id	sid	tid	token	lemma	pos
alusur-6996	1	1	al-ʿuṣūr	al-ʿuṣūr	PROPN
alusur-6996	1	2	al	al	PROPN
alusur-6996	1	3	-	-	PUNCT
alusur-6996	1	4	wusṭā	wusṭā	NOUN
alusur-6996	1	5	25	25	NUM
alusur-6996	1	6	(	(	PUNCT
alusur-6996	1	7	2017	2017	NUM
alusur-6996	1	8	):	):	PUNCT
alusur-6996	1	9	1	1	NUM
alusur-6996	1	10	-	-	SYM
alusur-6996	1	11	13	13	NUM
alusur-6996	1	12	1.1	1.1	NUM
alusur-6996	1	13	summary	summary	NOUN
alusur-6996	1	14	of	of	ADP
alusur-6996	1	15	results	result	NOUN
alusur-6996	1	16	the	the	DET
alusur-6996	1	17	open	open	ADJ
alusur-6996	1	18	islamicate	islamicate	ADJ
alusur-6996	1	19	texts	text	NOUN
alusur-6996	1	20	initiative	initiative	NOUN
alusur-6996	1	21	(	(	PUNCT
alusur-6996	1	22	openiti	openiti	PROPN
alusur-6996	1	23	)	)	PUNCT
alusur-6996	1	24	team1	team1	PROPN
alusur-6996	1	25	—	—	PUNCT
alusur-6996	1	26	building	build	VERB
alusur-6996	1	27	on	on	ADP
alusur-6996	1	28	the	the	DET
alusur-6996	1	29	foundational	foundational	ADJ
alusur-6996	1	30	opensource	opensource	PROPN
alusur-6996	1	31	ocr	ocr	PROPN
alusur-6996	1	32	work	work	NOUN
alusur-6996	1	33	of	of	ADP
alusur-6996	1	34	the	the	DET
alusur-6996	1	35	leipzig	leipzig	PROPN
alusur-6996	1	36	university	university	PROPN
alusur-6996	1	37	(	(	PUNCT
alusur-6996	1	38	lu	lu	PROPN
alusur-6996	1	39	)	)	PUNCT
alusur-6996	1	40	alexander	alexander	PROPN
alusur-6996	1	41	von	von	PROPN
alusur-6996	1	42	humboldt	humboldt	PROPN
alusur-6996	1	43	chair	chair	PROPN
alusur-6996	1	44	for	for	ADP
alusur-6996	1	45	digital	digital	ADJ
alusur-6996	1	46	humanities	humanity	NOUN
alusur-6996	1	47	—	—	PUNCT
alusur-6996	1	48	has	have	AUX
alusur-6996	1	49	achieved	achieve	VERB
alusur-6996	1	50	optical	optical	ADJ
alusur-6996	1	51	character	character	NOUN
alusur-6996	1	52	recognition	recognition	NOUN
alusur-6996	1	53	(	(	PUNCT
alusur-6996	1	54	ocr	ocr	ADJ
alusur-6996	1	55	)	)	PUNCT
alusur-6996	1	56	accuracy	accuracy	NOUN
alusur-6996	1	57	rates	rate	NOUN
alusur-6996	1	58	for	for	ADP
alusur-6996	1	59	printed	print	VERB
alusur-6996	1	60	classical	classical	ADJ
alusur-6996	1	61	arabic	arabic	ADJ
alusur-6996	1	62	-	-	PUNCT
alusur-6996	1	63	script	script	NOUN
alusur-6996	1	64	texts	text	NOUN
alusur-6996	1	65	in	in	ADP
alusur-6996	1	66	the	the	DET
alusur-6996	1	67	high	high	ADJ
alusur-6996	1	68	nineties	ninety	NOUN
alusur-6996	1	69	.	.	PUNCT
alusur-6996	2	1	these	these	DET
alusur-6996	2	2	numbers	number	NOUN
alusur-6996	2	3	are	be	AUX
alusur-6996	2	4	based	base	VERB
alusur-6996	2	5	on	on	ADP
alusur-6996	2	6	our	our	PRON
alusur-6996	2	7	tests	test	NOUN
alusur-6996	2	8	of	of	ADP
alusur-6996	2	9	seven	seven	NUM
alusur-6996	2	10	different	different	ADJ
alusur-6996	2	11	arabic	arabic	ADJ
alusur-6996	2	12	-	-	PUNCT
alusur-6996	2	13	script	script	NOUN
alusur-6996	2	14	texts	text	NOUN
alusur-6996	2	15	of	of	ADP
alusur-6996	2	16	varying	vary	VERB
alusur-6996	2	17	quality	quality	NOUN
alusur-6996	2	18	and	and	CCONJ
alusur-6996	2	19	typefaces	typeface	NOUN
alusur-6996	2	20	,	,	PUNCT
alusur-6996	2	21	totaling	total	VERB
alusur-6996	2	22	over	over	ADP
alusur-6996	2	23	7,000	7,000	NUM
alusur-6996	2	24	lines	line	NOUN
alusur-6996	2	25	(	(	PUNCT
alusur-6996	2	26	~400	~400	NOUN
alusur-6996	2	27	pages	page	NOUN
alusur-6996	2	28	,	,	PUNCT
alusur-6996	2	29	87,000	87,000	NUM
alusur-6996	2	30	words	word	NOUN
alusur-6996	2	31	;	;	PUNCT
alusur-6996	2	32	see	see	VERB
alusur-6996	2	33	table	table	NOUN
alusur-6996	2	34	1	1	NUM
alusur-6996	2	35	for	for	ADP
alusur-6996	2	36	full	full	ADJ
alusur-6996	2	37	details	detail	NOUN
alusur-6996	2	38	)	)	PUNCT
alusur-6996	2	39	.	.	PUNCT
alusur-6996	3	1	these	these	DET
alusur-6996	3	2	accuracy	accuracy	NOUN
alusur-6996	3	3	rates	rate	NOUN
alusur-6996	3	4	not	not	PART
alusur-6996	3	5	only	only	ADV
alusur-6996	3	6	represent	represent	VERB
alusur-6996	3	7	a	a	DET
alusur-6996	3	8	distinct	distinct	ADJ
alusur-6996	3	9	improvement	improvement	NOUN
alusur-6996	3	10	over	over	ADP
alusur-6996	3	11	the	the	DET
alusur-6996	3	12	actual2	actual2	ADJ
alusur-6996	3	13	accuracy	accuracy	NOUN
alusur-6996	3	14	rates	rate	NOUN
alusur-6996	3	15	of	of	ADP
alusur-6996	3	16	the	the	DET
alusur-6996	3	17	various	various	ADJ
alusur-6996	3	18	proprietary	proprietary	ADJ
alusur-6996	3	19	ocr	ocr	ADJ
alusur-6996	3	20	options	option	NOUN
alusur-6996	3	21	for	for	ADP
alusur-6996	3	22	printed	print	VERB
alusur-6996	3	23	classical	classical	ADJ
alusur-6996	3	24	arabic	arabic	ADJ
alusur-6996	3	25	-	-	PUNCT
alusur-6996	3	26	script	script	NOUN
alusur-6996	3	27	texts	text	NOUN
alusur-6996	3	28	,	,	PUNCT
alusur-6996	3	29	but	but	CCONJ
alusur-6996	3	30	,	,	PUNCT
alusur-6996	3	31	equally	equally	ADV
alusur-6996	3	32	important	important	ADJ
alusur-6996	3	33	,	,	PUNCT
alusur-6996	3	34	they	they	PRON
alusur-6996	3	35	are	be	AUX
alusur-6996	3	36	produced	produce	VERB
alusur-6996	3	37	using	use	VERB
alusur-6996	3	38	an	an	DET
alusur-6996	3	39	open	open	ADJ
alusur-6996	3	40	-	-	PUNCT
alusur-6996	3	41	source	source	NOUN
alusur-6996	3	42	ocr	ocr	PROPN
alusur-6996	3	43	software	software	NOUN
alusur-6996	3	44	called	call	VERB
alusur-6996	3	45	kraken	kraken	PROPN
alusur-6996	3	46	(	(	PUNCT
alusur-6996	3	47	developed	develop	VERB
alusur-6996	3	48	by	by	ADP
alusur-6996	3	49	benjamin	benjamin	PROPN
alusur-6996	3	50	kiessling	kiessling	PROPN
alusur-6996	3	51	,	,	PUNCT
alusur-6996	3	52	lu	lu	PROPN
alusur-6996	3	53	)	)	PUNCT
alusur-6996	3	54	,	,	PUNCT
alusur-6996	4	1	1	1	X
alusur-6996	4	2	.	.	PUNCT
alusur-6996	5	1	the	the	DET
alusur-6996	5	2	co	co	NOUN
alusur-6996	5	3	-	-	NOUN
alusur-6996	5	4	pis	pis	NOUN
alusur-6996	5	5	of	of	ADP
alusur-6996	5	6	the	the	DET
alusur-6996	5	7	open	open	ADJ
alusur-6996	5	8	islamicate	islamicate	ADJ
alusur-6996	5	9	texts	text	NOUN
alusur-6996	5	10	initiative	initiative	NOUN
alusur-6996	5	11	(	(	PUNCT
alusur-6996	5	12	openiti	openiti	PROPN
alusur-6996	5	13	)	)	PUNCT
alusur-6996	5	14	are	be	AUX
alusur-6996	5	15	sarah	sarah	PROPN
alusur-6996	5	16	bowen	bowen	PROPN
alusur-6996	5	17	savant	savant	NOUN
alusur-6996	5	18	(	(	PUNCT
alusur-6996	5	19	aga	aga	PROPN
alusur-6996	5	20	khan	khan	PROPN
alusur-6996	5	21	university	university	PROPN
alusur-6996	5	22	,	,	PUNCT
alusur-6996	5	23	institute	institute	NOUN
alusur-6996	5	24	for	for	ADP
alusur-6996	5	25	the	the	DET
alusur-6996	5	26	study	study	NOUN
alusur-6996	5	27	of	of	ADP
alusur-6996	5	28	muslim	muslim	ADJ
alusur-6996	5	29	civilisations	civilisation	NOUN
alusur-6996	5	30	,	,	PUNCT
alusur-6996	5	31	london	london	PROPN
alusur-6996	5	32	;	;	PUNCT
alusur-6996	5	33	sarah.savant@aku.edu	sarah.savant@aku.edu	PROPN
alusur-6996	5	34	)	)	PUNCT
alusur-6996	5	35	,	,	PUNCT
alusur-6996	5	36	maxim	maxim	NOUN
alusur-6996	5	37	g.	g.	PROPN
alusur-6996	5	38	romanov	romanov	PROPN
alusur-6996	5	39	(	(	PUNCT
alusur-6996	5	40	leipzig	leipzig	PROPN
alusur-6996	5	41	university	university	PROPN
alusur-6996	5	42	through	through	ADP
alusur-6996	5	43	june	june	PROPN
alusur-6996	5	44	2017	2017	NUM
alusur-6996	5	45	;	;	PUNCT
alusur-6996	5	46	now	now	ADV
alusur-6996	5	47	university	university	NOUN
alusur-6996	5	48	of	of	ADP
alusur-6996	5	49	vienna	vienna	PROPN
alusur-6996	5	50	;	;	PUNCT
alusur-6996	5	51	maxim.romanov@univie.ac.at	maxim.romanov@univie.ac.at	NUM
alusur-6996	5	52	)	)	PUNCT
alusur-6996	5	53	,	,	PUNCT
alusur-6996	5	54	and	and	CCONJ
alusur-6996	5	55	matthew	matthew	PROPN
alusur-6996	5	56	thomas	thomas	PROPN
alusur-6996	5	57	miller	miller	PROPN
alusur-6996	5	58	(	(	PUNCT
alusur-6996	5	59	roshan	roshan	PROPN
alusur-6996	5	60	institute	institute	PROPN
alusur-6996	5	61	for	for	ADP
alusur-6996	5	62	persian	persian	PROPN
alusur-6996	5	63	studies	study	NOUN
alusur-6996	5	64	,	,	PUNCT
alusur-6996	5	65	university	university	PROPN
alusur-6996	5	66	of	of	ADP
alusur-6996	5	67	maryland	maryland	PROPN
alusur-6996	5	68	,	,	PUNCT
alusur-6996	5	69	college	college	NOUN
alusur-6996	5	70	park	park	NOUN
alusur-6996	5	71	;	;	PUNCT
alusur-6996	5	72	mtmiller@umd.edu	mtmiller@umd.edu	PROPN
alusur-6996	5	73	)	)	PUNCT
alusur-6996	5	74	.	.	PUNCT
alusur-6996	6	1	benjamin	benjamin	PROPN
alusur-6996	6	2	kiessling	kiessling	PROPN
alusur-6996	6	3	can	can	AUX
alusur-6996	6	4	be	be	AUX
alusur-6996	6	5	contacted	contact	VERB
alusur-6996	6	6	at	at	ADP
alusur-6996	6	7	mittagessen@l.unchti.me	mittagessen@l.unchti.me	PROPN
alusur-6996	6	8	.	.	PUNCT
alusur-6996	7	1	2	2	X
alusur-6996	7	2	.	.	X
alusur-6996	7	3	proprietary	proprietary	ADJ
alusur-6996	7	4	ocr	ocr	ADJ
alusur-6996	7	5	programs	program	NOUN
alusur-6996	7	6	for	for	ADP
alusur-6996	7	7	persian	persian	ADJ
alusur-6996	7	8	and	and	CCONJ
alusur-6996	7	9	arabic	arabic	ADJ
alusur-6996	7	10	(	(	PUNCT
alusur-6996	7	11	e.g.	e.g.	ADV
alusur-6996	7	12	,	,	PUNCT
alusur-6996	7	13	sakhr	sakhr	NOUN
alusur-6996	7	14	’s	’s	PART
alusur-6996	7	15	automatic	automatic	ADJ
alusur-6996	7	16	reader	reader	NOUN
alusur-6996	7	17	,	,	PUNCT
alusur-6996	7	18	abbyy	abbyy	PROPN
alusur-6996	7	19	finereader	finereader	PROPN
alusur-6996	7	20	,	,	PUNCT
alusur-6996	7	21	readiris	readiris	PROPN
alusur-6996	7	22	)	)	PUNCT
alusur-6996	7	23	overpromise	overpromise	VERB
alusur-6996	7	24	the	the	DET
alusur-6996	7	25	level	level	NOUN
alusur-6996	7	26	of	of	ADP
alusur-6996	7	27	accuracy	accuracy	NOUN
alusur-6996	7	28	they	they	PRON
alusur-6996	7	29	deliver	deliver	VERB
alusur-6996	7	30	in	in	ADP
alusur-6996	7	31	practice	practice	NOUN
alusur-6996	7	32	when	when	SCONJ
alusur-6996	7	33	used	use	VERB
alusur-6996	7	34	on	on	ADP
alusur-6996	7	35	classical	classical	ADJ
alusur-6996	7	36	texts	text	NOUN
alusur-6996	7	37	(	(	PUNCT
alusur-6996	7	38	in	in	ADP
alusur-6996	7	39	particular	particular	ADJ
alusur-6996	7	40	,	,	PUNCT
alusur-6996	7	41	highly	highly	ADV
alusur-6996	7	42	vocalized	vocalized	ADJ
alusur-6996	7	43	texts	text	NOUN
alusur-6996	7	44	)	)	PUNCT
alusur-6996	7	45	.	.	PUNCT
alusur-6996	8	1	these	these	DET
alusur-6996	8	2	companies	company	NOUN
alusur-6996	8	3	claim	claim	VERB
alusur-6996	8	4	that	that	SCONJ
alusur-6996	8	5	they	they	PRON
alusur-6996	8	6	provide	provide	VERB
alusur-6996	8	7	accuracy	accuracy	NOUN
alusur-6996	8	8	rates	rate	NOUN
alusur-6996	8	9	in	in	ADP
alusur-6996	8	10	the	the	DET
alusur-6996	8	11	high	high	ADJ
alusur-6996	8	12	90	90	NUM
alusur-6996	8	13	percentages	percentage	NOUN
alusur-6996	8	14	(	(	PUNCT
alusur-6996	8	15	e.g.	e.g.	ADV
alusur-6996	8	16	,	,	PUNCT
alusur-6996	8	17	sakhr	sakhr	NOUN
alusur-6996	8	18	claims	claim	NOUN
alusur-6996	8	19	99.8	99.8	NUM
alusur-6996	8	20	%	%	NOUN
alusur-6996	8	21	accuracy	accuracy	NOUN
alusur-6996	8	22	for	for	ADP
alusur-6996	8	23	high	high	ADJ
alusur-6996	8	24	-	-	PUNCT
alusur-6996	8	25	quality	quality	NOUN
alusur-6996	8	26	documents	document	NOUN
alusur-6996	8	27	)	)	PUNCT
alusur-6996	8	28	.	.	PUNCT
alusur-6996	9	1	this	this	PRON
alusur-6996	9	2	may	may	AUX
alusur-6996	9	3	be	be	AUX
alusur-6996	9	4	the	the	DET
alusur-6996	9	5	case	case	NOUN
alusur-6996	9	6	for	for	ADP
alusur-6996	9	7	texts	text	NOUN
alusur-6996	9	8	with	with	ADP
alusur-6996	9	9	simplified	simplified	ADJ
alusur-6996	9	10	typesets	typeset	NOUN
alusur-6996	9	11	and	and	CCONJ
alusur-6996	9	12	no	no	DET
alusur-6996	9	13	short	short	ADJ
alusur-6996	9	14	vowels	vowel	NOUN
alusur-6996	9	15	;	;	PUNCT
alusur-6996	9	16	however	however	ADV
alusur-6996	9	17	,	,	PUNCT
alusur-6996	9	18	our	our	PRON
alusur-6996	9	19	tests	test	NOUN
alusur-6996	9	20	of	of	ADP
alusur-6996	9	21	abbyy	abbyy	PROPN
alusur-6996	9	22	finereader	finereader	NOUN
alusur-6996	9	23	and	and	CCONJ
alusur-6996	9	24	readiris	readiris	VERB
alusur-6996	9	25	on	on	ADP
alusur-6996	9	26	high	high	ADJ
alusur-6996	9	27	-	-	PUNCT
alusur-6996	9	28	quality	quality	NOUN
alusur-6996	9	29	scans	scan	NOUN
alusur-6996	9	30	of	of	ADP
alusur-6996	9	31	classical	classical	ADJ
alusur-6996	9	32	texts	text	NOUN
alusur-6996	9	33	turned	turn	VERB
alusur-6996	9	34	out	out	ADP
alusur-6996	9	35	accuracy	accuracy	NOUN
alusur-6996	9	36	rates	rate	NOUN
alusur-6996	9	37	of	of	ADP
alusur-6996	9	38	between	between	ADP
alusur-6996	9	39	65	65	NUM
alusur-6996	9	40	%	%	NOUN
alusur-6996	9	41	and	and	CCONJ
alusur-6996	9	42	75	75	NUM
alusur-6996	9	43	%	%	NOUN
alusur-6996	9	44	.	.	PUNCT
alusur-6996	10	1	sakhr	sakhr	NOUN
alusur-6996	10	2	software	software	PROPN
alusur-6996	10	3	was	be	AUX
alusur-6996	10	4	not	not	PART
alusur-6996	10	5	available	available	ADJ
alusur-6996	10	6	to	to	ADP
alusur-6996	10	7	us	we	PRON
alusur-6996	10	8	,	,	PUNCT
alusur-6996	10	9	as	as	SCONJ
alusur-6996	10	10	the	the	DET
alusur-6996	10	11	developers	developer	NOUN
alusur-6996	10	12	offer	offer	VERB
alusur-6996	10	13	no	no	DET
alusur-6996	10	14	trial	trial	NOUN
alusur-6996	10	15	versions	version	NOUN
alusur-6996	10	16	and	and	CCONJ
alusur-6996	10	17	it	it	PRON
alusur-6996	10	18	is	be	AUX
alusur-6996	10	19	the	the	DET
alusur-6996	10	20	most	most	ADV
alusur-6996	10	21	expensive	expensive	ADJ
alusur-6996	10	22	commercial	commercial	ADJ
alusur-6996	10	23	ocr	ocr	ADJ
alusur-6996	10	24	solution	solution	NOUN
alusur-6996	10	25	for	for	ADP
alusur-6996	10	26	arabic	arabic	ADJ
alusur-6996	10	27	.	.	PUNCT
alusur-6996	11	1	moreover	moreover	ADV
alusur-6996	11	2	,	,	PUNCT
alusur-6996	11	3	since	since	SCONJ
alusur-6996	11	4	these	these	DET
alusur-6996	11	5	programs	program	NOUN
alusur-6996	11	6	are	be	AUX
alusur-6996	11	7	not	not	PART
alusur-6996	11	8	open	open	ADJ
alusur-6996	11	9	-	-	PUNCT
alusur-6996	11	10	source	source	NOUN
alusur-6996	11	11	and	and	CCONJ
alusur-6996	11	12	offer	offer	VERB
alusur-6996	11	13	only	only	ADV
alusur-6996	11	14	limited	limited	ADJ
alusur-6996	11	15	trainability	trainability	NOUN
alusur-6996	11	16	(	(	PUNCT
alusur-6996	11	17	and	and	CCONJ
alusur-6996	11	18	created	create	VERB
alusur-6996	11	19	training	training	NOUN
alusur-6996	11	20	data	datum	NOUN
alusur-6996	11	21	can	can	AUX
alusur-6996	11	22	not	not	PART
alusur-6996	11	23	be	be	AUX
alusur-6996	11	24	reused	reuse	VERB
alusur-6996	11	25	)	)	PUNCT
alusur-6996	11	26	,	,	PUNCT
alusur-6996	11	27	their	their	PRON
alusur-6996	11	28	costs	cost	NOUN
alusur-6996	11	29	are	be	AUX
alusur-6996	11	30	prohibitive	prohibitive	ADJ
alusur-6996	11	31	for	for	ADP
alusur-6996	11	32	most	most	ADJ
alusur-6996	11	33	students	student	NOUN
alusur-6996	11	34	and	and	CCONJ
alusur-6996	11	35	scholars	scholar	NOUN
alusur-6996	11	36	and	and	CCONJ
alusur-6996	11	37	they	they	PRON
alusur-6996	11	38	can	can	AUX
alusur-6996	11	39	not	not	PART
alusur-6996	11	40	be	be	AUX
alusur-6996	11	41	modified	modify	VERB
alusur-6996	11	42	according	accord	VERB
alusur-6996	11	43	to	to	ADP
alusur-6996	11	44	the	the	DET
alusur-6996	11	45	interests	interest	NOUN
alusur-6996	11	46	and	and	CCONJ
alusur-6996	11	47	needs	need	NOUN
alusur-6996	11	48	of	of	ADP
alusur-6996	11	49	the	the	DET
alusur-6996	11	50	academic	academic	ADJ
alusur-6996	11	51	community	community	NOUN
alusur-6996	11	52	or	or	CCONJ
alusur-6996	11	53	the	the	DET
alusur-6996	11	54	public	public	NOUN
alusur-6996	11	55	at	at	ADP
alusur-6996	11	56	large	large	ADJ
alusur-6996	11	57	.	.	PUNCT
alusur-6996	12	1	most	most	ADV
alusur-6996	12	2	importantly	importantly	ADV
alusur-6996	12	3	,	,	PUNCT
alusur-6996	12	4	they	they	PRON
alusur-6996	12	5	have	have	VERB
alusur-6996	12	6	no	no	DET
alusur-6996	12	7	web	web	NOUN
alusur-6996	12	8	interfaces	interface	NOUN
alusur-6996	12	9	that	that	PRON
alusur-6996	12	10	would	would	AUX
alusur-6996	12	11	enable	enable	VERB
alusur-6996	12	12	the	the	DET
alusur-6996	12	13	production	production	NOUN
alusur-6996	12	14	of	of	ADP
alusur-6996	12	15	wider	wide	ADJ
alusur-6996	12	16	,	,	PUNCT
alusur-6996	12	17	user	user	NOUN
alusur-6996	12	18	-	-	PUNCT
alusur-6996	12	19	generated	generate	VERB
alusur-6996	12	20	collections	collection	NOUN
alusur-6996	12	21	.	.	PUNCT
alusur-6996	13	1	important	important	ADJ
alusur-6996	13	2	new	new	ADJ
alusur-6996	13	3	developments	development	NOUN
alusur-6996	13	4	in	in	ADP
alusur-6996	13	5	arabographic	arabographic	ADJ
alusur-6996	13	6	optical	optical	ADJ
alusur-6996	13	7	character	character	NOUN
alusur-6996	13	8	recognition	recognition	NOUN
alusur-6996	13	9	(	(	PUNCT
alusur-6996	13	10	ocr	ocr	PROPN
alusur-6996	13	11	)	)	PUNCT
alusur-6996	13	12	benjamin	benjamin	PROPN
alusur-6996	13	13	kiessling	kiessling	PROPN
alusur-6996	13	14	,	,	PUNCT
alusur-6996	13	15	matthew	matthew	PROPN
alusur-6996	13	16	thomas	thomas	PROPN
alusur-6996	13	17	miller	miller	PROPN
alusur-6996	13	18	,	,	PUNCT
alusur-6996	13	19	maxim	maxim	NOUN
alusur-6996	13	20	g.	g.	PROPN
alusur-6996	13	21	romanov	romanov	PROPN
alusur-6996	13	22	,	,	PUNCT
alusur-6996	13	23	&	&	CCONJ
alusur-6996	13	24	sarah	sarah	PROPN
alusur-6996	13	25	bowen	bowen	PROPN
alusur-6996	13	26	savant	savant	PROPN
alusur-6996	13	27	https://github.com/mittagessen/kraken	https://github.com/mittagessen/kraken	PROPN
alusur-6996	13	28	mailto	mailto	PROPN
alusur-6996	13	29	:	:	PUNCT
alusur-6996	13	30	sarah.savant%40aku.edu?subject=	sarah.savant%40aku.edu?subject=	PROPN
alusur-6996	13	31	mailto	mailto	PROPN
alusur-6996	13	32	:	:	PUNCT
alusur-6996	13	33	maxim.romanov%40univie.ac.at?subject=	maxim.romanov%40univie.ac.at?subject=	ADJ
alusur-6996	13	34	mailto	mailto	NOUN
alusur-6996	13	35	:	:	PUNCT
alusur-6996	13	36	mtmiller%40umd.edu?subject=	mtmiller%40umd.edu?subject=	PROPN
alusur-6996	13	37	mailto	mailto	PROPN
alusur-6996	13	38	:	:	PUNCT
alusur-6996	13	39	mittagessen%40l.unchti.me?subject=	mittagessen%40l.unchti.me?subject=	ADJ
alusur-6996	13	40	2	2	NUM
alusur-6996	13	41	•	•	NOUN
alusur-6996	13	42	kiessling	kiessle	VERB
alusur-6996	13	43	,	,	PUNCT
alusur-6996	13	44	miller	miller	PROPN
alusur-6996	13	45	,	,	PUNCT
alusur-6996	13	46	romanov	romanov	PROPN
alusur-6996	13	47	,	,	PUNCT
alusur-6996	13	48	&	&	CCONJ
alusur-6996	13	49	savant	savant	PROPN
alusur-6996	13	50	al-ʿuṣūr	al-ʿuṣūr	PROPN
alusur-6996	13	51	al	al	PROPN
alusur-6996	13	52	-	-	PUNCT
alusur-6996	13	53	wusṭā	wusṭā	NOUN
alusur-6996	13	54	25	25	NUM
alusur-6996	13	55	(	(	PUNCT
alusur-6996	13	56	2017	2017	NUM
alusur-6996	13	57	)	)	PUNCT
alusur-6996	13	58	thus	thus	ADV
alusur-6996	13	59	enabling	enable	VERB
alusur-6996	13	60	us	we	PRON
alusur-6996	13	61	to	to	PART
alusur-6996	13	62	make	make	VERB
alusur-6996	13	63	this	this	DET
alusur-6996	13	64	arabic	arabic	ADJ
alusur-6996	13	65	-	-	PUNCT
alusur-6996	13	66	script	script	NOUN
alusur-6996	13	67	ocr	ocr	ADJ
alusur-6996	13	68	technology	technology	NOUN
alusur-6996	13	69	freely	freely	ADV
alusur-6996	13	70	available	available	ADJ
