	X
ijassa-659	449	2	0.90±0.030	0.90±0.030	NOUN
ijassa-659	450	1	0.89±0.035	0.89±0.035	PUNCT
ijassa-659	451	1	0.86±0.025	0.86±0.025	PUNCT
ijassa-659	452	1	0.78±0.035	0.78±0.035	INTJ
ijassa-659	452	2	0.67±0.010	0.67±0.010	NOUN
ijassa-659	452	3	breast	breast	NOUN
ijassa-659	452	4	proposed	propose	VERB
ijassa-659	452	5	method	method	PROPN
ijassa-659	452	6	npu	npu	PROPN
ijassa-659	452	7	asc	asc	PROPN
ijassa-659	452	8	min	min	PROPN
ijassa-659	452	9	-	-	PROPN
ijassa-659	452	10	max	max	ADJ
ijassa-659	452	11	random	random	ADJ
ijassa-659	452	12	0.81±0.038	0.81±0.038	PROPN
ijassa-659	453	1	0.81±0.021	0.81±0.021	PUNCT
ijassa-659	454	1	0.77±0.054	0.77±0.054	PUNCT
ijassa-659	455	1	0.71±0.084	0.71±0.084	NOUN
ijassa-659	455	2	0.66±0.005	0.66±0.005	PUNCT
ijassa-659	456	1	0.83±0.038	0.83±0.038	PUNCT
ijassa-659	456	2	0.82±0.022	0.82±0.022	PUNCT
ijassa-659	457	1	0.78±0.054	0.78±0.054	PUNCT
ijassa-659	458	1	0.71±0.084	0.71±0.084	NOUN
ijassa-659	458	2	0.65±0.005	0.65±0.005	PUNCT
ijassa-659	459	1	0.85±0.039	0.85±0.039	NOUN
ijassa-659	459	2	0.83±00.22	0.83±00.22	NUM
ijassa-659	459	3	0.79±0.055	0.79±0.055	X
ijassa-659	460	1	0.74±0.088	0.74±0.088	NOUN
ijassa-659	460	2	0.60±0.006	0.60±0.006	NOUN
ijassa-659	460	3	0.87±0.040	0.87±0.040	PUNCT
ijassa-659	461	1	0.84±0.023	0.84±0.023	NOUN
ijassa-659	461	2	0.81±0.056	0.81±0.056	PROPN
ijassa-659	461	3	0.75±0.087	0.75±0.087	NOUN
ijassa-659	461	4	0.59±0.007	0.59±0.007	VERB
ijassa-659	462	1	0.89±0.042	0.89±0.042	NOUN
ijassa-659	462	2	0.87±0.024	0.87±0.024	NOUN
ijassa-659	462	3	0.83±0.057	0.83±0.057	X
ijassa-659	462	4	0.76±0.087	0.76±0.087	NOUN
ijassa-659	462	5	0.57±0.008	0.57±0.008	PUNCT
ijassa-659	463	1	0.90±0.042	0.90±0.042	PROPN
ijassa-659	463	2	0.88±0.024	0.88±0.024	NOUN
ijassa-659	463	3	0.84±0.057	0.84±0.057	PUNCT
ijassa-659	463	4	0.77±0.089	0.77±0.089	NOUN
ijassa-659	463	5	0.55±0.008	0.55±0.008	NOUN
ijassa-659	463	6	yeast	yeast	NOUN
ijassa-659	463	7	proposed	propose	VERB
ijassa-659	463	8	method	method	PROPN
ijassa-659	463	9	npu	npu	PROPN
ijassa-659	463	10	asc	asc	PROPN
ijassa-659	463	11	min	min	PROPN
ijassa-659	463	12	-	-	PROPN
ijassa-659	463	13	max	max	ADJ
ijassa-659	463	14	random	random	PROPN
ijassa-659	464	1	0.82±0.043	0.82±0.043	PUNCT
ijassa-659	465	1	0.84±0.051	0.84±0.051	PUNCT
ijassa-659	465	2	0.80±0.066	0.80±0.066	PUNCT
ijassa-659	465	3	0.80±0.074	0.80±0.074	PROPN
ijassa-659	466	1	0.73±0.022	0.73±0.022	X
ijassa-659	467	1	0.84±0.045	0.84±0.045	X
ijassa-659	467	2	0.86±0.052	0.86±0.052	X
ijassa-659	467	3	0.83±0.066	0.83±0.066	PUNCT
ijassa-659	468	1	0.83±0.074	0.83±0.074	X
ijassa-659	468	2	0.77±0.022	0.77±0.022	PUNCT
ijassa-659	469	1	0.87±0.049	0.87±0.049	PUNCT
ijassa-659	469	2	0.87±00.52	0.87±00.52	NOUN
ijassa-659	469	3	0.84±0.064	0.84±0.064	NUM
ijassa-659	470	1	0.84±0.078	0.84±0.078	PROPN
ijassa-659	470	2	0.79±0.023	0.79±0.023	PUNCT
ijassa-659	470	3	0.90±0.045	0.90±0.045	PUNCT
ijassa-659	470	4	0.90±0.023	0.90±0.023	PROPN
ijassa-659	471	1	0.85±0.067	0.85±0.067	NUM
ijassa-659	471	2	0.87±0.067	0.87±0.067	PROPN
ijassa-659	471	3	0.82±0.014	0.82±0.014	NOUN
ijassa-659	471	4	0.92±0.067	0.92±0.067	PUNCT
ijassa-659	472	1	0.91±0.024	0.91±0.024	NOUN
ijassa-659	472	2	0.86±0.067	0.86±0.067	X
ijassa-659	472	3	0.88±0.067	0.88±0.067	NOUN
ijassa-659	473	1	0.84±0.013	0.84±0.013	NOUN
ijassa-659	473	2	0.93±0.078	0.93±0.078	PROPN
ijassa-659	473	3	0.92±0.024	0.92±0.024	PUNCT
ijassa-659	474	1	0.87±0.068	0.87±0.068	PRON
ijassa-659	474	2	0.89±0.069	0.89±0.069	PROPN
ijassa-659	474	3	0.86±0.014	0.86±0.014	PROPN
ijassa-659	474	4	segment	segment	NOUN
ijassa-659	474	5	proposed	propose	VERB
ijassa-659	474	6	method	method	PROPN
ijassa-659	474	7	npu	npu	PROPN
ijassa-659	474	8	asc	asc	PROPN
ijassa-659	474	9	min	min	PROPN
ijassa-659	474	10	-	-	PROPN
ijassa-659	474	11	max	max	PROPN
ijassa-659	474	12	random	random	NOUN
ijassa-659	475	1	0.60±0.035	0.60±0.035	X
ijassa-659	476	1	0.58±0.017	0.58±0.017	X
ijassa-659	476	2	0.61±0.034	0.61±0.034	PUNCT
ijassa-659	477	1	0.57±0.045	0.57±0.045	PRON
ijassa-659	478	1	0.52±0.012	0.52±0.012	NOUN
ijassa-659	478	2	0.61±0.037	0.61±0.037	NOUN
ijassa-659	478	3	0.59±0.017	0.59±0.017	PUNCT
ijassa-659	478	4	0.62±0.028	0.62±0.028	PROPN
ijassa-659	478	5	0.58±0.045	0.58±0.045	NOUN
ijassa-659	478	6	0.54±0.012	0.54±0.012	NOUN
ijassa-659	478	7	0.63±0.059	0.63±0.059	PUNCT
ijassa-659	479	1	0.60±0.018	0.60±0.018	NOUN
ijassa-659	479	2	0.62±0.030	0.62±0.030	NUM
ijassa-659	480	1	0.58±0.045	0.58±0.045	NOUN
ijassa-659	480	2	0.53±0.013	0.53±0.013	X
ijassa-659	480	3	0.65±0.024	0.65±0.024	PUNCT
ijassa-659	481	1	0.60±0.032	0.60±0.032	PUNCT
ijassa-659	481	2	0.62±0.004	0.62±0.004	PUNCT
ijassa-659	482	1	0.59±0.025	0.59±0.025	NOUN
ijassa-659	483	1	0.51±0.025	0.51±0.025	NOUN
ijassa-659	484	1	0.67±0.037	0.67±0.037	INTJ
ijassa-659	485	1	0.62±0.053	0.62±0.053	INTJ
ijassa-659	486	1	0.62±0.021	0.62±0.021	NOUN
ijassa-659	487	1	0.60±0.055	0.60±0.055	NUM
ijassa-659	487	2	0.50±0.014	0.50±0.014	PROPN
ijassa-659	487	3	0.69±0.026	0.69±0.026	PUNCT
ijassa-659	488	1	0.63±0.035	0.63±0.035	PROPN
ijassa-659	488	2	0.63±0.054	0.63±0.054	PUNCT
ijassa-659	489	1	0.60±0.067	0.60±0.067	NOUN
ijassa-659	490	1	0.48±0.024	0.48±0.024	DET
ijassa-659	490	2	digit-389	digit-389	NOUN
ijassa-659	490	3	proposed	propose	VERB
ijassa-659	490	4	method	method	PROPN
ijassa-659	490	5	npu	npu	PROPN
ijassa-659	490	6	asc	asc	PROPN
ijassa-659	490	7	min	min	PROPN
ijassa-659	490	8	-	-	PROPN
ijassa-659	490	9	max	max	ADJ
ijassa-659	490	10	random	random	ADJ
ijassa-659	490	11	0.80±0.055	0.80±0.055	NOUN
ijassa-659	490	12	0.81±0.001	0.81±0.001	PUNCT
ijassa-659	491	1	0.78±0.026	0.78±0.026	PUNCT
ijassa-659	491	2	0.76±0.044	0.76±0.044	X
ijassa-659	492	1	0.70±0.012	0.70±0.012	PUNCT
ijassa-659	493	1	0.83±0.055	0.83±0.055	PRON
ijassa-659	493	2	0.83±0.002	0.83±0.002	PUNCT
ijassa-659	494	1	0.81±0.026	0.81±0.026	PUNCT
ijassa-659	494	2	0.79±0.044	0.79±0.044	X
ijassa-659	495	1	0.70±0.012	0.70±0.012	ADJ
ijassa-659	495	2	0.86±0.056	0.86±0.056	PROPN
ijassa-659	495	3	0.85±0.002	0.85±0.002	PUNCT
ijassa-659	496	1	0.83±0.025	0.83±0.025	ADP
ijassa-659	496	2	0.81±0.048	0.81±0.048	PROPN
ijassa-659	496	3	0.74±0.013	0.74±0.013	NOUN
ijassa-659	496	4	0.88±0.055	0.88±0.055	NOUN
ijassa-659	496	5	0.86±0.003	0.86±0.003	PUNCT
ijassa-659	496	6	0.85±0.027	0.85±0.027	PUNCT
ijassa-659	497	1	0.82±0.047	0.82±0.047	PRON
ijassa-659	497	2	0.77±0.014	0.77±0.014	PUNCT
ijassa-659	498	1	0.91±0.057	0.91±0.057	NUM
ijassa-659	498	2	0.89±0.004	0.89±0.004	PUNCT
ijassa-659	498	3	0.86±0.027	0.86±0.027	PUNCT
ijassa-659	499	1	0.84±0.047	0.84±0.047	NOUN
ijassa-659	499	2	0.78±0.013	0.78±0.013	NOUN
ijassa-659	499	3	0.92±0.055	0.92±0.055	PUNCT
ijassa-659	500	1	0.91±0.004	0.91±0.004	PUNCT
ijassa-659	500	2	0.88±0.028	0.88±0.028	NUM
ijassa-659	500	3	0.86±0.049	0.86±0.049	PRON
ijassa-659	500	4	0.82±0.014	0.82±0.014	PROPN
ijassa-659	500	5	magic	magic	NOUN
ijassa-659	500	6	proposed	propose	VERB
ijassa-659	500	7	method	method	PROPN
ijassa-659	500	8	npu	npu	PROPN
ijassa-659	500	9	asc	asc	PROPN
ijassa-659	500	10	min	min	PROPN
ijassa-659	500	11	-	-	PROPN
ijassa-659	500	12	max	max	ADJ
ijassa-659	500	13	random	random	NOUN
ijassa-659	501	1	0.64±0.033	0.64±0.033	PUNCT
ijassa-659	501	2	0.58±0.007	0.58±0.007	PUNCT
ijassa-659	502	1	0.57±0.024	0.57±0.024	NOUN
ijassa-659	502	2	0.56±0.005	0.56±0.005	NUM
ijassa-659	503	1	0.48±0.012	0.48±0.012	NUM
ijassa-659	504	1	0.70±0.035	0.70±0.035	PUNCT
ijassa-659	505	1	0.62±0.007	0.62±0.007	INTJ
ijassa-659	506	1	0.58±0.028	0.58±0.028	NOUN
ijassa-659	506	2	0.58±0.005	0.58±0.005	PUNCT
ijassa-659	507	1	0.51±0.012	0.51±0.012	NOUN
ijassa-659	507	2	0.72±0.035	0.72±0.035	PUNCT
ijassa-659	508	1	0.66±0.008	0.66±0.008	PUNCT
ijassa-659	508	2	0.61±0.030	0.61±0.030	NOUN
ijassa-659	508	3	0.58±0.005	0.58±0.005	PUNCT
ijassa-659	509	1	0.55±0.013	0.55±0.013	PUNCT
ijassa-659	509	2	0.74±0.036	0.74±0.036	PUNCT
ijassa-659	510	1	0.67±0.012	0.67±0.012	PROPN
ijassa-659	510	2	0.64±0.011	0.64±0.011	PROPN
ijassa-659	511	1	0.61±0.004	0.61±0.004	PUNCT
ijassa-659	511	2	0.51±0.014	0.51±0.014	PUNCT
ijassa-659	512	1	0.75±0.037	0.75±0.037	X
ijassa-659	512	2	0.68±0.013	0.68±0.013	PUNCT
ijassa-659	513	1	0.66±0.021	0.66±0.021	INTJ
ijassa-659	514	1	0.62±0.005	0.62±0.005	NOUN
ijassa-659	514	2	0.49±0.014	0.49±0.014	PUNCT
ijassa-659	515	1	0.77±0.038	0.77±0.038	NUM
ijassa-659	516	1	0.69±0.012	0.69±0.012	PROPN
ijassa-659	516	2	0.67±0.022	0.67±0.022	PROPN
ijassa-659	517	1	0.62±0.008	0.62±0.008	X
ijassa-659	518	1	0.47±0.015	0.47±0.015	NOUN
ijassa-659	518	2	4.5.2	4.5.2	NUM
ijassa-659	518	3	efficiency	efficiency	NOUN
ijassa-659	518	4	and	and	CCONJ
ijassa-659	518	5	scalability	scalability	NOUN
ijassa-659	518	6	analysis	analysis	NOUN
ijassa-659	518	7	one	one	NUM
ijassa-659	518	8	of	of	ADP
ijassa-659	518	9	the	the	DET
ijassa-659	518	10	primary	primary	ADJ
ijassa-659	518	11	motivations	motivation	NOUN
ijassa-659	518	12	of	of	ADP
ijassa-659	518	13	the	the	DET
ijassa-659	518	14	proposed	propose	VERB
ijassa-659	518	15	active	active	ADJ
ijassa-659	518	16	learning	learning	NOUN
ijassa-659	518	17	method	method	NOUN
ijassa-659	518	18	is	be	AUX
ijassa-659	518	19	to	to	PART
ijassa-659	518	20	reduce	reduce	VERB
ijassa-659	518	21	the	the	DET
ijassa-659	518	22	amount	amount	NOUN
ijassa-659	518	23	of	of	ADP
ijassa-659	518	24	computation	computation	NOUN
ijassa-659	518	25	in	in	ADP
ijassa-659	518	26	iterative	iterative	ADJ
ijassa-659	518	27	active	active	ADJ
ijassa-659	518	28	learning	learning	NOUN
ijassa-659	518	29	.	.	PUNCT
ijassa-659	519	1	figure	figure	VERB
ijassa-659	519	2	4	4	NUM
ijassa-659	519	3	shows	show	VERB
ijassa-659	519	4	the	the	DET
ijassa-659	519	5	average	average	ADJ
ijassa-659	519	6	cpu	cpu	NOUN
ijassa-659	519	7	time	time	NOUN
ijassa-659	519	8	of	of	ADP
ijassa-659	519	9	the	the	DET
ijassa-659	519	10	proposed	propose	VERB
ijassa-659	519	11	selection	selection	NOUN
ijassa-659	519	12	method	method	NOUN
ijassa-659	519	13	(	(	PUNCT
ijassa-659	519	14	using	use	VERB
ijassa-659	519	15	batch	batch	NOUN
ijassa-659	519	16	size	size	NOUN
ijassa-659	519	17	value	value	NOUN
ijassa-659	519	18	b	b	NOUN
ijassa-659	519	19	=	=	NOUN
ijassa-659	519	20	10	10	NUM
ijassa-659	519	21	)	)	PUNCT
ijassa-659	519	22	with	with	ADP
ijassa-659	519	23	the	the	DET
ijassa-659	519	24	baseline	baseline	NOUN
ijassa-659	519	25	methods	method	NOUN
ijassa-659	519	26	on	on	ADP
ijassa-659	519	27	a	a	DET
ijassa-659	519	28	3.5	3.5	NUM
ijassa-659	519	29	ghz	ghz	NOUN
ijassa-659	519	30	intel	intel	NOUN
ijassa-659	519	31	core	core	NOUN
ijassa-659	519	32	i5	i5	NOUN
ijassa-659	519	33	and	and	CCONJ
ijassa-659	519	34	4	4	NUM
ijassa-659	519	35	gb	gb	NOUN
ijassa-659	519	36	main	main	ADJ
ijassa-659	519	37	memory	memory	NOUN
ijassa-659	519	38	.	.	PUNCT
ijassa-659	520	1	the	the	DET
ijassa-659	520	2	horizontal	horizontal	ADJ
ijassa-659	520	3	axis	axis	NOUN
ijassa-659	520	4	indicates	indicate	VERB
ijassa-659	520	5	the	the	DET
ijassa-659	520	6	total	total	ADJ
ijassa-659	520	7	number	number	NOUN
ijassa-659	520	8	of	of	ADP
ijassa-659	520	9	queries	query	NOUN
ijassa-659	520	10	and	and	CCONJ
ijassa-659	520	11	the	the	DET
ijassa-659	520	12	vertical	vertical	ADJ
ijassa-659	520	13	axis	axis	NOUN
ijassa-659	520	14	shows	show	VERB
ijassa-659	520	15	the	the	DET
ijassa-659	520	16	cpu	cpu	NOUN
ijassa-659	520	17	running	running	NOUN
ijassa-659	520	18	time	time	NOUN
ijassa-659	520	19	(	(	PUNCT
ijassa-659	520	20	in	in	ADP
ijassa-659	520	21	seconds	second	NOUN
ijassa-659	520	22	)	)	PUNCT
ijassa-659	520	23	.	.	PUNCT
ijassa-659	521	1	from	from	ADP
ijassa-659	521	2	the	the	DET
ijassa-659	521	3	results	result	NOUN
ijassa-659	521	4	,	,	PUNCT
ijassa-659	521	5	we	we	PRON
ijassa-659	521	6	clearly	clearly	ADV
ijassa-659	521	7	see	see	VERB
ijassa-659	521	8	that	that	SCONJ
ijassa-659	521	9	the	the	DET
ijassa-659	521	10	proposed	propose	VERB
ijassa-659	521	11	method	method	NOUN
ijassa-659	521	12	is	be	AUX