alusur-6996	13	71	to	to	ADP
alusur-6996	13	72	the	the	DET
alusur-6996	13	73	broader	broad	ADJ
alusur-6996	13	74	islamicate	islamicate	ADJ
alusur-6996	13	75	,	,	PUNCT
alusur-6996	13	76	persian	persian	ADJ
alusur-6996	13	77	,	,	PUNCT
alusur-6996	13	78	and	and	CCONJ
alusur-6996	13	79	arabic	arabic	ADJ
alusur-6996	13	80	studies	study	NOUN
alusur-6996	13	81	communities	community	NOUN
alusur-6996	13	82	in	in	ADP
alusur-6996	13	83	the	the	DET
alusur-6996	13	84	near	near	ADJ
alusur-6996	13	85	future	future	NOUN
alusur-6996	13	86	.	.	PUNCT
alusur-6996	14	1	in	in	ADP
alusur-6996	14	2	the	the	DET
alusur-6996	14	3	process	process	NOUN
alusur-6996	14	4	we	we	PRON
alusur-6996	14	5	also	also	ADV
alusur-6996	14	6	generated	generate	VERB
alusur-6996	14	7	over	over	ADP
alusur-6996	14	8	7,000	7,000	NUM
alusur-6996	14	9	lines	line	NOUN
alusur-6996	14	10	of	of	ADP
alusur-6996	14	11	“	"	PUNCT
alusur-6996	14	12	gold	gold	ADJ
alusur-6996	14	13	standard	standard	NOUN
alusur-6996	14	14	”	"	PUNCT
alusur-6996	14	15	(	(	PUNCT
alusur-6996	14	16	double	double	ADV
alusur-6996	14	17	-	-	PUNCT
alusur-6996	14	18	checked	check	VERB
alusur-6996	14	19	)	)	PUNCT
alusur-6996	14	20	training	training	NOUN
alusur-6996	14	21	data	datum	NOUN
alusur-6996	14	22	that	that	PRON
alusur-6996	14	23	can	can	AUX
alusur-6996	14	24	be	be	AUX
alusur-6996	14	25	used	use	VERB
alusur-6996	14	26	by	by	ADP
alusur-6996	14	27	others	other	NOUN
alusur-6996	14	28	for	for	ADP
alusur-6996	14	29	arabic	arabic	ADJ
alusur-6996	14	30	-	-	PUNCT
alusur-6996	14	31	script	script	NOUN
alusur-6996	14	32	ocr	ocr	ADJ
alusur-6996	14	33	training	training	NOUN
alusur-6996	14	34	and	and	CCONJ
alusur-6996	14	35	testing	testing	NOUN
alusur-6996	14	36	purposes.3	purposes.3	PROPN
alusur-6996	14	37	1.2	1.2	NUM
alusur-6996	14	38	ocr	ocr	NOUN
alusur-6996	14	39	and	and	CCONJ
alusur-6996	14	40	its	its	PRON
alusur-6996	14	41	importance	importance	NOUN
alusur-6996	14	42	for	for	ADP
alusur-6996	14	43	islamicate	islamicate	ADJ
alusur-6996	14	44	studies	study	NOUN
alusur-6996	14	45	fields	field	NOUN
alusur-6996	14	46	although	although	SCONJ
alusur-6996	14	47	there	there	PRON
alusur-6996	14	48	is	be	VERB
alusur-6996	14	49	a	a	DET
alusur-6996	14	50	wealth	wealth	NOUN
alusur-6996	14	51	of	of	ADP
alusur-6996	14	52	digital	digital	ADJ
alusur-6996	14	53	persian	persian	ADJ
alusur-6996	14	54	and	and	CCONJ
alusur-6996	14	55	arabic	arabic	ADJ
alusur-6996	14	56	texts	text	NOUN
alusur-6996	14	57	currently	currently	ADV
alusur-6996	14	58	available	available	ADJ
alusur-6996	14	59	in	in	ADP
alusur-6996	14	60	various	various	ADJ
alusur-6996	14	61	open	open	ADJ
alusur-6996	14	62	-	-	PUNCT
alusur-6996	14	63	access	access	NOUN
alusur-6996	14	64	online	online	ADJ
alusur-6996	14	65	repositories,4	repositories,4	NOUN
alusur-6996	14	66	these	these	DET
alusur-6996	14	67	collections	collection	NOUN
alusur-6996	14	68	are	be	AUX
alusur-6996	14	69	not	not	PART
alusur-6996	14	70	representative	representative	NOUN
alusur-6996	14	71	of	of	ADP
alusur-6996	14	72	the	the	DET
alusur-6996	14	73	textual	textual	ADJ
alusur-6996	14	74	traditions	tradition	NOUN
alusur-6996	14	75	in	in	ADP
alusur-6996	14	76	their	their	PRON
alusur-6996	14	77	chronological	chronological	ADJ
alusur-6996	14	78	,	,	PUNCT
alusur-6996	14	79	geographical	geographical	ADJ
alusur-6996	14	80	,	,	PUNCT
alusur-6996	14	81	and	and	CCONJ
alusur-6996	14	82	generic	generic	ADJ
alusur-6996	14	83	spread	spread	NOUN
alusur-6996	14	84	.	.	PUNCT
alusur-6996	15	1	the	the	DET
alusur-6996	15	2	existing	exist	VERB
alusur-6996	15	3	persian	persian	ADJ
alusur-6996	15	4	collections	collection	NOUN
alusur-6996	15	5	,	,	PUNCT
alusur-6996	15	6	for	for	ADP
alusur-6996	15	7	example	example	NOUN
alusur-6996	15	8	,	,	PUNCT
alusur-6996	15	9	are	be	AUX
alusur-6996	15	10	significantly	significantly	ADV
alusur-6996	15	11	smaller	small	ADJ
alusur-6996	15	12	than	than	ADP
alusur-6996	15	13	the	the	DET
alusur-6996	15	14	arabic	arabic	ADJ
alusur-6996	15	15	collections	collection	NOUN
alusur-6996	15	16	and	and	CCONJ
alusur-6996	15	17	lack	lack	VERB
alusur-6996	15	18	prose	prose	NOUN
alusur-6996	15	19	chronicles	chronicle	NOUN
alusur-6996	15	20	and	and	CCONJ
alusur-6996	15	21	philosophical	philosophical	ADJ
alusur-6996	15	22	,	,	PUNCT
alusur-6996	15	23	mystical	mystical	ADJ
alusur-6996	15	24	,	,	PUNCT
alusur-6996	15	25	and	and	CCONJ
alusur-6996	15	26	scientific	scientific	ADJ
alusur-6996	15	27	treatises	treatise	NOUN
alusur-6996	15	28	.	.	PUNCT
alusur-6996	16	1	the	the	DET
alusur-6996	16	2	arabic	arabic	ADJ
alusur-6996	16	3	collections	collection	NOUN
alusur-6996	16	4	would	would	AUX
alusur-6996	16	5	more	more	ADV
alusur-6996	16	6	fully	fully	ADV
alusur-6996	16	7	represent	represent	VERB
alusur-6996	16	8	the	the	DET
alusur-6996	16	9	arabic	arabic	ADJ
alusur-6996	16	10	literary	literary	ADJ
alusur-6996	16	11	tradition	tradition	NOUN
alusur-6996	16	12	if	if	SCONJ
alusur-6996	16	13	they	they	PRON
alusur-6996	16	14	had	have	VERB
alusur-6996	16	15	more	more	ADV
alusur-6996	16	16	scientific	scientific	ADJ
alusur-6996	16	17	and	and	CCONJ
alusur-6996	16	18	philosophical	philosophical	ADJ
alusur-6996	16	19	texts	text	NOUN
alusur-6996	16	20	and	and	CCONJ
alusur-6996	16	21	texts	text	NOUN
alusur-6996	16	22	written	write	VERB
alusur-6996	16	23	by	by	ADP
alusur-6996	16	24	representatives	representative	NOUN
alusur-6996	16	25	of	of	ADP
alusur-6996	16	26	smaller	small	ADJ
alusur-6996	16	27	arabicspeaking	arabicspeake	VERB
alusur-6996	16	28	religious	religious	ADJ
alusur-6996	16	29	communities	community	NOUN
alusur-6996	16	30	.	.	PUNCT
alusur-6996	17	1	moreover	moreover	ADV
alusur-6996	17	2	,	,	PUNCT
alusur-6996	17	3	the	the	DET
alusur-6996	17	4	selection	selection	NOUN
alusur-6996	17	5	of	of	ADP
alusur-6996	17	6	texts	text	NOUN
alusur-6996	17	7	for	for	ADP
alusur-6996	17	8	both	both	CCONJ
alusur-6996	17	9	persian	persian	ADJ
alusur-6996	17	10	and	and	CCONJ
alusur-6996	17	11	arabic	arabic	ADJ
alusur-6996	17	12	digital	digital	ADJ
alusur-6996	17	13	collections	collection	NOUN
alusur-6996	17	14	reflects	reflect	VERB
alusur-6996	17	15	the	the	DET
alusur-6996	17	16	contemporary	contemporary	ADJ
alusur-6996	17	17	ideological	ideological	NOUN
alusur-6996	17	18	,	,	PUNCT
alusur-6996	17	19	aesthetic	aesthetic	ADJ
alusur-6996	17	20	,	,	PUNCT
alusur-6996	17	21	and	and	CCONJ
alusur-6996	17	22	communal	communal	ADJ
alusur-6996	17	23	commitments	commitment	NOUN
alusur-6996	17	24	of	of	ADP
alusur-6996	17	25	their	their	PRON
alusur-6996	17	26	creators	creator	NOUN
alusur-6996	17	27	and	and	CCONJ
alusur-6996	17	28	funders	funder	NOUN
alusur-6996	17	29	.	.	PUNCT
alusur-6996	18	1	while	while	SCONJ
alusur-6996	18	2	these	these	DET
alusur-6996	18	3	shortcomings	shortcoming	NOUN
alusur-6996	18	4	of	of	ADP
alusur-6996	18	5	the	the	DET
alusur-6996	18	6	existing	exist	VERB
alusur-6996	18	7	persian	persian	NOUN
alusur-6996	18	8	and	and	CCONJ
alusur-6996	18	9	arabic	arabic	ADJ
alusur-6996	18	10	digital	digital	ADJ
alusur-6996	18	11	collections	collection	NOUN
alusur-6996	18	12	are	be	AUX
alusur-6996	18	13	well	well	ADV
alusur-6996	18	14	known	know	VERB
alusur-6996	18	15	,	,	PUNCT
alusur-6996	18	16	the	the	DET
alusur-6996	18	17	production	production	NOUN
alusur-6996	18	18	of	of	ADP
alusur-6996	18	19	larger	large	ADJ
alusur-6996	18	20	and	and	CCONJ
alusur-6996	18	21	more	more	ADV
alusur-6996	18	22	representative	representative	ADJ
alusur-6996	18	23	digital	digital	ADJ
alusur-6996	18	24	islamicate	islamicate	NOUN
alusur-6996	18	25	corpora	corpora	PROPN
alusur-6996	18	26	has	have	AUX
alusur-6996	18	27	been	be	AUX
alusur-6996	18	28	stymied	stymie	VERB
alusur-6996	18	29	for	for	ADP
alusur-6996	18	30	decades	decade	NOUN
alusur-6996	18	31	by	by	ADP
alusur-6996	18	32	the	the	DET
alusur-6996	18	33	lack	lack	NOUN
alusur-6996	18	34	of	of	ADP
alusur-6996	18	35	accurate	accurate	ADJ
alusur-6996	18	36	and	and	CCONJ
alusur-6996	18	37	affordable	affordable	ADJ
alusur-6996	18	38	ocr	ocr	PROPN
alusur-6996	18	39	software.5	software.5	VERB
alusur-6996	18	40	3	3	NUM
alusur-6996	18	41	.	.	PUNCT
alusur-6996	19	1	this	this	DET
alusur-6996	19	2	gold	gold	ADJ
alusur-6996	19	3	standard	standard	ADJ
alusur-6996	19	4	data	datum	NOUN
alusur-6996	19	5	is	be	AUX
alusur-6996	19	6	available	available	ADJ
alusur-6996	19	7	at	at	ADP
alusur-6996	19	8	:	:	PUNCT
alusur-6996	19	9	https://github.com/openiti/ocr_gs_data	https://github.com/openiti/ocr_gs_data	X
alusur-6996	19	10	.	.	PUNCT
alusur-6996	19	11	4	4	X
alusur-6996	19	12	.	.	X
alusur-6996	19	13	collecting	collect	VERB
alusur-6996	19	14	and	and	CCONJ
alusur-6996	19	15	rendering	render	VERB
alusur-6996	19	16	these	these	DET
alusur-6996	19	17	texts	text	NOUN
alusur-6996	19	18	useful	useful	ADJ
alusur-6996	19	19	for	for	ADP
alusur-6996	19	20	computational	computational	ADJ
alusur-6996	19	21	textual	textual	ADJ
alusur-6996	19	22	analysis	analysis	NOUN
alusur-6996	19	23	(	(	PUNCT
alusur-6996	19	24	through	through	ADP
alusur-6996	19	25	,	,	PUNCT
alusur-6996	19	26	for	for	ADP
alusur-6996	19	27	example	example	NOUN
alusur-6996	19	28	,	,	PUNCT
alusur-6996	19	29	adding	add	VERB
alusur-6996	19	30	scholarly	scholarly	ADJ
alusur-6996	19	31	metadata	metadata	NOUN
alusur-6996	19	32	and	and	CCONJ
alusur-6996	19	33	making	make	VERB
alusur-6996	19	34	them	they	PRON
alusur-6996	19	35	machine	machine	NOUN
alusur-6996	19	36	-	-	PUNCT
alusur-6996	19	37	actionable	actionable	ADJ
alusur-6996	19	38	)	)	PUNCT
alusur-6996	19	39	is	be	AUX
alusur-6996	19	40	a	a	DET
alusur-6996	19	41	somewhat	somewhat	ADV
alusur-6996	19	42	separate	separate	ADJ
alusur-6996	19	43	but	but	CCONJ
alusur-6996	19	44	deeply	deeply	ADV
alusur-6996	19	45	interrelated	interrelated	ADJ
alusur-6996	19	46	project	project	NOUN
alusur-6996	19	47	that	that	PRON
alusur-6996	19	48	the	the	DET
alusur-6996	19	49	openiti	openiti	NOUN
alusur-6996	19	50	is	be	AUX
alusur-6996	19	51	currently	currently	ADV
alusur-6996	19	52	working	work	VERB
alusur-6996	19	53	on	on	ADP
alusur-6996	19	54	as	as	ADV
alusur-6996	19	55	well	well	ADV
alusur-6996	19	56	.	.	PUNCT
alusur-6996	20	1	5	5	X
alusur-6996	20	2	.	.	X
alusur-6996	20	3	see	see	AUX
alusur-6996	20	4	footnote	footnote	NOUN
alusur-6996	20	5	2	2	NUM
alusur-6996	20	6	for	for	ADP
alusur-6996	20	7	more	more	ADJ
alusur-6996	20	8	details	detail	NOUN
alusur-6996	20	9	.	.	PUNCT
alusur-6996	21	1	table	table	NOUN
alusur-6996	21	2	1	1	NUM
alusur-6996	21	3	:	:	PUNCT
alusur-6996	21	4	description	description	NOUN
alusur-6996	21	5	of	of	ADP
alusur-6996	21	6	data	datum	NOUN
alusur-6996	21	7	https://github.com/openiti/ocr_gs_data	https://github.com/openiti/ocr_gs_data	X
alusur-6996	21	8	al-ʿuṣūr	al-ʿuṣūr	PROPN
alusur-6996	21	9	al	al	PROPN
alusur-6996	21	10	-	-	PUNCT
alusur-6996	21	11	wusṭā	wusṭā	NOUN
alusur-6996	21	12	25	25	NUM
alusur-6996	21	13	(	(	PUNCT
alusur-6996	21	14	2017	2017	NUM
alusur-6996	21	15	)	)	PUNCT
alusur-6996	21	16	new	new	ADJ
alusur-6996	21	17	developments	development	NOUN
alusur-6996	21	18	in	in	ADP
alusur-6996	21	19	arabographic	arabographic	ADJ
alusur-6996	21	20	optical	optical	ADJ
alusur-6996	21	21	character	character	NOUN
alusur-6996	21	22	recognition	recognition	NOUN
alusur-6996	21	23	•	•	NOUN
alusur-6996	21	24	3	3	NUM
alusur-6996	21	25	ocr	ocr	NOUN
alusur-6996	21	26	programs	program	NOUN
alusur-6996	21	27	,	,	PUNCT
alusur-6996	21	28	in	in	ADP
alusur-6996	21	29	the	the	DET
alusur-6996	21	30	simplest	simple	ADJ
alusur-6996	21	31	terms	term	NOUN
alusur-6996	21	32	,	,	PUNCT
alusur-6996	21	33	take	take	VERB
alusur-6996	21	34	an	an	DET
alusur-6996	21	35	image	image	NOUN
alusur-6996	21	36	of	of	ADP
alusur-6996	21	37	a	a	DET
alusur-6996	21	38	text	text	NOUN
alusur-6996	21	39	,	,	PUNCT
alusur-6996	21	40	such	such	ADJ
alusur-6996	21	41	as	as	ADP
alusur-6996	21	42	a	a	DET
alusur-6996	21	43	scan	scan	NOUN
alusur-6996	21	44	of	of	ADP
alusur-6996	21	45	a	a	DET
alusur-6996	21	46	printed	print	VERB
alusur-6996	21	47	book	book	NOUN
alusur-6996	21	48	,	,	PUNCT
alusur-6996	21	49	and	and	CCONJ
alusur-6996	21	50	extract	extract	VERB
alusur-6996	21	51	the	the	DET
alusur-6996	21	52	text	text	NOUN
alusur-6996	21	53	,	,	PUNCT
alusur-6996	21	54	converting	convert	VERB
alusur-6996	21	55	the	the	DET
alusur-6996	21	56	image	image	NOUN
alusur-6996	21	57	of	of	ADP
alusur-6996	21	58	the	the	DET
alusur-6996	21	59	text	text	NOUN
alusur-6996	21	60	into	into	ADP
alusur-6996	21	61	a	a	DET
alusur-6996	21	62	digital	digital	ADJ
alusur-6996	21	63	text	text	NOUN
alusur-6996	21	64	that	that	PRON
alusur-6996	21	65	then	then	ADV
alusur-6996	21	66	can	can	AUX
alusur-6996	21	67	be	be	AUX
alusur-6996	21	68	edited	edit	VERB
alusur-6996	21	69	,	,	PUNCT
alusur-6996	21	70	searched	search	VERB
alusur-6996	21	71	,	,	PUNCT
alusur-6996	21	72	computationally	computationally	ADV
alusur-6996	21	73	analyzed	analyze	VERB
alusur-6996	21	74	,	,	PUNCT
alusur-6996	21	75	etc	etc	X
alusur-6996	21	76	.	.	X
alusur-6996	22	1	ocr	ocr	PROPN
alusur-6996	22	2	’s	’s	PART
alusur-6996	22	3	automation	automation	NOUN
alusur-6996	22	4	of	of	ADP
alusur-6996	22	5	the	the	DET
alusur-6996	22	6	process	process	NOUN
alusur-6996	22	7	of	of	ADP
alusur-6996	22	8	transforming	transform	VERB
alusur-6996	22	9	a	a	DET
alusur-6996	22	10	printed	print	VERB
alusur-6996	22	11	book	book	NOUN
alusur-6996	22	12	into	into	ADP
alusur-6996	22	13	a	a	DET
alusur-6996	22	14	digital	digital	ADJ
alusur-6996	22	15	text	text	NOUN
alusur-6996	22	16	exponentially	exponentially	ADV
alusur-6996	22	17	reduces	reduce	VERB
alusur-6996	22	18	the	the	DET
alusur-6996	22	19	effort	effort	NOUN
alusur-6996	22	20	,	,	PUNCT
alusur-6996	22	21	cost	cost	NOUN
alusur-6996	22	22	,	,	PUNCT
alusur-6996	22	23	and	and	CCONJ
alusur-6996	22	24	time	time	NOUN
alusur-6996	22	25	necessary	necessary	ADJ
alusur-6996	22	26	for	for	ADP
alusur-6996	22	27	the	the	DET
alusur-6996	22	28	production	production	NOUN
alusur-6996	22	29	of	of	ADP
alusur-6996	22	30	digital	digital	ADJ
alusur-6996	22	31	corpora	corpora	PROPN
alusur-6996	22	32	(	(	PUNCT
alusur-6996	22	33	as	as	ADP
alusur-6996	22	34	compared	compare	VERB
alusur-6996	22	35	to	to	ADP
alusur-6996	22	36	the	the	DET
alusur-6996	22	37	alternative	alternative	ADJ
alusur-6996	22	38	option	option	NOUN
alusur-6996	22	39	for	for	ADP
alusur-6996	22	40	producing	produce	VERB
alusur-6996	22	41	high	high	ADJ
alusur-6996	22	42	-	-	PUNCT
alusur-6996	22	43	quality	quality	NOUN
alusur-6996	22	44	digital	digital	ADJ
alusur-6996	22	45	texts	text	NOUN
alusur-6996	22	46	:	:	PUNCT
alusur-6996	22	47	i.e.	i.e.	X
alusur-6996	22	48	,	,	PUNCT
alusur-6996	22	49	paying	pay	VERB
alusur-6996	22	50	multiple	multiple	ADJ
alusur-6996	22	51	individuals	individual	NOUN
alusur-6996	22	52	to	to	PART
alusur-6996	22	53	transcribe	transcribe	VERB
alusur-6996	22	54	,	,	PUNCT
alusur-6996	22	55	or	or	CCONJ
alusur-6996	22	56	“	"	PUNCT
alusur-6996	22	57	double	double	ADJ
alusur-6996	22	58	key	key	NOUN
alusur-6996	22	59	,	,	PUNCT
alusur-6996	22	60	”	"	PUNCT
alusur-6996	22	61	entire	entire	ADJ
alusur-6996	22	62	volumes	volume	NOUN
alusur-6996	22	63	of	of	ADP
alusur-6996	22	64	printed	print	VERB
alusur-6996	22	65	texts	text	NOUN
alusur-6996	22	66	)	)	PUNCT
alusur-6996	22	67	.	.	PUNCT
alusur-6996	23	1	ocr	ocr	PROPN
alusur-6996	23	2	,	,	PUNCT
alusur-6996	23	3	in	in	ADP
alusur-6996	23	4	short	short	ADJ
alusur-6996	23	5	,	,	PUNCT
alusur-6996	23	6	is	be	AUX
alusur-6996	23	7	essential	essential	ADJ
alusur-6996	23	8	for	for	ADP
alusur-6996	23	9	the	the	DET
alusur-6996	23	10	digitization	digitization	NOUN
alusur-6996	23	11	of	of	ADP
alusur-6996	23	12	large	large	ADJ
alusur-6996	23	13	collections	collection	NOUN
alusur-6996	23	14	of	of	ADP
alusur-6996	23	15	printed	print	VERB
alusur-6996	23	16	texts	text	NOUN
alusur-6996	23	17	—	—	PUNCT
alusur-6996	23	18	a	a	DET
alusur-6996	23	19	project	project	NOUN
alusur-6996	23	20	that	that	PRON
alusur-6996	23	21	to	to	ADP
alusur-6996	23	22	date	date	NOUN
alusur-6996	23	23	has	have	AUX
alusur-6996	23	24	remained	remain	VERB
alusur-6996	23	25	unrealized	unrealized	ADJ
alusur-6996	23	26	in	in	ADP
alusur-6996	23	27	persian	persian	PROPN
alusur-6996	23	28	,	,	PUNCT
alusur-6996	23	29	arabic	arabic	ADJ
alusur-6996	23	30	,	,	PUNCT
alusur-6996	23	31	and	and	CCONJ
alusur-6996	23	32	other	other	ADJ
alusur-6996	23	33	islamicate	islamicate	VERB
alusur-6996	23	34	languages.6	languages.6	PROPN
alusur-6996	23	35	the	the	DET
alusur-6996	23	36	specific	specific	ADJ
alusur-6996	23	37	type	type	NOUN
alusur-6996	23	38	of	of	ADP
alusur-6996	23	39	ocr	ocr	ADJ
alusur-6996	23	40	software	software	NOUN
alusur-6996	23	41	that	that	PRON
alusur-6996	23	42	we	we	PRON
alusur-6996	23	43	employed	employ	VERB
alusur-6996	23	44	in	in	ADP
alusur-6996	23	45	our	our	PRON
alusur-6996	23	46	tests	test	NOUN
alusur-6996	23	47	is	be	AUX
alusur-6996	23	48	an	an	DET
alusur-6996	23	49	open	open	ADJ
alusur-6996	23	50	-	-	PUNCT
alusur-6996	23	51	source	source	NOUN
alusur-6996	23	52	ocr	ocr	PROPN
alusur-6996	23	53	program	program	NOUN
alusur-6996	23	54	called	call	VERB
alusur-6996	23	55	kraken	kraken	NOUN
alusur-6996	23	56	(	(	PUNCT
alusur-6996	23	57	more	more	ADV
alusur-6996	23	58	specifically	specifically	ADV
alusur-6996	23	59	,	,	PUNCT
alusur-6996	23	60	kraken	kraken	VERB
alusur-6996	23	61	ibn	ibn	PROPN
alusur-6996	23	62	ocropus	ocropus	PROPN
alusur-6996	23	63	,	,	PUNCT
alusur-6996	23	64	see	see	VERB
alusur-6996	23	65	figure	figure	NOUN
alusur-6996	23	66	1	1	NUM
alusur-6996	23	67	)	)	PUNCT
alusur-6996	23	68	,	,	PUNCT
alusur-6996	23	69	which	which	PRON
alusur-6996	23	70	was	be	AUX
alusur-6996	23	71	developed	develop	VERB
alusur-6996	23	72	by	by	ADP
alusur-6996	23	73	benjamin	benjamin	PROPN
alusur-6996	23	74	kiessling	kiessle	VERB
alusur-6996	23	75	at	at	ADP
alusur-6996	23	76	leipzig	leipzig	PROPN
alusur-6996	23	77	university	university	PROPN
alusur-6996	23	78	’s	’s	PART
alusur-6996	23	79	alexander	alexander	PROPN
alusur-6996	23	80	von	von	PROPN
alusur-6996	23	81	humboldt	humboldt	PROPN
alusur-6996	23	82	chair	chair	PROPN
alusur-6996	23	83	for	for	ADP
alusur-6996	23	84	digital	digital	ADJ
alusur-6996	23	85	humanities	humanity	NOUN
alusur-6996	23	86	.	.	PUNCT
alusur-6996	24	1	unlike	unlike	ADP
alusur-6996	24	2	more	more	ADV
alusur-6996	24	3	traditional	traditional	ADJ
alusur-6996	24	4	ocr	ocr	ADJ
alusur-6996	24	5	approaches	approach	NOUN
alusur-6996	24	6	,	,	PUNCT
alusur-6996	24	7	kraken	kraken	PROPN
alusur-6996	24	8	relies	rely	VERB
alusur-6996	24	9	on	on	ADP
alusur-6996	24	10	a	a	DET
alusur-6996	24	11	neural	neural	ADJ
alusur-6996	24	12	network	network	NOUN
alusur-6996	24	13	—	—	PUNCT
alusur-6996	24	14	which	which	PRON
alusur-6996	24	15	mimics	mimic	VERB
alusur-6996	24	16	the	the	DET
alusur-6996	24	17	way	way	NOUN
alusur-6996	24	18	we	we	PRON
alusur-6996	24	19	learn	learn	VERB
alusur-6996	24	20	—	—	PUNCT
alusur-6996	24	21	to	to	PART
alusur-6996	24	22	recognize	recognize	VERB
alusur-6996	24	23	letters	letter	NOUN
alusur-6996	24	24	in	in	ADP
alusur-6996	24	25	the	the	DET
alusur-6996	24	26	images	image	NOUN
alusur-6996	24	27	of	of	ADP
alusur-6996	24	28	entire	entire	ADJ
alusur-6996	24	29	lines	line	NOUN
alusur-6996	24	30	of	of	ADP
alusur-6996	24	31	text	text	NOUN
alusur-6996	24	32	without	without	ADP
alusur-6996	24	33	attempting	attempt	VERB
alusur-6996	24	34	first	first	ADV
alusur-6996	24	35	to	to	ADP
alusur-6996	24	36	segment	segment	VERB
alusur-6996	24	37	lines	line	NOUN
alusur-6996	24	38	into	into	ADP
alusur-6996	24	39	words	word	NOUN
alusur-6996	24	40	and	and	CCONJ
alusur-6996	24	41	then	then	ADV
alusur-6996	24	42	words	word	NOUN
alusur-6996	24	43	into	into	ADP
alusur-6996	24	44	letters	letter	NOUN
alusur-6996	24	45	.	.	PUNCT
alusur-6996	25	1	this	this	DET
alusur-6996	25	2	segmentation	segmentation	NOUN
alusur-6996	25	3	step	step	NOUN
alusur-6996	25	4	—	—	PUNCT
alusur-6996	25	5	a	a	DET
alusur-6996	25	6	mainstream	mainstream	NOUN
alusur-6996	25	7	ocr	ocr	NOUN
alusur-6996	25	8	approach	approach	NOUN
alusur-6996	25	9	that	that	PRON
alusur-6996	25	10	persistently	persistently	ADV
alusur-6996	25	11	performs	perform	VERB
alusur-6996	25	12	poorly	poorly	ADV
alusur-6996	25	13	on	on	ADP
alusur-6996	25	14	connected	connected	ADJ
alusur-6996	25	15	scripts	script	NOUN
alusur-6996	25	16	—	—	PUNCT
alusur-6996	25	17	is	be	AUX
alusur-6996	25	18	thus	thus	ADV
alusur-6996	25	19	completely	completely	ADV
alusur-6996	25	20	removed	remove	VERB
alusur-6996	25	21	from	from	ADP
alusur-6996	25	22	the	the	DET
alusur-6996	25	23	process	process	NOUN
alusur-6996	25	24	,	,	PUNCT
alusur-6996	25	25	making	make	VERB
alusur-6996	25	26	kraken	kraken	NOUN
alusur-6996	25	27	uniquely	uniquely	ADV
alusur-6996	25	28	powerful	powerful	ADJ
alusur-6996	25	29	for	for	ADP
alusur-6996	25	30	dealing	deal	VERB
alusur-6996	25	31	with	with	ADP
alusur-6996	25	32	the	the	DET
alusur-6996	25	33	diverse	diverse	ADJ
alusur-6996	25	34	variety	variety	NOUN
alusur-6996	25	35	of	of	ADP
alusur-6996	25	36	ligatures	ligature	NOUN
alusur-6996	25	37	in	in	ADP
alusur-6996	25	38	connected	connected	ADJ
alusur-6996	25	39	arabic	arabic	ADJ
alusur-6996	25	40	script	script	NOUN
alusur-6996	25	41	(	(	PUNCT
alusur-6996	25	42	see	see	VERB
alusur-6996	25	43	section	section	NOUN
alusur-6996	25	44	4.1	4.1	NUM
alusur-6996	25	45	for	for	ADP
alusur-6996	25	46	more	more	ADJ
alusur-6996	25	47	technical	technical	ADJ
alusur-6996	25	48	details	detail	NOUN
alusur-6996	25	49	)	)	PUNCT
alusur-6996	25	50	.	.	PUNCT
alusur-6996	26	1	2.1	2.1	NUM
alusur-6996	26	2	initial	initial	ADJ
alusur-6996	26	3	ocr	ocr	ADJ
alusur-6996	26	4	tests	test	NOUN
alusur-6996	26	5	we	we	PRON
alusur-6996	26	6	began	begin	VERB
alusur-6996	26	7	our	our	PRON
alusur-6996	26	8	experiments	experiment	NOUN
alusur-6996	26	9	by	by	ADP
alusur-6996	26	10	using	use	VERB
alusur-6996	26	11	kraken	kraken	NOUN
alusur-6996	26	12	to	to	PART
alusur-6996	26	13	train	train	VERB
alusur-6996	26	14	a	a	DET
alusur-6996	26	15	model7	model7	NOUN
alusur-6996	26	16	on	on	ADP
alusur-6996	26	17	high	high	ADJ
alusur-6996	26	18	-	-	PUNCT
alusur-6996	26	19	quality8	quality8	NOUN
alusur-6996	26	20	scans	scan	NOUN
alusur-6996	26	21	of	of	ADP
alusur-6996	26	22	1,466	1,466	NUM
alusur-6996	26	23	lines	line	NOUN
alusur-6996	26	24	of	of	ADP
alusur-6996	26	25	ibn	ibn	PROPN
alusur-6996	26	26	al	al	PROPN
alusur-6996	26	27	-	-	PUNCT
alusur-6996	26	28	faqīh	faqīh	PROPN
alusur-6996	26	29	’s	’s	PART
alusur-6996	26	30	kitāb	kitāb	ADJ
alusur-6996	26	31	al	al	PROPN
alusur-6996	26	32	-	-	PUNCT
alusur-6996	26	33	buldān	buldān	NOUN
alusur-6996	26	34	(	(	PUNCT
alusur-6996	26	35	work	work	NOUN
alusur-6996	26	36	#	#	SYM
alusur-6996	26	37	0	0	NUM
alusur-6996	26	38	)	)	PUNCT
alusur-6996	26	39	.	.	PUNCT
alusur-6996	27	1	we	we	PRON
alusur-6996	27	2	first	first	ADV
alusur-6996	27	3	generated	generate	VERB
alusur-6996	27	4	training	training	NOUN
alusur-6996	27	5	data	datum	NOUN
alusur-6996	27	6	(	(	PUNCT
alusur-6996	27	7	line	line	NOUN
alusur-6996	27	8	transcriptions	transcription	NOUN
alusur-6996	27	9	)	)	PUNCT
alusur-6996	27	10	for	for	ADP
alusur-6996	27	11	all	all	PRON
alusur-6996	27	12	of	of	ADP
alusur-6996	27	13	these	these	DET
alusur-6996	27	14	lines	line	NOUN
alusur-6996	27	15	,	,	PUNCT
alusur-6996	27	16	double	double	ADV
alusur-6996	27	17	checked	check	VERB
alusur-6996	27	18	them	they	PRON
alusur-6996	27	19	(	(	PUNCT
alusur-6996	27	20	creating	create	VERB
alusur-6996	27	21	so	so	ADV
alusur-6996	27	22	-	-	PUNCT
alusur-6996	27	23	called	call	VERB
alusur-6996	27	24	“	"	PUNCT
alusur-6996	27	25	gold	gold	ADJ
alusur-6996	27	26	standard	standard	NOUN
alusur-6996	27	27	”	"	PUNCT
alusur-6996	27	28	data	datum	NOUN
alusur-6996	27	29	)	)	PUNCT
alusur-6996	27	30	,	,	PUNCT
alusur-6996	27	31	trained	train	VERB
alusur-6996	27	32	the	the	DET
alusur-6996	27	33	model	model	NOUN
alusur-6996	27	34	,	,	PUNCT
alusur-6996	27	35	and	and	CCONJ
alusur-6996	27	36	,	,	PUNCT
alusur-6996	27	37	finally	finally	ADV
alusur-6996	27	38	,	,	PUNCT
alusur-6996	27	39	tested	test	VERB
alusur-6996	27	40	its	its	PRON
alusur-6996	27	41	ability	ability	NOUN
alusur-6996	27	42	to	to	PART
alusur-6996	27	43	accurately	accurately	ADV
alusur-6996	27	44	recognize	recognize	VERB
alusur-6996	27	45	and	and	CCONJ
alusur-6996	27	46	extract	extract	VERB
alusur-6996	27	47	the	the	DET
alusur-6996	27	48	6	6	NUM
alusur-6996	27	49	.	.	PUNCT
alusur-6996	28	1	kraken	kraken	PROPN
alusur-6996	28	2	’s	’s	PART
alusur-6996	28	3	logo	logo	NOUN
alusur-6996	28	4	,	,	PUNCT
alusur-6996	28	5	kraken	kraken	NOUN
alusur-6996	28	6	ibn	ibn	PROPN
alusur-6996	28	7	ocropus	ocropus	NOUN
alusur-6996	28	8	,	,	PUNCT
alusur-6996	28	9	is	be	AUX
alusur-6996	28	10	based	base	VERB
alusur-6996	28	11	on	on	ADP
alusur-6996	28	12	a	a	DET
alusur-6996	28	13	depiction	depiction	NOUN
alusur-6996	28	14	of	of	ADP
alusur-6996	28	15	an	an	DET
alusur-6996	28	16	octopus	octopus	NOUN
alusur-6996	28	17	from	from	ADP
alusur-6996	28	18	a	a	DET
alusur-6996	28	19	manuscript	manuscript	NOUN
alusur-6996	28	20	of	of	ADP
alusur-6996	28	21	kitāb	kitāb	PROPN
alusur-6996	28	22	al	al	PROPN
alusur-6996	28	23	-	-	PUNCT
alusur-6996	28	24	ḥashāʾish	ḥashāʾish	PROPN
alusur-6996	28	25	fī	fī	PROPN
alusur-6996	28	26	hāyūlā	hāyūlā	PROPN
alusur-6996	28	27	al-ʿilāj	al-ʿilāj	PROPN
alusur-6996	28	28	al	al	PROPN
alusur-6996	28	29	-	-	PUNCT
alusur-6996	28	30	ṭibbī	ṭibbī	PROPN
alusur-6996	28	31	(	(	PUNCT
alusur-6996	28	32	leiden	leiden	ADJ
alusur-6996	28	33	,	,	PUNCT
alusur-6996	28	34	ub	ub	ADP
alusur-6996	28	35	:	:	PUNCT
alusur-6996	28	36	or	or	CCONJ
alusur-6996	28	37	.	.	PUNCT
alusur-6996	28	38	289	289	NUM
alusur-6996	28	39	)	)	PUNCT
alusur-6996	28	40	;	;	PUNCT
alusur-6996	28	41	special	special	ADJ
alusur-6996	28	42	thanks	thank	NOUN
alusur-6996	28	43	to	to	ADP
alusur-6996	28	44	emily	emily	ADV
alusur-6996	28	45	selove	selove	VERB
alusur-6996	28	46	for	for	ADP
alusur-6996	28	47	help	help	NOUN
alusur-6996	28	48	with	with	ADP
alusur-6996	28	49	finding	find	VERB
alusur-6996	28	50	an	an	DET
alusur-6996	28	51	octopus	octopus	NOUN
alusur-6996	28	52	in	in	ADP
alusur-6996	28	53	the	the	DET
alusur-6996	28	54	depths	depth	NOUN
alusur-6996	28	55	of	of	ADP
alusur-6996	28	56	the	the	DET
alusur-6996	28	57	islamicate	islamicate	ADJ
alusur-6996	28	58	manuscript	manuscript	NOUN
alusur-6996	28	59	traditions	tradition	NOUN
alusur-6996	28	60	.	.	PUNCT
alusur-6996	29	1	7	7	X
alusur-6996	29	2	.	.	PUNCT
alusur-6996	29	3	“	"	PUNCT
alusur-6996	29	4	training	train	VERB
alusur-6996	29	5	a	a	DET
alusur-6996	29	6	model	model	NOUN
alusur-6996	29	7	”	"	PUNCT
alusur-6996	29	8	is	be	AUX
alusur-6996	29	9	a	a	DET
alusur-6996	29	10	general	general	ADJ
alusur-6996	29	11	term	term	NOUN
alusur-6996	29	12	used	use	VERB
alusur-6996	29	13	in	in	ADP
alusur-6996	29	14	machine	machine	NOUN
alusur-6996	29	15	learning	learn	VERB
alusur-6996	29	16	for	for	ADP
alusur-6996	29	17	training	train	VERB
alusur-6996	29	18	a	a	DET
alusur-6996	29	19	program	program	NOUN
alusur-6996	29	20	to	to	PART
alusur-6996	29	21	recognize	recognize	VERB
alusur-6996	29	22	certain	certain	ADJ
alusur-6996	29	23	patterns	pattern	NOUN
alusur-6996	29	24	in	in	ADP
alusur-6996	29	25	data	datum	NOUN
alusur-6996	29	26	.	.	PUNCT
alusur-6996	30	1	in	in	ADP
alusur-6996	30	2	the	the	DET
alusur-6996	30	3	context	context	NOUN
alusur-6996	30	4	of	of	ADP
alusur-6996	30	5	ocr	ocr	ADJ
alusur-6996	30	6	work	work	NOUN
alusur-6996	30	7	,	,	PUNCT
alusur-6996	30	8	it	it	PRON
alusur-6996	30	9	refers	refer	VERB
alusur-6996	30	10	to	to	ADP
alusur-6996	30	11	teaching	teach	VERB
alusur-6996	30	12	the	the	DET
alusur-6996	30	13	ocr	ocr	ADJ
alusur-6996	30	14	software	software	NOUN
alusur-6996	30	15	to	to	PART
alusur-6996	30	16	recognize	recognize	VERB
alusur-6996	30	17	a	a	DET
alusur-6996	30	18	particular	particular	ADJ
alusur-6996	30	19	script	script	NOUN
alusur-6996	30	20	or	or	CCONJ
alusur-6996	30	21	typeface	typeface	NOUN
alusur-6996	30	22	—	—	PUNCT
alusur-6996	30	23	a	a	DET
alusur-6996	30	24	process	process	NOUN
alusur-6996	30	25	that	that	PRON
alusur-6996	30	26	only	only	ADV
alusur-6996	30	27	requires	require	VERB
alusur-6996	30	28	time	time	NOUN
alusur-6996	30	29	and	and	CCONJ
alusur-6996	30	30	computing	computing	NOUN
alusur-6996	30	31	power	power	NOUN
alusur-6996	30	32	.	.	PUNCT
alusur-6996	31	1	in	in	ADP
alusur-6996	31	2	our	our	PRON
alusur-6996	31	3	case	case	NOUN
alusur-6996	31	4	,	,	PUNCT
alusur-6996	31	5	this	this	DET
alusur-6996	31	6	process	process	NOUN
alusur-6996	31	7	required	require	VERB
alusur-6996	31	8	1	1	NUM
alusur-6996	31	9	computer	computer	NOUN
alusur-6996	31	10	core	core	NOUN
alusur-6996	31	11	and	and	CCONJ
alusur-6996	31	12	approximately	approximately	ADV
alusur-6996	31	13	24	24	NUM
alusur-6996	31	14	hours	hour	NOUN
alusur-6996	31	15	per	per	ADP
alusur-6996	31	16	model	model	NOUN
alusur-6996	31	17	.	.	PUNCT
alusur-6996	32	1	8	8	X
alusur-6996	32	2	.	.	PUNCT
alusur-6996	32	3	“	"	PUNCT
alusur-6996	32	4	high	high	ADJ
alusur-6996	32	5	quality	quality	NOUN
alusur-6996	32	6	”	"	PUNCT
alusur-6996	32	7	here	here	ADV
alusur-6996	32	8	means	mean	VERB
alusur-6996	32	9	300	300	NUM
alusur-6996	32	10	dpi	dpi	NOUN
alusur-6996	32	11	color	color	NOUN
alusur-6996	32	12	or	or	CCONJ
alusur-6996	32	13	grayscale	grayscale	NOUN
alusur-6996	32	14	images	image	NOUN
alusur-6996	32	15	.	.	PUNCT
alusur-6996	33	1	before	before	ADP
alusur-6996	33	2	the	the	DET
alusur-6996	33	3	actual	actual	ADJ
alusur-6996	33	4	process	process	NOUN
alusur-6996	33	5	of	of	ADP
alusur-6996	33	6	ocr	ocr	NOUN
alusur-6996	33	7	,	,	PUNCT
alusur-6996	33	8	these	these	DET
alusur-6996	33	9	images	image	NOUN
alusur-6996	33	10	must	must	AUX
alusur-6996	33	11	be	be	AUX
alusur-6996	33	12	binarized	binarize	VERB
alusur-6996	33	13	—	—	PUNCT
alusur-6996	33	14	i.e.	i.e.	X
alusur-6996	33	15	converted	convert	VERB
alusur-6996	33	16	into	into	ADP
alusur-6996	33	17	black	black	NOUN
alusur-6996	33	18	-	-	PUNCT
alusur-6996	33	19	and	and	CCONJ
alusur-6996	33	20	-	-	PUNCT
alusur-6996	33	21	white	white	ADJ
alusur-6996	33	22	images	image	NOUN
alusur-6996	33	23	.	.	PUNCT
alusur-6996	34	1	if	if	SCONJ
alusur-6996	34	2	binarization	binarization	NOUN
alusur-6996	34	3	is	be	AUX
alusur-6996	34	4	not	not	PART
alusur-6996	34	5	performed	perform	VERB
alusur-6996	34	6	properly	properly	ADV
alusur-6996	34	7	,	,	PUNCT
alusur-6996	34	8	a	a	DET
alusur-6996	34	9	lot	lot	NOUN
alusur-6996	34	10	of	of	ADP
alusur-6996	34	11	information	information	NOUN
alusur-6996	34	12	is	be	AUX
alusur-6996	34	13	lost	lose	VERB
alusur-6996	34	14	from	from	ADP
alusur-6996	34	15	the	the	DET
alusur-6996	34	16	image	image	NOUN
alusur-6996	34	17	,	,	PUNCT
alusur-6996	34	18	negatively	negatively	ADV
alusur-6996	34	19	affecting	affect	VERB
alusur-6996	34	20	the	the	DET
alusur-6996	34	21	accuracy	accuracy	NOUN
alusur-6996	34	22	of	of	ADP
alusur-6996	34	23	the	the	DET
alusur-6996	34	24	ocr	ocr	ADJ
alusur-6996	34	25	output	output	NOUN
alusur-6996	34	26	.	.	PUNCT
alusur-6996	35	1	for	for	ADP
alusur-6996	35	2	this	this	DET
alusur-6996	35	3	reason	reason	NOUN
alusur-6996	35	4	,	,	PUNCT
alusur-6996	35	5	for	for	ADP
alusur-6996	35	6	best	good	ADJ
alusur-6996	35	7	results	result	NOUN
alusur-6996	35	8	,	,	PUNCT
alusur-6996	35	9	one	one	PRON
alusur-6996	35	10	should	should	AUX
alusur-6996	35	11	avoid	avoid	VERB
alusur-6996	35	12	using	use	VERB
alusur-6996	35	13	pre	pre	ADJ
alusur-6996	35	14	-	-	ADJ
alusur-6996	35	15	binarized	binarized	ADJ
alusur-6996	35	16	images	image	NOUN
alusur-6996	35	17	(	(	PUNCT
alusur-6996	35	18	i.e.	i.e.	X
alusur-6996	35	19	,	,	PUNCT
alusur-6996	35	20	images	image	NOUN
alusur-6996	35	21	that	that	PRON
alusur-6996	35	22	were	be	AUX
alusur-6996	35	23	already	already	ADV
alusur-6996	35	24	converted	convert	VERB
alusur-6996	35	25	to	to	ADP
alusur-6996	35	26	black	black	ADJ
alusur-6996	35	27	and	and	CCONJ
alusur-6996	35	28	white	white	ADJ
alusur-6996	35	29	during	during	ADP
alusur-6996	35	30	the	the	DET
alusur-6996	35	31	scanning	scanning	NOUN
alusur-6996	35	32	process	process	NOUN
alusur-6996	35	33	,	,	PUNCT
alusur-6996	35	34	usually	usually	ADV
alusur-6996	35	35	for	for	ADP
alusur-6996	35	36	size	size	NOUN
alusur-6996	35	37	reduction	reduction	NOUN
alusur-6996	35	38	,	,	PUNCT
alusur-6996	35	39	which	which	PRON
alusur-6996	35	40	results	result	VERB
alusur-6996	35	41	in	in	ADP
alusur-6996	35	42	some	some	DET
alusur-6996	35	43	degradation	degradation	NOUN
alusur-6996	35	44	of	of	ADP
alusur-6996	35	45	quality	quality	NOUN
alusur-6996	35	46	and	and	CCONJ
alusur-6996	35	47	the	the	DET
alusur-6996	35	48	loss	loss	NOUN
alusur-6996	35	49	of	of	ADP
alusur-6996	35	50	information	information	NOUN
alusur-6996	35	51	)	)	PUNCT
alusur-6996	35	52	.	.	PUNCT
alusur-6996	36	1	figure	figure	NOUN
alusur-6996	36	2	1	1	NUM
alusur-6996	36	3	:	:	PUNCT
alusur-6996	36	4	kraken	kraken	NOUN
alusur-6996	36	5	ibn	ibn	PROPN
alusur-6996	36	6	ocropus6	ocropus6	NOUN
alusur-6996	36	7	4	4	NUM
alusur-6996	36	8	•	•	NOUN
alusur-6996	36	9	kiessling	kiessle	VERB
alusur-6996	36	10	,	,	PUNCT
alusur-6996	36	11	miller	miller	PROPN
alusur-6996	36	12	,	,	PUNCT
alusur-6996	36	13	romanov	romanov	PROPN
alusur-6996	36	14	,	,	PUNCT
alusur-6996	36	15	&	&	CCONJ
alusur-6996	36	16	savant	savant	PROPN
alusur-6996	36	17	al-ʿuṣūr	al-ʿuṣūr	PROPN
alusur-6996	36	18	al	al	PROPN
alusur-6996	36	19	-	-	PUNCT
alusur-6996	36	20	wusṭā	wusṭā	NOUN
alusur-6996	36	21	25	25	NUM
alusur-6996	36	22	(	(	PUNCT
alusur-6996	36	23	2017	2017	NUM
alusur-6996	36	24	)	)	PUNCT
alusur-6996	36	25	text	text	NOUN
alusur-6996	36	26	.	.	PUNCT
alusur-6996	37	1	the	the	DET
alusur-6996	37	2	results	result	NOUN
alusur-6996	37	3	were	be	AUX
alusur-6996	37	4	impressive	impressive	ADJ
alusur-6996	37	5	,	,	PUNCT
alusur-6996	37	6	reaching	reach	VERB
alusur-6996	37	7	97.56	97.56	NUM
alusur-6996	37	8	%	%	NOUN
alusur-6996	37	9	accuracy	accuracy	NOUN
alusur-6996	37	10	for	for	ADP
alusur-6996	37	11	the	the	DET
alusur-6996	37	12	entire	entire	ADJ
alusur-6996	37	13	text	text	NOUN
alusur-6996	37	14	and	and	CCONJ
alusur-6996	37	15	an	an	DET
alusur-6996	37	16	even	even	ADV
alusur-6996	37	17	more	more	ADV
alusur-6996	37	18	impressive	impressive	ADJ
alusur-6996	37	19	99.68	99.68	NUM
alusur-6996	37	20	%	%	NOUN
alusur-6996	37	21	accuracy	accuracy	NOUN
alusur-6996	37	22	rate	rate	NOUN
alusur-6996	37	23	on	on	ADP
alusur-6996	37	24	the	the	DET
alusur-6996	37	25	arabic	arabic	ADJ
alusur-6996	37	26	script	script	NOUN
alusur-6996	37	27	alone	alone	ADV
alusur-6996	37	28	(	(	PUNCT
alusur-6996	37	29	i.e.	i.e.	X
alusur-6996	37	30	,	,	PUNCT
alusur-6996	37	31	when	when	SCONJ
alusur-6996	37	32	errors	error	NOUN
alusur-6996	37	33	related	relate	VERB
alusur-6996	37	34	to	to	ADP
alusur-6996	37	35	punctuation	punctuation	NOUN
alusur-6996	37	36	and	and	CCONJ
alusur-6996	37	37	spaces	space	NOUN
alusur-6996	37	38	were	be	AUX
alusur-6996	37	39	removed	remove	VERB
alusur-6996	37	40	from	from	ADP
alusur-6996	37	41	consideration	consideration	NOUN
alusur-6996	37	42	;	;	PUNCT
alusur-6996	37	43	such	such	ADJ
alusur-6996	37	44	non	non	ADJ
alusur-6996	37	45	-	-	ADJ
alusur-6996	37	46	script	script	ADJ
alusur-6996	37	47	errors	error	NOUN
alusur-6996	37	48	are	be	AUX
alusur-6996	37	49	easy	easy	ADJ
alusur-6996	37	50	to	to	PART
alusur-6996	37	51	fix	fix	VERB
alusur-6996	37	52	in	in	ADP
alusur-6996	37	53	the	the	DET
alusur-6996	37	54	post	post	ADJ
alusur-6996	37	55	-	-	ADJ
alusur-6996	37	56	correction	correction	ADJ
alusur-6996	37	57	phase	phase	NOUN
alusur-6996	37	58	and	and	CCONJ
alusur-6996	37	59	,	,	PUNCT
alusur-6996	37	60	in	in	ADP
alusur-6996	37	61	many	many	ADJ
alusur-6996	37	62	cases	case	NOUN
alusur-6996	37	63	,	,	PUNCT
alusur-6996	37	64	this	this	DET
alusur-6996	37	65	correction	correction	NOUN
alusur-6996	37	66	process	process	NOUN
alusur-6996	37	67	for	for	ADP
alusur-6996	37	68	non	non	ADJ
alusur-6996	37	69	-	-	ADJ
alusur-6996	37	70	script	script	ADJ
alusur-6996	37	71	errors	error	NOUN
alusur-6996	37	72	can	can	AUX
alusur-6996	37	73	be	be	AUX
alusur-6996	37	74	automated	automate	VERB
alusur-6996	37	75	)	)	PUNCT
alusur-6996	37	76	.	.	PUNCT
alusur-6996	38	1	see	see	VERB
alusur-6996	38	2	table	table	NOUN
alusur-6996	38	3	2	2	NUM
alusur-6996	38	4	,	,	PUNCT
alusur-6996	38	5	row	row	VERB
alusur-6996	38	6	#	#	SYM
alusur-6996	38	7	0	0	NUM
alusur-6996	38	8	for	for	ADP
alusur-6996	38	9	full	full	ADJ
alusur-6996	38	10	details.9	details.9	PROPN
alusur-6996	38	11	these	these	DET
alusur-6996	38	12	numbers	number	NOUN
alusur-6996	38	13	were	be	AUX
alusur-6996	38	14	so	so	ADV
alusur-6996	38	15	impressive	impressive	ADJ
alusur-6996	38	16	that	that	SCONJ
alusur-6996	38	17	we	we	PRON
alusur-6996	38	18	decided	decide	VERB
alusur-6996	38	19	to	to	PART
alusur-6996	38	20	expand	expand	VERB
alusur-6996	38	21	our	our	PRON
alusur-6996	38	22	study	study	NOUN
alusur-6996	38	23	and	and	CCONJ
alusur-6996	38	24	use	use	VERB
alusur-6996	38	25	the	the	DET