ijassa-659	521	13	significantly	significantly	ADV
ijassa-659	521	14	more	more	ADV
ijassa-659	521	15	efficient	efficient	ADJ
ijassa-659	521	16	and	and	CCONJ
ijassa-659	521	17	scalable	scalable	ADJ
ijassa-659	521	18	than	than	ADP
ijassa-659	521	19	other	other	ADJ
ijassa-659	521	20	methods	method	NOUN
ijassa-659	521	21	.	.	PUNCT
ijassa-659	522	1	as	as	SCONJ
ijassa-659	522	2	we	we	PRON
ijassa-659	522	3	observe	observe	VERB
ijassa-659	522	4	,	,	PUNCT
ijassa-659	522	5	when	when	SCONJ
ijassa-659	522	6	the	the	DET
ijassa-659	522	7	dataset	dataset	NOUN
ijassa-659	522	8	size	size	NOUN
ijassa-659	522	9	increases	increase	NOUN
ijassa-659	522	10	,	,	PUNCT
ijassa-659	522	11	the	the	DET
ijassa-659	522	12	time	time	NOUN
ijassa-659	522	13	cost	cost	NOUN
ijassa-659	522	14	of	of	ADP
ijassa-659	522	15	npu	npu	ADJ
ijassa-659	522	16	and	and	CCONJ
ijassa-659	522	17	asc	asc	PROPN
ijassa-659	522	18	increases	increase	VERB
ijassa-659	522	19	dramatically	dramatically	ADV
ijassa-659	522	20	,	,	PUNCT
ijassa-659	522	21	while	while	SCONJ
ijassa-659	522	22	the	the	DET
ijassa-659	522	23	proposed	propose	VERB
ijassa-659	522	24	method	method	NOUN
ijassa-659	522	25	increases	increase	VERB
ijassa-659	522	26	linearly	linearly	ADV
ijassa-659	522	27	.	.	PUNCT
ijassa-659	523	1	specifically	specifically	ADV
ijassa-659	523	2	,	,	PUNCT
ijassa-659	523	3	on	on	ADP
ijassa-659	523	4	the	the	DET
ijassa-659	523	5	magic	magic	NOUN
ijassa-659	523	6	dataset	dataset	VERB
ijassa-659	523	7	with	with	ADP
ijassa-659	523	8	150	150	NUM
ijassa-659	523	9	queries	query	NOUN
ijassa-659	523	10	,	,	PUNCT
ijassa-659	523	11	npu	npu	ADJ
ijassa-659	523	12	takes	take	VERB
ijassa-659	523	13	about	about	ADV
ijassa-659	523	14	240	240	NUM
ijassa-659	523	15	seconds	second	NOUN
ijassa-659	523	16	and	and	CCONJ
ijassa-659	523	17	asc	asc	PROPN
ijassa-659	523	18	takes	take	VERB
ijassa-659	523	19	about	about	ADV
ijassa-659	523	20	260	260	NUM
ijassa-659	523	21	seconds	second	NOUN
ijassa-659	523	22	,	,	PUNCT
ijassa-659	523	23	while	while	SCONJ
ijassa-659	523	24	the	the	DET
ijassa-659	523	25	proposed	propose	VERB
ijassa-659	523	26	method	method	NOUN
ijassa-659	523	27	needs	need	VERB
ijassa-659	523	28	only	only	ADV
ijassa-659	523	29	about	about	ADV
ijassa-659	523	30	24	24	NUM
ijassa-659	523	31	second	second	ADJ
ijassa-659	523	32	.	.	PUNCT
ijassa-659	524	1	hence	hence	ADV
ijassa-659	524	2	,	,	PUNCT
ijassa-659	524	3	we	we	PRON
ijassa-659	524	4	can	can	AUX
ijassa-659	524	5	conclude	conclude	VERB
ijassa-659	524	6	that	that	SCONJ
ijassa-659	524	7	the	the	DET
ijassa-659	524	8	proposed	propose	VERB
ijassa-659	524	9	method	method	NOUN
ijassa-659	524	10	is	be	AUX
ijassa-659	524	11	more	more	ADV
ijassa-659	524	12	efficient	efficient	ADJ
ijassa-659	524	13	and	and	CCONJ
ijassa-659	524	14	scalable	scalable	ADJ
ijassa-659	524	15	than	than	ADP
ijassa-659	524	16	other	other	ADJ
ijassa-659	524	17	methods	method	NOUN
ijassa-659	524	18	for	for	ADP
ijassa-659	524	19	large	large	ADJ
ijassa-659	524	20	applications	application	NOUN
ijassa-659	524	21	.	.	PUNCT
ijassa-659	525	1	finally	finally	ADV
ijassa-659	525	2	,	,	PUNCT
ijassa-659	525	3	as	as	SCONJ
ijassa-659	525	4	indicated	indicate	VERB
ijassa-659	525	5	in	in	ADP
ijassa-659	525	6	figure	figure	NOUN
ijassa-659	525	7	4	4	NUM
ijassa-659	525	8	,	,	PUNCT
ijassa-659	525	9	the	the	DET
ijassa-659	525	10	time	time	NOUN
ijassa-659	525	11	cost	cost	NOUN
ijassa-659	525	12	of	of	ADP
ijassa-659	525	13	the	the	DET
ijassa-659	525	14	proposed	propose	VERB
ijassa-659	525	15	method	method	NOUN
ijassa-659	525	16	increases	increase	VERB
ijassa-659	525	17	moderately	moderately	ADV
ijassa-659	525	18	as	as	ADP
ijassa-659	525	19	the	the	DET
ijassa-659	525	20	number	number	NOUN
ijassa-659	525	21	improving	improve	VERB
ijassa-659	525	22	semi	semi	ADJ
ijassa-659	525	23	-	-	ADJ
ijassa-659	525	24	supervised	supervised	ADJ
ijassa-659	525	25	clustering	clustering	ADJ
ijassa-659	525	26	algorithms	algorithm	NOUN
ijassa-659	525	27	with	with	ADP
ijassa-659	525	28	active	active	ADJ
ijassa-659	525	29	query	query	NOUN
ijassa-659	525	30	selection	selection	NOUN
ijassa-659	525	31	39	39	NUM
ijassa-659	525	32	copyright	copyright	NOUN
ijassa-659	525	33	©	©	PROPN
ijassa-659	525	34	2019	2019	NUM
ijassa-659	525	35	assa	assa	PROPN
ijassa-659	525	36	adv	adv	PROPN
ijassa-659	525	37	.	.	PUNCT
ijassa-659	526	1	in	in	ADP
ijassa-659	526	2	systems	system	NOUN
ijassa-659	526	3	science	science	NOUN
ijassa-659	526	4	and	and	CCONJ
ijassa-659	526	5	appl	appl	NOUN
ijassa-659	526	6	.	.	PUNCT
ijassa-659	527	1	(	(	PUNCT
ijassa-659	527	2	2019	2019	NUM
ijassa-659	527	3	)	)	PUNCT
ijassa-659	527	4	of	of	ADP
ijassa-659	527	5	queries	query	NOUN
ijassa-659	527	6	increases	increase	NOUN
ijassa-659	527	7	.	.	PUNCT
ijassa-659	528	1	while	while	SCONJ
ijassa-659	528	2	,	,	PUNCT
ijassa-659	528	3	the	the	DET
ijassa-659	528	4	time	time	NOUN
ijassa-659	528	5	cost	cost	NOUN
ijassa-659	528	6	of	of	ADP
ijassa-659	528	7	other	other	ADJ
ijassa-659	528	8	methods	method	NOUN
ijassa-659	528	9	increases	increase	VERB
ijassa-659	528	10	tremendously	tremendously	ADV
ijassa-659	528	11	as	as	ADP
ijassa-659	528	12	the	the	DET
ijassa-659	528	13	number	number	NOUN
ijassa-659	528	14	of	of	ADP
ijassa-659	528	15	queries	query	NOUN
ijassa-659	528	16	increases	increase	NOUN
ijassa-659	528	17	.	.	PUNCT
ijassa-659	529	1	generally	generally	ADV
ijassa-659	529	2	,	,	PUNCT
ijassa-659	529	3	the	the	DET
ijassa-659	529	4	results	result	NOUN
ijassa-659	529	5	show	show	VERB
ijassa-659	529	6	that	that	SCONJ
ijassa-659	529	7	the	the	DET
ijassa-659	529	8	proposed	propose	VERB
ijassa-659	529	9	method	method	NOUN
ijassa-659	529	10	is	be	AUX
ijassa-659	529	11	slower	slow	ADJ
ijassa-659	529	12	than	than	ADP
ijassa-659	529	13	random	random	ADJ
ijassa-659	529	14	method	method	NOUN
ijassa-659	529	15	that	that	PRON
ijassa-659	529	16	selects	select	VERB
ijassa-659	529	17	set	set	VERB
ijassa-659	529	18	of	of	ADP
ijassa-659	529	19	constraints	constraint	NOUN
ijassa-659	529	20	randomly	randomly	ADV
ijassa-659	529	21	without	without	ADP
ijassa-659	529	22	performing	perform	VERB
ijassa-659	529	23	additional	additional	ADJ
ijassa-659	529	24	cpu	cpu	NOUN
ijassa-659	529	25	time	time	NOUN
ijassa-659	529	26	for	for	ADP
ijassa-659	529	27	selecting	select	VERB
ijassa-659	529	28	constraints	constraint	NOUN
ijassa-659	529	29	.	.	PUNCT
ijassa-659	530	1	protein	protein	NOUN
ijassa-659	530	2	heart	heart	NOUN
ijassa-659	530	3	ionosphere	ionosphere	NOUN
ijassa-659	530	4	breast	breast	NOUN
ijassa-659	530	5	yeast	yeast	NOUN
ijassa-659	530	6	segment	segment	NOUN
ijassa-659	530	7	0	0	NUM
ijassa-659	530	8	5	5	NUM
ijassa-659	530	9	10	10	NUM
ijassa-659	530	10	15	15	NUM
ijassa-659	530	11	20	20	NUM
ijassa-659	530	12	25	25	NUM
ijassa-659	530	13	25	25	NUM
ijassa-659	530	14	50	50	NUM
ijassa-659	530	15	75	75	NUM
ijassa-659	530	16	100	100	NUM
ijassa-659	530	17	125	125	NUM
ijassa-659	530	18	150	150	NUM
ijassa-659	530	19	c	c	NOUN
ijassa-659	530	20	p	p	X
ijassa-659	530	21	u	u	X
ijassa-659	530	22	t	t	X
ijassa-659	531	1	i	i	PRON
ijassa-659	531	2	m	m	VERB
ijassa-659	531	3	e	e	VERB
ijassa-659	531	4	(	(	PUNCT
ijassa-659	531	5	se	se	X
ijassa-659	531	6	co	co	X
ijassa-659	531	7	n	n	PROPN
ijassa-659	531	8	d	d	PROPN
ijassa-659	531	9	s	s	NOUN
ijassa-659	531	10	)	)	PUNCT
ijassa-659	531	11	number	number	NOUN
ijassa-659	531	12	of	of	ADP
ijassa-659	531	13	queries	query	NOUN
ijassa-659	531	14	our	our	PRON
ijassa-659	531	15	method	method	PROPN
ijassa-659	531	16	npu	npu	PROPN
ijassa-659	531	17	asc	asc	PROPN
ijassa-659	531	18	min	min	PROPN
ijassa-659	531	19	-	-	PROPN
ijassa-659	531	20	max	max	ADJ
ijassa-659	531	21	random	random	ADJ
ijassa-659	531	22	0	0	NUM
ijassa-659	531	23	5	5	NUM
ijassa-659	531	24	10	10	NUM
ijassa-659	531	25	15	15	NUM
ijassa-659	531	26	20	20	NUM
ijassa-659	531	27	25	25	NUM
ijassa-659	531	28	25	25	NUM
ijassa-659	531	29	50	50	NUM
ijassa-659	531	30	75	75	NUM
ijassa-659	531	31	100	100	NUM
ijassa-659	531	32	125	125	NUM
ijassa-659	531	33	150	150	NUM
ijassa-659	531	34	c	c	NOUN
ijassa-659	531	35	p	p	X
ijassa-659	531	36	u	u	X
ijassa-659	531	37	t	t	X
ijassa-659	532	1	i	i	PRON
ijassa-659	532	2	m	m	VERB
ijassa-659	532	3	e	e	VERB
ijassa-659	532	4	(	(	PUNCT
ijassa-659	532	5	se	se	X
ijassa-659	532	6	co	co	X
ijassa-659	532	7	n	n	PROPN
ijassa-659	532	8	d	d	PROPN
ijassa-659	532	9	s	s	NOUN
ijassa-659	532	10	)	)	PUNCT
ijassa-659	532	11	number	number	NOUN
ijassa-659	532	12	of	of	ADP
ijassa-659	532	13	queries	query	NOUN
ijassa-659	532	14	our	our	PRON
ijassa-659	532	15	method	method	PROPN
ijassa-659	532	16	npu	npu	PROPN
ijassa-659	532	17	asc	asc	PROPN
ijassa-659	532	18	min	min	PROPN
ijassa-659	532	19	-	-	PROPN
ijassa-659	532	20	max	max	ADJ
ijassa-659	532	21	random	random	ADJ
ijassa-659	532	22	0	0	NUM
ijassa-659	532	23	5	5	NUM
ijassa-659	532	24	10	10	NUM
ijassa-659	532	25	15	15	NUM
ijassa-659	532	26	20	20	NUM
ijassa-659	532	27	25	25	NUM
ijassa-659	532	28	30	30	NUM
ijassa-659	532	29	35	35	NUM
ijassa-659	532	30	25	25	NUM
ijassa-659	532	31	50	50	NUM
ijassa-659	532	32	75	75	NUM
ijassa-659	532	33	100	100	NUM
ijassa-659	532	34	125	125	NUM
ijassa-659	532	35	150	150	NUM
ijassa-659	532	36	c	c	NOUN
ijassa-659	532	37	p	p	X
ijassa-659	532	38	u	u	X
ijassa-659	532	39	t	t	X
ijassa-659	533	1	i	i	PRON
ijassa-659	533	2	m	m	VERB
ijassa-659	533	3	e	e	VERB
ijassa-659	533	4	(	(	PUNCT
ijassa-659	533	5	se	se	X
ijassa-659	533	6	co	co	X
ijassa-659	533	7	n	n	PROPN
ijassa-659	533	8	d	d	PROPN
ijassa-659	533	9	s	s	NOUN
ijassa-659	533	10	)	)	PUNCT
ijassa-659	533	11	number	number	NOUN
ijassa-659	533	12	of	of	ADP
ijassa-659	533	13	queries	query	NOUN
ijassa-659	533	14	our	our	PRON
ijassa-659	533	15	method	method	PROPN
ijassa-659	533	16	npu	npu	PROPN
ijassa-659	533	17	asc	asc	PROPN
ijassa-659	533	18	min	min	PROPN
ijassa-659	533	19	-	-	PROPN
ijassa-659	533	20	max	max	ADJ
ijassa-659	533	21	random	random	ADJ
ijassa-659	533	22	0	0	NUM
ijassa-659	533	23	10	10	NUM
ijassa-659	533	24	20	20	NUM
ijassa-659	533	25	30	30	NUM
ijassa-659	533	26	40	40	NUM
ijassa-659	533	27	50	50	NUM
ijassa-659	533	28	60	60	NUM
ijassa-659	533	29	25	25	NUM
ijassa-659	533	30	50	50	NUM
ijassa-659	533	31	75	75	NUM
ijassa-659	533	32	100	100	NUM
ijassa-659	533	33	125	125	NUM
ijassa-659	533	34	150	150	NUM
ijassa-659	533	35	c	c	NOUN
ijassa-659	533	36	p	p	X
ijassa-659	533	37	u	u	X
ijassa-659	533	38	t	t	X
ijassa-659	534	1	i	i	PRON
ijassa-659	534	2	m	m	VERB
ijassa-659	534	3	e	e	VERB
ijassa-659	534	4	(	(	PUNCT
ijassa-659	534	5	se	se	X
ijassa-659	534	6	co	co	X
ijassa-659	534	7	n	n	PROPN
ijassa-659	534	8	d	d	PROPN
ijassa-659	534	9	s	s	NOUN
ijassa-659	534	10	)	)	PUNCT
ijassa-659	534	11	number	number	NOUN
ijassa-659	534	12	of	of	ADP
ijassa-659	534	13	queries	query	NOUN
ijassa-659	534	14	our	our	PRON
ijassa-659	534	15	method	method	PROPN
ijassa-659	534	16	npu	npu	PROPN
ijassa-659	534	17	asc	asc	PROPN
ijassa-659	534	18	min	min	PROPN
ijassa-659	534	19	-	-	PROPN
ijassa-659	534	20	max	max	ADJ
ijassa-659	534	21	random	random	ADJ
ijassa-659	534	22	0	0	NUM
ijassa-659	534	23	20	20	NUM
ijassa-659	534	24	40	40	NUM
ijassa-659	534	25	60	60	NUM
ijassa-659	534	26	80	80	NUM
ijassa-659	534	27	100	100	NUM
ijassa-659	534	28	25	25	NUM
ijassa-659	534	29	50	50	NUM
ijassa-659	534	30	75	75	NUM
ijassa-659	534	31	100	100	NUM
ijassa-659	534	32	125	125	NUM
ijassa-659	534	33	150	150	NUM
ijassa-659	534	34	c	c	NOUN
ijassa-659	534	35	p	p	X
ijassa-659	534	36	u	u	X
ijassa-659	534	37	t	t	X
ijassa-659	535	1	i	i	PRON
ijassa-659	535	2	m	m	VERB
ijassa-659	535	3	e	e	VERB
ijassa-659	535	4	(	(	PUNCT
ijassa-659	535	5	se	se	X
ijassa-659	535	6	co	co	X
ijassa-659	535	7	n	n	PROPN
ijassa-659	535	8	d	d	PROPN
ijassa-659	535	9	s	s	NOUN
ijassa-659	535	10	)	)	PUNCT
ijassa-659	535	11	number	number	NOUN
ijassa-659	535	12	of	of	ADP
ijassa-659	535	13	queries	query	NOUN
ijassa-659	535	14	our	our	PRON
ijassa-659	535	15	method	method	PROPN
ijassa-659	535	16	npu	npu	PROPN
ijassa-659	535	17	asc	asc	PROPN
ijassa-659	535	18	min	min	PROPN
ijassa-659	535	19	-	-	PROPN
ijassa-659	535	20	max	max	ADJ
ijassa-659	535	21	random	random	ADJ
ijassa-659	535	22	0	0	NUM
ijassa-659	535	23	20	20	NUM
ijassa-659	535	24	40	40	NUM
ijassa-659	535	25	60	60	NUM
ijassa-659	535	26	80	80	NUM
ijassa-659	535	27	100	100	NUM
ijassa-659	535	28	120	120	NUM
ijassa-659	535	29	25	25	NUM
ijassa-659	535	30	50	50	NUM
ijassa-659	535	31	75	75	NUM
ijassa-659	535	32	100	100	NUM
ijassa-659	535	33	125	125	NUM
ijassa-659	535	34	150	150	NUM
ijassa-659	535	35	c	c	NOUN
ijassa-659	535	36	p	p	X
ijassa-659	535	37	u	u	X
ijassa-659	535	38	t	t	X
ijassa-659	536	1	i	i	PRON
ijassa-659	536	2	m	m	VERB
ijassa-659	536	3	e	e	VERB
ijassa-659	536	4	(	(	PUNCT
ijassa-659	536	5	se	se	X
ijassa-659	536	6	co	co	X
ijassa-659	536	7	n	n	PROPN
ijassa-659	536	8	d	d	PROPN
ijassa-659	536	9	s	s	NOUN
ijassa-659	536	10	)	)	PUNCT
ijassa-659	536	11	number	number	NOUN
ijassa-659	536	12	of	of	ADP
ijassa-659	536	13	queries	query	NOUN
ijassa-659	536	14	our	our	PRON
ijassa-659	536	15	method	method	PROPN
ijassa-659	536	16	npu	npu	PROPN
ijassa-659	536	17	asc	asc	PROPN
ijassa-659	536	18	min	min	PROPN
ijassa-659	536	19	-	-	PROPN
ijassa-659	536	20	max	max	ADJ
ijassa-659	536	21	random	random	ADJ
ijassa-659	536	22	40	40	NUM
ijassa-659	536	23	w.	w.	NOUN
ijassa-659	536	24	atwa	atwa	PROPN
ijassa-659	536	25	,	,	PUNCT
ijassa-659	536	26	m.	m.	NOUN
ijassa-659	536	27	emam	emam	PROPN
ijassa-659	536	28	copyright	copyright	NOUN
ijassa-659	536	29	©	©	PROPN
ijassa-659	536	30	2019	2019	NUM
ijassa-659	536	31	assa	assa	PROPN
ijassa-659	536	32	adv	adv	PROPN
ijassa-659	536	33	.	.	PUNCT
ijassa-659	537	1	in	in	ADP
ijassa-659	537	2	systems	system	NOUN
ijassa-659	537	3	science	science	NOUN
ijassa-659	537	4	and	and	CCONJ
ijassa-659	537	5	appl	appl	NOUN
ijassa-659	537	6	.	.	PUNCT
ijassa-659	538	1	(	(	PUNCT
ijassa-659	538	2	2019	2019	NUM
ijassa-659	538	3	)	)	PUNCT
ijassa-659	538	4	digit-389	digit-389	ADJ
ijassa-659	538	5	magic	magic	PROPN
ijassa-659	538	6	fig	fig	NOUN
ijassa-659	538	7	.	.	PUNCT
ijassa-659	539	1	4	4	NUM
ijassa-659	539	2	:	:	PUNCT
ijassa-659	539	3	comparison	comparison	NOUN
ijassa-659	539	4	of	of	ADP
ijassa-659	539	5	average	average	ADJ
ijassa-659	539	6	run	run	NOUN
ijassa-659	539	7	time	time	NOUN
ijassa-659	539	8	over	over	ADP
ijassa-659	539	9	different	different	ADJ
ijassa-659	539	10	number	number	NOUN
ijassa-659	539	11	of	of	ADP
ijassa-659	539	12	queries	query	NOUN
ijassa-659	539	13	with	with	ADP
ijassa-659	539	14	batch	batch	NOUN
ijassa-659	539	15	size	size	NOUN
ijassa-659	539	16	value	value	NOUN
ijassa-659	539	17	b	b	NOUN
ijassa-659	539	18	=	=	SYM
ijassa-659	539	19	10	10	NUM
ijassa-659	539	20	.	.	PUNCT
ijassa-659	540	1	4.3.3	4.3.3	NUM
ijassa-659	540	2	analysis	analysis	NOUN
ijassa-659	540	3	of	of	ADP
ijassa-659	540	4	different	different	ADJ
ijassa-659	540	5	batch	batch	NOUN
ijassa-659	540	6	size	size	NOUN
ijassa-659	540	7	values	value	NOUN
ijassa-659	540	8	in	in	ADP
ijassa-659	540	9	this	this	DET
ijassa-659	540	10	section	section	NOUN
ijassa-659	540	11	,	,	PUNCT
ijassa-659	540	12	we	we	PRON
ijassa-659	540	13	carried	carry	VERB
ijassa-659	540	14	out	out	ADP
ijassa-659	540	15	an	an	DET
ijassa-659	540	16	analysis	analysis	NOUN
ijassa-659	540	17	of	of	ADP
ijassa-659	540	18	the	the	DET
ijassa-659	540	19	proposed	propose	VERB
ijassa-659	540	20	active	active	ADJ
ijassa-659	540	21	learning	learning	NOUN
ijassa-659	540	22	with	with	ADP
ijassa-659	540	23	varying	vary	VERB
ijassa-659	540	24	the	the	DET
ijassa-659	540	25	value	value	NOUN
ijassa-659	540	26	of	of	ADP
ijassa-659	540	27	the	the	DET
ijassa-659	540	28	batch	batch	NOUN
ijassa-659	540	29	size	size	NOUN
ijassa-659	540	30	b.	b.	PROPN
ijassa-659	540	31	figure	figure	NOUN
ijassa-659	540	32	5	5	NUM
ijassa-659	540	33	shows	show	VERB
ijassa-659	540	34	the	the	DET
ijassa-659	540	35	performance	performance	NOUN
ijassa-659	540	36	versus	versus	ADP
ijassa-659	540	37	the	the	DET
ijassa-659	540	38	number	number	NOUN
ijassa-659	540	39	of	of	ADP
ijassa-659	540	40	queries	query	NOUN
ijassa-659	540	41	on	on	ADP
ijassa-659	540	42	the	the	DET
ijassa-659	540	43	datasets	dataset	NOUN
ijassa-659	540	44	with	with	ADP
ijassa-659	540	45	mpckmeans	mpckmean	NOUN
ijassa-659	540	46	algorithm	algorithm	NOUN
ijassa-659	540	47	.	.	PUNCT
ijassa-659	541	1	from	from	ADP
ijassa-659	541	2	figure	figure	NOUN
ijassa-659	541	3	5	5	NUM
ijassa-659	541	4	,	,	PUNCT
ijassa-659	541	5	selecting	select	VERB
ijassa-659	541	6	small	small	ADJ
ijassa-659	541	7	b	b	PROPN
ijassa-659	541	8	values	value	NOUN
ijassa-659	541	9	results	result	VERB
ijassa-659	541	10	in	in	ADP
ijassa-659	541	11	similar	similar	ADJ
ijassa-659	541	12	(	(	PUNCT
ijassa-659	541	13	or	or	CCONJ
ijassa-659	541	14	better	well	ADJ
ijassa-659	541	15	)	)	PUNCT
ijassa-659	541	16	performance	performance	NOUN
ijassa-659	541	17	compared	compare	VERB
ijassa-659	541	18	to	to	ADP
ijassa-659	541	19	those	those	PRON
ijassa-659	541	20	obtained	obtain	VERB
ijassa-659	541	21	selecting	select	VERB
ijassa-659	541	22	only	only	ADV
ijassa-659	541	23	one	one	NUM
ijassa-659	541	24	instance	instance	NOUN
ijassa-659	541	25	.	.	PUNCT
ijassa-659	542	1	on	on	ADP
ijassa-659	542	2	the	the	DET
ijassa-659	542	3	contrary	contrary	NOUN
ijassa-659	542	4	,	,	PUNCT
ijassa-659	542	5	high	high	ADJ
ijassa-659	542	6	b	b	PROPN
ijassa-659	542	7	values	value	NOUN
ijassa-659	542	8	decrease	decrease	VERB
ijassa-659	542	9	the	the	DET
ijassa-659	542	10	performance	performance	NOUN
ijassa-659	542	11	without	without	ADP
ijassa-659	542	12	decreasing	decrease	VERB
ijassa-659	542	13	the	the	DET
ijassa-659	542	14	computational	computational	ADJ
ijassa-659	542	15	time	time	NOUN
ijassa-659	542	16	if	if	SCONJ
ijassa-659	542	17	compared	compare	VERB
ijassa-659	542	18	to	to	ADP
ijassa-659	542	19	small	small	ADJ
ijassa-659	542	20	b	b	PROPN
ijassa-659	542	21	values	value	NOUN
ijassa-659	542	22	.	.	PUNCT
ijassa-659	543	1	figure	figure	NOUN
ijassa-659	543	2	6	6	NUM
ijassa-659	543	3	shows	show	VERB
ijassa-659	543	4	the	the	DET
ijassa-659	543	5	computational	computational	ADJ
ijassa-659	543	6	time	time	NOUN
ijassa-659	543	7	taken	take	VERB
ijassa-659	543	8	for	for	ADP
ijassa-659	543	9	different	different	ADJ
ijassa-659	543	10	b	b	NOUN
ijassa-659	543	11	values	value	NOUN
ijassa-659	543	12	on	on	ADP
ijassa-659	543	13	both	both	PRON
ijassa-659	543	14	yeast	yeast	NOUN
ijassa-659	543	15	and	and	CCONJ
ijassa-659	543	16	magic	magic	ADJ
ijassa-659	543	17	datasets	dataset	NOUN
ijassa-659	543	18	.	.	PUNCT
ijassa-659	544	1	from	from	ADP
ijassa-659	544	2	figure	figure	NOUN
ijassa-659	544	3	6	6	NUM
ijassa-659	544	4	,	,	PUNCT
ijassa-659	544	5	it	it	PRON
ijassa-659	544	6	could	could	AUX
ijassa-659	544	7	be	be	AUX
ijassa-659	544	8	noticed	notice	VERB
ijassa-659	544	9	that	that	SCONJ
ijassa-659	544	10	the	the	DET
ijassa-659	544	11	largest	large	ADJ
ijassa-659	544	12	learning	learning	NOUN
ijassa-659	544	13	time	time	NOUN
ijassa-659	544	14	is	be	AUX
ijassa-659	544	15	obtained	obtain	VERB
ijassa-659	544	16	in	in	ADP
ijassa-659	544	17	the	the	DET
ijassa-659	544	18	case	case	NOUN
ijassa-659	544	19	where	where	SCONJ
ijassa-659	544	20	one	one	NUM
ijassa-659	544	21	instance	instance	NOUN
ijassa-659	544	22	is	be	AUX
ijassa-659	544	23	selected	select	VERB
ijassa-659	544	24	(	(	PUNCT
ijassa-659	544	25	i.e.	i.e.	X
ijassa-659	544	26	b	b	X
ijassa-659	544	27	=	=	SYM
ijassa-659	544	28	1	1	NUM
ijassa-659	544	29	)	)	PUNCT
ijassa-659	544	30	.	.	PUNCT
ijassa-659	545	1	also	also	ADV
ijassa-659	545	2	,	,	PUNCT
ijassa-659	545	3	we	we	PRON
ijassa-659	545	4	can	can	AUX
ijassa-659	545	5	notice	notice	VERB
ijassa-659	545	6	that	that	SCONJ
ijassa-659	545	7	the	the	DET
ijassa-659	545	8	cpu	cpu	NOUN
ijassa-659	545	9	time	time	NOUN
ijassa-659	545	10	is	be	AUX
ijassa-659	545	11	increased	increase	VERB
ijassa-659	545	12	as	as	SCONJ
ijassa-659	545	13	the	the	DET
ijassa-659	545	14	value	value	NOUN
ijassa-659	545	15	of	of	ADP
ijassa-659	545	16	batch	batch	NOUN
ijassa-659	545	17	size	size	NOUN
ijassa-659	545	18	b	b	NOUN
ijassa-659	545	19	is	be	AUX
ijassa-659	545	20	increased	increase	VERB
ijassa-659	545	21	.	.	PUNCT
ijassa-659	546	1	therefore	therefore	ADV
ijassa-659	546	2	,	,	PUNCT
ijassa-659	546	3	it	it	PRON
ijassa-659	546	4	could	could	AUX
ijassa-659	546	5	be	be	AUX
ijassa-659	546	6	interesting	interesting	ADJ
ijassa-659	546	7	to	to	PART
ijassa-659	546	8	automatically	automatically	ADV
ijassa-659	546	9	identify	identify	VERB
ijassa-659	546	10	the	the	DET
ijassa-659	546	11	best	good	ADJ
ijassa-659	546	12	value	value	NOUN
ijassa-659	546	13	of	of	ADP
ijassa-659	546	14	the	the	DET
ijassa-659	546	15	batch	batch	NOUN
ijassa-659	546	16	size	size	NOUN
ijassa-659	546	17	b	b	NOUN
ijassa-659	546	18	that	that	PRON
ijassa-659	546	19	can	can	AUX
ijassa-659	546	20	achieve	achieve	VERB
ijassa-659	546	21	the	the	DET
ijassa-659	546	22	best	good	ADJ
ijassa-659	546	23	clustering	clustering	ADJ
ijassa-659	546	24	performance	performance	NOUN
ijassa-659	546	25	.	.	PUNCT
ijassa-659	547	1	(	(	PUNCT
ijassa-659	547	2	a	a	X
ijassa-659	547	3	)	)	PUNCT
ijassa-659	547	4	protein	protein	NOUN
ijassa-659	547	5	(	(	PUNCT
ijassa-659	547	6	b	b	NOUN
ijassa-659	547	7	)	)	PUNCT
ijassa-659	547	8	heart	heart	NOUN
ijassa-659	547	9	(	(	PUNCT
ijassa-659	547	10	c	c	NOUN
ijassa-659	547	11	)	)	PUNCT
ijassa-659	547	12	ionosphere	ionosphere	ADV
ijassa-659	547	13	(	(	PUNCT
ijassa-659	547	14	d	d	NOUN
ijassa-659	547	15	)	)	PUNCT
ijassa-659	547	16	breast	breast	NOUN
ijassa-659	547	17	(	(	PUNCT
ijassa-659	547	18	e	e	NOUN
ijassa-659	547	19	)	)	PUNCT
ijassa-659	547	20	yeast	yeast	NOUN
ijassa-659	547	21	(	(	PUNCT
ijassa-659	547	22	f	f	X
ijassa-659	547	23	)	)	PUNCT
ijassa-659	547	24	segment	segment	NOUN
ijassa-659	547	25	0	0	NUM
ijassa-659	547	26	20	20	NUM
ijassa-659	547	27	40	40	NUM
ijassa-659	547	28	60	60	NUM
ijassa-659	547	29	80	80	NUM
ijassa-659	547	30	100	100	NUM
ijassa-659	547	31	120	120	NUM
ijassa-659	547	32	140	140	NUM
ijassa-659	547	33	160	160	NUM
ijassa-659	547	34	180	180	NUM
ijassa-659	547	35	25	25	NUM
ijassa-659	547	36	50	50	NUM
ijassa-659	547	37	75	75	NUM
ijassa-659	547	38	100	100	NUM
ijassa-659	547	39	125	125	NUM
ijassa-659	547	40	150	150	NUM
ijassa-659	547	41	c	c	NOUN
ijassa-659	547	42	p	p	X
ijassa-659	547	43	u	u	X
ijassa-659	547	44	t	t	X
ijassa-659	547	45	i	i	PRON
ijassa-659	547	46	m	m	VERB
ijassa-659	547	47	e	e	VERB
ijassa-659	547	48	(	(	PUNCT
ijassa-659	547	49	se	se	X
ijassa-659	547	50	co	co	X
ijassa-659	547	51	n	n	PROPN
ijassa-659	547	52	d	d	PROPN
ijassa-659	547	53	s	s	NOUN
ijassa-659	547	54	)	)	PUNCT
ijassa-659	547	55	number	number	NOUN
ijassa-659	547	56	of	of	ADP
ijassa-659	547	57	queries	query	NOUN
ijassa-659	547	58	our	our	PRON
ijassa-659	547	59	method	method	PROPN
ijassa-659	547	60	npu	npu	PROPN
ijassa-659	547	61	asc	asc	PROPN
ijassa-659	547	62	min	min	PROPN
ijassa-659	547	63	-	-	PROPN
ijassa-659	547	64	max	max	ADJ
ijassa-659	547	65	random	random	ADJ
ijassa-659	547	66	0	0	NUM
ijassa-659	547	67	50	50	NUM
ijassa-659	547	68	100	100	NUM
ijassa-659	547	69	150	150	NUM
ijassa-659	547	70	200	200	NUM
ijassa-659	547	71	250	250	NUM
ijassa-659	547	72	300	300	NUM
ijassa-659	547	73	25	25	NUM
ijassa-659	547	74	50	50	NUM
ijassa-659	547	75	75	75	NUM
ijassa-659	547	76	100	100	NUM
ijassa-659	547	77	125	125	NUM
ijassa-659	547	78	150	150	NUM
ijassa-659	547	79	c	c	NOUN
ijassa-659	547	80	p	p	X
ijassa-659	547	81	u	u	X
ijassa-659	547	82	t	t	X
ijassa-659	547	83	i	i	PRON
ijassa-659	547	84	m	m	VERB
ijassa-659	547	85	e	e	VERB
ijassa-659	547	86	(	(	PUNCT
ijassa-659	547	87	se	se	X
ijassa-659	547	88	co	co	X
ijassa-659	547	89	n	n	PROPN
ijassa-659	547	90	d	d	PROPN
ijassa-659	547	91	s	s	NOUN
ijassa-659	547	92	)	)	PUNCT
ijassa-659	547	93	number	number	NOUN
ijassa-659	547	94	of	of	ADP
ijassa-659	547	95	queries	query	NOUN