alusur-6996	38	26	model	model	NOUN
alusur-6996	38	27	built	build	VERB
alusur-6996	38	28	on	on	ADP
alusur-6996	38	29	the	the	DET
alusur-6996	38	30	text	text	NOUN
alusur-6996	38	31	of	of	ADP
alusur-6996	38	32	ibn	ibn	PROPN
alusur-6996	38	33	al	al	PROPN
alusur-6996	38	34	-	-	PUNCT
alusur-6996	38	35	faqīh	faqīh	PROPN
alusur-6996	38	36	’s	’s	PART
alusur-6996	38	37	kitāb	kitāb	ADJ
alusur-6996	38	38	al	al	PROPN
alusur-6996	38	39	-	-	PUNCT
alusur-6996	38	40	buldān	buldān	NOUN
alusur-6996	38	41	(	(	PUNCT
alusur-6996	38	42	work	work	NOUN
alusur-6996	38	43	#	#	SYM
alusur-6996	38	44	0	0	NUM
alusur-6996	38	45	)	)	PUNCT
alusur-6996	38	46	to	to	AUX
alusur-6996	38	47	ocr	ocr	VERB
alusur-6996	38	48	six	six	NUM
alusur-6996	38	49	other	other	ADJ
alusur-6996	38	50	texts	text	NOUN
alusur-6996	38	51	.	.	PUNCT
alusur-6996	39	1	we	we	PRON
alusur-6996	39	2	deliberately	deliberately	ADV
alusur-6996	39	3	selected	select	VERB
alusur-6996	39	4	texts	text	NOUN
alusur-6996	39	5	that	that	PRON
alusur-6996	39	6	were	be	AUX
alusur-6996	39	7	different	different	ADJ
alusur-6996	39	8	from	from	ADP
alusur-6996	39	9	ibn	ibn	PROPN
alusur-6996	39	10	al	al	PROPN
alusur-6996	39	11	-	-	PUNCT
alusur-6996	39	12	faqīh	faqīh	PROPN
alusur-6996	39	13	’s	’s	PART
alusur-6996	39	14	original	original	ADJ
alusur-6996	39	15	text	text	NOUN
alusur-6996	39	16	in	in	ADP
alusur-6996	39	17	terms	term	NOUN
alusur-6996	39	18	of	of	ADP
alusur-6996	39	19	both	both	CCONJ
alusur-6996	39	20	their	their	PRON
alusur-6996	39	21	arabic	arabic	ADJ
alusur-6996	39	22	typeface	typeface	NOUN
alusur-6996	39	23	,	,	PUNCT
alusur-6996	39	24	editorial	editorial	ADJ
alusur-6996	39	25	orthographic	orthographic	ADJ
alusur-6996	39	26	conventions	convention	NOUN
alusur-6996	39	27	,	,	PUNCT
alusur-6996	39	28	and	and	CCONJ
alusur-6996	39	29	image	image	NOUN
alusur-6996	39	30	quality	quality	NOUN
alusur-6996	39	31	.	.	PUNCT
alusur-6996	40	1	these	these	DET
alusur-6996	40	2	texts	text	NOUN
alusur-6996	40	3	represent	represent	VERB
alusur-6996	40	4	at	at	ADV
alusur-6996	40	5	least	least	ADV
alusur-6996	40	6	two	two	NUM
alusur-6996	40	7	different	different	ADJ
alusur-6996	40	8	typefaces	typeface	NOUN
alusur-6996	40	9	(	(	PUNCT
alusur-6996	40	10	within	within	ADP
alusur-6996	40	11	which	which	PRON
alusur-6996	40	12	there	there	PRON
alusur-6996	40	13	are	be	VERB
alusur-6996	40	14	noticeable	noticeable	ADJ
alusur-6996	40	15	variations	variation	NOUN
alusur-6996	40	16	of	of	ADP
alusur-6996	40	17	font	font	NOUN
alusur-6996	40	18	,	,	PUNCT
alusur-6996	40	19	spacing	spacing	NOUN
alusur-6996	40	20	,	,	PUNCT
alusur-6996	40	21	and	and	CCONJ
alusur-6996	40	22	ligature	ligature	NOUN
alusur-6996	40	23	styles	style	NOUN
alusur-6996	40	24	)	)	PUNCT
alusur-6996	40	25	,	,	PUNCT
alusur-6996	40	26	and	and	CCONJ
alusur-6996	40	27	four	four	NUM
alusur-6996	40	28	of	of	ADP
alusur-6996	40	29	the	the	DET
alusur-6996	40	30	texts	text	NOUN
alusur-6996	40	31	were	be	AUX
alusur-6996	40	32	high	high	ADJ
alusur-6996	40	33	-	-	PUNCT
alusur-6996	40	34	quality	quality	NOUN
alusur-6996	40	35	scans	scan	NOUN
alusur-6996	40	36	while	while	SCONJ
alusur-6996	40	37	the	the	DET
alusur-6996	40	38	other	other	ADJ
alusur-6996	40	39	two	two	NUM
alusur-6996	40	40	were	be	AUX
alusur-6996	40	41	low	low	ADJ
alusur-6996	40	42	-	-	PUNCT
alusur-6996	40	43	quality	quality	NOUN
alusur-6996	40	44	scans	scan	NOUN
alusur-6996	40	45	downloaded	download	VERB
alusur-6996	40	46	from	from	ADP
alusur-6996	40	47	www.archive.org	www.archive.org	NUM
alusur-6996	40	48	(	(	PUNCT
alusur-6996	40	49	via	via	ADP
alusur-6996	40	50	http://waqfeya.com/).10	http://waqfeya.com/).10	NOUN
alusur-6996	40	51	when	when	SCONJ
alusur-6996	40	52	looking	look	VERB
alusur-6996	40	53	at	at	ADP
alusur-6996	40	54	the	the	DET
alusur-6996	40	55	results	result	NOUN
alusur-6996	40	56	in	in	ADP
alusur-6996	40	57	table	table	NOUN
alusur-6996	40	58	2	2	NUM
alusur-6996	40	59	,	,	PUNCT
alusur-6996	40	60	it	it	PRON
alusur-6996	40	61	is	be	AUX
alusur-6996	40	62	important	important	ADJ
alusur-6996	40	63	that	that	SCONJ
alusur-6996	40	64	the	the	DET
alusur-6996	40	65	reader	reader	NOUN
alusur-6996	40	66	notes	note	VERB
alusur-6996	40	67	that	that	PRON
alusur-6996	40	68	works	work	VERB
alusur-6996	40	69	#	#	SYM
alusur-6996	40	70	1	1	NUM
alusur-6996	40	71	-	-	SYM
alusur-6996	40	72	6	6	NUM
alusur-6996	40	73	are	be	AUX
alusur-6996	40	74	“	"	PUNCT
alusur-6996	40	75	testing	test	VERB
alusur-6996	40	76	”	"	PUNCT
alusur-6996	40	77	data	datum	NOUN
alusur-6996	40	78	.	.	PUNCT
alusur-6996	41	1	that	that	PRON
alusur-6996	41	2	is	be	AUX
alusur-6996	41	3	,	,	PUNCT
alusur-6996	41	4	these	these	DET
alusur-6996	41	5	accuracy	accuracy	NOUN
alusur-6996	41	6	results	result	NOUN
alusur-6996	41	7	were	be	AUX
alusur-6996	41	8	achieved	achieve	VERB
alusur-6996	41	9	by	by	ADP
alusur-6996	41	10	utilizing	utilize	VERB
alusur-6996	41	11	a	a	DET
alusur-6996	41	12	model	model	NOUN
alusur-6996	41	13	built	build	VERB
alusur-6996	41	14	on	on	ADP
alusur-6996	41	15	the	the	DET
alusur-6996	41	16	text	text	NOUN
alusur-6996	41	17	of	of	ADP
alusur-6996	41	18	work	work	NOUN
alusur-6996	41	19	#	#	SYM
alusur-6996	41	20	0	0	NUM
alusur-6996	41	21	to	to	PART
alusur-6996	41	22	perform	perform	VERB
alusur-6996	41	23	ocr	ocr	NOUN
alusur-6996	41	24	on	on	ADP
alusur-6996	41	25	these	these	DET
alusur-6996	41	26	other	other	ADJ
alusur-6996	41	27	texts	text	NOUN
alusur-6996	41	28	.	.	PUNCT
alusur-6996	42	1	for	for	ADP
alusur-6996	42	2	this	this	DET
alusur-6996	42	3	reason	reason	NOUN
alusur-6996	42	4	,	,	PUNCT
alusur-6996	42	5	it	it	PRON
alusur-6996	42	6	is	be	AUX
alusur-6996	42	7	not	not	PART
alusur-6996	42	8	surprising	surprising	ADJ
alusur-6996	42	9	9	9	NUM
alusur-6996	42	10	.	.	PUNCT
alusur-6996	43	1	we	we	PRON
alusur-6996	43	2	have	have	AUX
alusur-6996	43	3	also	also	ADV
alusur-6996	43	4	experimented	experiment	VERB
alusur-6996	43	5	with	with	ADP
alusur-6996	43	6	the	the	DET
alusur-6996	43	7	internal	internal	ADJ
alusur-6996	43	8	configuration	configuration	NOUN
alusur-6996	43	9	of	of	ADP
alusur-6996	43	10	our	our	PRON
alusur-6996	43	11	models	model	NOUN
alusur-6996	43	12	:	:	PUNCT
alusur-6996	43	13	more	more	ADV
alusur-6996	43	14	extensive	extensive	ADJ
alusur-6996	43	15	models	model	NOUN
alusur-6996	43	16	,	,	PUNCT
alusur-6996	43	17	containing	contain	VERB
alusur-6996	43	18	200	200	NUM
alusur-6996	43	19	nodes	node	NOUN
alusur-6996	43	20	in	in	ADP
alusur-6996	43	21	the	the	DET
alusur-6996	43	22	hidden	hidden	ADJ
alusur-6996	43	23	middle	middle	ADJ
alusur-6996	43	24	layer	layer	NOUN
alusur-6996	43	25	,	,	PUNCT
alusur-6996	43	26	showed	show	VERB
alusur-6996	43	27	slightly	slightly	ADV
alusur-6996	43	28	higher	high	ADJ
alusur-6996	43	29	accuracy	accuracy	NOUN
alusur-6996	43	30	in	in	ADP
alusur-6996	43	31	most	most	ADJ
alusur-6996	43	32	cases	case	NOUN
alusur-6996	43	33	(	(	PUNCT
alusur-6996	43	34	works	work	VERB
alusur-6996	43	35	#	#	SYM
alusur-6996	43	36	3	3	NUM
alusur-6996	43	37	-	-	SYM
alusur-6996	43	38	4	4	NUM
alusur-6996	43	39	were	be	AUX
alusur-6996	43	40	an	an	DET
alusur-6996	43	41	exception	exception	NOUN
alusur-6996	43	42	to	to	ADP
alusur-6996	43	43	this	this	DET
alusur-6996	43	44	pattern	pattern	NOUN
alusur-6996	43	45	)	)	PUNCT
alusur-6996	43	46	,	,	PUNCT
alusur-6996	43	47	but	but	CCONJ
alusur-6996	43	48	it	it	PRON
alusur-6996	43	49	took	take	VERB
alusur-6996	43	50	twice	twice	ADV
alusur-6996	43	51	as	as	ADV
alusur-6996	43	52	long	long	ADJ
alusur-6996	43	53	to	to	PART
alusur-6996	43	54	train	train	VERB
alusur-6996	43	55	the	the	DET
alusur-6996	43	56	model	model	NOUN
alusur-6996	43	57	and	and	CCONJ
alusur-6996	43	58	the	the	DET
alusur-6996	43	59	ocr	ocr	ADJ
alusur-6996	43	60	process	process	NOUN
alusur-6996	43	61	using	use	VERB
alusur-6996	43	62	the	the	DET
alusur-6996	43	63	larger	large	ADJ
alusur-6996	43	64	model	model	NOUN
alusur-6996	43	65	also	also	ADV
alusur-6996	43	66	takes	take	VERB
alusur-6996	43	67	more	more	ADJ
alusur-6996	43	68	time	time	NOUN
alusur-6996	43	69	.	.	PUNCT
alusur-6996	44	1	10	10	NUM
alusur-6996	44	2	.	.	PUNCT
alusur-6996	45	1	“	"	PUNCT
alusur-6996	45	2	low	low	ADJ
alusur-6996	45	3	-	-	PUNCT
alusur-6996	45	4	quality	quality	NOUN
alusur-6996	45	5	”	"	PUNCT
alusur-6996	45	6	here	here	ADV
alusur-6996	45	7	means	mean	VERB
alusur-6996	45	8	200	200	NUM
alusur-6996	45	9	dpi	dpi	NOUN
alusur-6996	45	10	,	,	PUNCT
alusur-6996	45	11	black	black	ADJ
alusur-6996	45	12	and	and	CCONJ
alusur-6996	45	13	white	white	ADJ
alusur-6996	45	14	,	,	PUNCT
alusur-6996	45	15	pre	pre	ADJ
alusur-6996	45	16	-	-	ADJ
alusur-6996	45	17	binarized	binarized	ADJ
alusur-6996	45	18	images	image	NOUN
alusur-6996	45	19	.	.	PUNCT
alusur-6996	46	1	in	in	ADP
alusur-6996	46	2	short	short	ADJ
alusur-6996	46	3	,	,	PUNCT
alusur-6996	46	4	the	the	DET
alusur-6996	46	5	standard	standard	ADJ
alusur-6996	46	6	quality	quality	NOUN
alusur-6996	46	7	of	of	ADP
alusur-6996	46	8	most	most	ADJ
alusur-6996	46	9	scans	scan	NOUN
alusur-6996	46	10	available	available	ADJ
alusur-6996	46	11	on	on	ADP
alusur-6996	46	12	the	the	DET
alusur-6996	46	13	internet	internet	NOUN
alusur-6996	46	14	,	,	PUNCT
alusur-6996	46	15	which	which	PRON
alusur-6996	46	16	are	be	AUX
alusur-6996	46	17	the	the	DET
alusur-6996	46	18	product	product	NOUN
alusur-6996	46	19	of	of	ADP
alusur-6996	46	20	scanners	scanner	NOUN
alusur-6996	46	21	that	that	PRON
alusur-6996	46	22	prioritize	prioritize	VERB
alusur-6996	46	23	smaller	small	ADJ
alusur-6996	46	24	size	size	NOUN
alusur-6996	46	25	and	and	CCONJ
alusur-6996	46	26	speed	speed	NOUN
alusur-6996	46	27	of	of	ADP
alusur-6996	46	28	scanning	scanning	NOUN
alusur-6996	46	29	for	for	ADP
alusur-6996	46	30	online	online	ADJ
alusur-6996	46	31	sharing	sharing	NOUN
alusur-6996	46	32	(	(	PUNCT
alusur-6996	46	33	i.e.	i.e.	X
alusur-6996	46	34	,	,	PUNCT
alusur-6996	46	35	in	in	ADP
alusur-6996	46	36	contrast	contrast	NOUN
alusur-6996	46	37	to	to	ADP
alusur-6996	46	38	high	high	ADJ
alusur-6996	46	39	-	-	PUNCT
alusur-6996	46	40	quality	quality	NOUN
alusur-6996	46	41	scans	scan	NOUN
alusur-6996	46	42	that	that	PRON
alusur-6996	46	43	are	be	AUX
alusur-6996	46	44	produced	produce	VERB
alusur-6996	46	45	for	for	ADP
alusur-6996	46	46	long	long	ADJ
alusur-6996	46	47	-	-	PUNCT
alusur-6996	46	48	term	term	NOUN
alusur-6996	46	49	preservation	preservation	NOUN
alusur-6996	46	50	)	)	PUNCT
alusur-6996	46	51	.	.	PUNCT
alusur-6996	47	1	table	table	NOUN
alusur-6996	47	2	2	2	NUM
alusur-6996	47	3	:	:	PUNCT
alusur-6996	47	4	accuracy	accuracy	NOUN
alusur-6996	47	5	rates	rate	NOUN
alusur-6996	47	6	in	in	ADP
alusur-6996	47	7	tests	test	NOUN
alusur-6996	47	8	of	of	ADP
alusur-6996	47	9	our	our	PRON
alusur-6996	47	10	custom	custom	NOUN
alusur-6996	47	11	model	model	NOUN
alusur-6996	47	12	http://www.archive.org	http://www.archive.org	PROPN
alusur-6996	47	13	http://waqfeya.com/	http://waqfeya.com/	PRON
alusur-6996	47	14	al-ʿuṣūr	al-ʿuṣūr	ADJ
alusur-6996	47	15	al	al	PROPN
alusur-6996	47	16	-	-	PUNCT
alusur-6996	47	17	wusṭā	wusṭā	NOUN
alusur-6996	47	18	25	25	NUM
alusur-6996	47	19	(	(	PUNCT
alusur-6996	47	20	2017	2017	NUM
alusur-6996	47	21	)	)	PUNCT
alusur-6996	47	22	new	new	ADJ
alusur-6996	47	23	developments	development	NOUN
alusur-6996	47	24	in	in	ADP
alusur-6996	47	25	arabographic	arabographic	ADJ
alusur-6996	47	26	optical	optical	ADJ
alusur-6996	47	27	character	character	NOUN
alusur-6996	47	28	recognition	recognition	NOUN
alusur-6996	47	29	•	•	ADP
alusur-6996	47	30	5	5	NUM
alusur-6996	47	31	that	that	SCONJ
alusur-6996	47	32	the	the	DET
alusur-6996	47	33	accuracy	accuracy	NOUN
alusur-6996	47	34	rates	rate	NOUN
alusur-6996	47	35	for	for	ADP
alusur-6996	47	36	works	work	NOUN
alusur-6996	47	37	#	#	SYM
alusur-6996	47	38	1	1	NUM
alusur-6996	47	39	-	-	SYM
alusur-6996	47	40	4	4	NUM
alusur-6996	47	41	are	be	AUX
alusur-6996	47	42	not	not	PART
alusur-6996	47	43	as	as	ADV
alusur-6996	47	44	high	high	ADJ
alusur-6996	47	45	as	as	ADP
alusur-6996	47	46	the	the	DET
alusur-6996	47	47	accuracy	accuracy	NOUN
alusur-6996	47	48	rates	rate	NOUN
alusur-6996	47	49	for	for	ADP
alusur-6996	47	50	the	the	DET
alusur-6996	47	51	training	training	NOUN
alusur-6996	47	52	text	text	NOUN
alusur-6996	47	53	,	,	PUNCT
alusur-6996	47	54	work	work	VERB
alusur-6996	47	55	#	#	SYM
alusur-6996	47	56	0	0	NUM
alusur-6996	47	57	.	.	PUNCT
alusur-6996	48	1	the	the	DET
alusur-6996	48	2	point	point	NOUN
alusur-6996	48	3	that	that	PRON
alusur-6996	48	4	is	be	AUX
alusur-6996	48	5	surprising	surprising	ADJ
alusur-6996	48	6	is	be	AUX
alusur-6996	48	7	that	that	SCONJ
alusur-6996	48	8	the	the	DET
alusur-6996	48	9	use	use	NOUN
alusur-6996	48	10	of	of	ADP
alusur-6996	48	11	the	the	DET
alusur-6996	48	12	work	work	NOUN
alusur-6996	48	13	#	#	SYM
alusur-6996	48	14	0	0	NUM
alusur-6996	48	15	-	-	PUNCT
alusur-6996	48	16	based	base	VERB
alusur-6996	48	17	model	model	NOUN
alusur-6996	48	18	on	on	ADP
alusur-6996	48	19	the	the	DET
alusur-6996	48	20	low	low	ADJ
alusur-6996	48	21	quality	quality	NOUN
alusur-6996	48	22	scans	scan	NOUN
alusur-6996	48	23	of	of	ADP
alusur-6996	48	24	works	work	NOUN
alusur-6996	48	25	#	#	SYM
alusur-6996	48	26	5	5	NUM
alusur-6996	48	27	-	-	SYM
alusur-6996	48	28	6	6	NUM
alusur-6996	48	29	achieved	achieve	VERB
alusur-6996	48	30	a	a	DET
alusur-6996	48	31	substantially	substantially	ADV
alusur-6996	48	32	higher	high	ADJ
alusur-6996	48	33	accuracy	accuracy	NOUN
alusur-6996	48	34	rate	rate	NOUN
alusur-6996	48	35	(	(	PUNCT
alusur-6996	48	36	97.61	97.61	NUM
alusur-6996	48	37	%	%	NOUN
alusur-6996	48	38	and	and	CCONJ
alusur-6996	48	39	97.8	97.8	NUM
alusur-6996	48	40	%	%	NOUN
alusur-6996	48	41	respectively	respectively	ADV
alusur-6996	48	42	on	on	ADP
alusur-6996	48	43	their	their	PRON
alusur-6996	48	44	arabic	arabic	ADJ
alusur-6996	48	45	script	script	NOUN
alusur-6996	48	46	alone	alone	ADV
alusur-6996	48	47	)	)	PUNCT
alusur-6996	48	48	than	than	ADP
alusur-6996	48	49	on	on	ADP
alusur-6996	48	50	the	the	DET
alusur-6996	48	51	high	high	ADJ
alusur-6996	48	52	-	-	PUNCT
alusur-6996	48	53	quality	quality	NOUN
alusur-6996	48	54	scans	scan	NOUN
alusur-6996	48	55	of	of	ADP
alusur-6996	48	56	works	work	NOUN
alusur-6996	48	57	#	#	SYM
alusur-6996	48	58	1	1	NUM
alusur-6996	48	59	-	-	SYM
alusur-6996	48	60	4	4	NUM
alusur-6996	48	61	.	.	PUNCT
alusur-6996	48	62	while	while	SCONJ
alusur-6996	48	63	these	these	DET
alusur-6996	48	64	higher	high	ADJ
alusur-6996	48	65	accuracy	accuracy	NOUN
alusur-6996	48	66	rates	rate	NOUN
alusur-6996	48	67	for	for	ADP
alusur-6996	48	68	works	work	NOUN
alusur-6996	48	69	#	#	SYM
alusur-6996	48	70	5	5	NUM
alusur-6996	48	71	-	-	SYM
alusur-6996	48	72	6	6	NUM
alusur-6996	48	73	are	be	AUX
alusur-6996	48	74	the	the	DET
alusur-6996	48	75	result	result	NOUN
alusur-6996	48	76	of	of	ADP
alusur-6996	48	77	a	a	DET
alusur-6996	48	78	closer	close	ADJ
alusur-6996	48	79	affinity	affinity	NOUN
alusur-6996	48	80	between	between	ADP
alusur-6996	48	81	their	their	PRON
alusur-6996	48	82	typefaces	typeface	NOUN
alusur-6996	48	83	and	and	CCONJ
alusur-6996	48	84	that	that	PRON
alusur-6996	48	85	of	of	ADP
alusur-6996	48	86	work	work	NOUN
alusur-6996	48	87	#	#	SYM
alusur-6996	48	88	0	0	NUM
alusur-6996	48	89	,	,	PUNCT
alusur-6996	48	90	it	it	PRON
alusur-6996	48	91	also	also	ADV
alusur-6996	48	92	indicates	indicate	VERB
alusur-6996	48	93	that	that	SCONJ
alusur-6996	48	94	the	the	DET
alusur-6996	48	95	distinction	distinction	NOUN
alusur-6996	48	96	between	between	ADP
alusur-6996	48	97	highand	highand	ADJ
alusur-6996	48	98	low	low	ADJ
alusur-6996	48	99	-	-	PUNCT
alusur-6996	48	100	quality	quality	NOUN
alusur-6996	48	101	images	image	NOUN
alusur-6996	48	102	is	be	AUX
alusur-6996	48	103	not	not	PART
alusur-6996	48	104	as	as	ADV
alusur-6996	48	105	important	important	ADJ
alusur-6996	48	106	for	for	ADP
alusur-6996	48	107	achieving	achieve	VERB
alusur-6996	48	108	high	high	ADJ
alusur-6996	48	109	accuracy	accuracy	NOUN
alusur-6996	48	110	rates	rate	NOUN
alusur-6996	48	111	with	with	ADP
alusur-6996	48	112	kraken	kraken	PROPN
alusur-6996	48	113	as	as	SCONJ
alusur-6996	48	114	we	we	PRON
alusur-6996	48	115	initially	initially	ADV
alusur-6996	48	116	believed	believe	VERB
alusur-6996	48	117	.	.	PUNCT
alusur-6996	49	1	in	in	ADP
alusur-6996	49	2	the	the	DET
alusur-6996	49	3	future	future	NOUN
alusur-6996	49	4	,	,	PUNCT
alusur-6996	49	5	this	this	PRON
alusur-6996	49	6	will	will	AUX
alusur-6996	49	7	help	help	VERB
alusur-6996	49	8	reduce	reduce	VERB
alusur-6996	49	9	substantially	substantially	ADV
alusur-6996	49	10	both	both	CCONJ
alusur-6996	49	11	the	the	DET
alusur-6996	49	12	total	total	ADJ
alusur-6996	49	13	length	length	NOUN
alusur-6996	49	14	of	of	ADP
alusur-6996	49	15	time	time	NOUN
alusur-6996	49	16	it	it	PRON
alusur-6996	49	17	takes	take	VERB
alusur-6996	49	18	to	to	ADP
alusur-6996	49	19	ocr	ocr	VERB
alusur-6996	49	20	a	a	DET
alusur-6996	49	21	work	work	NOUN
alusur-6996	49	22	and	and	CCONJ
alusur-6996	49	23	the	the	DET
alusur-6996	49	24	barriers	barrier	NOUN
alusur-6996	49	25	to	to	ADP
alusur-6996	49	26	entry	entry	NOUN
alusur-6996	49	27	for	for	ADP
alusur-6996	49	28	researchers	researcher	NOUN
alusur-6996	49	29	wanting	want	VERB
alusur-6996	49	30	to	to	PART
alusur-6996	49	31	ocr	ocr	VERB
alusur-6996	49	32	the	the	DET
alusur-6996	49	33	low	low	ADJ
alusur-6996	49	34	-	-	PUNCT
alusur-6996	49	35	quality	quality	NOUN
alusur-6996	49	36	scans	scan	NOUN
alusur-6996	49	37	they	they	PRON
alusur-6996	49	38	already	already	ADV
alusur-6996	49	39	possess	possess	VERB
alusur-6996	49	40	.	.	PUNCT
alusur-6996	50	1	the	the	DET
alusur-6996	50	2	decreased	decrease	VERB
alusur-6996	50	3	accuracy	accuracy	NOUN
alusur-6996	50	4	results	result	NOUN
alusur-6996	50	5	for	for	ADP
alusur-6996	50	6	works	work	NOUN
alusur-6996	50	7	#	#	SYM
alusur-6996	50	8	1	1	NUM
alusur-6996	50	9	-	-	SYM
alusur-6996	50	10	4	4	NUM
alusur-6996	50	11	are	be	AUX
alusur-6996	50	12	explainable	explainable	ADJ
alusur-6996	50	13	by	by	ADP
alusur-6996	50	14	a	a	DET
alusur-6996	50	15	few	few	ADJ
alusur-6996	50	16	factors	factor	NOUN
alusur-6996	50	17	:	:	PUNCT
alusur-6996	50	18	1	1	X
alusur-6996	50	19	)	)	PUNCT
alusur-6996	50	20	the	the	DET
alusur-6996	50	21	typeface	typeface	NOUN
alusur-6996	50	22	of	of	ADP
alusur-6996	50	23	works	work	NOUN
alusur-6996	50	24	#	#	SYM
alusur-6996	50	25	3	3	NUM
alusur-6996	50	26	-	-	SYM
alusur-6996	50	27	4	4	NUM
alusur-6996	50	28	is	be	AUX
alusur-6996	50	29	different	different	ADJ
alusur-6996	50	30	than	than	ADP
alusur-6996	50	31	work	work	VERB
alusur-6996	50	32	#	#	SYM
alusur-6996	50	33	0	0	NUM
alusur-6996	51	1	and	and	CCONJ
alusur-6996	51	2	it	it	PRON
alusur-6996	51	3	utilizes	utilize	VERB
alusur-6996	51	4	a	a	DET
alusur-6996	51	5	number	number	NOUN
alusur-6996	51	6	of	of	ADP
alusur-6996	51	7	ligatures	ligature	NOUN
alusur-6996	51	8	that	that	PRON
alusur-6996	51	9	are	be	AUX
alusur-6996	51	10	not	not	PART
alusur-6996	51	11	present	present	ADJ
alusur-6996	51	12	in	in	ADP
alusur-6996	51	13	the	the	DET
alusur-6996	51	14	typeface	typeface	NOUN
alusur-6996	51	15	of	of	ADP
alusur-6996	51	16	work	work	NOUN
alusur-6996	51	17	#	#	SYM
alusur-6996	51	18	0	0	NUM
alusur-6996	51	19	(	(	PUNCT
alusur-6996	51	20	for	for	ADP
alusur-6996	51	21	examples	example	NOUN
alusur-6996	51	22	,	,	PUNCT
alusur-6996	51	23	see	see	VERB
alusur-6996	51	24	table	table	NOUN
alusur-6996	51	25	3	3	NUM
alusur-6996	51	26	)	)	PUNCT
alusur-6996	51	27	.	.	PUNCT
alusur-6996	52	1	2	2	X
alusur-6996	52	2	)	)	PUNCT
alusur-6996	52	3	the	the	DET
alusur-6996	52	4	typefaces	typeface	NOUN
alusur-6996	52	5	of	of	ADP
alusur-6996	52	6	works	work	NOUN
alusur-6996	52	7	#	#	SYM
alusur-6996	52	8	1	1	NUM
alusur-6996	52	9	-	-	SYM
alusur-6996	52	10	2	2	NUM
alusur-6996	52	11	are	be	AUX
alusur-6996	52	12	very	very	ADV
alusur-6996	52	13	similar	similar	ADJ
alusur-6996	52	14	to	to	ADP
alusur-6996	52	15	that	that	PRON
alusur-6996	52	16	of	of	ADP
alusur-6996	52	17	#	#	SYM
alusur-6996	52	18	0	0	NUM
alusur-6996	52	19	,	,	PUNCT
alusur-6996	52	20	but	but	CCONJ
alusur-6996	52	21	they	they	PRON
alusur-6996	52	22	both	both	PRON
alusur-6996	52	23	have	have	VERB
alusur-6996	52	24	features	feature	NOUN
alusur-6996	52	25	that	that	PRON
alusur-6996	52	26	interfere	interfere	VERB
alusur-6996	52	27	with	with	ADP
alusur-6996	52	28	the	the	DET
alusur-6996	52	29	#	#	SYM
alusur-6996	52	30	0	0	NUM
alusur-6996	52	31	-	-	PUNCT
alusur-6996	52	32	based	base	VERB
alusur-6996	52	33	model	model	NOUN
alusur-6996	52	34	.	.	PUNCT
alusur-6996	53	1	#	#	SYM
alusur-6996	53	2	1	1	NUM
alusur-6996	53	3	actually	actually	ADV
alusur-6996	53	4	uses	use	VERB
alusur-6996	53	5	two	two	NUM
alusur-6996	53	6	different	different	ADJ
alusur-6996	53	7	fonts	font	NOUN
alusur-6996	53	8	,	,	PUNCT
alusur-6996	53	9	and	and	CCONJ
alusur-6996	53	10	the	the	DET
alusur-6996	53	11	length	length	NOUN
alusur-6996	53	12	of	of	ADP
alusur-6996	53	13	connections	connection	NOUN
alusur-6996	53	14	—	—	PUNCT
alusur-6996	53	15	kashīdas	kashīdas	NUM
alusur-6996	53	16	—	—	PUNCT
alusur-6996	53	17	between	between	ADP
alusur-6996	53	18	letters	letter	NOUN
alusur-6996	53	19	vary	vary	VERB
alusur-6996	53	20	dramatically	dramatically	ADV
alusur-6996	53	21	(	(	PUNCT
alusur-6996	53	22	visually	visually	ADV
alusur-6996	53	23	,	,	PUNCT
alusur-6996	53	24	one	one	PRON
alusur-6996	53	25	can	can	AUX
alusur-6996	53	26	say	say	VERB
alusur-6996	53	27	that	that	SCONJ
alusur-6996	53	28	these	these	DET
alusur-6996	53	29	connections	connection	NOUN
alusur-6996	53	30	vary	vary	VERB
alusur-6996	53	31	within	within	ADP
alusur-6996	53	32	the	the	DET
alusur-6996	53	33	range	range	NOUN
alusur-6996	53	34	of	of	ADP
alusur-6996	53	35	0.3	0.3	NUM
alusur-6996	53	36	kashīda	kashīda	PROPN
alusur-6996	53	37	to	to	ADP
alusur-6996	53	38	2	2	NUM
alusur-6996	53	39	kashīdas	kashīda	NOUN
alusur-6996	53	40	)	)	PUNCT
alusur-6996	53	41	,	,	PUNCT
alusur-6996	53	42	which	which	PRON
alusur-6996	53	43	is	be	AUX
alusur-6996	53	44	not	not	PART
alusur-6996	53	45	the	the	DET
alusur-6996	53	46	case	case	NOUN
alusur-6996	53	47	with	with	ADP
alusur-6996	53	48	#	#	SYM
alusur-6996	53	49	0	0	NUM
alusur-6996	53	50	,	,	PUNCT
alusur-6996	53	51	where	where	SCONJ
alusur-6996	53	52	letter	letter	NOUN
alusur-6996	53	53	spacing	space	VERB
alusur-6996	53	54	is	be	AUX
alusur-6996	53	55	very	very	ADV
alusur-6996	53	56	consistent	consistent	ADJ
alusur-6996	53	57	.	.	PUNCT
alusur-6996	54	1	table	table	NOUN
alusur-6996	54	2	3	3	NUM
alusur-6996	54	3	:	:	PUNCT
alusur-6996	54	4	ligature	ligature	VERB
alusur-6996	54	5	variations	variation	NOUN
alusur-6996	54	6	in	in	ADP
alusur-6996	54	7	typefaces	typeface	NOUN
alusur-6996	54	8	(	(	PUNCT
alusur-6996	54	9	the	the	DET
alusur-6996	54	10	table	table	NOUN
alusur-6996	54	11	highlights	highlight	VERB
alusur-6996	54	12	only	only	ADV
alusur-6996	54	13	a	a	DET
alusur-6996	54	14	few	few	ADJ
alusur-6996	54	15	striking	striking	ADJ
alusur-6996	54	16	differences	difference	NOUN
alusur-6996	54	17	and	and	CCONJ
alusur-6996	54	18	is	be	AUX
alusur-6996	54	19	not	not	PART
alusur-6996	54	20	meant	mean	VERB
alusur-6996	54	21	to	to	PART
alusur-6996	54	22	be	be	AUX
alusur-6996	54	23	comprehensive	comprehensive	ADJ
alusur-6996	54	24	;	;	PUNCT
alusur-6996	54	25	examples	example	NOUN
alusur-6996	54	26	similar	similar	ADJ
alusur-6996	54	27	to	to	ADP
alusur-6996	54	28	those	those	PRON
alusur-6996	54	29	of	of	ADP
alusur-6996	54	30	the	the	DET
alusur-6996	54	31	main	main	ADJ
alusur-6996	54	32	text	text	NOUN
alusur-6996	54	33	are	be	AUX
alusur-6996	54	34	“	"	PUNCT
alusur-6996	54	35	greyed	grey	VERB
alusur-6996	54	36	out	out	ADP
alusur-6996	54	37	”	"	PUNCT
alusur-6996	54	38	)	)	PUNCT
alusur-6996	54	39	6	6	NUM
alusur-6996	54	40	•	•	NOUN
alusur-6996	54	41	kiessling	kiessle	VERB
alusur-6996	54	42	,	,	PUNCT
alusur-6996	54	43	miller	miller	PROPN
alusur-6996	54	44	,	,	PUNCT
alusur-6996	54	45	romanov	romanov	PROPN
alusur-6996	54	46	,	,	PUNCT
alusur-6996	54	47	&	&	CCONJ
alusur-6996	54	48	savant	savant	PROPN
alusur-6996	54	49	al-ʿuṣūr	al-ʿuṣūr	PROPN
alusur-6996	54	50	al	al	PROPN
alusur-6996	54	51	-	-	PUNCT
alusur-6996	54	52	wusṭā	wusṭā	NOUN
alusur-6996	54	53	25	25	NUM
alusur-6996	54	54	(	(	PUNCT
alusur-6996	54	55	2017	2017	NUM
alusur-6996	54	56	)	)	PUNCT
alusur-6996	54	57	3	3	NUM
alusur-6996	54	58	)	)	PUNCT
alusur-6996	54	59	the	the	DET
alusur-6996	54	60	text	text	NOUN
alusur-6996	54	61	of	of	ADP
alusur-6996	54	62	work	work	NOUN
alusur-6996	54	63	#	#	SYM
alusur-6996	54	64	2	2	NUM
alusur-6996	54	65	is	be	AUX
alusur-6996	54	66	highly	highly	ADV
alusur-6996	54	67	vocalized	vocalize	VERB
alusur-6996	54	68	—	—	PUNCT
alusur-6996	54	69	it	it	PRON
alusur-6996	54	70	has	have	VERB
alusur-6996	54	71	more	more	ADJ
alusur-6996	54	72	ḥarakāt	ḥarakāt	NOUN
alusur-6996	54	73	than	than	ADP
alusur-6996	54	74	any	any	DET
alusur-6996	54	75	other	other	ADJ
alusur-6996	54	76	text	text	NOUN
alusur-6996	54	77	in	in	ADP
alusur-6996	54	78	the	the	DET
alusur-6996	54	79	sample	sample	NOUN
alusur-6996	54	80	(	(	PUNCT
alusur-6996	54	81	and	and	CCONJ
alusur-6996	54	82	especially	especially	ADV
alusur-6996	54	83	in	in	ADP
alusur-6996	54	84	comparison	comparison	NOUN
alusur-6996	54	85	with	with	ADP
alusur-6996	54	86	the	the	DET
alusur-6996	54	87	model	model	NOUN
alusur-6996	54	88	work	work	NOUN
alusur-6996	54	89	#	#	SYM
alusur-6996	54	90	0	0	NUM
alusur-6996	54	91	)	)	PUNCT
alusur-6996	54	92	.	.	PUNCT
alusur-6996	55	1	4	4	X
alusur-6996	55	2	)	)	PUNCT
alusur-6996	55	3	the	the	DET
alusur-6996	55	4	text	text	NOUN
alusur-6996	55	5	of	of	ADP
alusur-6996	55	6	work	work	NOUN
alusur-6996	55	7	#	#	SYM
alusur-6996	55	8	2	2	NUM
alusur-6996	55	9	also	also	ADV
alusur-6996	55	10	has	have	VERB
alusur-6996	55	11	very	very	ADV
alusur-6996	55	12	complex	complex	ADJ
alusur-6996	55	13	and	and	CCONJ
alusur-6996	55	14	excessive	excessive	ADJ
alusur-6996	55	15	punctuation	punctuation	NOUN
alusur-6996	55	16	with	with	ADP
alusur-6996	55	17	highly	highly	ADV
alusur-6996	55	18	inconsistent	inconsistent	ADJ
alusur-6996	55	19	spacing	spacing	NOUN
alusur-6996	55	20	.	.	PUNCT
alusur-6996	56	1	our	our	PRON
alusur-6996	56	2	#	#	SYM
alusur-6996	56	3	0	0	NUM
alusur-6996	56	4	-	-	PUNCT
alusur-6996	56	5	based	base	VERB
alusur-6996	56	6	model	model	NOUN
alusur-6996	56	7	could	could	AUX
alusur-6996	56	8	not	not	PART
alusur-6996	56	9	completely	completely	ADV
alusur-6996	56	10	handle	handle	VERB
alusur-6996	56	11	these	these	DET
alusur-6996	56	12	novel	novel	ADJ
alusur-6996	56	13	features	feature	NOUN
alusur-6996	56	14	in	in	ADP
alusur-6996	56	15	the	the	DET
alusur-6996	56	16	texts	text	NOUN
alusur-6996	56	17	of	of	ADP
alusur-6996	56	18	works	work	NOUN
alusur-6996	56	19	#	#	SYM
alusur-6996	56	20	1	1	NUM
alusur-6996	56	21	-	-	SYM
alusur-6996	56	22	4	4	NUM
alusur-6996	56	23	because	because	SCONJ
alusur-6996	56	24	it	it	PRON
alusur-6996	56	25	was	be	AUX
alusur-6996	56	26	not	not	PART
alusur-6996	56	27	trained	train	VERB
alusur-6996	56	28	to	to	PART
alusur-6996	56	29	do	do	VERB
alusur-6996	56	30	so	so	ADV
alusur-6996	56	31	.	.	PUNCT
alusur-6996	57	1	however	however	ADV
alusur-6996	57	2	,	,	PUNCT
alusur-6996	57	3	as	as	SCONJ
alusur-6996	57	4	the	the	DET
alusur-6996	57	5	results	result	NOUN
alusur-6996	57	6	in	in	ADP
alusur-6996	57	7	table	table	NOUN
alusur-6996	57	8	4	4	NUM
alusur-6996	57	9	of	of	ADP
alusur-6996	57	10	the	the	DET
alusur-6996	57	11	following	follow	VERB
alusur-6996	57	12	section	section	NOUN
alusur-6996	57	13	show	show	NOUN
alusur-6996	57	14	,	,	PUNCT
alusur-6996	57	15	new	new	ADJ
alusur-6996	57	16	models	model	NOUN
alusur-6996	57	17	can	can	AUX
alusur-6996	57	18	be	be	AUX
alusur-6996	57	19	trained	train	VERB
alusur-6996	57	20	to	to	PART
alusur-6996	57	21	handle	handle	VERB
alusur-6996	57	22	these	these	DET
alusur-6996	57	23	issues	issue	NOUN
alusur-6996	57	24	successfully	successfully	ADV
alusur-6996	57	25	.	.	PUNCT
alusur-6996	58	1	2.2	2.2	NUM
alusur-6996	58	2	round	round	NOUN
alusur-6996	58	3	#	#	SYM
alusur-6996	58	4	2	2	NUM
alusur-6996	58	5	tests	test	NOUN
alusur-6996	58	6	:	:	PUNCT
alusur-6996	59	1	training	train	VERB
alusur-6996	59	2	new	new	ADJ
alusur-6996	59	3	models	model	NOUN
alusur-6996	59	4	the	the	DET
alusur-6996	59	5	most	most	ADV
alusur-6996	59	6	important	important	ADJ
alusur-6996	59	7	advantage	advantage	NOUN
alusur-6996	59	8	of	of	ADP
alusur-6996	59	9	kraken	kraken	NOUN
alusur-6996	59	10	is	be	AUX
alusur-6996	59	11	that	that	SCONJ
alusur-6996	59	12	its	its	PRON
alusur-6996	59	13	workflow	workflow	NOUN
alusur-6996	59	14	allows	allow	VERB
alusur-6996	59	15	one	one	NUM
alusur-6996	59	16	to	to	PART
alusur-6996	59	17	train	train	VERB
alusur-6996	59	18	new	new	ADJ
alusur-6996	59	19	models	model	NOUN
alusur-6996	59	20	relatively	relatively	ADV
alusur-6996	59	21	easily	easily	ADV
alusur-6996	59	22	,	,	PUNCT
alusur-6996	59	23	including	include	VERB
alusur-6996	59	24	text	text	NOUN
alusur-6996	59	25	-	-	PUNCT
alusur-6996	59	26	specific	specific	ADJ
alusur-6996	59	27	ones	one	NOUN
alusur-6996	59	28	.	.	PUNCT
alusur-6996	60	1	in	in	ADP
alusur-6996	60	2	a	a	DET
alusur-6996	60	3	nutshell	nutshell	NOUN
alusur-6996	60	4	,	,	PUNCT
alusur-6996	60	5	the	the	DET
alusur-6996	60	6	training	training	NOUN
alusur-6996	60	7	process	process	NOUN
alusur-6996	60	8	requires	require	VERB
alusur-6996	60	9	a	a	DET
alusur-6996	60	10	transcription	transcription	NOUN
alusur-6996	60	11	of	of	ADP
alusur-6996	60	12	approximately	approximately	ADV
alusur-6996	60	13	1,000	1,000	NUM
alusur-6996	60	14	lines	line	NOUN
alusur-6996	60	15	(	(	PUNCT
alusur-6996	60	16	the	the	DET
alusur-6996	60	17	number	number	NOUN
alusur-6996	60	18	will	will	AUX
alusur-6996	60	19	vary	vary	VERB
alusur-6996	60	20	depending	depend	VERB
alusur-6996	60	21	on	on	ADP
alusur-6996	60	22	the	the	DET
alusur-6996	60	23	complexity	complexity	NOUN
alusur-6996	60	24	of	of	ADP
alusur-6996	60	25	the	the	DET
alusur-6996	60	26	typeface	typeface	NOUN
alusur-6996	60	27	)	)	PUNCT
alusur-6996	60	28	aligned	align	VERB
alusur-6996	60	29	with	with	ADP
alusur-6996	60	30	images	image	NOUN
alusur-6996	60	31	of	of	ADP
alusur-6996	60	32	these	these	DET
alusur-6996	60	33	lines	line	NOUN
alusur-6996	60	34	as	as	SCONJ
alusur-6996	60	35	they	they	PRON
alusur-6996	60	36	appear	appear	VERB
alusur-6996	60	37	in	in	ADP
alusur-6996	60	38	the	the	DET
alusur-6996	60	39	printed	print	VERB
alusur-6996	60	40	edition	edition	NOUN
alusur-6996	60	41	.	.	PUNCT
alusur-6996	61	1	the	the	DET
alusur-6996	61	2	training	training	NOUN
alusur-6996	61	3	itself	itself	PRON
alusur-6996	61	4	takes	take	VERB
alusur-6996	61	5	12	12	NUM
alusur-6996	61	6	-	-	SYM
alusur-6996	61	7	24	24	NUM
alusur-6996	61	8	hours	hour	NOUN
alusur-6996	61	9	.	.	PUNCT
alusur-6996	62	1	it	it	PRON
alusur-6996	62	2	is	be	AUX
alusur-6996	62	3	performed	perform	VERB
alusur-6996	62	4	by	by	ADP
alusur-6996	62	5	a	a	DET
alusur-6996	62	6	machine	machine	NOUN
alusur-6996	62	7	without	without	ADP
alusur-6996	62	8	human	human	ADJ
alusur-6996	62	9	involvement	involvement	NOUN
alusur-6996	62	10	and	and	CCONJ
alusur-6996	62	11	multiple	multiple	ADJ
alusur-6996	62	12	models	model	NOUN
alusur-6996	62	13	can	can	AUX
alusur-6996	62	14	be	be	AUX
alusur-6996	62	15	trained	train	VERB
alusur-6996	62	16	simultaneously	simultaneously	ADV
alusur-6996	62	17	.	.	PUNCT
alusur-6996	63	1	kraken	kraken	NOUN
alusur-6996	63	2	includes	include	VERB
alusur-6996	63	3	tools	tool	NOUN
alusur-6996	63	4	for	for	ADP
alusur-6996	63	5	the	the	DET
alusur-6996	63	6	production	production	NOUN
alusur-6996	63	7	of	of	ADP
alusur-6996	63	8	transcription	transcription	NOUN
alusur-6996	63	9	forms	form	NOUN
alusur-6996	63	10	(	(	PUNCT
alusur-6996	63	11	see	see	VERB
alusur-6996	63	12	figure	figure	NOUN
alusur-6996	63	13	2	2	NUM
alusur-6996	63	14	above	above	ADV
alusur-6996	63	15	)	)	PUNCT
alusur-6996	63	16	and	and	CCONJ
alusur-6996	63	17	the	the	DET
alusur-6996	63	18	data	datum	NOUN
alusur-6996	63	19	supplied	supply	VERB
alusur-6996	63	20	through	through	ADP
alusur-6996	63	21	these	these	DET
alusur-6996	63	22	forms	form	NOUN
alusur-6996	63	23	is	be	AUX
alusur-6996	63	24	then	then	ADV
alusur-6996	63	25	used	use	VERB
alusur-6996	63	26	to	to	PART
alusur-6996	63	27	train	train	VERB
alusur-6996	63	28	a	a	DET
alusur-6996	63	29	new	new	ADJ
alusur-6996	63	30	model	model	NOUN
alusur-6996	63	31	.	.	PUNCT
alusur-6996	64	1	since	since	SCONJ
alusur-6996	64	2	there	there	PRON
alusur-6996	64	3	are	be	VERB
alusur-6996	64	4	a	a	DET
alusur-6996	64	5	great	great	ADJ
alusur-6996	64	6	number	number	NOUN
alusur-6996	64	7	of	of	ADP
alusur-6996	64	8	arabic	arabic	ADJ
alusur-6996	64	9	-	-	PUNCT
alusur-6996	64	10	script	script	NOUN
alusur-6996	64	11	texts	text	NOUN
alusur-6996	64	12	that	that	PRON
alusur-6996	64	13	have	have	AUX
alusur-6996	64	14	already	already	ADV
alusur-6996	64	15	been	be	AUX
alusur-6996	64	16	converted	convert	VERB
alusur-6996	64	17	into	into	ADP
alusur-6996	64	18	digital	digital	ADJ
alusur-6996	64	19	texts	text	NOUN
alusur-6996	64	20	,	,	PUNCT
alusur-6996	64	21	one	one	PRON
alusur-6996	64	22	can	can	AUX
alusur-6996	64	23	use	use	VERB
alusur-6996	64	24	these	these	PRON
alusur-6996	64	25	to	to	PART
alusur-6996	64	26	fill	fill	VERB
alusur-6996	64	27	in	in	ADP
alusur-6996	64	28	the	the	DET
alusur-6996	64	29	forms	form	NOUN
alusur-6996	64	30	quickly	quickly	ADV
alusur-6996	64	31	by	by	ADP
alusur-6996	64	32	copying	copy	VERB
alusur-6996	64	33	and	and	CCONJ
alusur-6996	64	34	pasting	pasting	NOUN
alusur-6996	64	35	from	from	ADP
alusur-6996	64	36	them	they	PRON
alusur-6996	64	37	into	into	ADP
alusur-6996	64	38	the	the	DET
alusur-6996	64	39	forms	form	NOUN
alusur-6996	64	40	(	(	PUNCT
alusur-6996	64	41	rather	rather	ADV
alusur-6996	64	42	than	than	ADP
alusur-6996	64	43	transcribing	transcribe	VERB
alusur-6996	64	44	directly	directly	ADV
alusur-6996	64	45	from	from	ADP
alusur-6996	64	46	the	the	DET
alusur-6996	64	47	printed	print	VERB
alusur-6996	64	48	texts	text	NOUN
alusur-6996	64	49	)	)	PUNCT
alusur-6996	64	50	and	and	CCONJ
alusur-6996	64	51	then	then	ADV
alusur-6996	64	52	double	double	ADV
alusur-6996	64	53	-	-	PUNCT
alusur-6996	64	54	checking	check	VERB
alusur-6996	64	55	the	the	DET
alusur-6996	64	56	forms	form	NOUN
alusur-6996	64	57	for	for	ADP
alusur-6996	64	58	accuracy	accuracy	NOUN
alusur-6996	64	59	.	.	PUNCT
alusur-6996	65	1	this	this	PRON
alusur-6996	65	2	was	be	AUX
alusur-6996	65	3	what	what	PRON
alusur-6996	65	4	we	we	PRON
alusur-6996	65	5	did	do	VERB
alusur-6996	65	6	,	,	PUNCT
alusur-6996	65	7	and	and	CCONJ
alusur-6996	65	8	it	it	PRON
alusur-6996	65	9	saved	save	VERB
alusur-6996	65	10	us	we	PRON
alusur-6996	65	11	a	a	DET
alusur-6996	65	12	lot	lot	NOUN
alusur-6996	65	13	of	of	ADP
alusur-6996	65	14	time.11	time.11	NOUN
alusur-6996	65	15	11	11	NUM
alusur-6996	65	16	.	.	PUNCT
alusur-6996	66	1	we	we	PRON
alusur-6996	66	2	are	be	AUX
alusur-6996	66	3	also	also	ADV
alusur-6996	66	4	currently	currently	ADV
alusur-6996	66	5	working	work	VERB
alusur-6996	66	6	on	on	ADP
alusur-6996	66	7	an	an	DET
alusur-6996	66	8	even	even	ADV
alusur-6996	66	9	more	more	ADJ
alusur-6996	66	10	user	user	NOUN
alusur-6996	66	11	-	-	PUNCT
alusur-6996	66	12	friendly	friendly	ADJ
alusur-6996	66	13	interface	interface	NOUN
alusur-6996	66	14	for	for	ADP
alusur-6996	66	15	training	training	NOUN
alusur-6996	66	16	data	data	NOUN
alusur-6996	66	17	generation	generation	NOUN
alusur-6996	66	18	.	.	PUNCT
alusur-6996	67	1	please	please	INTJ
alusur-6996	67	2	see	see	VERB
alusur-6996	67	3	section	section	NOUN
alusur-6996	67	4	3.1	3.1	NUM
alusur-6996	67	5	for	for	ADP
alusur-6996	67	6	more	more	ADJ
alusur-6996	67	7	information	information	NOUN
alusur-6996	67	8	.	.	PUNCT
alusur-6996	68	1	table	table	NOUN
alusur-6996	68	2	4	4	NUM
alusur-6996	68	3	:	:	PUNCT
alusur-6996	68	4	accuracy	accuracy	NOUN
alusur-6996	68	5	rates	rate	NOUN
alusur-6996	68	6	in	in	ADP
alusur-6996	68	7	text	text	NOUN
alusur-6996	68	8	-	-	PUNCT
alusur-6996	68	9	specific	specific	ADJ
alusur-6996	68	10	models	model	NOUN
alusur-6996	68	11	al-ʿuṣūr	al-ʿuṣūr	PROPN
alusur-6996	68	12	al	al	PROPN
alusur-6996	68	13	-	-	PUNCT
alusur-6996	68	14	wusṭā	wusṭā	NOUN
alusur-6996	68	15	25	25	NUM
alusur-6996	68	16	(	(	PUNCT
alusur-6996	68	17	2017	2017	NUM
alusur-6996	68	18	)	)	PUNCT
alusur-6996	68	19	new	new	ADJ
alusur-6996	68	20	developments	development	NOUN
alusur-6996	68	21	in	in	ADP
alusur-6996	68	22	arabographic	arabographic	ADJ
alusur-6996	68	23	optical	optical	ADJ
alusur-6996	68	24	character	character	NOUN
alusur-6996	68	25	recognition	recognition	NOUN
alusur-6996	68	26	•	•	ADP
alusur-6996	68	27	7	7	NUM
alusur-6996	68	28	the	the	DET
alusur-6996	68	29	importance	importance	NOUN
alusur-6996	68	30	of	of	ADP
alusur-6996	68	31	kraken	kraken	NOUN
alusur-6996	68	32	’s	’s	PART
alusur-6996	68	33	ability	ability	NOUN
alusur-6996	68	34	to	to	PART
alusur-6996	68	35	quickly	quickly	ADV
alusur-6996	68	36	train	train	VERB
alusur-6996	68	37	new	new	ADJ