ijassa-659	547	96	our	our	PRON
ijassa-659	547	97	method	method	PROPN
ijassa-659	547	98	npu	npu	PROPN
ijassa-659	547	99	asc	asc	PROPN
ijassa-659	547	100	min	min	PROPN
ijassa-659	547	101	-	-	PROPN
ijassa-659	547	102	max	max	ADJ
ijassa-659	547	103	random	random	ADJ
ijassa-659	547	104	0	0	NUM
ijassa-659	547	105	0,1	0,1	NUM
ijassa-659	547	106	0,2	0,2	NUM
ijassa-659	547	107	0,3	0,3	NUM
ijassa-659	547	108	0,4	0,4	NUM
ijassa-659	547	109	0,5	0,5	NUM
ijassa-659	547	110	0,6	0,6	NUM
ijassa-659	547	111	0,7	0,7	NUM
ijassa-659	547	112	0	0	NUM
ijassa-659	547	113	25	25	NUM
ijassa-659	547	114	50	50	NUM
ijassa-659	547	115	75	75	NUM
ijassa-659	547	116	100	100	NUM
ijassa-659	547	117	125	125	NUM
ijassa-659	547	118	150	150	NUM
ijassa-659	547	119	n	n	NOUN
ijassa-659	547	120	m	m	VERB
ijassa-659	547	121	i	i	PRON
ijassa-659	547	122	v	v	NUM
ijassa-659	547	123	al	al	PROPN
ijassa-659	547	124	u	u	NOUN
ijassa-659	547	125	e	e	NOUN
ijassa-659	547	126	number	number	NOUN
ijassa-659	547	127	of	of	ADP
ijassa-659	547	128	queries	query	NOUN
ijassa-659	547	129	b=1	b=1	PUNCT
ijassa-659	547	130	b=10	b=10	PUNCT
ijassa-659	547	131	b=50	b=50	PROPN
ijassa-659	547	132	b=100	b=100	PROPN
ijassa-659	547	133	0	0	NUM
ijassa-659	547	134	0,2	0,2	NUM
ijassa-659	547	135	0,4	0,4	NUM
ijassa-659	547	136	0,6	0,6	NUM
ijassa-659	547	137	0,8	0,8	NUM
ijassa-659	547	138	0	0	NUM
ijassa-659	547	139	25	25	NUM
ijassa-659	547	140	50	50	NUM
ijassa-659	547	141	75	75	NUM
ijassa-659	547	142	100	100	NUM
ijassa-659	547	143	125	125	NUM
ijassa-659	547	144	150	150	NUM
ijassa-659	547	145	n	n	NOUN
ijassa-659	547	146	m	m	VERB
ijassa-659	547	147	i	i	PRON
ijassa-659	547	148	v	v	NUM
ijassa-659	547	149	al	al	PROPN
ijassa-659	547	150	u	u	NOUN
ijassa-659	547	151	e	e	NOUN
ijassa-659	547	152	number	number	NOUN
ijassa-659	547	153	of	of	ADP
ijassa-659	547	154	queries	query	NOUN
ijassa-659	547	155	b=1	b=1	PUNCT
ijassa-659	547	156	b=10	b=10	PUNCT
ijassa-659	547	157	b=50	b=50	PROPN
ijassa-659	547	158	b=100	b=100	PROPN
ijassa-659	547	159	0,5	0,5	NUM
ijassa-659	547	160	0,55	0,55	NOUN
ijassa-659	547	161	0,6	0,6	NUM
ijassa-659	547	162	0,65	0,65	NUM
ijassa-659	547	163	0,7	0,7	NUM
ijassa-659	547	164	0,75	0,75	NUM
ijassa-659	547	165	0,8	0,8	NUM
ijassa-659	547	166	0,85	0,85	SYM
ijassa-659	547	167	0	0	NUM
ijassa-659	547	168	25	25	NUM
ijassa-659	547	169	50	50	NUM
ijassa-659	547	170	75	75	NUM
ijassa-659	547	171	100	100	NUM
ijassa-659	547	172	125	125	NUM
ijassa-659	547	173	150	150	NUM
ijassa-659	547	174	n	n	NOUN
ijassa-659	547	175	m	m	VERB
ijassa-659	547	176	i	i	PRON
ijassa-659	547	177	v	v	NUM
ijassa-659	547	178	al	al	PROPN
ijassa-659	547	179	u	u	NOUN
ijassa-659	547	180	e	e	NOUN
ijassa-659	547	181	number	number	NOUN
ijassa-659	547	182	of	of	ADP
ijassa-659	547	183	queries	query	NOUN
ijassa-659	547	184	b=1	b=1	PUNCT
ijassa-659	547	185	b=10	b=10	PUNCT
ijassa-659	547	186	b=50	b=50	PROPN
ijassa-659	547	187	b=100	b=100	PROPN
ijassa-659	547	188	0,7	0,7	NUM
ijassa-659	547	189	0,75	0,75	NUM
ijassa-659	547	190	0,8	0,8	NUM
ijassa-659	547	191	0,85	0,85	PUNCT
ijassa-659	547	192	0,9	0,9	NUM
ijassa-659	547	193	0,95	0,95	SYM
ijassa-659	547	194	1	1	NUM
ijassa-659	547	195	0	0	NUM
ijassa-659	547	196	25	25	NUM
ijassa-659	547	197	50	50	NUM
ijassa-659	547	198	75	75	NUM
ijassa-659	547	199	100	100	NUM
ijassa-659	547	200	125	125	NUM
ijassa-659	547	201	150	150	NUM
ijassa-659	547	202	n	n	NOUN
ijassa-659	547	203	m	m	VERB
ijassa-659	547	204	i	i	PRON
ijassa-659	547	205	v	v	NUM
ijassa-659	547	206	al	al	PROPN
ijassa-659	547	207	u	u	NOUN
ijassa-659	547	208	e	e	NOUN
ijassa-659	547	209	number	number	NOUN
ijassa-659	547	210	of	of	ADP
ijassa-659	547	211	queries	query	NOUN
ijassa-659	547	212	b=1	b=1	PUNCT
ijassa-659	547	213	b=10	b=10	PUNCT
ijassa-659	547	214	b=50	b=50	PROPN
ijassa-659	547	215	b=100	b=100	PROPN
ijassa-659	547	216	0,75	0,75	NOUN
ijassa-659	547	217	0,8	0,8	NUM
ijassa-659	547	218	0,85	0,85	PUNCT
ijassa-659	547	219	0,9	0,9	NUM
ijassa-659	547	220	0,95	0,95	SYM
ijassa-659	547	221	0	0	NUM
ijassa-659	547	222	25	25	NUM
ijassa-659	547	223	50	50	NUM
ijassa-659	547	224	75	75	NUM
ijassa-659	547	225	100	100	NUM
ijassa-659	547	226	125	125	NUM
ijassa-659	547	227	150	150	NUM
ijassa-659	547	228	n	n	NOUN
ijassa-659	547	229	m	m	VERB
ijassa-659	547	230	i	i	PRON
ijassa-659	547	231	v	v	NUM
ijassa-659	547	232	al	al	PROPN
ijassa-659	547	233	u	u	NOUN
ijassa-659	547	234	e	e	NOUN
ijassa-659	547	235	number	number	NOUN
ijassa-659	547	236	of	of	ADP
ijassa-659	547	237	queries	query	NOUN
ijassa-659	547	238	b=1	b=1	PUNCT
ijassa-659	547	239	b=10	b=10	PUNCT
ijassa-659	547	240	b=50	b=50	PROPN
ijassa-659	547	241	b=100	b=100	PROPN
ijassa-659	547	242	0,6	0,6	NUM
ijassa-659	547	243	0,65	0,65	NUM
ijassa-659	547	244	0,7	0,7	NUM
ijassa-659	547	245	0,75	0,75	NUM
ijassa-659	547	246	0,8	0,8	NOUN
ijassa-659	547	247	0	0	NUM
ijassa-659	547	248	25	25	NUM
ijassa-659	547	249	50	50	NUM
ijassa-659	547	250	75	75	NUM
ijassa-659	547	251	100	100	NUM
ijassa-659	547	252	125	125	NUM
ijassa-659	547	253	150	150	NUM
ijassa-659	547	254	n	n	NOUN
ijassa-659	547	255	m	m	VERB
ijassa-659	547	256	i	i	PRON
ijassa-659	547	257	v	v	NUM
ijassa-659	547	258	al	al	PROPN
ijassa-659	547	259	u	u	NOUN
ijassa-659	547	260	e	e	NOUN
ijassa-659	547	261	number	number	NOUN
ijassa-659	547	262	of	of	ADP
ijassa-659	547	263	queries	query	NOUN
ijassa-659	547	264	b=1	b=1	PUNCT
ijassa-659	547	265	b=10	b=10	ADP
ijassa-659	547	266	b=50	b=50	PROPN
ijassa-659	547	267	b=100	b=100	PROPN
ijassa-659	547	268	improving	improve	VERB
ijassa-659	547	269	semi	semi	ADJ
ijassa-659	547	270	-	-	ADJ
ijassa-659	547	271	supervised	supervised	ADJ
ijassa-659	547	272	clustering	clustering	ADJ
ijassa-659	547	273	algorithms	algorithm	NOUN
ijassa-659	547	274	with	with	ADP
ijassa-659	547	275	active	active	ADJ
ijassa-659	547	276	query	query	NOUN
ijassa-659	547	277	selection	selection	NOUN
ijassa-659	547	278	41	41	NUM
ijassa-659	547	279	copyright	copyright	NOUN
ijassa-659	547	280	©	©	PROPN
ijassa-659	547	281	2019	2019	NUM
ijassa-659	547	282	assa	assa	PROPN
ijassa-659	547	283	adv	adv	PROPN
ijassa-659	547	284	.	.	PUNCT
ijassa-659	548	1	in	in	ADP
ijassa-659	548	2	systems	system	NOUN
ijassa-659	548	3	science	science	NOUN
ijassa-659	548	4	and	and	CCONJ
ijassa-659	548	5	appl	appl	NOUN
ijassa-659	548	6	.	.	PUNCT
ijassa-659	549	1	(	(	PUNCT
ijassa-659	549	2	2019	2019	NUM
ijassa-659	549	3	)	)	PUNCT
ijassa-659	549	4	(	(	PUNCT
ijassa-659	549	5	g	g	NOUN
ijassa-659	549	6	)	)	PUNCT
ijassa-659	549	7	digit-389	digit-389	NOUN
ijassa-659	549	8	(	(	PUNCT
ijassa-659	549	9	h	h	NOUN
ijassa-659	549	10	)	)	PUNCT
ijassa-659	549	11	magic	magic	NOUN
ijassa-659	549	12	fig.5	fig.5	PROPN
ijassa-659	549	13	:	:	PUNCT
ijassa-659	549	14	clustering	cluster	VERB
ijassa-659	549	15	performance	performance	NOUN
ijassa-659	549	16	versus	versus	ADP
ijassa-659	549	17	different	different	ADJ
ijassa-659	549	18	batch	batch	NOUN
ijassa-659	549	19	size	size	NOUN
ijassa-659	549	20	(	(	PUNCT
ijassa-659	549	21	a	a	NOUN
ijassa-659	549	22	)	)	PUNCT
ijassa-659	549	23	protein	protein	NOUN
ijassa-659	549	24	(	(	PUNCT
ijassa-659	549	25	b	b	NOUN
ijassa-659	549	26	)	)	PUNCT
ijassa-659	549	27	heart	heart	NOUN
ijassa-659	549	28	(	(	PUNCT
ijassa-659	549	29	c	c	NOUN
ijassa-659	549	30	)	)	PUNCT
ijassa-659	549	31	ionosphere	ionosphere	ADV
ijassa-659	549	32	(	(	PUNCT
ijassa-659	549	33	d	d	NOUN
ijassa-659	549	34	)	)	PUNCT
ijassa-659	549	35	breast	breast	NOUN
ijassa-659	549	36	(	(	PUNCT
ijassa-659	549	37	e	e	NOUN
ijassa-659	549	38	)	)	PUNCT
ijassa-659	549	39	yeast	yeast	NOUN
ijassa-659	549	40	(	(	PUNCT
ijassa-659	549	41	f	f	X
ijassa-659	549	42	)	)	PUNCT
ijassa-659	549	43	segment	segment	NOUN
ijassa-659	549	44	(	(	PUNCT
ijassa-659	549	45	g	g	NOUN
ijassa-659	549	46	)	)	PUNCT
ijassa-659	549	47	digit-389	digit-389	NOUN
ijassa-659	549	48	(	(	PUNCT
ijassa-659	549	49	h	h	NOUN
ijassa-659	549	50	)	)	PUNCT
ijassa-659	549	51	magic	magic	ADJ
ijassa-659	549	52	fig	fig	NOUN
ijassa-659	549	53	.	.	PUNCT
ijassa-659	550	1	6	6	NUM
ijassa-659	550	2	:	:	PUNCT
ijassa-659	550	3	cpu	cpu	VERB
ijassa-659	550	4	time	time	NOUN
ijassa-659	550	5	versus	versus	ADP
ijassa-659	550	6	different	different	ADJ
ijassa-659	550	7	batch	batch	NOUN
ijassa-659	550	8	size	size	NOUN
ijassa-659	550	9	5	5	NUM
ijassa-659	550	10	.	.	PUNCT
ijassa-659	550	11	conclusion	conclusion	NOUN
ijassa-659	550	12	and	and	CCONJ
ijassa-659	550	13	future	future	ADJ
ijassa-659	550	14	work	work	NOUN
ijassa-659	550	15	identifying	identify	VERB
ijassa-659	550	16	the	the	DET
ijassa-659	550	17	most	most	ADV
ijassa-659	550	18	beneficial	beneficial	ADJ
ijassa-659	550	19	set	set	NOUN
ijassa-659	550	20	of	of	ADP
ijassa-659	550	21	clustering	clustering	ADJ
ijassa-659	550	22	constraints	constraint	NOUN
ijassa-659	550	23	was	be	AUX
ijassa-659	550	24	considered	consider	VERB
ijassa-659	550	25	in	in	ADP
ijassa-659	550	26	this	this	DET
ijassa-659	550	27	paper	paper	NOUN
ijassa-659	550	28	.	.	PUNCT
ijassa-659	551	1	we	we	PRON
ijassa-659	551	2	present	present	VERB
ijassa-659	551	3	an	an	DET
ijassa-659	551	4	iterative	iterative	NOUN
ijassa-659	551	5	active	active	ADJ
ijassa-659	551	6	learning	learning	NOUN
ijassa-659	551	7	method	method	NOUN
ijassa-659	551	8	that	that	PRON
ijassa-659	551	9	selects	select	VERB
ijassa-659	551	10	a	a	DET
ijassa-659	551	11	batch	batch	NOUN
ijassa-659	551	12	of	of	ADP
ijassa-659	551	13	query	query	NOUN
ijassa-659	551	14	instances	instance	NOUN
ijassa-659	551	15	from	from	ADP
ijassa-659	551	16	the	the	DET
ijassa-659	551	17	0,65	0,65	NUM
ijassa-659	551	18	0,7	0,7	NUM
ijassa-659	551	19	0,75	0,75	NUM
ijassa-659	551	20	0,8	0,8	NUM
ijassa-659	551	21	0,85	0,85	SYM
ijassa-659	551	22	0	0	NUM
ijassa-659	551	23	25	25	NUM
ijassa-659	551	24	50	50	NUM
ijassa-659	551	25	75	75	NUM
ijassa-659	551	26	100	100	NUM
ijassa-659	551	27	125	125	NUM
ijassa-659	551	28	150	150	NUM
ijassa-659	551	29	n	n	NOUN
ijassa-659	551	30	m	m	VERB
ijassa-659	551	31	i	i	PRON
ijassa-659	551	32	v	v	NUM
ijassa-659	551	33	al	al	PROPN
ijassa-659	551	34	u	u	NOUN
ijassa-659	551	35	e	e	NOUN
ijassa-659	551	36	number	number	NOUN
ijassa-659	551	37	of	of	ADP
ijassa-659	551	38	queries	query	NOUN
ijassa-659	551	39	b=1	b=1	PUNCT
ijassa-659	551	40	b=10	b=10	PUNCT
ijassa-659	551	41	b=50	b=50	PROPN
ijassa-659	551	42	b=100	b=100	PROPN
ijassa-659	551	43	0,35	0,35	NUM
ijassa-659	552	1	0,4	0,4	NUM
ijassa-659	552	2	0,45	0,45	NUM
ijassa-659	552	3	0,5	0,5	NUM
ijassa-659	552	4	0,55	0,55	NOUN
ijassa-659	552	5	0,6	0,6	NUM
ijassa-659	552	6	0,65	0,65	NUM
ijassa-659	553	1	0,7	0,7	NUM
ijassa-659	553	2	0	0	NUM
ijassa-659	553	3	25	25	NUM
ijassa-659	553	4	50	50	NUM
ijassa-659	553	5	75	75	NUM
ijassa-659	553	6	100	100	NUM
ijassa-659	553	7	125	125	NUM
ijassa-659	553	8	150	150	NUM
ijassa-659	553	9	n	n	NOUN
ijassa-659	553	10	m	m	VERB
ijassa-659	553	11	i	i	PRON
ijassa-659	553	12	v	v	NUM
ijassa-659	553	13	al	al	PROPN
ijassa-659	553	14	u	u	NOUN
ijassa-659	553	15	e	e	NOUN
ijassa-659	553	16	number	number	NOUN
ijassa-659	553	17	of	of	ADP
ijassa-659	553	18	queries	query	NOUN
ijassa-659	553	19	b=1	b=1	PUNCT
ijassa-659	553	20	b=10	b=10	PUNCT
ijassa-659	553	21	b=50	b=50	NOUN
ijassa-659	553	22	0	0	NUM
ijassa-659	553	23	10	10	NUM
ijassa-659	553	24	20	20	NUM
ijassa-659	553	25	30	30	NUM
ijassa-659	553	26	40	40	NUM
ijassa-659	553	27	50	50	NUM
ijassa-659	553	28	0	0	NUM
ijassa-659	553	29	25	25	NUM
ijassa-659	553	30	50	50	NUM
ijassa-659	553	31	75	75	NUM
ijassa-659	553	32	100	100	NUM
ijassa-659	553	33	125	125	NUM
ijassa-659	553	34	150	150	NUM
ijassa-659	553	35	c	c	NOUN
ijassa-659	553	36	p	p	X
ijassa-659	553	37	u	u	X
ijassa-659	553	38	t	t	X
ijassa-659	554	1	i	i	PRON
ijassa-659	554	2	m	m	VERB
ijassa-659	554	3	e	e	VERB
ijassa-659	554	4	(	(	PUNCT
ijassa-659	554	5	se	se	X
ijassa-659	554	6	co	co	X
ijassa-659	554	7	n	n	PROPN
ijassa-659	554	8	d	d	PROPN
ijassa-659	554	9	s	s	NOUN
ijassa-659	554	10	)	)	PUNCT
ijassa-659	554	11	number	number	NOUN
ijassa-659	554	12	of	of	ADP
ijassa-659	554	13	queries	query	NOUN
ijassa-659	554	14	b=1	b=1	PUNCT
ijassa-659	554	15	b=10	b=10	PUNCT
ijassa-659	554	16	b=50	b=50	PROPN
ijassa-659	554	17	b=100	b=100	PROPN
ijassa-659	554	18	0	0	NUM
ijassa-659	554	19	5	5	NUM
ijassa-659	554	20	10	10	NUM
ijassa-659	554	21	15	15	NUM
ijassa-659	554	22	20	20	NUM
ijassa-659	554	23	25	25	NUM
ijassa-659	554	24	0	0	NUM
ijassa-659	554	25	25	25	NUM
ijassa-659	554	26	50	50	NUM
ijassa-659	554	27	75	75	NUM
ijassa-659	554	28	100	100	NUM
ijassa-659	554	29	125	125	NUM
ijassa-659	554	30	150	150	NUM
ijassa-659	554	31	c	c	NOUN
ijassa-659	554	32	p	p	X
ijassa-659	554	33	u	u	X
ijassa-659	554	34	t	t	X
ijassa-659	555	1	i	i	PRON
ijassa-659	555	2	m	m	VERB
ijassa-659	555	3	e	e	VERB
ijassa-659	555	4	(	(	PUNCT
ijassa-659	555	5	se	se	X
ijassa-659	555	6	co	co	X
ijassa-659	555	7	n	n	PROPN
ijassa-659	555	8	d	d	PROPN
ijassa-659	555	9	s	s	NOUN
ijassa-659	555	10	)	)	PUNCT
ijassa-659	555	11	number	number	NOUN
ijassa-659	555	12	of	of	ADP
ijassa-659	555	13	queries	query	NOUN
ijassa-659	555	14	b=1	b=1	PUNCT
ijassa-659	555	15	b=10	b=10	PUNCT
ijassa-659	555	16	b=50	b=50	PROPN
ijassa-659	555	17	b=100	b=100	PROPN
ijassa-659	555	18	0	0	NUM
ijassa-659	555	19	5	5	NUM
ijassa-659	555	20	10	10	NUM
ijassa-659	555	21	15	15	NUM
ijassa-659	555	22	20	20	NUM
ijassa-659	555	23	25	25	NUM
ijassa-659	555	24	30	30	NUM
ijassa-659	555	25	35	35	NUM
ijassa-659	555	26	40	40	NUM
ijassa-659	555	27	0	0	NUM
ijassa-659	555	28	25	25	NUM
ijassa-659	555	29	50	50	NUM
ijassa-659	555	30	75	75	NUM
ijassa-659	555	31	100	100	NUM
ijassa-659	555	32	125	125	NUM
ijassa-659	555	33	150	150	NUM
ijassa-659	555	34	c	c	NOUN
ijassa-659	555	35	p	p	X
ijassa-659	555	36	u	u	X
ijassa-659	555	37	t	t	X
ijassa-659	556	1	i	i	PRON
ijassa-659	556	2	m	m	VERB
ijassa-659	556	3	e	e	VERB
ijassa-659	556	4	(	(	PUNCT
ijassa-659	556	5	se	se	X
ijassa-659	556	6	co	co	X
ijassa-659	556	7	n	n	PROPN
ijassa-659	556	8	d	d	PROPN
ijassa-659	556	9	s	s	NOUN
ijassa-659	556	10	)	)	PUNCT
ijassa-659	556	11	number	number	NOUN
ijassa-659	556	12	of	of	ADP
ijassa-659	556	13	queries	query	NOUN
ijassa-659	556	14	b=1	b=1	PUNCT
ijassa-659	556	15	b=10	b=10	PUNCT
ijassa-659	556	16	b=50	b=50	PROPN
ijassa-659	556	17	b=100	b=100	PROPN
ijassa-659	556	18	0	0	NUM
ijassa-659	556	19	20	20	NUM
ijassa-659	556	20	40	40	NUM
ijassa-659	556	21	60	60	NUM
ijassa-659	556	22	80	80	NUM
ijassa-659	556	23	0	0	NUM
ijassa-659	556	24	25	25	NUM
ijassa-659	556	25	50	50	NUM
ijassa-659	556	26	75	75	NUM
ijassa-659	556	27	100	100	NUM
ijassa-659	556	28	125	125	NUM
ijassa-659	556	29	150	150	NUM
ijassa-659	556	30	c	c	NOUN
ijassa-659	556	31	p	p	X
ijassa-659	556	32	u	u	X
ijassa-659	556	33	t	t	X
ijassa-659	557	1	i	i	PRON
ijassa-659	557	2	m	m	VERB
ijassa-659	557	3	e	e	VERB
ijassa-659	557	4	(	(	PUNCT
ijassa-659	557	5	se	se	X
ijassa-659	557	6	co	co	X
ijassa-659	557	7	n	n	PROPN
ijassa-659	557	8	d	d	PROPN
ijassa-659	557	9	s	s	NOUN
ijassa-659	557	10	)	)	PUNCT
ijassa-659	557	11	number	number	NOUN
ijassa-659	557	12	of	of	ADP
ijassa-659	557	13	queries	query	NOUN
ijassa-659	557	14	b=1	b=1	PUNCT
ijassa-659	557	15	b=10	b=10	PUNCT
ijassa-659	557	16	b=50	b=50	PROPN
ijassa-659	557	17	b=100	b=100	PROPN
ijassa-659	557	18	0	0	NUM
ijassa-659	557	19	20	20	NUM
ijassa-659	557	20	40	40	NUM
ijassa-659	557	21	60	60	NUM
ijassa-659	557	22	80	80	NUM
ijassa-659	557	23	100	100	NUM
ijassa-659	557	24	120	120	NUM
ijassa-659	557	25	0	0	NUM
ijassa-659	557	26	25	25	NUM
ijassa-659	557	27	50	50	NUM
ijassa-659	557	28	75	75	NUM
ijassa-659	557	29	100	100	NUM
ijassa-659	557	30	125	125	NUM
ijassa-659	557	31	150	150	NUM
ijassa-659	557	32	c	c	NOUN
ijassa-659	557	33	p	p	X
ijassa-659	557	34	u	u	X
ijassa-659	557	35	t	t	X
ijassa-659	558	1	i	i	PRON
ijassa-659	558	2	m	m	VERB
ijassa-659	558	3	e	e	VERB
ijassa-659	558	4	(	(	PUNCT
ijassa-659	558	5	se	se	X
ijassa-659	558	6	co	co	X
ijassa-659	558	7	n	n	PROPN
ijassa-659	558	8	d	d	PROPN
ijassa-659	558	9	s	s	NOUN
ijassa-659	558	10	)	)	PUNCT
ijassa-659	558	11	number	number	NOUN
ijassa-659	558	12	of	of	ADP
ijassa-659	558	13	queries	query	NOUN
ijassa-659	558	14	b=1	b=1	PUNCT
ijassa-659	558	15	b=10	b=10	PUNCT
ijassa-659	558	16	b=50	b=50	PROPN
ijassa-659	558	17	b=100	b=100	PROPN
ijassa-659	558	18	0	0	NUM
ijassa-659	558	19	20	20	NUM
ijassa-659	558	20	40	40	NUM
ijassa-659	558	21	60	60	NUM
ijassa-659	558	22	80	80	NUM
ijassa-659	558	23	100	100	NUM
ijassa-659	558	24	120	120	NUM
ijassa-659	558	25	140	140	NUM
ijassa-659	558	26	160	160	NUM
ijassa-659	558	27	0	0	NUM
ijassa-659	558	28	25	25	NUM
ijassa-659	558	29	50	50	NUM
ijassa-659	558	30	75	75	NUM
ijassa-659	558	31	100	100	NUM
ijassa-659	558	32	125	125	NUM
ijassa-659	558	33	150	150	NUM
ijassa-659	558	34	c	c	NOUN
ijassa-659	558	35	p	p	X
ijassa-659	558	36	u	u	X
ijassa-659	558	37	t	t	X
ijassa-659	559	1	i	i	PRON
ijassa-659	559	2	m	m	VERB
ijassa-659	559	3	e	e	VERB
ijassa-659	559	4	(	(	PUNCT
ijassa-659	559	5	se	se	X
ijassa-659	559	6	co	co	X
ijassa-659	559	7	n	n	PROPN
ijassa-659	559	8	d	d	PROPN
ijassa-659	559	9	s	s	NOUN
ijassa-659	559	10	)	)	PUNCT
ijassa-659	559	11	number	number	NOUN
ijassa-659	559	12	of	of	ADP
ijassa-659	559	13	queries	query	NOUN
ijassa-659	559	14	b=1	b=1	PUNCT
ijassa-659	559	15	b=10	b=10	PUNCT
ijassa-659	559	16	b=50	b=50	PROPN
ijassa-659	559	17	b=100	b=100	PROPN
ijassa-659	559	18	0	0	NUM
ijassa-659	559	19	20	20	NUM
ijassa-659	559	20	40	40	NUM
ijassa-659	559	21	60	60	NUM
ijassa-659	559	22	80	80	NUM
ijassa-659	559	23	100	100	NUM
ijassa-659	559	24	120	120	NUM
ijassa-659	559	25	140	140	NUM
ijassa-659	559	26	0	0	NUM
ijassa-659	559	27	25	25	NUM
ijassa-659	559	28	50	50	NUM
ijassa-659	559	29	75	75	NUM
ijassa-659	559	30	100	100	NUM
ijassa-659	559	31	125	125	NUM
ijassa-659	559	32	150	150	NUM
ijassa-659	559	33	c	c	NOUN
ijassa-659	559	34	p	p	X
ijassa-659	559	35	u	u	X
ijassa-659	559	36	t	t	X
ijassa-659	560	1	i	i	PRON
ijassa-659	560	2	m	m	VERB
ijassa-659	560	3	e	e	VERB
ijassa-659	560	4	(	(	PUNCT
ijassa-659	560	5	se	se	X
ijassa-659	560	6	co	co	X
ijassa-659	560	7	n	n	PROPN
ijassa-659	560	8	d	d	PROPN
ijassa-659	560	9	s	s	NOUN
ijassa-659	560	10	)	)	PUNCT
ijassa-659	560	11	number	number	NOUN
ijassa-659	560	12	of	of	ADP
ijassa-659	560	13	queries	query	NOUN
ijassa-659	560	14	b=1	b=1	PUNCT
ijassa-659	560	15	b=10	b=10	PUNCT
ijassa-659	560	16	b=50	b=50	PROPN
ijassa-659	560	17	b=100	b=100	PROPN
ijassa-659	560	18	0	0	NUM
ijassa-659	560	19	20	20	NUM
ijassa-659	560	20	40	40	NUM
ijassa-659	560	21	60	60	NUM
ijassa-659	560	22	80	80	NUM
ijassa-659	560	23	100	100	NUM
ijassa-659	560	24	120	120	NUM
ijassa-659	560	25	140	140	NUM
ijassa-659	560	26	160	160	NUM
ijassa-659	560	27	180	180	NUM
ijassa-659	560	28	0	0	NUM
ijassa-659	560	29	25	25	NUM
ijassa-659	560	30	50	50	NUM
ijassa-659	560	31	75	75	NUM
ijassa-659	560	32	100	100	NUM
ijassa-659	560	33	125	125	NUM
ijassa-659	560	34	150	150	NUM
ijassa-659	560	35	c	c	NOUN
ijassa-659	560	36	p	p	X
ijassa-659	560	37	u	u	X
ijassa-659	560	38	t	t	X
ijassa-659	561	1	i	i	PRON
ijassa-659	561	2	m	m	VERB
ijassa-659	561	3	e	e	VERB
ijassa-659	561	4	(	(	PUNCT
ijassa-659	561	5	se	se	X
ijassa-659	561	6	co	co	X
ijassa-659	561	7	n	n	PROPN
ijassa-659	561	8	d	d	PROPN
ijassa-659	561	9	s	s	NOUN
ijassa-659	561	10	)	)	PUNCT
ijassa-659	561	11	number	number	NOUN
ijassa-659	561	12	of	of	ADP
ijassa-659	561	13	queries	query	NOUN
ijassa-659	561	14	b=1	b=1	PUNCT
ijassa-659	561	15	b=10	b=10	PROPN
ijassa-659	561	16	b=50	b=50	PROPN
ijassa-659	561	17	b=100	b=100	PROPN
ijassa-659	561	18	42	42	NUM
ijassa-659	561	19	w.	w.	NOUN
ijassa-659	561	20	atwa	atwa	PROPN
ijassa-659	561	21	,	,	PUNCT
ijassa-659	561	22	m.	m.	NOUN
ijassa-659	561	23	emam	emam	PROPN
ijassa-659	561	24	copyright	copyright	NOUN
ijassa-659	561	25	©	©	PROPN
ijassa-659	561	26	2019	2019	NUM
ijassa-659	561	27	assa	assa	PROPN
ijassa-659	561	28	adv	adv	PROPN
ijassa-659	561	29	.	.	PUNCT
ijassa-659	562	1	in	in	ADP
ijassa-659	562	2	systems	system	NOUN
ijassa-659	562	3	science	science	NOUN
ijassa-659	562	4	and	and	CCONJ
ijassa-659	562	5	appl	appl	NOUN
ijassa-659	562	6	.	.	PUNCT
ijassa-659	563	1	(	(	PUNCT
ijassa-659	563	2	2019	2019	NUM
ijassa-659	563	3	)	)	PUNCT
ijassa-659	563	4	unlabeled	unlabele	VERB
ijassa-659	563	5	data	datum	NOUN
ijassa-659	563	6	so	so	SCONJ
ijassa-659	563	7	that	that	SCONJ
ijassa-659	563	8	the	the	DET
ijassa-659	563	9	marginal	marginal	ADJ
ijassa-659	563	10	probability	probability	NOUN
ijassa-659	563	11	distribution	distribution	NOUN
ijassa-659	563	12	represented	represent	VERB
ijassa-659	563	13	by	by	ADP
ijassa-659	563	14	the	the	DET
ijassa-659	563	15	labeled	label	VERB
ijassa-659	563	16	data	datum	NOUN
ijassa-659	563	17	after	after	ADP
ijassa-659	563	18	annotation	annotation	NOUN
ijassa-659	563	19	,	,	PUNCT
ijassa-659	563	20	is	be	AUX
ijassa-659	563	21	similar	similar	ADJ
ijassa-659	563	22	to	to	ADP
ijassa-659	563	23	the	the	DET
ijassa-659	563	24	marginal	marginal	ADJ
ijassa-659	563	25	probability	probability	NOUN
ijassa-659	563	26	distribution	distribution	NOUN
ijassa-659	563	27	represented	represent	VERB
ijassa-659	563	28	by	by	ADP
ijassa-659	563	29	the	the	DET
ijassa-659	563	30	unlabeled	unlabeled	ADJ
ijassa-659	563	31	data	datum	NOUN
ijassa-659	563	32	.	.	PUNCT
ijassa-659	564	1	moreover	moreover	ADV
ijassa-659	564	2	,	,	PUNCT
ijassa-659	564	3	incrementally	incrementally	ADV
ijassa-659	564	4	expands	expand	VERB
ijassa-659	564	5	the	the	DET
ijassa-659	564	6	neighborhoods	neighborhood	NOUN
ijassa-659	564	7	by	by	ADP
ijassa-659	564	8	using	use	VERB
ijassa-659	564	9	the	the	DET
ijassa-659	564	10	selected	select	VERB
ijassa-659	564	11	queries	query	NOUN
ijassa-659	564	12	.	.	PUNCT
ijassa-659	565	1	experiments	experiment	NOUN
ijassa-659	565	2	carried	carry	VERB
ijassa-659	565	3	out	out	ADP
ijassa-659	565	4	on	on	ADP
ijassa-659	565	5	different	different	ADJ
ijassa-659	565	6	real	real	ADJ
ijassa-659	565	7	datasets	dataset	NOUN
ijassa-659	565	8	show	show	VERB
ijassa-659	565	9	that	that	SCONJ
ijassa-659	565	10	the	the	DET
ijassa-659	565	11	constraints	constraint	NOUN
ijassa-659	565	12	selected	select	VERB
ijassa-659	565	13	by	by	ADP
ijassa-659	565	14	the	the	DET
ijassa-659	565	15	proposed	propose	VERB
ijassa-659	565	16	active	active	ADJ
ijassa-659	565	17	learning	learning	NOUN
ijassa-659	565	18	process	process	NOUN
ijassa-659	565	19	are	be	AUX
ijassa-659	565	20	generally	generally	ADV
ijassa-659	565	21	more	more	ADV
ijassa-659	565	22	beneficial	beneficial	ADJ
ijassa-659	565	23	for	for	ADP
ijassa-659	565	24	constraint	constraint	NOUN
ijassa-659	565	25	-	-	PUNCT
ijassa-659	565	26	based	base	VERB
ijassa-659	565	27	clustering	clustering	ADJ
ijassa-659	565	28	algorithms	algorithm	NOUN
ijassa-659	565	29	than	than	ADP
ijassa-659	565	30	those	those	PRON
ijassa-659	565	31	provided	provide	VERB
ijassa-659	565	32	by	by	ADP
ijassa-659	565	33	the	the	DET
ijassa-659	565	34	npu	npu	PROPN
ijassa-659	565	35	and	and	CCONJ
ijassa-659	565	36	asc	asc	PROPN
ijassa-659	565	37	methods	method	NOUN
ijassa-659	565	38	and	and	CCONJ
ijassa-659	565	39	achieving	achieve	VERB
ijassa-659	565	40	high	high	ADJ
ijassa-659	565	41	clustering	clustering	NOUN
ijassa-659	565	42	performance	performance	NOUN
ijassa-659	565	43	with	with	ADP
ijassa-659	565	44	minimizing	minimize	VERB
ijassa-659	565	45	the	the	DET
ijassa-659	565	46	amount	amount	NOUN
ijassa-659	565	47	of	of	ADP
ijassa-659	565	48	computation	computation	NOUN
ijassa-659	565	49	for	for	ADP
ijassa-659	565	50	selecting	select	VERB
ijassa-659	565	51	the	the	DET
ijassa-659	565	52	active	active	ADJ
ijassa-659	565	53	constraints	constraint	NOUN
ijassa-659	565	54	.	.	PUNCT
ijassa-659	566	1	however	however	ADV
ijassa-659	566	2	,	,	PUNCT
ijassa-659	566	3	the	the	DET
ijassa-659	566	4	efficiency	efficiency	NOUN
ijassa-659	566	5	of	of	ADP
ijassa-659	566	6	the	the	DET
ijassa-659	566	7	proposed	propose	VERB
ijassa-659	566	8	method	method	NOUN
ijassa-659	566	9	strongly	strongly	ADV