alusur-6996	68	38	models	model	NOUN
alusur-6996	68	39	is	be	AUX
alusur-6996	68	40	illustrated	illustrate	VERB
alusur-6996	68	41	clearly	clearly	ADV
alusur-6996	68	42	by	by	ADP
alusur-6996	68	43	its	its	PRON
alusur-6996	68	44	performance	performance	NOUN
alusur-6996	68	45	on	on	ADP
alusur-6996	68	46	works	work	NOUN
alusur-6996	68	47	#	#	SYM
alusur-6996	68	48	1	1	NUM
alusur-6996	68	49	-	-	SYM
alusur-6996	68	50	4	4	NUM
alusur-6996	68	51	.	.	PUNCT
alusur-6996	69	1	when	when	SCONJ
alusur-6996	69	2	using	use	VERB
alusur-6996	69	3	the	the	DET
alusur-6996	69	4	model	model	NOUN
alusur-6996	69	5	built	build	VERB
alusur-6996	69	6	on	on	ADP
alusur-6996	69	7	work	work	NOUN
alusur-6996	69	8	#	#	SYM
alusur-6996	69	9	0	0	NUM
alusur-6996	69	10	in	in	ADP
alusur-6996	69	11	our	our	PRON
alusur-6996	69	12	initial	initial	ADJ
alusur-6996	69	13	round	round	NOUN
alusur-6996	69	14	of	of	ADP
alusur-6996	69	15	testing	testing	NOUN
alusur-6996	69	16	,	,	PUNCT
alusur-6996	69	17	we	we	PRON
alusur-6996	69	18	were	be	AUX
alusur-6996	69	19	only	only	ADV
alusur-6996	69	20	able	able	ADJ
alusur-6996	69	21	to	to	PART
alusur-6996	69	22	achieve	achieve	VERB
alusur-6996	69	23	accuracy	accuracy	NOUN
alusur-6996	69	24	rates	rate	NOUN
alusur-6996	69	25	ranging	range	VERB
alusur-6996	69	26	from	from	ADP
alusur-6996	69	27	the	the	DET
alusur-6996	69	28	low	low	ADJ
alusur-6996	69	29	seventies	seventy	NOUN
alusur-6996	69	30	to	to	ADP
alusur-6996	69	31	low	low	ADJ
alusur-6996	69	32	nineties	ninety	NOUN
alusur-6996	69	33	on	on	ADP
alusur-6996	69	34	these	these	DET
alusur-6996	69	35	texts	text	NOUN
alusur-6996	69	36	(	(	PUNCT
alusur-6996	69	37	see	see	VERB
alusur-6996	69	38	table	table	NOUN
alusur-6996	69	39	2	2	NUM
alusur-6996	69	40	)	)	PUNCT
alusur-6996	69	41	.	.	PUNCT
alusur-6996	70	1	however	however	ADV
alusur-6996	70	2	,	,	PUNCT
alusur-6996	70	3	when	when	SCONJ
alusur-6996	70	4	we	we	PRON
alusur-6996	70	5	trained	train	VERB
alusur-6996	70	6	models	model	NOUN
alusur-6996	70	7	on	on	ADP
alusur-6996	70	8	works	work	NOUN
alusur-6996	70	9	#	#	SYM
alusur-6996	70	10	1	1	NUM
alusur-6996	70	11	-	-	SYM
alusur-6996	70	12	4	4	NUM
alusur-6996	70	13	specifically	specifically	ADV
alusur-6996	70	14	in	in	ADP
alusur-6996	70	15	our	our	PRON
alusur-6996	70	16	second	second	ADJ
alusur-6996	70	17	round	round	NOUN
alusur-6996	70	18	of	of	ADP
alusur-6996	70	19	testing	testing	NOUN
alusur-6996	70	20	,	,	PUNCT
alusur-6996	70	21	the	the	DET
alusur-6996	70	22	accuracy	accuracy	NOUN
alusur-6996	70	23	rates	rate	NOUN
alusur-6996	70	24	for	for	ADP
alusur-6996	70	25	these	these	DET
alusur-6996	70	26	texts	text	NOUN
alusur-6996	70	27	substantially	substantially	ADV
alusur-6996	70	28	improved	improve	VERB
alusur-6996	70	29	,	,	PUNCT
alusur-6996	70	30	reaching	reach	VERB
alusur-6996	70	31	into	into	ADP
alusur-6996	70	32	the	the	DET
alusur-6996	70	33	high	high	ADJ
alusur-6996	70	34	nineties	ninety	NOUN
alusur-6996	70	35	(	(	PUNCT
alusur-6996	70	36	see	see	VERB
alusur-6996	70	37	full	full	ADJ
alusur-6996	70	38	results	result	NOUN
alusur-6996	70	39	in	in	ADP
alusur-6996	70	40	table	table	NOUN
alusur-6996	70	41	4	4	NUM
alusur-6996	70	42	above	above	ADV
alusur-6996	70	43	)	)	PUNCT
alusur-6996	70	44	.	.	PUNCT
alusur-6996	71	1	the	the	DET
alusur-6996	71	2	accuracy	accuracy	NOUN
alusur-6996	71	3	results	result	VERB
alusur-6996	71	4	for	for	ADP
alusur-6996	71	5	work	work	NOUN
alusur-6996	71	6	#	#	SYM
alusur-6996	71	7	4	4	NUM
alusur-6996	71	8	,	,	PUNCT
alusur-6996	71	9	for	for	ADP
alusur-6996	71	10	example	example	NOUN
alusur-6996	71	11	,	,	PUNCT
alusur-6996	71	12	improved	improve	VERB
alusur-6996	71	13	from	from	ADP
alusur-6996	71	14	83.42	83.42	NUM
alusur-6996	71	15	%	%	NOUN
alusur-6996	71	16	on	on	ADP
alusur-6996	71	17	arabic	arabic	ADJ
alusur-6996	71	18	script	script	NOUN
alusur-6996	71	19	alone	alone	ADV
alusur-6996	71	20	in	in	ADP
alusur-6996	71	21	our	our	PRON
alusur-6996	71	22	first	first	ADJ
alusur-6996	71	23	work	work	NOUN
alusur-6996	71	24	#	#	SYM
alusur-6996	71	25	0	0	NUM
alusur-6996	71	26	-	-	PUNCT
alusur-6996	71	27	based	base	VERB
alusur-6996	71	28	model	model	NOUN
alusur-6996	71	29	tests	test	NOUN
alusur-6996	71	30	to	to	ADP
alusur-6996	71	31	99.18	99.18	NUM
alusur-6996	71	32	%	%	NOUN
alusur-6996	71	33	accuracy	accuracy	NOUN
alusur-6996	71	34	when	when	SCONJ
alusur-6996	71	35	we	we	PRON
alusur-6996	71	36	trained	train	VERB
alusur-6996	71	37	a	a	DET
alusur-6996	71	38	mode	mode	NOUN
alusur-6996	71	39	on	on	ADP
alusur-6996	71	40	this	this	DET
alusur-6996	71	41	text	text	NOUN
alusur-6996	71	42	.	.	PUNCT
alusur-6996	72	1	the	the	DET
alusur-6996	72	2	accuracy	accuracy	NOUN
alusur-6996	72	3	rates	rate	NOUN
alusur-6996	72	4	for	for	ADP
alusur-6996	72	5	works	work	NOUN
alusur-6996	72	6	#	#	SYM
alusur-6996	72	7	1	1	NUM
alusur-6996	72	8	-	-	SYM
alusur-6996	72	9	3	3	NUM
alusur-6996	72	10	similarly	similarly	ADV
alusur-6996	72	11	improved	improve	VERB
alusur-6996	72	12	,	,	PUNCT
alusur-6996	72	13	increasing	increase	VERB
alusur-6996	72	14	from	from	ADP
alusur-6996	72	15	90.90	90.90	NUM
alusur-6996	72	16	%	%	NOUN
alusur-6996	72	17	to	to	ADP
alusur-6996	72	18	97.71	97.71	NUM
alusur-6996	72	19	%	%	NOUN
alusur-6996	72	20	,	,	PUNCT
alusur-6996	72	21	87.90	87.90	NUM
alusur-6996	72	22	%	%	NOUN
alusur-6996	72	23	to	to	ADP
alusur-6996	72	24	98.47	98.47	NUM
alusur-6996	72	25	%	%	NOUN
alusur-6996	72	26	,	,	PUNCT
alusur-6996	72	27	and	and	CCONJ
alusur-6996	72	28	72.78	72.78	NUM
alusur-6996	72	29	%	%	NOUN
alusur-6996	72	30	to	to	ADP
alusur-6996	72	31	97.59	97.59	NUM
alusur-6996	72	32	%	%	NOUN
alusur-6996	72	33	respectively	respectively	ADV
alusur-6996	72	34	.	.	PUNCT
alusur-6996	73	1	(	(	PUNCT
alusur-6996	73	2	see	see	VERB
alusur-6996	73	3	appendix	appendix	NOUN
alusur-6996	73	4	for	for	ADP
alusur-6996	73	5	the	the	DET
alusur-6996	73	6	accuracy	accuracy	NOUN
alusur-6996	73	7	rates	rate	NOUN
alusur-6996	73	8	of	of	ADP
alusur-6996	73	9	these	these	DET
alusur-6996	73	10	new	new	ADJ
alusur-6996	73	11	models	model	NOUN
alusur-6996	73	12	on	on	ADP
alusur-6996	73	13	all	all	DET
alusur-6996	73	14	other	other	ADJ
alusur-6996	73	15	texts	text	NOUN
alusur-6996	73	16	as	as	ADV
alusur-6996	73	17	well	well	ADV
alusur-6996	73	18	.	.	PUNCT
alusur-6996	73	19	)	)	PUNCT
alusur-6996	74	1	these	these	DET
alusur-6996	74	2	accuracy	accuracy	NOUN
alusur-6996	74	3	rates	rate	NOUN
alusur-6996	74	4	for	for	ADP
alusur-6996	74	5	arabic	arabic	ADJ
alusur-6996	74	6	-	-	PUNCT
alusur-6996	74	7	script	script	NOUN
alusur-6996	74	8	recognition	recognition	NOUN
alusur-6996	74	9	are	be	AUX
alusur-6996	74	10	already	already	ADV
alusur-6996	74	11	impressively	impressively	ADV
alusur-6996	74	12	high	high	ADJ
alusur-6996	74	13	,	,	PUNCT
alusur-6996	74	14	but	but	CCONJ
alusur-6996	74	15	we	we	PRON
alusur-6996	74	16	actually	actually	ADV
alusur-6996	74	17	believe	believe	VERB
alusur-6996	74	18	that	that	SCONJ
alusur-6996	74	19	they	they	PRON
alusur-6996	74	20	can	can	AUX
alusur-6996	74	21	be	be	AUX
alusur-6996	74	22	improved	improve	VERB
alusur-6996	74	23	further	far	ADV
alusur-6996	74	24	with	with	ADP
alusur-6996	74	25	larger	large	ADJ
alusur-6996	74	26	training	training	NOUN
alusur-6996	74	27	data	datum	NOUN
alusur-6996	74	28	sets	set	NOUN
alusur-6996	74	29	.	.	PUNCT
alusur-6996	75	1	although	although	SCONJ
alusur-6996	75	2	the	the	DET
alusur-6996	75	3	process	process	NOUN
alusur-6996	75	4	of	of	ADP
alusur-6996	75	5	training	train	VERB
alusur-6996	75	6	a	a	DET
alusur-6996	75	7	new	new	ADJ
alusur-6996	75	8	model	model	NOUN
alusur-6996	75	9	for	for	ADP
alusur-6996	75	10	a	a	DET
alusur-6996	75	11	new	new	ADJ
alusur-6996	75	12	text	text	NOUN
alusur-6996	75	13	/	/	SYM
alusur-6996	75	14	typeface	typeface	NOUN
alusur-6996	75	15	does	do	AUX
alusur-6996	75	16	require	require	VERB
alusur-6996	75	17	some	some	DET
alusur-6996	75	18	effort	effort	NOUN
alusur-6996	75	19	,	,	PUNCT
alusur-6996	75	20	the	the	DET
alusur-6996	75	21	only	only	ADJ
alusur-6996	75	22	time	time	NOUN
alusur-6996	75	23	-	-	PUNCT
alusur-6996	75	24	consuming	consume	VERB
alusur-6996	75	25	component	component	NOUN
alusur-6996	75	26	is	be	AUX
alusur-6996	75	27	the	the	DET
alusur-6996	75	28	generation	generation	NOUN
alusur-6996	75	29	of	of	ADP
alusur-6996	75	30	the	the	DET
alusur-6996	75	31	~1,000	~1,000	PROPN
alusur-6996	75	32	lines	line	NOUN
alusur-6996	75	33	of	of	ADP
alusur-6996	75	34	gold	gold	NOUN
alusur-6996	75	35	standard	standard	ADJ
alusur-6996	75	36	training	training	NOUN
alusur-6996	75	37	data	datum	NOUN
alusur-6996	75	38	.	.	PUNCT
alusur-6996	76	1	as	as	SCONJ
alusur-6996	76	2	we	we	PRON
alusur-6996	76	3	develop	develop	VERB
alusur-6996	76	4	the	the	DET
alusur-6996	76	5	openiti	openiti	PROPN
alusur-6996	76	6	ocr	ocr	PROPN
alusur-6996	76	7	project	project	NOUN
alusur-6996	76	8	we	we	PRON
alusur-6996	76	9	will	will	AUX
alusur-6996	76	10	address	address	VERB
alusur-6996	76	11	the	the	DET
alusur-6996	76	12	issue	issue	NOUN
alusur-6996	76	13	of	of	ADP
alusur-6996	76	14	the	the	DET
alusur-6996	76	15	need	need	NOUN
alusur-6996	76	16	for	for	ADP
alusur-6996	76	17	multiple	multiple	ADJ
alusur-6996	76	18	models12	models12	NOUN
alusur-6996	76	19	through	through	ADP
alusur-6996	76	20	a	a	DET
alusur-6996	76	21	two	two	NUM
alusur-6996	76	22	-	-	PUNCT
alusur-6996	76	23	pronged	pronged	ADJ
alusur-6996	76	24	strategy	strategy	NOUN
alusur-6996	76	25	.	.	PUNCT
alusur-6996	77	1	first	first	ADV
alusur-6996	77	2	,	,	PUNCT
alusur-6996	77	3	we	we	PRON
alusur-6996	77	4	will	will	AUX
alusur-6996	77	5	try	try	VERB
alusur-6996	77	6	to	to	PART
alusur-6996	77	7	train	train	VERB
alusur-6996	77	8	generalized	generalized	ADJ
alusur-6996	77	9	models	model	NOUN
alusur-6996	77	10	for	for	ADP
alusur-6996	77	11	each	each	DET
alusur-6996	77	12	script	script	NOUN
alusur-6996	77	13	,	,	PUNCT
alusur-6996	77	14	periodically	periodically	ADV
alusur-6996	77	15	adding	add	VERB
alusur-6996	77	16	new	new	ADJ
alusur-6996	77	17	features	feature	NOUN
alusur-6996	77	18	that	that	SCONJ
alusur-6996	77	19	the	the	DET
alusur-6996	77	20	model	model	NOUN
alusur-6996	77	21	has	have	AUX
alusur-6996	77	22	not	not	PART
alusur-6996	77	23	“	"	PUNCT
alusur-6996	77	24	seen	see	VERB
alusur-6996	77	25	”	"	PUNCT
alusur-6996	77	26	before	before	ADV
alusur-6996	77	27	.	.	PUNCT
alusur-6996	78	1	secondly	secondly	ADV
alusur-6996	78	2	,	,	PUNCT
alusur-6996	78	3	we	we	PRON
alusur-6996	78	4	will	will	AUX
alusur-6996	78	5	train	train	VERB
alusur-6996	78	6	individual	individual	ADJ
alusur-6996	78	7	models	model	NOUN
alusur-6996	78	8	for	for	ADP
alusur-6996	78	9	distinct	distinct	ADJ
alusur-6996	78	10	typefaces	typeface	NOUN
alusur-6996	78	11	and	and	CCONJ
alusur-6996	78	12	editorial	editorial	ADJ
alusur-6996	78	13	styles	style	NOUN
alusur-6996	78	14	(	(	PUNCT
alusur-6996	78	15	which	which	PRON
alusur-6996	78	16	sometimes	sometimes	ADV
alusur-6996	78	17	vary	vary	VERB
alusur-6996	78	18	in	in	ADP
alusur-6996	78	19	their	their	PRON
alusur-6996	78	20	use	use	NOUN
alusur-6996	78	21	of	of	ADP
alusur-6996	78	22	vocalization	vocalization	NOUN
alusur-6996	78	23	,	,	PUNCT
alusur-6996	78	24	fonts	font	NOUN
alusur-6996	78	25	,	,	PUNCT
alusur-6996	78	26	spacing	spacing	NOUN
alusur-6996	78	27	,	,	PUNCT
alusur-6996	78	28	and	and	CCONJ
alusur-6996	78	29	punctuation	punctuation	NOUN
alusur-6996	78	30	)	)	PUNCT
alusur-6996	78	31	,	,	PUNCT
alusur-6996	78	32	12	12	NUM
alusur-6996	78	33	.	.	PUNCT
alusur-6996	79	1	generalized	generalized	ADJ
alusur-6996	79	2	models	model	NOUN
alusur-6996	79	3	achieve	achieve	VERB
alusur-6996	79	4	acceptable	acceptable	ADJ
alusur-6996	79	5	accuracy	accuracy	NOUN
alusur-6996	79	6	across	across	ADP
alusur-6996	79	7	a	a	DET
alusur-6996	79	8	wide	wide	ADJ
alusur-6996	79	9	range	range	NOUN
alusur-6996	79	10	of	of	ADP
alusur-6996	79	11	fonts	font	NOUN
alusur-6996	79	12	by	by	ADP
alusur-6996	79	13	incorporating	incorporate	VERB
alusur-6996	79	14	features	feature	NOUN
alusur-6996	79	15	of	of	ADP
alusur-6996	79	16	a	a	DET
alusur-6996	79	17	variety	variety	NOUN
alusur-6996	79	18	of	of	ADP
alusur-6996	79	19	typefaces	typeface	NOUN
alusur-6996	79	20	during	during	ADP
alusur-6996	79	21	training	training	NOUN
alusur-6996	79	22	,	,	PUNCT
alusur-6996	79	23	allowing	allow	VERB
alusur-6996	79	24	them	they	PRON
alusur-6996	79	25	to	to	PART
alusur-6996	79	26	be	be	AUX
alusur-6996	79	27	used	use	VERB
alusur-6996	79	28	for	for	ADP
alusur-6996	79	29	most	most	ADJ
alusur-6996	79	30	texts	text	NOUN
alusur-6996	79	31	with	with	ADP
alusur-6996	79	32	common	common	ADJ
alusur-6996	79	33	typefaces	typeface	NOUN
alusur-6996	79	34	.	.	PUNCT
alusur-6996	80	1	figure	figure	NOUN
alusur-6996	80	2	2	2	NUM
alusur-6996	80	3	:	:	PUNCT
alusur-6996	80	4	kraken	kraken	PROPN
alusur-6996	80	5	’s	’s	PART
alusur-6996	80	6	transcription	transcription	NOUN
alusur-6996	80	7	interface	interface	NOUN
alusur-6996	80	8	8	8	NUM
alusur-6996	80	9	•	•	NOUN
alusur-6996	80	10	kiessling	kiessle	VERB
alusur-6996	80	11	,	,	PUNCT
alusur-6996	80	12	miller	miller	PROPN
alusur-6996	80	13	,	,	PUNCT
alusur-6996	80	14	romanov	romanov	PROPN
alusur-6996	80	15	,	,	PUNCT
alusur-6996	80	16	&	&	CCONJ
alusur-6996	80	17	savant	savant	PROPN
alusur-6996	80	18	al-ʿuṣūr	al-ʿuṣūr	PROPN
alusur-6996	80	19	al	al	PROPN
alusur-6996	80	20	-	-	PUNCT
alusur-6996	80	21	wusṭā	wusṭā	NOUN
alusur-6996	80	22	25	25	NUM
alusur-6996	80	23	(	(	PUNCT
alusur-6996	80	24	2017	2017	NUM
alusur-6996	80	25	)	)	PUNCT
alusur-6996	80	26	producing	produce	VERB
alusur-6996	80	27	a	a	DET
alusur-6996	80	28	library	library	NOUN
alusur-6996	80	29	of	of	ADP
alusur-6996	80	30	ocr	ocr	ADJ
alusur-6996	80	31	models	model	NOUN
alusur-6996	80	32	that	that	PRON
alusur-6996	80	33	gradually	gradually	ADV
alusur-6996	80	34	will	will	AUX
alusur-6996	80	35	cover	cover	VERB
alusur-6996	80	36	all	all	DET
alusur-6996	80	37	major	major	ADJ
alusur-6996	80	38	typefaces	typeface	NOUN
alusur-6996	80	39	and	and	CCONJ
alusur-6996	80	40	editorial	editorial	ADJ
alusur-6996	80	41	styles	style	NOUN
alusur-6996	80	42	used	use	VERB
alusur-6996	80	43	in	in	ADP
alusur-6996	80	44	modern	modern	ADJ
alusur-6996	80	45	arabic	arabic	ADJ
alusur-6996	80	46	-	-	PUNCT
alusur-6996	80	47	script	script	NOUN
alusur-6996	80	48	printing	printing	NOUN
alusur-6996	80	49	.	.	PUNCT
alusur-6996	81	1	there	there	PRON
alusur-6996	81	2	certainly	certainly	ADV
alusur-6996	81	3	are	be	AUX
alusur-6996	81	4	numerous	numerous	ADJ
alusur-6996	81	5	arabic	arabic	ADJ
alusur-6996	81	6	-	-	PUNCT
alusur-6996	81	7	script	script	NOUN
alusur-6996	81	8	typefaces	typeface	NOUN
alusur-6996	81	9	and	and	CCONJ
alusur-6996	81	10	editorial	editorial	ADJ
alusur-6996	81	11	styles	style	NOUN
alusur-6996	81	12	that	that	PRON
alusur-6996	81	13	have	have	AUX
alusur-6996	81	14	been	be	AUX
alusur-6996	81	15	used	use	VERB
alusur-6996	81	16	throughout	throughout	ADP
alusur-6996	81	17	the	the	DET
alusur-6996	81	18	last	last	ADJ
alusur-6996	81	19	century	century	NOUN
alusur-6996	81	20	and	and	CCONJ
alusur-6996	81	21	a	a	DET
alusur-6996	81	22	half	half	NOUN
alusur-6996	81	23	of	of	ADP
alusur-6996	81	24	arabic	arabic	ADJ
alusur-6996	81	25	-	-	PUNCT
alusur-6996	81	26	script	script	NOUN
alusur-6996	81	27	printing	printing	NOUN
alusur-6996	81	28	,	,	PUNCT
alusur-6996	81	29	but	but	CCONJ
alusur-6996	81	30	ultimately	ultimately	ADV
alusur-6996	81	31	the	the	DET
alusur-6996	81	32	number	number	NOUN
alusur-6996	81	33	is	be	AUX
alusur-6996	81	34	finite	finite	ADJ
alusur-6996	81	35	and	and	CCONJ
alusur-6996	81	36	definitely	definitely	ADV
alusur-6996	81	37	not	not	PART
alusur-6996	81	38	so	so	ADV
alusur-6996	81	39	numerous	numerous	ADJ
alusur-6996	81	40	as	as	SCONJ
alusur-6996	81	41	to	to	PART
alusur-6996	81	42	make	make	VERB
alusur-6996	81	43	it	it	PRON
alusur-6996	81	44	impossible	impossible	ADJ
alusur-6996	81	45	to	to	PART
alusur-6996	81	46	create	create	VERB
alusur-6996	81	47	models	model	NOUN
alusur-6996	81	48	for	for	ADP
alusur-6996	81	49	each	each	PRON
alusur-6996	81	50	over	over	ADP
alusur-6996	81	51	the	the	DET
alusur-6996	81	52	long	long	ADJ
alusur-6996	81	53	term	term	NOUN
alusur-6996	81	54	.	.	PUNCT
alusur-6996	82	1	3.1	3.1	NUM
alusur-6996	82	2	conclusions	conclusion	NOUN
alusur-6996	82	3	and	and	CCONJ
alusur-6996	82	4	next	next	ADJ
alusur-6996	82	5	steps	step	NOUN
alusur-6996	82	6	for	for	ADP
alusur-6996	82	7	the	the	DET
alusur-6996	82	8	openiti	openiti	PROPN
alusur-6996	82	9	ocr	ocr	PROPN
alusur-6996	82	10	project	project	NOUN
alusur-6996	82	11	the	the	DET
alusur-6996	82	12	two	two	NUM
alusur-6996	82	13	rounds	round	NOUN
alusur-6996	82	14	of	of	ADP
alusur-6996	82	15	testing	testing	NOUN
alusur-6996	82	16	presented	present	VERB
alusur-6996	82	17	here	here	ADV
alusur-6996	82	18	indicate	indicate	VERB
alusur-6996	82	19	that	that	SCONJ
alusur-6996	82	20	with	with	ADP
alusur-6996	82	21	a	a	DET
alusur-6996	82	22	fairly	fairly	ADV
alusur-6996	82	23	modest	modest	ADJ
alusur-6996	82	24	amount	amount	NOUN
alusur-6996	82	25	of	of	ADP
alusur-6996	82	26	gold	gold	NOUN
alusur-6996	82	27	standard	standard	ADJ
alusur-6996	82	28	training	training	NOUN
alusur-6996	82	29	data	datum	NOUN
alusur-6996	82	30	(	(	PUNCT
alusur-6996	82	31	~800–1,000	~800–1,000	NUM
alusur-6996	82	32	lines	line	NOUN
alusur-6996	82	33	)	)	PUNCT
alusur-6996	82	34	,	,	PUNCT
alusur-6996	82	35	kraken	kraken	PROPN
alusur-6996	82	36	is	be	AUX
alusur-6996	82	37	consistently	consistently	ADV
alusur-6996	82	38	able	able	ADJ
alusur-6996	82	39	to	to	PART
alusur-6996	82	40	produce	produce	VERB
alusur-6996	82	41	ocr	ocr	ADJ
alusur-6996	82	42	results	result	NOUN
alusur-6996	82	43	for	for	ADP
alusur-6996	82	44	arabic	arabic	ADJ
alusur-6996	82	45	-	-	PUNCT
alusur-6996	82	46	script	script	NOUN
alusur-6996	82	47	documents	document	NOUN
alusur-6996	82	48	that	that	PRON
alusur-6996	82	49	achieve	achieve	VERB
alusur-6996	82	50	accuracy	accuracy	NOUN
alusur-6996	82	51	rates	rate	NOUN
alusur-6996	82	52	in	in	ADP
alusur-6996	82	53	the	the	DET
alusur-6996	82	54	high	high	ADJ
alusur-6996	82	55	nineties	ninety	NOUN
alusur-6996	82	56	.	.	PUNCT
alusur-6996	83	1	in	in	ADP
alusur-6996	83	2	some	some	DET
alusur-6996	83	3	cases	case	NOUN
alusur-6996	83	4	,	,	PUNCT
alusur-6996	83	5	such	such	ADJ
alusur-6996	83	6	as	as	ADP
alusur-6996	83	7	works	work	NOUN
alusur-6996	83	8	#	#	SYM
alusur-6996	83	9	5	5	NUM
alusur-6996	83	10	-	-	SYM
alusur-6996	83	11	6	6	NUM
alusur-6996	83	12	,	,	PUNCT
alusur-6996	83	13	achieving	achieve	VERB
alusur-6996	83	14	ocr	ocr	ADJ
alusur-6996	83	15	accuracy	accuracy	NOUN
alusur-6996	83	16	rates	rate	NOUN
alusur-6996	83	17	of	of	ADP
alusur-6996	83	18	up	up	ADP
alusur-6996	83	19	to	to	PART
alusur-6996	83	20	97.5	97.5	NUM
alusur-6996	83	21	%	%	NOUN
alusur-6996	83	22	does	do	AUX
alusur-6996	83	23	not	not	PART
alusur-6996	83	24	even	even	ADV
alusur-6996	83	25	require	require	VERB
alusur-6996	83	26	training	train	VERB
alusur-6996	83	27	a	a	DET
alusur-6996	83	28	new	new	ADJ
alusur-6996	83	29	model	model	NOUN
alusur-6996	83	30	on	on	ADP
alusur-6996	83	31	that	that	DET
alusur-6996	83	32	text	text	NOUN
alusur-6996	83	33	.	.	PUNCT
alusur-6996	84	1	however	however	ADV
alusur-6996	84	2	,	,	PUNCT
alusur-6996	84	3	in	in	ADP
alusur-6996	84	4	other	other	ADJ
alusur-6996	84	5	cases	case	NOUN
alusur-6996	84	6	,	,	PUNCT
alusur-6996	84	7	such	such	ADJ
alusur-6996	84	8	as	as	ADP
alusur-6996	84	9	works	work	NOUN
alusur-6996	84	10	#	#	SYM
alusur-6996	84	11	1	1	NUM
alusur-6996	84	12	-	-	SYM
alusur-6996	84	13	4	4	NUM
alusur-6996	84	14	,	,	PUNCT
alusur-6996	84	15	achieving	achieve	VERB
alusur-6996	84	16	high	high	ADJ
alusur-6996	84	17	levels	level	NOUN
alusur-6996	84	18	of	of	ADP
alusur-6996	84	19	ocr	ocr	ADJ
alusur-6996	84	20	accuracy	accuracy	NOUN
alusur-6996	84	21	does	do	AUX
alusur-6996	84	22	require	require	VERB
alusur-6996	84	23	training	train	VERB
alusur-6996	84	24	a	a	DET
alusur-6996	84	25	model	model	NOUN
alusur-6996	84	26	specific	specific	ADJ
alusur-6996	84	27	to	to	ADP
alusur-6996	84	28	that	that	DET
alusur-6996	84	29	typeface	typeface	NOUN
alusur-6996	84	30	,	,	PUNCT
alusur-6996	84	31	and	and	CCONJ
alusur-6996	84	32	,	,	PUNCT
alusur-6996	84	33	in	in	ADP
alusur-6996	84	34	some	some	DET
alusur-6996	84	35	select	select	ADJ
alusur-6996	84	36	cases	case	NOUN
alusur-6996	84	37	of	of	ADP
alusur-6996	84	38	texts	text	NOUN
alusur-6996	84	39	with	with	ADP
alusur-6996	84	40	similar	similar	ADJ
alusur-6996	84	41	typefaces	typeface	NOUN
alusur-6996	84	42	but	but	CCONJ
alusur-6996	84	43	different	different	ADJ
alusur-6996	84	44	styles	style	NOUN
alusur-6996	84	45	of	of	ADP
alusur-6996	84	46	vocalization	vocalization	NOUN
alusur-6996	84	47	,	,	PUNCT
alusur-6996	84	48	font	font	NOUN
alusur-6996	84	49	variations	variation	NOUN
alusur-6996	84	50	,	,	PUNCT
alusur-6996	84	51	and	and	CCONJ
alusur-6996	84	52	punctuation	punctuation	NOUN
alusur-6996	84	53	patterns	pattern	NOUN
alusur-6996	84	54	(	(	PUNCT
alusur-6996	84	55	e.g.	e.g.	ADV
alusur-6996	84	56	,	,	PUNCT
alusur-6996	84	57	works	work	VERB
alusur-6996	84	58	#	#	SYM
alusur-6996	84	59	1	1	NUM
alusur-6996	84	60	-	-	SYM
alusur-6996	84	61	2	2	NUM
alusur-6996	84	62	)	)	PUNCT
alusur-6996	84	63	,	,	PUNCT
alusur-6996	84	64	training	train	VERB
alusur-6996	84	65	a	a	DET
alusur-6996	84	66	model	model	NOUN
alusur-6996	84	67	for	for	ADP
alusur-6996	84	68	the	the	DET
alusur-6996	84	69	peculiarities	peculiarity	NOUN
alusur-6996	84	70	of	of	ADP
alusur-6996	84	71	a	a	DET
alusur-6996	84	72	particular	particular	ADJ
alusur-6996	84	73	edition	edition	NOUN
alusur-6996	84	74	.	.	PUNCT
alusur-6996	85	1	in	in	ADP
alusur-6996	85	2	the	the	DET
alusur-6996	85	3	near	near	ADJ
alusur-6996	85	4	future	future	NOUN
alusur-6996	85	5	we	we	PRON
alusur-6996	85	6	are	be	AUX
alusur-6996	85	7	planning	plan	VERB
alusur-6996	85	8	to	to	PART
alusur-6996	85	9	release	release	VERB
alusur-6996	85	10	a	a	DET
alusur-6996	85	11	new	new	ADJ
alusur-6996	85	12	web	web	NOUN
alusur-6996	85	13	-	-	NOUN
alusur-6996	85	14	interface	interface	NOUN
alusur-6996	85	15	powered	power	VERB
alusur-6996	85	16	by	by	ADP
alusur-6996	85	17	the	the	DET
alusur-6996	85	18	microtask	microtask	NOUN
alusur-6996	85	19	platform	platform	NOUN
alusur-6996	85	20	pybossa	pybossa	NOUN
alusur-6996	85	21	that	that	PRON
alusur-6996	85	22	will	will	AUX
alusur-6996	85	23	enable	enable	VERB
alusur-6996	85	24	more	more	ADJ
alusur-6996	85	25	user	user	NOUN
alusur-6996	85	26	-	-	PUNCT
alusur-6996	85	27	friendly	friendly	ADJ
alusur-6996	85	28	generation	generation	NOUN
alusur-6996	85	29	of	of	ADP
alusur-6996	85	30	training	training	NOUN
alusur-6996	85	31	data	datum	NOUN
alusur-6996	85	32	and	and	CCONJ
alusur-6996	85	33	the	the	DET
alusur-6996	85	34	post	post	NOUN
alusur-6996	85	35	-	-	NOUN
alusur-6996	85	36	correction	correction	NOUN
alusur-6996	85	37	of	of	ADP
alusur-6996	85	38	the	the	DET
alusur-6996	85	39	ocr	ocr	ADJ
alusur-6996	85	40	output	output	NOUN
alusur-6996	85	41	(	(	PUNCT
alusur-6996	85	42	see	see	VERB
alusur-6996	85	43	,	,	PUNCT
alusur-6996	85	44	figure	figure	VERB
alusur-6996	85	45	3	3	NUM
alusur-6996	85	46	below	below	ADV
alusur-6996	85	47	and	and	CCONJ
alusur-6996	85	48	the	the	DET
alusur-6996	85	49	ocr	ocr	ADJ
alusur-6996	85	50	section	section	NOUN
alusur-6996	85	51	of	of	ADP
alusur-6996	85	52	the	the	DET
alusur-6996	85	53	openiti	openiti	NOUN
alusur-6996	85	54	figure	figure	NOUN
alusur-6996	85	55	3	3	NUM
alusur-6996	85	56	:	:	PUNCT
alusur-6996	85	57	web	web	NOUN
alusur-6996	85	58	-	-	PUNCT
alusur-6996	85	59	based	base	VERB
alusur-6996	85	60	ocr	ocr	ADJ
alusur-6996	85	61	pipeline	pipeline	NOUN
alusur-6996	85	62	flowchart	flowchart	PROPN
alusur-6996	85	63	al-ʿuṣūr	al-ʿuṣūr	PROPN
alusur-6996	85	64	al	al	PROPN
alusur-6996	85	65	-	-	PUNCT
alusur-6996	85	66	wusṭā	wusṭā	NOUN
alusur-6996	85	67	25	25	NUM
alusur-6996	85	68	(	(	PUNCT
alusur-6996	85	69	2017	2017	NUM
alusur-6996	85	70	)	)	PUNCT
alusur-6996	85	71	new	new	ADJ
alusur-6996	85	72	developments	development	NOUN
alusur-6996	85	73	in	in	ADP
alusur-6996	85	74	arabographic	arabographic	ADJ
alusur-6996	85	75	optical	optical	ADJ
alusur-6996	85	76	character	character	NOUN
alusur-6996	85	77	recognition	recognition	NOUN
alusur-6996	85	78	•	•	NOUN
alusur-6996	85	79	9	9	NUM
alusur-6996	85	80	website	website	NOUN
alusur-6996	85	81	)	)	PUNCT
alusur-6996	85	82	.	.	PUNCT
alusur-6996	86	1	new	new	ADJ
alusur-6996	86	2	data	datum	NOUN
alusur-6996	86	3	supplied	supply	VERB
alusur-6996	86	4	by	by	ADP
alusur-6996	86	5	users	user	NOUN
alusur-6996	86	6	will	will	AUX
alusur-6996	86	7	allow	allow	VERB
alusur-6996	86	8	us	we	PRON
alusur-6996	86	9	to	to	PART
alusur-6996	86	10	train	train	VERB
alusur-6996	86	11	additional	additional	ADJ
alusur-6996	86	12	typeface	typeface	NOUN
alusur-6996	86	13	-	-	PUNCT
alusur-6996	86	14	specific	specific	ADJ
alusur-6996	86	15	models	model	NOUN
alusur-6996	86	16	and	and	CCONJ
alusur-6996	86	17	improve	improve	VERB
alusur-6996	86	18	the	the	DET
alusur-6996	86	19	overall	overall	ADJ
alusur-6996	86	20	accuracy	accuracy	NOUN
alusur-6996	86	21	of	of	ADP
alusur-6996	86	22	kraken	kraken	NOUN
alusur-6996	86	23	on	on	ADP
alusur-6996	86	24	other	other	ADJ
alusur-6996	86	25	typefaces	typeface	NOUN
alusur-6996	86	26	through	through	ADP
alusur-6996	86	27	the	the	DET
alusur-6996	86	28	development	development	NOUN
alusur-6996	86	29	of	of	ADP
alusur-6996	86	30	generalized	generalized	ADJ
alusur-6996	86	31	language	language	NOUN
alusur-6996	86	32	models	model	NOUN
alusur-6996	86	33	.	.	PUNCT
alusur-6996	87	1	(	(	PUNCT
alusur-6996	87	2	it	it	PRON
alusur-6996	87	3	should	should	AUX
alusur-6996	87	4	be	be	AUX
alusur-6996	87	5	stressed	stress	VERB
alusur-6996	87	6	,	,	PUNCT
alusur-6996	87	7	however	however	ADV
alusur-6996	87	8	,	,	PUNCT
alusur-6996	87	9	that	that	SCONJ
alusur-6996	87	10	training	train	VERB
alusur-6996	87	11	edition	edition	NOUN
alusur-6996	87	12	-	-	PUNCT
alusur-6996	87	13	specific	specific	ADJ
alusur-6996	87	14	models	model	NOUN
alusur-6996	87	15	is	be	AUX
alusur-6996	87	16	quite	quite	ADV
alusur-6996	87	17	valuable	valuable	ADJ
alusur-6996	87	18	,	,	PUNCT
alusur-6996	87	19	as	as	SCONJ
alusur-6996	87	20	there	there	PRON
alusur-6996	87	21	are	be	VERB
alusur-6996	87	22	a	a	DET
alusur-6996	87	23	number	number	NOUN
alusur-6996	87	24	of	of	ADP
alusur-6996	87	25	multivolume	multivolume	ADJ
alusur-6996	87	26	books	book	NOUN
alusur-6996	87	27	—	—	PUNCT
alusur-6996	87	28	often	often	ADV
alusur-6996	87	29	with	with	ADP
alusur-6996	87	30	over	over	ADP
alusur-6996	87	31	a	a	DET
alusur-6996	87	32	dozen	dozen	NOUN
alusur-6996	87	33	volumes	volume	NOUN
alusur-6996	87	34	per	per	ADP
alusur-6996	87	35	text	text	NOUN
alusur-6996	87	36	—	—	PUNCT
alusur-6996	87	37	that	that	PRON
alusur-6996	87	38	need	need	VERB
alusur-6996	87	39	to	to	PART
alusur-6996	87	40	be	be	AUX
alusur-6996	87	41	converted	convert	VERB
alusur-6996	87	42	into	into	ADP
alusur-6996	87	43	proper	proper	ADJ
alusur-6996	87	44	digital	digital	ADJ
alusur-6996	87	45	editions	edition	NOUN
alusur-6996	87	46	)	)	PUNCT
alusur-6996	87	47	.	.	PUNCT
alusur-6996	88	1	furthermore	furthermore	ADV
alusur-6996	88	2	,	,	PUNCT
alusur-6996	88	3	in	in	ADP
alusur-6996	88	4	collaboration	collaboration	NOUN
alusur-6996	88	5	with	with	ADP
alusur-6996	88	6	several	several	ADJ
alusur-6996	88	7	colleagues	colleague	NOUN
alusur-6996	88	8	,	,	PUNCT
alusur-6996	88	9	we	we	PRON
alusur-6996	88	10	are	be	AUX
alusur-6996	88	11	also	also	ADV
alusur-6996	88	12	currently	currently	ADV
alusur-6996	88	13	testing	test	VERB
alusur-6996	88	14	kraken	kraken	NOUN
alusur-6996	88	15	on	on	ADP
alusur-6996	88	16	various	various	ADJ
alusur-6996	88	17	persian	persian	ADJ
alusur-6996	88	18	,	,	PUNCT
alusur-6996	88	19	hebrew	hebrew	ADJ
alusur-6996	88	20	,	,	PUNCT
alusur-6996	88	21	and	and	CCONJ
alusur-6996	88	22	syriac	syriac	ADJ
alusur-6996	88	23	typefaces	typeface	NOUN
alusur-6996	88	24	(	(	PUNCT
alusur-6996	88	25	results	result	NOUN
alusur-6996	88	26	forthcoming	forthcoming	ADJ
alusur-6996	88	27	spring	spring	NOUN
alusur-6996	88	28	2018	2018	NUM
alusur-6996	88	29	)	)	PUNCT
alusur-6996	88	30	and	and	CCONJ
alusur-6996	88	31	persian	persian	ADJ
alusur-6996	88	32	and	and	CCONJ
alusur-6996	88	33	arabic	arabic	ADJ
alusur-6996	88	34	manuscripts	manuscript	NOUN
alusur-6996	88	35	(	(	PUNCT
alusur-6996	88	36	results	result	NOUN
alusur-6996	88	37	forthcoming	forthcoming	ADJ
alusur-6996	88	38	summer	summer	NOUN
alusur-6996	88	39	2018	2018	NUM
alusur-6996	88	40	)	)	PUNCT
alusur-6996	88	41	.	.	PUNCT
alusur-6996	89	1	we	we	PRON
alusur-6996	89	2	plan	plan	VERB
alusur-6996	89	3	to	to	PART
alusur-6996	89	4	also	also	ADV
alusur-6996	89	5	train	train	VERB
alusur-6996	89	6	models	model	NOUN
alusur-6996	89	7	for	for	ADP
alusur-6996	89	8	other	other	ADJ
alusur-6996	89	9	islamicate	islamicate	ADJ
alusur-6996	89	10	languages	language	NOUN
alusur-6996	89	11	(	(	PUNCT
alusur-6996	89	12	in	in	ADP
alusur-6996	89	13	particular	particular	ADJ
alusur-6996	89	14	,	,	PUNCT
alusur-6996	89	15	ottoman	ottoman	NOUN
alusur-6996	89	16	turkish	turkish	NOUN
alusur-6996	89	17	and	and	CCONJ
alusur-6996	89	18	urdu	urdu	PROPN
alusur-6996	89	19	)	)	PUNCT
alusur-6996	89	20	as	as	ADV
alusur-6996	89	21	soon	soon	ADV
alusur-6996	89	22	as	as	SCONJ
alusur-6996	89	23	we	we	PRON
alusur-6996	89	24	can	can	AUX
alusur-6996	89	25	find	find	VERB
alusur-6996	89	26	experts	expert	NOUN
alusur-6996	89	27	in	in	ADP
alusur-6996	89	28	these	these	DET
alusur-6996	89	29	languages	language	NOUN
alusur-6996	89	30	who	who	PRON
alusur-6996	89	31	are	be	AUX
alusur-6996	89	32	willing	willing	ADJ
alusur-6996	89	33	to	to	PART
alusur-6996	89	34	collaborate	collaborate	VERB
alusur-6996	89	35	with	with	ADP
alusur-6996	89	36	us	we	PRON
alusur-6996	89	37	in	in	ADP
alusur-6996	89	38	training	training	NOUN
alusur-6996	89	39	data	datum	NOUN
alusur-6996	89	40	generation.13	generation.13	VERB
alusur-6996	89	41	in	in	ADP
alusur-6996	89	42	the	the	DET
alusur-6996	89	43	long	long	ADJ
alusur-6996	89	44	term	term	NOUN
alusur-6996	89	45	,	,	PUNCT
alusur-6996	89	46	our	our	PRON
alusur-6996	89	47	hope	hope	NOUN
alusur-6996	89	48	is	be	AUX
alusur-6996	89	49	that	that	SCONJ
alusur-6996	89	50	an	an	DET
alusur-6996	89	51	easy	easy	ADJ
alusur-6996	89	52	-	-	PUNCT
alusur-6996	89	53	to	to	ADP
alusur-6996	89	54	-	-	PUNCT
alusur-6996	89	55	use	use	VERB
alusur-6996	89	56	and	and	CCONJ
alusur-6996	89	57	effective	effective	ADJ
alusur-6996	89	58	ocr	ocr	ADJ
alusur-6996	89	59	pipeline	pipeline	NOUN
alusur-6996	89	60	will	will	AUX
alusur-6996	89	61	give	give	VERB
alusur-6996	89	62	us	we	PRON
alusur-6996	89	63	the	the	DET
alusur-6996	89	64	tools	tool	NOUN
alusur-6996	89	65	we	we	PRON
alusur-6996	89	66	,	,	PUNCT
alusur-6996	89	67	as	as	ADP
alusur-6996	89	68	a	a	DET
alusur-6996	89	69	field	field	NOUN
alusur-6996	89	70	,	,	PUNCT
alusur-6996	89	71	collectively	collectively	ADV
alusur-6996	89	72	need	need	VERB
alusur-6996	89	73	to	to	PART
alusur-6996	89	74	significantly	significantly	ADV
alusur-6996	89	75	enrich	enrich	VERB
alusur-6996	89	76	our	our	PRON
alusur-6996	89	77	collection	collection	NOUN
alusur-6996	89	78	of	of	ADP
alusur-6996	89	79	digital	digital	ADJ
alusur-6996	89	80	persian	persian	ADJ
alusur-6996	89	81	and	and	CCONJ
alusur-6996	89	82	arabic	arabic	ADJ
alusur-6996	89	83	texts	text	NOUN
alusur-6996	89	84	and	and	CCONJ
alusur-6996	89	85	thereby	thereby	ADV
alusur-6996	89	86	enable	enable	VERB
alusur-6996	89	87	us	we	PRON
alusur-6996	89	88	to	to	PART
alusur-6996	89	89	understand	understand	VERB
alusur-6996	89	90	better	well	ADJ
alusur-6996	89	91	the	the	DET
alusur-6996	89	92	cultural	cultural	ADJ
alusur-6996	89	93	heritage	heritage	NOUN
alusur-6996	89	94	of	of	ADP
alusur-6996	89	95	the	the	DET
alusur-6996	89	96	middle	middle	PROPN
alusur-6996	89	97	east	east	PROPN
alusur-6996	89	98	as	as	SCONJ
alusur-6996	89	99	reflected	reflect	VERB
alusur-6996	89	100	in	in	ADP
alusur-6996	89	101	its	its	PRON
alusur-6996	89	102	literary	literary	ADJ
alusur-6996	89	103	traditions	tradition	NOUN
alusur-6996	89	104	.	.	PUNCT
alusur-6996	90	1	ocr	ocr	PROPN
alusur-6996	90	2	,	,	PUNCT
alusur-6996	90	3	though	though	ADV
alusur-6996	90	4	,	,	PUNCT
alusur-6996	90	5	should	should	AUX
alusur-6996	90	6	not	not	PART
alusur-6996	90	7	be	be	AUX
alusur-6996	90	8	interpreted	interpret	VERB
alusur-6996	90	9	as	as	ADP
alusur-6996	90	10	a	a	DET
alusur-6996	90	11	magic	magic	ADJ
alusur-6996	90	12	bullet	bullet	NOUN
alusur-6996	90	13	.	.	PUNCT
alusur-6996	91	1	we	we	PRON
alusur-6996	91	2	must	must	AUX
alusur-6996	91	3	also	also	ADV
alusur-6996	91	4	cultivate	cultivate	VERB
alusur-6996	91	5	a	a	DET
alusur-6996	91	6	community	community	NOUN
alusur-6996	91	7	of	of	ADP
alusur-6996	91	8	users	user	NOUN
alusur-6996	91	9	and	and	CCONJ
alusur-6996	91	10	secure	secure	VERB
alusur-6996	91	11	long	long	ADJ
alusur-6996	91	12	-	-	PUNCT
alusur-6996	91	13	term	term	NOUN
alusur-6996	91	14	funding	funding	NOUN
alusur-6996	91	15	in	in	ADP
alusur-6996	91	16	order	order	NOUN
alusur-6996	91	17	to	to	PART
alusur-6996	91	18	make	make	VERB
alusur-6996	91	19	this	this	DET
alusur-6996	91	20	project	project	NOUN
alusur-6996	91	21	sustainable	sustainable	ADJ
alusur-6996	91	22	and	and	CCONJ
alusur-6996	91	23	develop	develop	VERB
alusur-6996	91	24	these	these	DET
alusur-6996	91	25	collections	collection	NOUN
alusur-6996	91	26	of	of	ADP
alusur-6996	91	27	digital	digital	ADJ
alusur-6996	91	28	texts	text	NOUN
alusur-6996	91	29	into	into	ADP
alusur-6996	91	30	a	a	DET
alusur-6996	91	31	representative	representative	ADJ
alusur-6996	91	32	islamicate	islamicate	ADJ
alusur-6996	91	33	corpus	corpus	NOUN
alusur-6996	91	34	—	—	PUNCT
alusur-6996	91	35	a	a	DET
alusur-6996	91	36	laborious	laborious	ADJ
alusur-6996	91	37	process	process	NOUN
alusur-6996	91	38	which	which	PRON
alusur-6996	91	39	involves	involve	VERB
alusur-6996	91	40	the	the	DET
alusur-6996	91	41	expert	expert	ADJ
alusur-6996	91	42	selection	selection	NOUN
alusur-6996	91	43	of	of	ADP
alusur-6996	91	44	new	new	ADJ
alusur-6996	91	45	works	work	NOUN
alusur-6996	91	46	and	and	CCONJ
alusur-6996	91	47	the	the	DET
alusur-6996	91	48	creation	creation	NOUN
alusur-6996	91	49	and	and	CCONJ
alusur-6996	91	50	curation	curation	NOUN
alusur-6996	91	51	of	of	ADP
alusur-6996	91	52	scholarly	scholarly	ADJ
alusur-6996	91	53	metadata	metadata	NOUN
alusur-6996	91	54	.	.	PUNCT
alusur-6996	92	1	however	however	ADV
alusur-6996	92	2	,	,	PUNCT
alusur-6996	92	3	at	at	ADP
alusur-6996	92	4	the	the	DET
alusur-6996	92	5	same	same	ADJ
alusur-6996	92	6	time	time	NOUN
alusur-6996	92	7	,	,	PUNCT
alusur-6996	92	8	the	the	DET
alusur-6996	92	9	possibilities	possibility	NOUN
alusur-6996	92	10	that	that	PRON
alusur-6996	92	11	an	an	DET
alusur-6996	92	12	effective	effective	ADJ
alusur-6996	92	13	open	open	ADJ
alusur-6996	92	14	-	-	PUNCT
alusur-6996	92	15	source	source	NOUN
alusur-6996	92	16	arabic	arabic	ADJ
alusur-6996	92	17	-	-	PUNCT
alusur-6996	92	18	script	script	NOUN
alusur-6996	92	19	ocr	ocr	PROPN
alusur-6996	92	20	program	program	NOUN
alusur-6996	92	21	will	will	AUX
alusur-6996	92	22	open	open	VERB
alusur-6996	92	23	up	up	ADP
alusur-6996	92	24	for	for	ADP
alusur-6996	92	25	islamicate	islamicate	ADJ
alusur-6996	92	26	studies	study	NOUN
alusur-6996	92	27	are	be	AUX
alusur-6996	92	28	difficult	difficult	ADJ
alusur-6996	92	29	to	to	PART
alusur-6996	92	30	overstate	overstate	VERB
alusur-6996	92	31	.	.	PUNCT
alusur-6996	93	1	in	in	ADP
alusur-6996	93	2	addition	addition	NOUN
alusur-6996	93	3	to	to	ADP
alusur-6996	93	4	rendering	render	VERB
alusur-6996	93	5	hundreds	hundred	NOUN
alusur-6996	93	6	,	,	PUNCT
alusur-6996	93	7	even	even	ADV
alusur-6996	93	8	thousands	thousand	NOUN
alusur-6996	93	9	,	,	PUNCT
alusur-6996	93	10	of	of	ADP
alusur-6996	93	11	new	new	ADJ
alusur-6996	93	12	texts	text	NOUN
alusur-6996	93	13	full	full	ADJ
alusur-6996	93	14	-	-	PUNCT
alusur-6996	93	15	text	text	NOUN
alusur-6996	93	16	searchable	searchable	NOUN
alusur-6996	93	17	,	,	PUNCT
alusur-6996	93	18	scholars	scholar	NOUN
alusur-6996	93	19	will	will	AUX
alusur-6996	93	20	be	be	AUX
alusur-6996	93	21	able	able	ADJ
alusur-6996	93	22	to	to	PART
alusur-6996	93	23	employ	employ	VERB
alusur-6996	93	24	computational	computational	ADJ
alusur-6996	93	25	modes	mode	NOUN
alusur-6996	93	26	of	of	ADP
alusur-6996	93	27	text	text	NOUN
alusur-6996	93	28	analysis	analysis	NOUN
alusur-6996	93	29	(	(	PUNCT
alusur-6996	93	30	e.g.	e.g.	ADV
alusur-6996	93	31	,	,	PUNCT
alusur-6996	93	32	text	text	NOUN
alusur-6996	93	33	re	re	NOUN
alusur-6996	93	34	-	-	NOUN
alusur-6996	93	35	use	use	VERB
alusur-6996	93	36	,	,	PUNCT
alusur-6996	93	37	topic	topic	NOUN
alusur-6996	93	38	modeling	modeling	NOUN
alusur-6996	93	39	,	,	PUNCT
alusur-6996	93	40	stylometric	stylometric	ADJ
alusur-6996	93	41	analysis	analysis	NOUN
alusur-6996	93	42	)	)	PUNCT
alusur-6996	93	43	on	on	ADP
alusur-6996	93	44	a	a	DET
alusur-6996	93	45	body	body	NOUN