ijassa-659	566	10	depends	depend	VERB
ijassa-659	566	11	on	on	ADP
ijassa-659	566	12	the	the	DET
ijassa-659	566	13	value	value	NOUN
ijassa-659	566	14	of	of	ADP
ijassa-659	566	15	the	the	DET
ijassa-659	566	16	batch	batch	NOUN
ijassa-659	566	17	size	size	NOUN
ijassa-659	566	18	b.	b.	NOUN
ijassa-659	566	19	in	in	ADP
ijassa-659	566	20	future	future	ADJ
ijassa-659	566	21	work	work	NOUN
ijassa-659	566	22	,	,	PUNCT
ijassa-659	566	23	we	we	PRON
ijassa-659	566	24	are	be	AUX
ijassa-659	566	25	interesting	interesting	ADJ
ijassa-659	566	26	to	to	PART
ijassa-659	566	27	automatically	automatically	ADV
ijassa-659	566	28	identify	identify	VERB
ijassa-659	566	29	the	the	DET
ijassa-659	566	30	best	good	ADJ
ijassa-659	566	31	value	value	NOUN
ijassa-659	566	32	of	of	ADP
ijassa-659	566	33	the	the	DET
ijassa-659	566	34	batch	batch	NOUN
ijassa-659	566	35	size	size	NOUN
ijassa-659	566	36	b	b	NOUN
ijassa-659	566	37	that	that	PRON
ijassa-659	566	38	achieves	achieve	VERB
ijassa-659	566	39	the	the	DET
ijassa-659	566	40	best	good	ADJ
ijassa-659	566	41	clustering	clustering	ADJ
ijassa-659	566	42	performance	performance	NOUN
ijassa-659	566	43	.	.	PUNCT
ijassa-659	567	1	also	also	ADV
ijassa-659	567	2	,	,	PUNCT
ijassa-659	567	3	the	the	DET
ijassa-659	567	4	problem	problem	NOUN
ijassa-659	567	5	of	of	ADP
ijassa-659	567	6	clustering	cluster	VERB
ijassa-659	567	7	the	the	DET
ijassa-659	567	8	big	big	ADJ
ijassa-659	567	9	data	datum	NOUN
ijassa-659	567	10	with	with	ADP
ijassa-659	567	11	an	an	DET
ijassa-659	567	12	incrementally	incrementally	ADV
ijassa-659	567	13	growing	grow	VERB
ijassa-659	567	14	constraint	constraint	NOUN
ijassa-659	567	15	set	set	NOUN
ijassa-659	567	16	.	.	PUNCT
ijassa-659	568	1	to	to	PART
ijassa-659	568	2	address	address	VERB
ijassa-659	568	3	this	this	DET
ijassa-659	568	4	problem	problem	NOUN
ijassa-659	568	5	,	,	PUNCT
ijassa-659	568	6	we	we	PRON
ijassa-659	568	7	interest	interest	VERB
ijassa-659	568	8	to	to	PART
ijassa-659	568	9	consider	consider	VERB
ijassa-659	568	10	an	an	DET
ijassa-659	568	11	incremental	incremental	ADJ
ijassa-659	568	12	semi	semi	ADJ
ijassa-659	568	13	-	-	ADJ
ijassa-659	568	14	supervised	supervised	ADJ
ijassa-659	568	15	clustering	clustering	NOUN
ijassa-659	568	16	method	method	NOUN
ijassa-659	568	17	.	.	PUNCT
ijassa-659	569	1	references	reference	NOUN
ijassa-659	569	2	1	1	NUM
ijassa-659	569	3	.	.	PUNCT
ijassa-659	569	4	basu	basu	PROPN
ijassa-659	569	5	,	,	PUNCT
ijassa-659	569	6	s.	s.	PROPN
ijassa-659	569	7	,	,	PUNCT
ijassa-659	569	8	banerjee	banerjee	PROPN
ijassa-659	569	9	,	,	PUNCT
ijassa-659	569	10	a.	a.	PROPN
ijassa-659	569	11	,	,	PUNCT
ijassa-659	569	12	&	&	CCONJ
ijassa-659	569	13	mooney	mooney	PROPN
ijassa-659	569	14	,	,	PUNCT
ijassa-659	569	15	r.	r.	PROPN
ijassa-659	569	16	j.	j.	PROPN
ijassa-659	569	17	(	(	PUNCT
ijassa-659	569	18	2004	2004	NUM
ijassa-659	569	19	,	,	PUNCT
ijassa-659	569	20	april	april	PROPN
ijassa-659	569	21	)	)	PUNCT
ijassa-659	569	22	.	.	PUNCT
ijassa-659	570	1	active	active	ADJ
ijassa-659	570	2	semi	semi	ADJ
ijassa-659	570	3	-	-	NOUN
ijassa-659	570	4	supervision	supervision	NOUN
ijassa-659	570	5	for	for	ADP
ijassa-659	570	6	pairwise	pairwise	NOUN
ijassa-659	570	7	constrained	constrain	VERB
ijassa-659	570	8	clustering	clustering	NOUN
ijassa-659	570	9	.	.	PUNCT
ijassa-659	571	1	in	in	ADP
ijassa-659	571	2	proceedings	proceeding	NOUN
ijassa-659	571	3	of	of	ADP
ijassa-659	571	4	the	the	DET
ijassa-659	571	5	2004	2004	NUM
ijassa-659	571	6	siam	siam	PROPN
ijassa-659	571	7	international	international	ADJ
ijassa-659	571	8	conference	conference	NOUN
ijassa-659	571	9	on	on	ADP
ijassa-659	571	10	data	datum	NOUN
ijassa-659	571	11	mining	mining	NOUN
ijassa-659	571	12	(	(	PUNCT
ijassa-659	571	13	pp	pp	ADJ
ijassa-659	571	14	.	.	PUNCT
ijassa-659	572	1	333	333	NUM
ijassa-659	572	2	-	-	SYM
ijassa-659	572	3	344	344	NUM
ijassa-659	572	4	)	)	PUNCT
ijassa-659	572	5	.	.	PUNCT
ijassa-659	573	1	society	society	NOUN
ijassa-659	573	2	for	for	ADP
ijassa-659	573	3	industrial	industrial	ADJ
ijassa-659	573	4	and	and	CCONJ
ijassa-659	573	5	applied	applied	ADJ
ijassa-659	573	6	mathematics	mathematic	NOUN
ijassa-659	573	7	.	.	PUNCT
ijassa-659	574	1	2	2	X
ijassa-659	574	2	.	.	X
ijassa-659	574	3	yu	yu	PROPN
ijassa-659	574	4	,	,	PUNCT
ijassa-659	574	5	c.	c.	PROPN
ijassa-659	574	6	,	,	PUNCT
ijassa-659	574	7	&	&	CCONJ
ijassa-659	574	8	hansen	hansen	PROPN
ijassa-659	574	9	,	,	PUNCT
ijassa-659	574	10	j.	j.	PROPN
ijassa-659	574	11	h.	h.	PROPN
ijassa-659	574	12	(	(	PUNCT
ijassa-659	574	13	2017	2017	NUM
ijassa-659	574	14	)	)	PUNCT
ijassa-659	574	15	.	.	PUNCT
ijassa-659	575	1	active	active	ADJ
ijassa-659	575	2	learning	learning	NOUN
ijassa-659	575	3	based	base	VERB
ijassa-659	575	4	constrained	constrain	VERB
ijassa-659	575	5	clustering	clustering	NOUN
ijassa-659	575	6	for	for	ADP
ijassa-659	575	7	speaker	speaker	NOUN
ijassa-659	575	8	diarization	diarization	NOUN
ijassa-659	575	9	.	.	PUNCT
ijassa-659	576	1	ieee	ieee	PROPN
ijassa-659	576	2	/	/	SYM
ijassa-659	576	3	acm	acm	PROPN
ijassa-659	576	4	transactions	transaction	NOUN
ijassa-659	576	5	on	on	ADP
ijassa-659	576	6	audio	audio	NOUN
ijassa-659	576	7	,	,	PUNCT
ijassa-659	576	8	speech	speech	NOUN
ijassa-659	576	9	,	,	PUNCT
ijassa-659	576	10	and	and	CCONJ
ijassa-659	576	11	language	language	NOUN
ijassa-659	576	12	processing	processing	NOUN
ijassa-659	576	13	,	,	PUNCT
ijassa-659	576	14	25(11	25(11	NUM
ijassa-659	576	15	)	)	PUNCT
ijassa-659	576	16	,	,	PUNCT
ijassa-659	576	17	2188	2188	NUM
ijassa-659	576	18	-	-	SYM
ijassa-659	576	19	2198	2198	NUM
ijassa-659	576	20	.	.	PUNCT
ijassa-659	577	1	3	3	X
ijassa-659	577	2	.	.	X
ijassa-659	577	3	van	van	NOUN
ijassa-659	577	4	craenendonck	craenendonck	NOUN
ijassa-659	577	5	,	,	PUNCT
ijassa-659	577	6	t.	t.	PROPN
ijassa-659	577	7	,	,	PUNCT
ijassa-659	577	8	&	&	CCONJ
ijassa-659	577	9	blockeel	blockeel	PROPN
ijassa-659	577	10	,	,	PUNCT
ijassa-659	577	11	h.	h.	PROPN
ijassa-659	577	12	(	(	PUNCT
ijassa-659	577	13	2017	2017	NUM
ijassa-659	577	14	)	)	PUNCT
ijassa-659	577	15	.	.	PUNCT
ijassa-659	578	1	constraint	constraint	NOUN
ijassa-659	578	2	-	-	PUNCT
ijassa-659	578	3	based	base	VERB
ijassa-659	578	4	clustering	clustering	ADJ
ijassa-659	578	5	selection	selection	NOUN
ijassa-659	578	6	.	.	PUNCT
ijassa-659	579	1	machine	machine	NOUN
ijassa-659	579	2	learning	learning	NOUN
ijassa-659	579	3	,	,	PUNCT
ijassa-659	579	4	106(9	106(9	NUM
ijassa-659	579	5	-	-	SYM
ijassa-659	579	6	10	10	NUM
ijassa-659	579	7	)	)	PUNCT
ijassa-659	579	8	,	,	PUNCT
ijassa-659	579	9	1497	1497	NUM
ijassa-659	579	10	-	-	SYM
ijassa-659	579	11	1521	1521	NUM
ijassa-659	579	12	.	.	PUNCT
ijassa-659	580	1	4	4	NUM
ijassa-659	580	2	.	.	X
ijassa-659	580	3	yu	yu	PROPN
ijassa-659	580	4	,	,	PUNCT
ijassa-659	580	5	z.	z.	PROPN
ijassa-659	580	6	,	,	PUNCT
ijassa-659	580	7	luo	luo	PROPN
ijassa-659	580	8	,	,	PUNCT
ijassa-659	580	9	p.	p.	NOUN
ijassa-659	580	10	,	,	PUNCT
ijassa-659	580	11	you	you	PRON
ijassa-659	580	12	,	,	PUNCT
ijassa-659	580	13	j.	j.	PROPN
ijassa-659	580	14	,	,	PUNCT
ijassa-659	580	15	wong	wong	PROPN
ijassa-659	580	16	,	,	PUNCT
ijassa-659	580	17	h.	h.	PROPN
ijassa-659	580	18	s.	s.	PROPN
ijassa-659	580	19	,	,	PUNCT
ijassa-659	580	20	leung	leung	PROPN
ijassa-659	580	21	,	,	PUNCT
ijassa-659	580	22	h.	h.	PROPN
ijassa-659	580	23	,	,	PUNCT
ijassa-659	580	24	wu	wu	PROPN
ijassa-659	580	25	,	,	PUNCT
ijassa-659	580	26	s.	s.	PROPN
ijassa-659	580	27	,	,	PUNCT
ijassa-659	580	28	&	&	CCONJ
ijassa-659	580	29	han	han	PROPN
ijassa-659	580	30	,	,	PUNCT
ijassa-659	580	31	g.	g.	PROPN
ijassa-659	580	32	(	(	PUNCT
ijassa-659	580	33	2015	2015	NUM
ijassa-659	580	34	)	)	PUNCT
ijassa-659	580	35	.	.	PUNCT
ijassa-659	581	1	incremental	incremental	ADJ
ijassa-659	581	2	semi	semi	ADJ
ijassa-659	581	3	-	-	ADJ
ijassa-659	581	4	supervised	supervised	ADJ
ijassa-659	581	5	clustering	clustering	NOUN
ijassa-659	581	6	ensemble	ensemble	ADJ
ijassa-659	581	7	for	for	ADP
ijassa-659	581	8	high	high	ADJ
ijassa-659	581	9	dimensional	dimensional	ADJ
ijassa-659	581	10	data	datum	NOUN
ijassa-659	581	11	clustering	clustering	NOUN
ijassa-659	581	12	.	.	PUNCT
ijassa-659	582	1	ieee	ieee	NOUN
ijassa-659	582	2	transactions	transaction	NOUN
ijassa-659	582	3	on	on	ADP
ijassa-659	582	4	knowledge	knowledge	NOUN
ijassa-659	582	5	and	and	CCONJ
ijassa-659	582	6	data	datum	NOUN
ijassa-659	582	7	engineering	engineering	NOUN
ijassa-659	582	8	,	,	PUNCT
ijassa-659	582	9	28(3	28(3	NUM
ijassa-659	582	10	)	)	PUNCT
ijassa-659	582	11	,	,	PUNCT
ijassa-659	582	12	701	701	NUM
ijassa-659	582	13	-	-	SYM
ijassa-659	582	14	714	714	NUM
ijassa-659	582	15	.	.	PUNCT
ijassa-659	583	1	5	5	NUM
ijassa-659	583	2	.	.	X
ijassa-659	583	3	borgwardt	borgwardt	PROPN
ijassa-659	583	4	,	,	PUNCT
ijassa-659	583	5	k.	k.	PROPN
ijassa-659	583	6	m.	m.	PROPN
ijassa-659	583	7	,	,	PUNCT
ijassa-659	583	8	gretton	gretton	PROPN
ijassa-659	583	9	,	,	PUNCT
ijassa-659	583	10	a.	a.	PROPN
ijassa-659	583	11	,	,	PUNCT
ijassa-659	583	12	rasch	rasch	PROPN
ijassa-659	583	13	,	,	PUNCT
ijassa-659	583	14	m.	m.	NOUN
ijassa-659	583	15	j.	j.	PROPN
ijassa-659	583	16	,	,	PUNCT
ijassa-659	583	17	kriegel	kriegel	PROPN
ijassa-659	583	18	,	,	PUNCT
ijassa-659	583	19	h.	h.	PROPN
ijassa-659	583	20	p.	p.	PROPN
ijassa-659	583	21	,	,	PUNCT
ijassa-659	583	22	schölkopf	schölkopf	VERB
ijassa-659	583	23	,	,	PUNCT
ijassa-659	583	24	b.	b.	PROPN
ijassa-659	583	25	,	,	PUNCT
ijassa-659	583	26	&	&	CCONJ
ijassa-659	583	27	smola	smola	PROPN
ijassa-659	583	28	,	,	PUNCT
ijassa-659	583	29	a.	a.	PROPN
ijassa-659	583	30	j.	j.	PROPN
ijassa-659	583	31	(	(	PUNCT
ijassa-659	583	32	2006	2006	NUM
ijassa-659	583	33	)	)	PUNCT
ijassa-659	583	34	.	.	PUNCT
ijassa-659	584	1	integrating	integrate	VERB
ijassa-659	584	2	structured	structured	ADJ
ijassa-659	584	3	biological	biological	ADJ
ijassa-659	584	4	data	datum	NOUN
ijassa-659	584	5	by	by	ADP
ijassa-659	584	6	kernel	kernel	PROPN
ijassa-659	584	7	maximum	maximum	PROPN
ijassa-659	584	8	mean	mean	PROPN
ijassa-659	584	9	discrepancy	discrepancy	NOUN
ijassa-659	584	10	.	.	PUNCT
ijassa-659	585	1	bioinformatics	bioinformatics	NOUN
ijassa-659	585	2	,	,	PUNCT
ijassa-659	585	3	22(14	22(14	PROPN
ijassa-659	585	4	)	)	PUNCT
ijassa-659	585	5	,	,	PUNCT
ijassa-659	585	6	e49	e49	NOUN
ijassa-659	585	7	-	-	PUNCT
ijassa-659	585	8	e57	e57	NOUN
ijassa-659	585	9	.	.	PUNCT
ijassa-659	586	1	6	6	NUM
ijassa-659	586	2	.	.	X
ijassa-659	586	3	sriperumbudur	sriperumbudur	PROPN
ijassa-659	586	4	,	,	PUNCT
ijassa-659	586	5	b.	b.	PROPN
ijassa-659	586	6	k.	k.	PROPN
ijassa-659	586	7	,	,	PUNCT
ijassa-659	586	8	gretton	gretton	PROPN
ijassa-659	586	9	,	,	PUNCT
ijassa-659	586	10	a.	a.	NOUN
ijassa-659	586	11	,	,	PUNCT
ijassa-659	586	12	fukumizu	fukumizu	PROPN
ijassa-659	586	13	,	,	PUNCT
ijassa-659	586	14	k.	k.	PROPN
ijassa-659	586	15	,	,	PUNCT
ijassa-659	586	16	schölkopf	schölkopf	VERB
ijassa-659	586	17	,	,	PUNCT
ijassa-659	586	18	b.	b.	PROPN
ijassa-659	586	19	,	,	PUNCT
ijassa-659	586	20	&	&	CCONJ
ijassa-659	586	21	lanckriet	lanckriet	PROPN
ijassa-659	586	22	,	,	PUNCT
ijassa-659	586	23	g.	g.	PROPN
ijassa-659	586	24	r.	r.	PROPN
ijassa-659	586	25	(	(	PUNCT
ijassa-659	586	26	2010	2010	NUM
ijassa-659	586	27	)	)	PUNCT
ijassa-659	586	28	.	.	PUNCT
ijassa-659	587	1	hilbert	hilbert	NOUN
ijassa-659	587	2	space	space	NOUN
ijassa-659	587	3	embeddings	embedding	NOUN
ijassa-659	587	4	and	and	CCONJ
ijassa-659	587	5	metrics	metric	NOUN
ijassa-659	587	6	on	on	ADP
ijassa-659	587	7	probability	probability	NOUN
ijassa-659	587	8	measures	measure	NOUN
ijassa-659	587	9	.	.	PUNCT
ijassa-659	588	1	journal	journal	NOUN
ijassa-659	588	2	of	of	ADP
ijassa-659	588	3	machine	machine	NOUN
ijassa-659	588	4	learning	learn	VERB
ijassa-659	588	5	research	research	NOUN
ijassa-659	588	6	,	,	PUNCT
ijassa-659	588	7	11(apr	11(apr	NOUN
ijassa-659	588	8	)	)	PUNCT
ijassa-659	588	9	,	,	PUNCT
ijassa-659	588	10	1517	1517	NUM
ijassa-659	588	11	-	-	SYM
ijassa-659	588	12	1561	1561	NUM
ijassa-659	588	13	.	.	PUNCT
ijassa-659	589	1	7	7	X
ijassa-659	589	2	.	.	X
ijassa-659	589	3	calma	calma	PROPN
ijassa-659	589	4	,	,	PUNCT
ijassa-659	589	5	a.	a.	NOUN
ijassa-659	589	6	,	,	PUNCT
ijassa-659	589	7	reitmaier	reitmaier	NOUN
ijassa-659	589	8	,	,	PUNCT
ijassa-659	589	9	t.	t.	PROPN
ijassa-659	589	10	,	,	PUNCT
ijassa-659	589	11	&	&	CCONJ
ijassa-659	589	12	sick	sick	ADJ
ijassa-659	589	13	,	,	PUNCT
ijassa-659	589	14	b.	b.	PROPN
ijassa-659	589	15	(	(	PUNCT
ijassa-659	589	16	2018	2018	NUM
ijassa-659	589	17	)	)	PUNCT
ijassa-659	589	18	.	.	PUNCT
ijassa-659	590	1	semi	semi	ADJ
ijassa-659	590	2	-	-	ADJ
ijassa-659	590	3	supervised	supervised	ADJ
ijassa-659	590	4	active	active	ADJ
ijassa-659	590	5	learning	learning	NOUN
ijassa-659	590	6	for	for	ADP
ijassa-659	590	7	support	support	NOUN
ijassa-659	590	8	vector	vector	NOUN
ijassa-659	590	9	machines	machine	NOUN
ijassa-659	590	10	:	:	PUNCT
ijassa-659	590	11	a	a	DET
ijassa-659	590	12	novel	novel	ADJ
ijassa-659	590	13	approach	approach	NOUN
ijassa-659	590	14	that	that	PRON
ijassa-659	590	15	exploits	exploit	VERB
ijassa-659	590	16	structure	structure	NOUN
ijassa-659	590	17	information	information	NOUN
ijassa-659	590	18	in	in	ADP
ijassa-659	590	19	data	data	PROPN
ijassa-659	590	20	.	.	PUNCT
ijassa-659	591	1	information	information	NOUN
ijassa-659	591	2	sciences	sciences	PROPN
ijassa-659	591	3	,	,	PUNCT
ijassa-659	591	4	456	456	NUM
ijassa-659	591	5	,	,	PUNCT
ijassa-659	591	6	13	13	NUM
ijassa-659	591	7	-	-	SYM
ijassa-659	591	8	33	33	NUM
ijassa-659	591	9	.	.	NOUN
ijassa-659	591	10	8	8	NUM
ijassa-659	591	11	.	.	X
ijassa-659	592	1	xiong	xiong	PROPN
ijassa-659	592	2	,	,	PUNCT
ijassa-659	592	3	c.	c.	PROPN
ijassa-659	592	4	,	,	PUNCT
ijassa-659	592	5	johnson	johnson	PROPN
ijassa-659	592	6	,	,	PUNCT
ijassa-659	592	7	d.	d.	PROPN
ijassa-659	592	8	m.	m.	PROPN
ijassa-659	592	9	,	,	PUNCT
ijassa-659	592	10	&	&	CCONJ
ijassa-659	592	11	corso	corso	PROPN
ijassa-659	592	12	,	,	PUNCT
ijassa-659	592	13	j.	j.	PROPN
ijassa-659	592	14	j.	j.	PROPN
ijassa-659	592	15	(	(	PUNCT
ijassa-659	592	16	2016	2016	NUM
ijassa-659	592	17	)	)	PUNCT
ijassa-659	592	18	.	.	PUNCT
ijassa-659	593	1	active	active	ADJ
ijassa-659	593	2	clustering	clustering	NOUN
ijassa-659	593	3	with	with	ADP
ijassa-659	593	4	model	model	NOUN
ijassa-659	593	5	-	-	PUNCT
ijassa-659	593	6	based	base	VERB
ijassa-659	593	7	uncertainty	uncertainty	NOUN
ijassa-659	593	8	reduction	reduction	NOUN
ijassa-659	593	9	.	.	PUNCT
ijassa-659	594	1	ieee	ieee	NOUN
ijassa-659	594	2	transactions	transaction	NOUN
ijassa-659	594	3	on	on	ADP
ijassa-659	594	4	pattern	pattern	NOUN
ijassa-659	594	5	analysis	analysis	NOUN
ijassa-659	594	6	and	and	CCONJ
ijassa-659	594	7	machine	machine	NOUN
ijassa-659	594	8	intelligence	intelligence	NOUN
ijassa-659	594	9	,	,	PUNCT
ijassa-659	594	10	39(1	39(1	NUM
ijassa-659	594	11	)	)	PUNCT
ijassa-659	594	12	,	,	PUNCT
ijassa-659	594	13	5	5	NUM
ijassa-659	594	14	-	-	SYM
ijassa-659	594	15	17	17	NUM
ijassa-659	594	16	.	.	PUNCT
ijassa-659	595	1	9	9	NUM
ijassa-659	595	2	.	.	X
ijassa-659	595	3	ngoc	ngoc	PROPN
ijassa-659	595	4	,	,	PUNCT
ijassa-659	595	5	m.	m.	NOUN
ijassa-659	595	6	t.	t.	PROPN
ijassa-659	595	7	,	,	PUNCT
ijassa-659	595	8	&	&	CCONJ
ijassa-659	595	9	park	park	PROPN
ijassa-659	595	10	,	,	PUNCT
ijassa-659	595	11	d.	d.	PROPN
ijassa-659	595	12	c.	c.	PROPN
ijassa-659	595	13	(	(	PUNCT
ijassa-659	595	14	2018	2018	NUM
ijassa-659	595	15	)	)	PUNCT
ijassa-659	595	16	.	.	PUNCT
ijassa-659	596	1	centroid	centroid	PROPN
ijassa-659	596	2	neural	neural	ADJ
ijassa-659	596	3	network	network	NOUN
ijassa-659	596	4	with	with	ADP
ijassa-659	596	5	pairwise	pairwise	NOUN
ijassa-659	596	6	constraints	constraint	NOUN
ijassa-659	596	7	for	for	ADP
ijassa-659	596	8	semi	semi	ADJ
ijassa-659	596	9	-	-	ADJ
ijassa-659	596	10	supervised	supervised	ADJ
ijassa-659	596	11	learning	learning	NOUN
ijassa-659	596	12	.	.	PUNCT
ijassa-659	597	1	neural	neural	ADJ
ijassa-659	597	2	processing	processing	NOUN
ijassa-659	597	3	letters	letter	NOUN
ijassa-659	597	4	,	,	PUNCT
ijassa-659	597	5	48(3	48(3	PROPN
ijassa-659	597	6	)	)	PUNCT
ijassa-659	597	7	,	,	PUNCT
ijassa-659	597	8	1721	1721	NUM
ijassa-659	597	9	-	-	SYM
ijassa-659	597	10	1747	1747	NUM
ijassa-659	597	11	.	.	PUNCT
ijassa-659	598	1	improving	improve	VERB
ijassa-659	598	2	semi	semi	ADJ
ijassa-659	598	3	-	-	ADJ
ijassa-659	598	4	supervised	supervised	ADJ
ijassa-659	598	5	clustering	clustering	ADJ
ijassa-659	598	6	algorithms	algorithm	NOUN
ijassa-659	598	7	with	with	ADP
ijassa-659	598	8	active	active	ADJ
ijassa-659	598	9	query	query	NOUN
ijassa-659	598	10	selection	selection	NOUN
ijassa-659	598	11	43	43	NUM
ijassa-659	598	12	copyright	copyright	NOUN
ijassa-659	598	13	©	©	PROPN
ijassa-659	598	14	2019	2019	NUM
ijassa-659	598	15	assa	assa	PROPN
ijassa-659	598	16	adv	adv	PROPN
ijassa-659	598	17	.	.	PUNCT
ijassa-659	599	1	in	in	ADP
ijassa-659	599	2	systems	system	NOUN
ijassa-659	599	3	science	science	NOUN
ijassa-659	599	4	and	and	CCONJ
ijassa-659	599	5	appl	appl	NOUN
ijassa-659	599	6	.	.	PUNCT
ijassa-659	600	1	(	(	PUNCT
ijassa-659	600	2	2019	2019	NUM
ijassa-659	600	3	)	)	PUNCT
ijassa-659	600	4	10	10	NUM
ijassa-659	600	5	.	.	PUNCT
ijassa-659	601	1	zeng	zeng	PROPN
ijassa-659	601	2	,	,	PUNCT
ijassa-659	601	3	h.	h.	PROPN
ijassa-659	601	4	,	,	PUNCT
ijassa-659	601	5	&	&	CCONJ
ijassa-659	601	6	cheung	cheung	PROPN
ijassa-659	601	7	,	,	PUNCT
ijassa-659	601	8	y.	y.	PROPN
ijassa-659	601	9	m.	m.	PROPN
ijassa-659	601	10	(	(	PUNCT
ijassa-659	601	11	2011	2011	NUM
ijassa-659	601	12	)	)	PUNCT
ijassa-659	601	13	.	.	PUNCT
ijassa-659	602	1	semi	semi	ADJ
ijassa-659	602	2	-	-	ADJ
ijassa-659	602	3	supervised	supervised	ADJ
ijassa-659	602	4	maximum	maximum	ADJ
ijassa-659	602	5	margin	margin	NOUN
ijassa-659	602	6	clustering	cluster	VERB
ijassa-659	602	7	with	with	ADP
ijassa-659	602	8	pairwise	pairwise	NOUN
ijassa-659	602	9	constraints	constraint	NOUN
ijassa-659	602	10	.	.	PUNCT
ijassa-659	603	1	ieee	ieee	NOUN
ijassa-659	603	2	transactions	transaction	NOUN
ijassa-659	603	3	on	on	ADP
ijassa-659	603	4	knowledge	knowledge	NOUN
ijassa-659	603	5	and	and	CCONJ
ijassa-659	603	6	data	datum	NOUN
ijassa-659	603	7	engineering	engineering	NOUN
ijassa-659	603	8	,	,	PUNCT
ijassa-659	603	9	24(5	24(5	NUM
ijassa-659	603	10	)	)	PUNCT
ijassa-659	603	11	,	,	PUNCT
ijassa-659	603	12	926	926	NUM
ijassa-659	603	13	-	-	SYM
ijassa-659	603	14	939	939	NUM
ijassa-659	603	15	.	.	PUNCT
ijassa-659	604	1	11	11	NUM
ijassa-659	604	2	.	.	X
ijassa-659	604	3	wang	wang	PROPN
ijassa-659	604	4	,	,	PUNCT
ijassa-659	604	5	g.	g.	PROPN
ijassa-659	604	6	,	,	PUNCT
ijassa-659	604	7	hwang	hwang	PROPN
ijassa-659	604	8	,	,	PUNCT
ijassa-659	604	9	j.	j.	PROPN
ijassa-659	604	10	n.	n.	PROPN
ijassa-659	604	11	,	,	PUNCT
ijassa-659	604	12	rose	rise	VERB
ijassa-659	604	13	,	,	PUNCT
ijassa-659	604	14	c.	c.	PROPN
ijassa-659	604	15	,	,	PUNCT
ijassa-659	604	16	&	&	CCONJ
ijassa-659	604	17	wallace	wallace	PROPN
ijassa-659	604	18	,	,	PUNCT
ijassa-659	604	19	f.	f.	PROPN
ijassa-659	604	20	(	(	PUNCT
ijassa-659	604	21	2018	2018	NUM
ijassa-659	604	22	)	)	PUNCT
ijassa-659	604	23	.	.	PUNCT
ijassa-659	605	1	uncertainty	uncertainty	NOUN
ijassa-659	605	2	-	-	PUNCT
ijassa-659	605	3	based	base	VERB
ijassa-659	605	4	active	active	ADJ
ijassa-659	605	5	learning	learning	NOUN
ijassa-659	605	6	via	via	ADP
ijassa-659	605	7	sparse	sparse	ADJ
ijassa-659	605	8	modeling	modeling	NOUN
ijassa-659	605	9	for	for	ADP
ijassa-659	605	10	image	image	NOUN
ijassa-659	605	11	classification	classification	NOUN
ijassa-659	605	12	.	.	PUNCT
ijassa-659	606	1	ieee	ieee	NOUN
ijassa-659	606	2	transactions	transaction	NOUN
ijassa-659	606	3	on	on	ADP
ijassa-659	606	4	image	image	NOUN
ijassa-659	606	5	processing	processing	NOUN
ijassa-659	606	6	,	,	PUNCT
ijassa-659	606	7	28(1	28(1	NOUN
ijassa-659	606	8	)	)	PUNCT
ijassa-659	606	9	,	,	PUNCT
ijassa-659	606	10	316	316	NUM
ijassa-659	606	11	-	-	SYM
ijassa-659	606	12	329	329	NUM
ijassa-659	606	13	.	.	PUNCT
ijassa-659	607	1	12	12	NUM
ijassa-659	607	2	.	.	PUNCT
ijassa-659	607	3	mallapragada	mallapragada	NOUN
ijassa-659	607	4	,	,	PUNCT
ijassa-659	607	5	p.	p.	PROPN
ijassa-659	607	6	k.	k.	PROPN
ijassa-659	607	7	,	,	PUNCT
ijassa-659	607	8	jin	jin	PROPN
ijassa-659	607	9	,	,	PUNCT
ijassa-659	607	10	r.	r.	PROPN
ijassa-659	607	11	,	,	PUNCT
ijassa-659	607	12	&	&	CCONJ
ijassa-659	607	13	jain	jain	PROPN
ijassa-659	607	14	,	,	PUNCT
ijassa-659	607	15	a.	a.	PROPN
ijassa-659	607	16	k.	k.	PROPN
ijassa-659	608	1	(	(	PUNCT
ijassa-659	608	2	2008	2008	NUM
ijassa-659	608	3	)	)	PUNCT
ijassa-659	608	4	.	.	PUNCT
ijassa-659	609	1	active	active	ADJ
ijassa-659	609	2	query	query	NOUN
ijassa-659	609	3	selection	selection	NOUN
ijassa-659	609	4	for	for	ADP
ijassa-659	609	5	semisupervised	semisupervised	ADJ
ijassa-659	609	6	clustering	clustering	NOUN
ijassa-659	609	7	.	.	PUNCT
ijassa-659	610	1	in	in	ADP
ijassa-659	610	2	2008	2008	NUM
ijassa-659	610	3	19th	19th	ADJ
ijassa-659	610	4	international	international	ADJ
ijassa-659	610	5	conference	conference	NOUN
ijassa-659	610	6	on	on	ADP
ijassa-659	610	7	pattern	pattern	NOUN
ijassa-659	610	8	recognition	recognition	NOUN
ijassa-659	610	9	,	,	PUNCT
ijassa-659	610	10	1	1	NUM
ijassa-659	610	11	-	-	SYM
ijassa-659	610	12	4	4	NUM
ijassa-659	610	13	.	.	NOUN
ijassa-659	610	14	13	13	NUM
ijassa-659	610	15	.	.	PUNCT
ijassa-659	611	1	xu	xu	PROPN
ijassa-659	611	2	,	,	PUNCT
ijassa-659	611	3	q.	q.	PROPN
ijassa-659	611	4	,	,	PUNCT
ijassa-659	611	5	&	&	CCONJ
ijassa-659	611	6	wagstaff	wagstaff	PROPN
ijassa-659	611	7	,	,	PUNCT
ijassa-659	611	8	k.	k.	PROPN
ijassa-659	611	9	l.	l.	PROPN
ijassa-659	611	10	(	(	PUNCT
ijassa-659	611	11	2005	2005	NUM
ijassa-659	611	12	)	)	PUNCT
ijassa-659	611	13	.	.	PUNCT
ijassa-659	612	1	active	active	ADJ
ijassa-659	612	2	constrained	constrain	VERB
ijassa-659	612	3	clustering	clustering	NOUN
ijassa-659	612	4	by	by	ADP
ijassa-659	612	5	examining	examine	VERB
ijassa-659	612	6	spectral	spectral	ADJ
ijassa-659	612	7	eigenvectors	eigenvector	NOUN
ijassa-659	612	8	.	.	PUNCT
ijassa-659	613	1	in	in	ADP
ijassa-659	613	2	international	international	ADJ
ijassa-659	613	3	conference	conference	NOUN
ijassa-659	613	4	on	on	ADP
ijassa-659	613	5	discovery	discovery	PROPN
ijassa-659	613	6	science	science	NOUN
ijassa-659	613	7	,	,	PUNCT
ijassa-659	613	8	294	294	NUM
ijassa-659	613	9	-	-	SYM
ijassa-659	613	10	307	307	NUM
ijassa-659	613	11	.	.	PUNCT
ijassa-659	613	12	14	14	NUM
ijassa-659	613	13	.	.	PUNCT
ijassa-659	614	1	wang	wang	PROPN
ijassa-659	614	2	,	,	PUNCT
ijassa-659	614	3	x.	x.	PROPN
ijassa-659	614	4	,	,	PUNCT
ijassa-659	614	5	&	&	CCONJ
ijassa-659	614	6	davidson	davidson	PROPN
ijassa-659	614	7	,	,	PUNCT
ijassa-659	614	8	i.	i.	PROPN
ijassa-659	614	9	(	(	PUNCT
ijassa-659	614	10	2010	2010	NUM
ijassa-659	614	11	)	)	PUNCT
ijassa-659	614	12	.	.	PUNCT
ijassa-659	615	1	active	active	ADJ
ijassa-659	615	2	spectral	spectral	ADJ
ijassa-659	615	3	clustering	clustering	NOUN
ijassa-659	615	4	.	.	PUNCT
ijassa-659	616	1	in	in	ADP
ijassa-659	616	2	2010	2010	NUM
ijassa-659	616	3	ieee	ieee	NOUN
ijassa-659	616	4	international	international	ADJ
ijassa-659	616	5	conference	conference	NOUN
ijassa-659	616	6	on	on	ADP
ijassa-659	616	7	data	datum	NOUN
ijassa-659	616	8	mining	mining	NOUN
ijassa-659	616	9	,	,	PUNCT
ijassa-659	616	10	561	561	NUM
ijassa-659	616	11	-	-	SYM
ijassa-659	616	12	568	568	NUM
ijassa-659	616	13	.	.	PUNCT
ijassa-659	616	14	15	15	NUM
ijassa-659	616	15	.	.	PUNCT
ijassa-659	617	1	vu	vu	X
ijassa-659	617	2	,	,	PUNCT
ijassa-659	617	3	v.	v.	ADP
ijassa-659	617	4	v.	v.	ADP
ijassa-659	617	5	,	,	PUNCT
ijassa-659	617	6	labroche	labroche	NOUN
ijassa-659	617	7	,	,	PUNCT
ijassa-659	617	8	n.	n.	NOUN
ijassa-659	617	9	,	,	PUNCT
ijassa-659	617	10	&	&	CCONJ
ijassa-659	617	11	bouchon	bouchon	PROPN
ijassa-659	617	12	-	-	PUNCT
ijassa-659	617	13	meunier	meunier	PROPN
ijassa-659	617	14	,	,	PUNCT
ijassa-659	617	15	b.	b.	PROPN
ijassa-659	617	16	(	(	PUNCT
ijassa-659	617	17	2012	2012	NUM
ijassa-659	617	18	)	)	PUNCT
ijassa-659	617	19	.	.	PUNCT
ijassa-659	618	1	improving	improve	VERB
ijassa-659	618	2	constrained	constrained	ADJ
ijassa-659	618	3	clustering	clustering	NOUN
ijassa-659	618	4	with	with	ADP
ijassa-659	618	5	active	active	ADJ
ijassa-659	618	6	query	query	NOUN
ijassa-659	618	7	selection	selection	NOUN
ijassa-659	618	8	.	.	PUNCT
ijassa-659	619	1	pattern	pattern	NOUN
ijassa-659	619	2	recognition	recognition	NOUN
ijassa-659	619	3	,	,	PUNCT
ijassa-659	619	4	45(4	45(4	NUM
ijassa-659	619	5	)	)	PUNCT
ijassa-659	619	6	,	,	PUNCT
ijassa-659	619	7	1749	1749	NUM
ijassa-659	619	8	-	-	SYM
ijassa-659	619	9	1758	1758	NUM
ijassa-659	619	10	.	.	PUNCT
ijassa-659	620	1	16	16	NUM
ijassa-659	620	2	.	.	PUNCT
ijassa-659	621	1	vu	vu	X
ijassa-659	621	2	,	,	PUNCT
ijassa-659	621	3	v.	v.	ADP
ijassa-659	621	4	v.	v.	ADP
ijassa-659	621	5	,	,	PUNCT
ijassa-659	621	6	labroche	labroche	NOUN
ijassa-659	621	7	,	,	PUNCT
ijassa-659	621	8	n.	n.	NOUN
ijassa-659	621	9	,	,	PUNCT
ijassa-659	621	10	&	&	CCONJ
ijassa-659	621	11	bouchon	bouchon	PROPN
ijassa-659	621	12	-	-	PUNCT
ijassa-659	621	13	meunier	meunier	PROPN
ijassa-659	621	14	,	,	PUNCT
ijassa-659	621	15	b.	b.	PROPN
ijassa-659	621	16	(	(	PUNCT
ijassa-659	621	17	2010	2010	NUM
ijassa-659	621	18	)	)	PUNCT
ijassa-659	621	19	.	.	PUNCT
ijassa-659	622	1	an	an	DET
ijassa-659	622	2	efficient	efficient	ADJ
ijassa-659	622	3	active	active	ADJ
ijassa-659	622	4	constraint	constraint	NOUN
ijassa-659	622	5	selection	selection	NOUN
ijassa-659	622	6	algorithm	algorithm	NOUN
ijassa-659	622	7	for	for	ADP
ijassa-659	622	8	clustering	clustering	NOUN
ijassa-659	622	9	.	.	PUNCT
ijassa-659	623	1	in	in	ADP
ijassa-659	623	2	2010	2010	NUM
ijassa-659	623	3	20th	20th	ADJ
ijassa-659	623	4	international	international	ADJ
ijassa-659	623	5	conference	conference	NOUN
ijassa-659	623	6	on	on	ADP
ijassa-659	623	7	pattern	pattern	NOUN
ijassa-659	623	8	recognition	recognition	NOUN
ijassa-659	623	9	,	,	PUNCT
ijassa-659	623	10	2969	2969	NUM
ijassa-659	623	11	-	-	SYM
ijassa-659	623	12	2972	2972	NUM
ijassa-659	623	13	.	.	PUNCT
ijassa-659	624	1	17	17	NUM
ijassa-659	624	2	.	.	PUNCT
ijassa-659	625	1	xiong	xiong	PROPN
ijassa-659	625	2	,	,	PUNCT
ijassa-659	625	3	s.	s.	PROPN
ijassa-659	625	4	,	,	PUNCT
ijassa-659	625	5	azimi	azimi	PROPN
ijassa-659	625	6	,	,	PUNCT
ijassa-659	625	7	j.	j.	PROPN
ijassa-659	625	8	,	,	PUNCT
ijassa-659	625	9	&	&	CCONJ
ijassa-659	625	10	fern	fern	NOUN
ijassa-659	625	11	,	,	PUNCT
ijassa-659	625	12	x.	x.	NOUN
ijassa-659	625	13	z.	z.	PROPN
ijassa-659	625	14	(	(	PUNCT
ijassa-659	625	15	2013	2013	NUM
ijassa-659	625	16	)	)	PUNCT
ijassa-659	625	17	.	.	PUNCT
ijassa-659	626	1	active	active	ADJ
ijassa-659	626	2	learning	learning	NOUN
ijassa-659	626	3	of	of	ADP
ijassa-659	626	4	constraints	constraint	NOUN
ijassa-659	626	5	for	for	ADP
ijassa-659	626	6	semisupervised	semisupervised	ADJ
ijassa-659	626	7	clustering	clustering	NOUN
ijassa-659	626	8	.	.	PUNCT
ijassa-659	627	1	ieee	ieee	NOUN
ijassa-659	627	2	transactions	transaction	NOUN
ijassa-659	627	3	on	on	ADP
ijassa-659	627	4	knowledge	knowledge	NOUN
ijassa-659	627	5	and	and	CCONJ
ijassa-659	627	6	data	datum	NOUN
ijassa-659	627	7	engineering	engineering	NOUN
ijassa-659	627	8	,	,	PUNCT
ijassa-659	627	9	26(1	26(1	NUM
ijassa-659	627	10	)	)	PUNCT
ijassa-659	627	11	,	,	PUNCT
ijassa-659	627	12	43	43	NUM
ijassa-659	627	13	-	-	SYM
ijassa-659	627	14	54	54	NUM
ijassa-659	627	15	.	.	PUNCT
ijassa-659	628	1	18	18	NUM
ijassa-659	628	2	.	.	PUNCT
ijassa-659	629	1	li	li	PROPN
ijassa-659	629	2	,	,	PUNCT
ijassa-659	629	3	y.	y.	PROPN
ijassa-659	629	4	,	,	PUNCT
ijassa-659	629	5	li	li	PROPN
ijassa-659	629	6	wang	wang	PROPN
ijassa-659	629	7	,	,	PUNCT
ijassa-659	629	8	y.	y.	PROPN
ijassa-659	629	9	,	,	PUNCT
ijassa-659	629	10	yu	yu	PROPN
ijassa-659	629	11	,	,	PUNCT
ijassa-659	629	12	d.	d.	PROPN
ijassa-659	629	13	j.	j.	PROPN
ijassa-659	629	14	,	,	PUNCT
ijassa-659	629	15	ning	ning	PROPN
ijassa-659	629	16	,	,	PUNCT
ijassa-659	629	17	y.	y.	PROPN
ijassa-659	629	18	,	,	PUNCT
ijassa-659	629	19	hu	hu	PROPN
ijassa-659	629	20	,	,	PUNCT
ijassa-659	629	21	p.	p.	NOUN
ijassa-659	629	22	,	,	PUNCT
ijassa-659	629	23	&	&	CCONJ
ijassa-659	629	24	zhao	zhao	PROPN
ijassa-659	629	25	,	,	PUNCT
ijassa-659	629	26	r.	r.	PROPN
ijassa-659	629	27	(	(	PUNCT
ijassa-659	629	28	2019	2019	NUM
ijassa-659	629	29	)	)	PUNCT
ijassa-659	629	30	.	.	PUNCT
ijassa-659	630	1	ascent	ascent	NOUN
ijassa-659	630	2	:	:	PUNCT
ijassa-659	630	3	active	active	ADJ
ijassa-659	630	4	supervision	supervision	NOUN
ijassa-659	630	5	for	for	ADP
ijassa-659	630	6	semi	semi	ADJ
ijassa-659	630	7	-	-	ADJ
ijassa-659	630	8	supervised	supervised	ADJ
ijassa-659	630	9	learning	learning	NOUN
ijassa-659	630	10	.	.	PUNCT
ijassa-659	631	1	ieee	ieee	NOUN
ijassa-659	631	2	transactions	transaction	NOUN
ijassa-659	631	3	on	on	ADP
ijassa-659	631	4	knowledge	knowledge	NOUN
ijassa-659	631	5	and	and	CCONJ
ijassa-659	631	6	data	datum	NOUN
ijassa-659	631	7	engineering	engineering	NOUN
ijassa-659	631	8	.	.	PUNCT
ijassa-659	632	1	19	19	NUM
ijassa-659	632	2	.	.	X
ijassa-659	632	3	bilenko	bilenko	ADJ
ijassa-659	632	4	,	,	PUNCT
ijassa-659	632	5	m.	m.	NOUN
ijassa-659	632	6	,	,	PUNCT
ijassa-659	632	7	basu	basu	PROPN
ijassa-659	632	8	,	,	PUNCT
ijassa-659	632	9	s.	s.	PROPN
ijassa-659	632	10	,	,	PUNCT
ijassa-659	632	11	&	&	CCONJ
ijassa-659	632	12	mooney	mooney	PROPN
ijassa-659	632	13	,	,	PUNCT
ijassa-659	632	14	r.	r.	PROPN
ijassa-659	632	15	j.	j.	PROPN
ijassa-659	632	16	(	(	PUNCT
ijassa-659	632	17	2004	2004	NUM
ijassa-659	632	18	,	,	PUNCT
ijassa-659	632	19	july	july	PROPN
ijassa-659	632	20	)	)	PUNCT
ijassa-659	632	21	.	.	PUNCT
ijassa-659	633	1	integrating	integrate	VERB
ijassa-659	633	2	constraints	constraint	NOUN
ijassa-659	633	3	and	and	CCONJ
ijassa-659	633	4	metric	metric	ADJ
ijassa-659	633	5	learning	learning	NOUN
ijassa-659	633	6	in	in	ADP
ijassa-659	633	7	semi	semi	ADJ
ijassa-659	633	8	-	-	ADJ
ijassa-659	633	9	supervised	supervised	ADJ
ijassa-659	633	10	clustering	clustering	NOUN
ijassa-659	633	11	.	.	PUNCT
ijassa-659	634	1	in	in	ADP
ijassa-659	634	2	proceedings	proceeding	NOUN
ijassa-659	634	3	of	of	ADP
ijassa-659	634	4	the	the	DET
ijassa-659	634	5	twenty	twenty	NUM
ijassa-659	634	6	-	-	PUNCT
ijassa-659	634	7	first	first	ADJ
ijassa-659	634	8	international	international	ADJ
ijassa-659	634	9	conference	conference	NOUN
ijassa-659	634	10	on	on	ADP
ijassa-659	634	11	machine	machine	NOUN
ijassa-659	634	12	learning	learning	NOUN
ijassa-659	634	13	(	(	PUNCT
ijassa-659	634	14	p.	p.	NOUN
ijassa-659	634	15	11	11	NUM
ijassa-659	634	16	)	)	PUNCT
ijassa-659	634	17	.	.	PUNCT
ijassa-659	635	1	acm	acm	PROPN
ijassa-659	635	2	.	.	PROPN
ijassa-659	636	1	20	20	NUM
ijassa-659	636	2	.	.	X
ijassa-659	636	3	davidson	davidson	PROPN
ijassa-659	636	4	,	,	PUNCT
ijassa-659	636	5	i.	i.	PROPN
ijassa-659	636	6	,	,	PUNCT
ijassa-659	636	7	&	&	CCONJ
ijassa-659	636	8	ravi	ravi	PROPN
ijassa-659	636	9	,	,	PUNCT
ijassa-659	636	10	s.	s.	PROPN
ijassa-659	636	11	s.	s.	PROPN
ijassa-659	636	12	(	(	PUNCT
ijassa-659	636	13	2005	2005	NUM
ijassa-659	636	14	)	)	PUNCT
ijassa-659	636	15	.	.	PUNCT
ijassa-659	637	1	agglomerative	agglomerative	ADJ
ijassa-659	637	2	hierarchical	hierarchical	ADJ
ijassa-659	637	3	clustering	clustering	NOUN
ijassa-659	637	4	with	with	ADP
ijassa-659	637	5	constraints	constraint	NOUN
ijassa-659	637	6	:	:	PUNCT
ijassa-659	637	7	theoretical	theoretical	ADJ
ijassa-659	637	8	and	and	CCONJ
ijassa-659	637	9	empirical	empirical	ADJ
ijassa-659	637	10	results	result	NOUN
ijassa-659	637	11	.	.	PUNCT
ijassa-659	638	1	in	in	ADP
ijassa-659	638	2	european	european	ADJ
ijassa-659	638	3	conference	conference	NOUN
ijassa-659	638	4	on	on	ADP
ijassa-659	638	5	principles	principle	NOUN
ijassa-659	638	6	of	of	ADP
ijassa-659	638	7	data	datum	NOUN
ijassa-659	638	8	mining	mining	NOUN
ijassa-659	638	9	and	and	CCONJ
ijassa-659	638	10	knowledge	knowledge	NOUN
ijassa-659	638	11	discovery	discovery	NOUN
ijassa-659	638	12	,	,	PUNCT
ijassa-659	638	13	59	59	NUM
ijassa-659	638	14	-	-	SYM
ijassa-659	638	15	70	70	NUM
ijassa-659	638	16	.	.	PUNCT
ijassa-659	638	17	21	21	NUM
ijassa-659	638	18	.	.	X
ijassa-659	638	19	cover	cover	VERB
ijassa-659	638	20	,	,	PUNCT
ijassa-659	638	21	t.	t.	NOUN
ijassa-659	638	22	m.	m.	NOUN
ijassa-659	638	23	,	,	PUNCT
ijassa-659	638	24	&	&	CCONJ
ijassa-659	638	25	thomas	thomas	PROPN
ijassa-659	638	26	,	,	PUNCT
ijassa-659	638	27	j.	j.	PROPN
ijassa-659	638	28	a.	a.	PROPN
ijassa-659	638	29	(	(	PUNCT
ijassa-659	638	30	2012	2012	NUM
ijassa-659	638	31	)	)	PUNCT
ijassa-659	638	32	.	.	PUNCT
ijassa-659	639	1	elements	element	NOUN
ijassa-659	639	2	of	of	ADP
ijassa-659	639	3	information	information	NOUN
ijassa-659	639	4	theory	theory	NOUN
ijassa-659	639	5	.	.	PUNCT
ijassa-659	640	1	john	john	PROPN
ijassa-659	640	2	wiley	wiley	PROPN
ijassa-659	640	3	&	&	CCONJ
ijassa-659	640	4	sons	son	NOUN
ijassa-659	640	5	.	.	PUNCT
ijassa-659	641	1	22	22	NUM
ijassa-659	641	2	.	.	PUNCT
ijassa-659	642	1	xiong	xiong	PROPN
ijassa-659	642	2	,	,	PUNCT
ijassa-659	642	3	s.	s.	PROPN
ijassa-659	642	4	,	,	PUNCT
ijassa-659	642	5	pei	pei	PROPN
ijassa-659	642	6	,	,	PUNCT
ijassa-659	642	7	y.	y.	PROPN
ijassa-659	642	8	,	,	PUNCT
ijassa-659	642	9	rosales	rosale	NOUN
ijassa-659	642	10	,	,	PUNCT
ijassa-659	642	11	r.	r.	PROPN
ijassa-659	642	12	,	,	PUNCT
ijassa-659	642	13	&	&	CCONJ
ijassa-659	642	14	fern	fern	NOUN
ijassa-659	642	15	,	,	PUNCT
ijassa-659	642	16	x.	x.	NOUN
ijassa-659	642	17	z.	z.	PROPN
ijassa-659	642	18	(	(	PUNCT
ijassa-659	642	19	2015	2015	NUM
ijassa-659	642	20	)	)	PUNCT
ijassa-659	642	21	.	.	PUNCT
ijassa-659	643	1	active	active	ADJ
ijassa-659	643	2	learning	learning	NOUN
ijassa-659	643	3	from	from	ADP
ijassa-659	643	4	relative	relative	ADJ
ijassa-659	643	5	comparisons	comparison	NOUN
ijassa-659	643	6	.	.	PUNCT
ijassa-659	644	1	ieee	ieee	NOUN
ijassa-659	644	2	transactions	transaction	NOUN
ijassa-659	644	3	on	on	ADP
ijassa-659	644	4	knowledge	knowledge	NOUN
ijassa-659	644	5	and	and	CCONJ
ijassa-659	644	6	data	datum	NOUN
ijassa-659	644	7	engineering	engineering	NOUN
ijassa-659	644	8	,	,	PUNCT
ijassa-659	644	9	27(12	27(12	NUM
ijassa-659	644	10	)	)	PUNCT
ijassa-659	644	11	,	,	PUNCT
ijassa-659	644	12	31663175	31663175	NUM
ijassa-659	644	13	.	.	PUNCT
ijassa-659	645	1	23	23	NUM
ijassa-659	645	2	.	.	X
ijassa-659	645	3	davidson	davidson	PROPN
ijassa-659	645	4	,	,	PUNCT
ijassa-659	645	5	i.	i.	PROPN
ijassa-659	645	6	,	,	PUNCT
ijassa-659	645	7	wagstaff	wagstaff	PROPN
ijassa-659	645	8	,	,	PUNCT
ijassa-659	645	9	k.	k.	PROPN
ijassa-659	645	10	l.	l.	PROPN
ijassa-659	645	11	,	,	PUNCT
ijassa-659	645	12	&	&	CCONJ
ijassa-659	645	13	basu	basu	PROPN
ijassa-659	645	14	,	,	PUNCT
ijassa-659	645	15	s.	s.	PROPN
ijassa-659	645	16	(	(	PUNCT
ijassa-659	645	17	2006	2006	NUM
ijassa-659	645	18	)	)	PUNCT
ijassa-659	645	19	.	.	PUNCT
ijassa-659	646	1	measuring	measure	VERB
ijassa-659	646	2	constraint	constraint	NOUN
ijassa-659	646	3	-	-	PUNCT
ijassa-659	646	4	set	set	VERB
ijassa-659	646	5	utility	utility	NOUN
ijassa-659	646	6	for	for	ADP
ijassa-659	646	7	partitional	partitional	ADJ
ijassa-659	646	8	clustering	clustering	ADJ
ijassa-659	646	9	algorithms	algorithm	NOUN
ijassa-659	646	10	.	.	PUNCT
ijassa-659	647	1	in	in	ADP
ijassa-659	647	2	european	european	ADJ
ijassa-659	647	3	conference	conference	NOUN
ijassa-659	647	4	on	on	ADP
ijassa-659	647	5	principles	principle	NOUN
ijassa-659	647	6	of	of	ADP
ijassa-659	647	7	data	datum	NOUN
ijassa-659	647	8	mining	mining	NOUN
ijassa-659	647	9	and	and	CCONJ
ijassa-659	647	10	knowledge	knowledge	NOUN
ijassa-659	647	11	discovery	discovery	NOUN
ijassa-659	647	12	,	,	PUNCT
ijassa-659	647	13	115	115	NUM
ijassa-659	647	14	-	-	SYM
ijassa-659	647	15	126	126	NUM
ijassa-659	647	16	.	.	PUNCT