alusur-6996	93	46	of	of	ADP
alusur-6996	93	47	material	material	NOUN
alusur-6996	93	48	much	much	ADV
alusur-6996	93	49	more	more	ADJ
alusur-6996	93	50	representative	representative	NOUN
alusur-6996	93	51	of	of	ADP
alusur-6996	93	52	the	the	DET
alusur-6996	93	53	historical	historical	ADJ
alusur-6996	93	54	tradition	tradition	NOUN
alusur-6996	93	55	than	than	ADP
alusur-6996	93	56	what	what	PRON
alusur-6996	93	57	we	we	PRON
alusur-6996	93	58	have	have	AUX
alusur-6996	93	59	at	at	ADP
alusur-6996	93	60	this	this	DET
alusur-6996	93	61	moment	moment	NOUN
alusur-6996	93	62	.	.	PUNCT
alusur-6996	94	1	the	the	DET
alusur-6996	94	2	full	full	ADJ
alusur-6996	94	3	impact	impact	NOUN
alusur-6996	94	4	of	of	ADP
alusur-6996	94	5	these	these	DET
alusur-6996	94	6	new	new	ADJ
alusur-6996	94	7	analytical	analytical	ADJ
alusur-6996	94	8	possibilities	possibility	NOUN
alusur-6996	94	9	and	and	CCONJ
alusur-6996	94	10	the	the	DET
alusur-6996	94	11	new	new	ADJ
alusur-6996	94	12	levels	level	NOUN
alusur-6996	94	13	of	of	ADP
alusur-6996	94	14	scale	scale	NOUN
alusur-6996	94	15	and	and	CCONJ
alusur-6996	94	16	specificity	specificity	NOUN
alusur-6996	94	17	in	in	ADP
alusur-6996	94	18	textual	textual	ADJ
alusur-6996	94	19	analysis	analysis	NOUN
alusur-6996	94	20	that	that	PRON
alusur-6996	94	21	they	they	PRON
alusur-6996	94	22	make	make	VERB
alusur-6996	94	23	possible	possible	ADJ
alusur-6996	94	24	are	be	AUX
alusur-6996	94	25	difficult	difficult	ADJ
alusur-6996	94	26	to	to	PART
alusur-6996	94	27	estimate	estimate	VERB
alusur-6996	94	28	at	at	ADP
alusur-6996	94	29	such	such	DET
alusur-6996	94	30	an	an	DET
alusur-6996	94	31	early	early	ADJ
alusur-6996	94	32	stage	stage	NOUN
alusur-6996	94	33	,	,	PUNCT
alusur-6996	94	34	but	but	CCONJ
alusur-6996	94	35	the	the	DET
alusur-6996	94	36	early	early	ADJ
alusur-6996	94	37	results	result	NOUN
alusur-6996	94	38	are	be	AUX
alusur-6996	94	39	promising	promise	VERB
alusur-6996	94	40	.	.	PUNCT
alusur-6996	95	1	4.1	4.1	NUM
alusur-6996	95	2	the	the	DET
alusur-6996	95	3	technical	technical	ADJ
alusur-6996	95	4	details	detail	NOUN
alusur-6996	95	5	:	:	PUNCT
alusur-6996	95	6	kraken	kraken	NOUN
alusur-6996	95	7	and	and	CCONJ
alusur-6996	95	8	its	its	PRON
alusur-6996	95	9	ocr	ocr	ADJ
alusur-6996	95	10	method	method	NOUN
alusur-6996	95	11	kraken	kraken	NOUN
alusur-6996	95	12	is	be	AUX
alusur-6996	95	13	the	the	DET
alusur-6996	95	14	open	open	ADJ
alusur-6996	95	15	-	-	PUNCT
alusur-6996	95	16	source	source	NOUN
alusur-6996	95	17	ocr	ocr	ADJ
alusur-6996	95	18	software	software	NOUN
alusur-6996	95	19	that	that	PRON
alusur-6996	95	20	we	we	PRON
alusur-6996	95	21	used	use	VERB
alusur-6996	95	22	in	in	ADP
alusur-6996	95	23	our	our	PRON
alusur-6996	95	24	tests	test	NOUN
alusur-6996	95	25	.	.	PUNCT
alusur-6996	96	1	developed	develop	VERB
alusur-6996	96	2	by	by	ADP
alusur-6996	96	3	benjamin	benjamin	PROPN
alusur-6996	96	4	kiessling	kiessle	VERB
alusur-6996	96	5	at	at	ADP
alusur-6996	96	6	lu	lu	PROPN
alusur-6996	96	7	’s	’s	PART
alusur-6996	96	8	alexander	alexander	PROPN
alusur-6996	96	9	von	von	PROPN
alusur-6996	96	10	humboldt	humboldt	PROPN
alusur-6996	96	11	chair	chair	PROPN
alusur-6996	96	12	for	for	ADP
alusur-6996	96	13	digital	digital	ADJ
alusur-6996	96	14	humanities	humanity	NOUN
alusur-6996	96	15	,	,	PUNCT
alusur-6996	96	16	kraken	kraken	PROPN
alusur-6996	96	17	is	be	AUX
alusur-6996	96	18	a	a	DET
alusur-6996	96	19	“	"	PUNCT
alusur-6996	96	20	fork”14	fork”14	PROPN
alusur-6996	96	21	of	of	ADP
alusur-6996	96	22	the	the	DET
alusur-6996	96	23	unmaintained	unmaintained	ADJ
alusur-6996	96	24	ocropus	ocropus	NOUN
alusur-6996	96	25	package15	package15	NOUN
alusur-6996	96	26	combined	combine	VERB
alusur-6996	96	27	with	with	ADP
alusur-6996	96	28	the	the	DET
alusur-6996	96	29	clstm	clstm	PROPN
alusur-6996	96	30	neural	neural	ADJ
alusur-6996	96	31	network	network	NOUN
alusur-6996	96	32	library.16	library.16	PROPN
alusur-6996	96	33	kraken	kraken	PROPN
alusur-6996	96	34	represents	represent	VERB
alusur-6996	96	35	a	a	DET
alusur-6996	96	36	substantial	substantial	ADJ
alusur-6996	96	37	improvement	improvement	NOUN
alusur-6996	96	38	over	over	ADP
alusur-6996	96	39	the	the	DET
alusur-6996	96	40	ocropus	ocropus	NOUN
alusur-6996	96	41	package	package	NOUN
alusur-6996	96	42	:	:	PUNCT
alusur-6996	96	43	its	its	PRON
alusur-6996	96	44	performance	performance	NOUN
alusur-6996	96	45	is	be	AUX
alusur-6996	96	46	dramatically	dramatically	ADV
alusur-6996	96	47	better	well	ADJ
alusur-6996	96	48	,	,	PUNCT
alusur-6996	96	49	it	it	PRON
alusur-6996	96	50	supports	support	VERB
alusur-6996	96	51	right	right	ADJ
alusur-6996	96	52	-	-	PUNCT
alusur-6996	96	53	to	to	ADP
alusur-6996	96	54	-	-	PUNCT
alusur-6996	96	55	left	leave	VERB
alusur-6996	96	56	scripts	script	NOUN
alusur-6996	96	57	and	and	CCONJ
alusur-6996	96	58	combined	combine	VERB
alusur-6996	96	59	ltr	ltr	PROPN
alusur-6996	96	60	/	/	SYM
alusur-6996	96	61	rtl	rtl	NOUN
alusur-6996	96	62	(	(	PUNCT
alusur-6996	96	63	bidi	bidi	PROPN
alusur-6996	96	64	)	)	PUNCT
alusur-6996	96	65	texts	text	NOUN
alusur-6996	96	66	,	,	PUNCT
alusur-6996	96	67	and	and	CCONJ
alusur-6996	96	68	it	it	PRON
alusur-6996	96	69	includes	include	VERB
alusur-6996	96	70	a	a	DET
alusur-6996	96	71	rudimentary	rudimentary	ADJ
alusur-6996	96	72	transcription	transcription	NOUN
alusur-6996	96	73	interface	interface	NOUN
alusur-6996	96	74	for	for	ADP
alusur-6996	96	75	offline	offline	ADJ
alusur-6996	96	76	use	use	NOUN
alusur-6996	96	77	.	.	PUNCT
alusur-6996	97	1	13	13	NUM
alusur-6996	97	2	.	.	PUNCT
alusur-6996	98	1	please	please	INTJ
alusur-6996	98	2	contact	contact	VERB
alusur-6996	98	3	us	we	PRON
alusur-6996	98	4	for	for	ADP
alusur-6996	98	5	more	more	ADJ
alusur-6996	98	6	details	detail	NOUN
alusur-6996	98	7	if	if	SCONJ
alusur-6996	98	8	you	you	PRON
alusur-6996	98	9	are	be	AUX
alusur-6996	98	10	interested	interested	ADJ
alusur-6996	98	11	in	in	ADP
alusur-6996	98	12	generating	generate	VERB
alusur-6996	98	13	1,000	1,000	NUM
alusur-6996	98	14	lines	line	NOUN
alusur-6996	98	15	of	of	ADP
alusur-6996	98	16	training	training	NOUN
alusur-6996	98	17	data	datum	NOUN
alusur-6996	98	18	for	for	ADP
alusur-6996	98	19	any	any	DET
alusur-6996	98	20	ottoman	ottoman	NOUN
alusur-6996	98	21	turkish	turkish	NOUN
alusur-6996	98	22	or	or	CCONJ
alusur-6996	98	23	urdu	urdu	PROPN
alusur-6996	98	24	typefaces	typeface	NOUN
alusur-6996	98	25	or	or	CCONJ
alusur-6996	98	26	a	a	DET
alusur-6996	98	27	specific	specific	ADJ
alusur-6996	98	28	arabic	arabic	ADJ
alusur-6996	98	29	or	or	CCONJ
alusur-6996	98	30	persian	persian	ADJ
alusur-6996	98	31	typeface	typeface	NOUN
alusur-6996	98	32	for	for	ADP
alusur-6996	98	33	which	which	PRON
alusur-6996	98	34	we	we	PRON
alusur-6996	98	35	do	do	AUX
alusur-6996	98	36	not	not	PART
alusur-6996	98	37	already	already	ADV
alusur-6996	98	38	have	have	VERB
alusur-6996	98	39	a	a	DET
alusur-6996	98	40	model	model	NOUN
alusur-6996	98	41	trained	train	VERB
alusur-6996	98	42	.	.	PUNCT
alusur-6996	99	1	14	14	NUM
alusur-6996	99	2	.	.	PUNCT
alusur-6996	99	3	“	"	PUNCT
alusur-6996	99	4	fork	fork	NOUN
alusur-6996	99	5	”	"	PUNCT
alusur-6996	99	6	is	be	AUX
alusur-6996	99	7	a	a	DET
alusur-6996	99	8	computer	computer	NOUN
alusur-6996	99	9	-	-	PUNCT
alusur-6996	99	10	science	science	NOUN
alusur-6996	99	11	term	term	NOUN
alusur-6996	99	12	for	for	ADP
alusur-6996	99	13	a	a	DET
alusur-6996	99	14	new	new	ADJ
alusur-6996	99	15	“	"	PUNCT
alusur-6996	99	16	branch	branch	NOUN
alusur-6996	99	17	”	"	PUNCT
alusur-6996	99	18	of	of	ADP
alusur-6996	99	19	independent	independent	ADJ
alusur-6996	99	20	development	development	NOUN
alusur-6996	99	21	that	that	PRON
alusur-6996	99	22	builds	build	VERB
alusur-6996	99	23	on	on	ADP
alusur-6996	99	24	an	an	DET
alusur-6996	99	25	existing	exist	VERB
alusur-6996	99	26	software	software	NOUN
alusur-6996	99	27	.	.	PUNCT
alusur-6996	100	1	15	15	X
alusur-6996	100	2	.	.	X
alusur-6996	101	1	for	for	ADP
alusur-6996	101	2	details	detail	NOUN
alusur-6996	101	3	,	,	PUNCT
alusur-6996	101	4	see	see	VERB
alusur-6996	101	5	:	:	PUNCT
alusur-6996	101	6	https://github.com/tmbdev/ocropy	https://github.com/tmbdev/ocropy	NOUN
alusur-6996	101	7	and	and	CCONJ
alusur-6996	101	8	https://en.wikipedia.org/wiki/ocropus	https://en.wikipedia.org/wiki/ocropus	PROPN
alusur-6996	101	9	16	16	NUM
alusur-6996	101	10	.	.	PUNCT
alusur-6996	102	1	see	see	VERB
alusur-6996	102	2	:	:	PUNCT
alusur-6996	102	3	https://github.com/tmbdev/clstm	https://github.com/tmbdev/clstm	NOUN
alusur-6996	102	4	https://github.com/tmbdev/ocropy	https://github.com/tmbdev/ocropy	PROPN
alusur-6996	102	5	https://en.wikipedia.org/wiki/ocropus	https://en.wikipedia.org/wiki/ocropus	PROPN
alusur-6996	102	6	https://github.com/tmbdev/clstm	https://github.com/tmbdev/clstm	NOUN
alusur-6996	102	7	10	10	NUM
alusur-6996	102	8	•	•	NOUN
alusur-6996	102	9	kiessling	kiessle	VERB
alusur-6996	102	10	,	,	PUNCT
alusur-6996	102	11	miller	miller	PROPN
alusur-6996	102	12	,	,	PUNCT
alusur-6996	102	13	romanov	romanov	PROPN
alusur-6996	102	14	,	,	PUNCT
alusur-6996	102	15	&	&	CCONJ
alusur-6996	102	16	savant	savant	PROPN
alusur-6996	102	17	al-ʿuṣūr	al-ʿuṣūr	PROPN
alusur-6996	102	18	al	al	PROPN
alusur-6996	102	19	-	-	PUNCT
alusur-6996	102	20	wusṭā	wusṭā	NOUN
alusur-6996	102	21	25	25	NUM
alusur-6996	102	22	(	(	PUNCT
alusur-6996	102	23	2017	2017	NUM
alusur-6996	102	24	)	)	PUNCT
alusur-6996	102	25	the	the	DET
alusur-6996	102	26	ocr	ocr	ADJ
alusur-6996	102	27	method	method	NOUN
alusur-6996	102	28	that	that	PRON
alusur-6996	102	29	powers	power	NOUN
alusur-6996	102	30	kraken	kraken	NOUN
alusur-6996	102	31	is	be	AUX
alusur-6996	102	32	based	base	VERB
alusur-6996	102	33	on	on	ADP
alusur-6996	102	34	a	a	DET
alusur-6996	102	35	long	long	ADJ
alusur-6996	102	36	short	short	ADJ
alusur-6996	102	37	-	-	PUNCT
alusur-6996	102	38	term	term	NOUN
alusur-6996	102	39	memory	memory	NOUN
alusur-6996	102	40	(	(	PUNCT
alusur-6996	102	41	hochreiter	hochreiter	PROPN
alusur-6996	102	42	and	and	CCONJ
alusur-6996	102	43	schmidhuber	schmidhuber	NOUN
alusur-6996	102	44	,	,	PUNCT
alusur-6996	102	45	1997	1997	NUM
alusur-6996	102	46	)	)	PUNCT
alusur-6996	102	47	recurrent	recurrent	ADJ
alusur-6996	102	48	neural	neural	ADJ
alusur-6996	102	49	network	network	NOUN
alusur-6996	102	50	utilizing	utilize	VERB
alusur-6996	102	51	the	the	DET
alusur-6996	102	52	connectionist	connectionist	ADJ
alusur-6996	102	53	temporal	temporal	ADJ
alusur-6996	102	54	classification	classification	NOUN
alusur-6996	102	55	objective	objective	ADJ
alusur-6996	102	56	function.17	function.17	NOUN
alusur-6996	102	57	in	in	ADP
alusur-6996	102	58	contrast	contrast	NOUN
alusur-6996	102	59	to	to	ADP
alusur-6996	102	60	other	other	ADJ
alusur-6996	102	61	systems	system	NOUN
alusur-6996	102	62	requiring	require	VERB
alusur-6996	102	63	character	character	NOUN
alusur-6996	102	64	-	-	PUNCT
alusur-6996	102	65	level	level	NOUN
alusur-6996	102	66	segmentation	segmentation	NOUN
alusur-6996	102	67	before	before	ADP
alusur-6996	102	68	classification	classification	NOUN
alusur-6996	102	69	,	,	PUNCT
alusur-6996	102	70	it	it	PRON
alusur-6996	102	71	is	be	AUX
alusur-6996	102	72	uniquely	uniquely	ADV
alusur-6996	102	73	suited	suited	ADJ
alusur-6996	102	74	for	for	ADP
alusur-6996	102	75	the	the	DET
alusur-6996	102	76	recognition	recognition	NOUN
alusur-6996	102	77	of	of	ADP
alusur-6996	102	78	connected	connect	VERB
alusur-6996	102	79	arabographic	arabographic	ADJ
alusur-6996	102	80	scripts	script	NOUN
alusur-6996	102	81	because	because	SCONJ
alusur-6996	102	82	the	the	DET
alusur-6996	102	83	objective	objective	ADJ
alusur-6996	102	84	function	function	NOUN
alusur-6996	102	85	used	use	VERB
alusur-6996	102	86	during	during	ADP
alusur-6996	102	87	training	training	NOUN
alusur-6996	102	88	is	be	AUX
alusur-6996	102	89	geared	gear	VERB
alusur-6996	102	90	towards	towards	ADP
alusur-6996	102	91	assigning	assign	VERB
alusur-6996	102	92	labels	label	NOUN
alusur-6996	102	93	—	—	PUNCT
alusur-6996	102	94	i.e.	i.e.	X
alusur-6996	102	95	,	,	PUNCT
alusur-6996	102	96	characters	character	NOUN
alusur-6996	102	97	/	/	SYM
alusur-6996	102	98	glyphs	glyph	NOUN
alusur-6996	102	99	—	—	PUNCT
alusur-6996	102	100	to	to	ADP
alusur-6996	102	101	regions	region	NOUN
alusur-6996	102	102	of	of	ADP
alusur-6996	102	103	unsegmented	unsegmented	ADJ
alusur-6996	102	104	input	input	NOUN
alusur-6996	102	105	data	datum	NOUN
alusur-6996	102	106	.	.	PUNCT
alusur-6996	103	1	the	the	DET
alusur-6996	103	2	system	system	NOUN
alusur-6996	103	3	works	work	VERB
alusur-6996	103	4	on	on	ADP
alusur-6996	103	5	unsegmented	unsegmented	ADJ
alusur-6996	103	6	data	datum	NOUN
alusur-6996	103	7	both	both	CCONJ
alusur-6996	103	8	during	during	ADP
alusur-6996	103	9	training	training	NOUN
alusur-6996	103	10	and	and	CCONJ
alusur-6996	103	11	recognition	recognition	NOUN
alusur-6996	103	12	—	—	PUNCT
alusur-6996	103	13	its	its	PRON
alusur-6996	103	14	base	base	NOUN
alusur-6996	103	15	unit	unit	NOUN
alusur-6996	103	16	is	be	AUX
alusur-6996	103	17	a	a	DET
alusur-6996	103	18	text	text	NOUN
alusur-6996	103	19	line	line	NOUN
alusur-6996	103	20	(	(	PUNCT
alusur-6996	103	21	line	line	NOUN
alusur-6996	103	22	recognizer	recognizer	NOUN
alusur-6996	103	23	)	)	PUNCT
alusur-6996	103	24	.	.	PUNCT
alusur-6996	104	1	for	for	ADP
alusur-6996	104	2	training	training	NOUN
alusur-6996	104	3	,	,	PUNCT
alusur-6996	104	4	a	a	DET
alusur-6996	104	5	number	number	NOUN
alusur-6996	104	6	of	of	ADP
alusur-6996	104	7	printed	print	VERB
alusur-6996	104	8	lines	line	NOUN
alusur-6996	104	9	have	have	VERB
alusur-6996	104	10	to	to	PART
alusur-6996	104	11	be	be	AUX
alusur-6996	104	12	transcribed	transcribe	VERB
alusur-6996	104	13	using	use	VERB
alusur-6996	104	14	a	a	DET
alusur-6996	104	15	simple	simple	ADJ
alusur-6996	104	16	html	html	NOUN
alusur-6996	104	17	transcription	transcription	NOUN
alusur-6996	104	18	interface	interface	NOUN
alusur-6996	104	19	(	(	PUNCT
alusur-6996	104	20	see	see	VERB
alusur-6996	104	21	figure	figure	NOUN
alusur-6996	104	22	2	2	NUM
alusur-6996	104	23	above	above	ADP
alusur-6996	104	24	)	)	PUNCT
alusur-6996	104	25	.	.	PUNCT
alusur-6996	105	1	the	the	DET
alusur-6996	105	2	total	total	ADJ
alusur-6996	105	3	amount	amount	NOUN
alusur-6996	105	4	of	of	ADP
alusur-6996	105	5	training	training	NOUN
alusur-6996	105	6	data	datum	NOUN
alusur-6996	105	7	(	(	PUNCT
alusur-6996	105	8	i.e.	i.e.	X
alusur-6996	105	9	,	,	PUNCT
alusur-6996	105	10	line	line	NOUN
alusur-6996	105	11	image	image	NOUN
alusur-6996	105	12	-	-	PUNCT
alusur-6996	105	13	text	text	NOUN
alusur-6996	105	14	pairs	pair	NOUN
alusur-6996	105	15	)	)	PUNCT
alusur-6996	105	16	required	require	VERB
alusur-6996	105	17	may	may	AUX
alusur-6996	105	18	vary	vary	VERB
alusur-6996	105	19	depending	depend	VERB
alusur-6996	105	20	on	on	ADP
alusur-6996	105	21	the	the	DET
alusur-6996	105	22	complexity	complexity	NOUN
alusur-6996	105	23	of	of	ADP
alusur-6996	105	24	the	the	DET
alusur-6996	105	25	typeface	typeface	NOUN
alusur-6996	105	26	and	and	CCONJ
alusur-6996	105	27	number	number	NOUN
alusur-6996	105	28	of	of	ADP
alusur-6996	105	29	glyphs	glyph	NOUN
alusur-6996	105	30	used	use	VERB
alusur-6996	105	31	by	by	ADP
alusur-6996	105	32	the	the	DET
alusur-6996	105	33	script	script	NOUN
alusur-6996	105	34	.	.	PUNCT
alusur-6996	106	1	acquisition	acquisition	NOUN
alusur-6996	106	2	of	of	ADP
alusur-6996	106	3	training	training	NOUN
alusur-6996	106	4	data	datum	NOUN
alusur-6996	106	5	can	can	AUX
alusur-6996	106	6	be	be	AUX
alusur-6996	106	7	optimized	optimize	VERB
alusur-6996	106	8	by	by	ADP
alusur-6996	106	9	line	line	NOUN
alusur-6996	106	10	-	-	PUNCT
alusur-6996	106	11	wise	wise	ADJ
alusur-6996	106	12	alignment	alignment	NOUN
alusur-6996	106	13	of	of	ADP
alusur-6996	106	14	existing	exist	VERB
alusur-6996	106	15	digital	digital	ADJ
alusur-6996	106	16	editions	edition	NOUN
alusur-6996	106	17	with	with	ADP
alusur-6996	106	18	printed	print	VERB
alusur-6996	106	19	lines	line	NOUN
alusur-6996	106	20	,	,	PUNCT
alusur-6996	106	21	although	although	SCONJ
alusur-6996	106	22	even	even	ADV
alusur-6996	106	23	wholesale	wholesale	ADJ
alusur-6996	106	24	transcription	transcription	NOUN
alusur-6996	106	25	is	be	AUX
alusur-6996	106	26	a	a	DET
alusur-6996	106	27	faster	fast	ADJ
alusur-6996	106	28	and	and	CCONJ
alusur-6996	106	29	relatively	relatively	ADV
alusur-6996	106	30	unskilled	unskilled	ADJ
alusur-6996	106	31	task	task	NOUN
alusur-6996	106	32	in	in	ADP
alusur-6996	106	33	comparison	comparison	NOUN
alusur-6996	106	34	to	to	ADP
alusur-6996	106	35	training	train	VERB
alusur-6996	106	36	data	datum	NOUN
alusur-6996	106	37	creation	creation	NOUN
alusur-6996	106	38	for	for	ADP
alusur-6996	106	39	other	other	ADJ
alusur-6996	106	40	systems	system	NOUN
alusur-6996	106	41	such	such	ADJ
alusur-6996	106	42	as	as	ADP
alusur-6996	106	43	tesseract.18	tesseract.18	PROPN
alusur-6996	106	44	our	our	PRON
alusur-6996	106	45	current	current	ADJ
alusur-6996	106	46	models	model	NOUN
alusur-6996	106	47	were	be	AUX
alusur-6996	106	48	trained	train	VERB
alusur-6996	106	49	on	on	ADP
alusur-6996	106	50	~1,000	~1,000	NUM
alusur-6996	106	51	pairs	pair	NOUN
alusur-6996	106	52	each	each	PRON
alusur-6996	106	53	,	,	PUNCT
alusur-6996	106	54	corresponding	correspond	VERB
alusur-6996	106	55	to	to	ADP
alusur-6996	106	56	~50	~50	NUM
alusur-6996	106	57	-	-	PUNCT
alusur-6996	106	58	60	60	NUM
alusur-6996	106	59	pages	page	NOUN
alusur-6996	106	60	of	of	ADP
alusur-6996	106	61	printed	print	VERB
alusur-6996	106	62	text	text	NOUN
alusur-6996	106	63	.	.	PUNCT
alusur-6996	107	1	models	model	NOUN
alusur-6996	107	2	are	be	AUX
alusur-6996	107	3	fairly	fairly	ADV
alusur-6996	107	4	typographically	typographically	ADV
alusur-6996	107	5	specific	specific	ADJ
alusur-6996	107	6	,	,	PUNCT
alusur-6996	107	7	the	the	DET
alusur-6996	107	8	most	most	ADV
alusur-6996	107	9	important	important	ADJ
alusur-6996	107	10	factor	factor	NOUN
alusur-6996	107	11	being	be	AUX
alusur-6996	107	12	fonts	font	NOUN
alusur-6996	107	13	and	and	CCONJ
alusur-6996	107	14	spacing	spacing	NOUN
alusur-6996	107	15	,	,	PUNCT
alusur-6996	107	16	although	although	SCONJ
alusur-6996	107	17	some	some	DET
alusur-6996	107	18	mismatch	mismatch	NOUN
alusur-6996	107	19	does	do	AUX
alusur-6996	107	20	not	not	PART
alusur-6996	107	21	degrade	degrade	VERB
alusur-6996	107	22	recognition	recognition	NOUN
alusur-6996	107	23	accuracy	accuracy	NOUN
alusur-6996	107	24	substantially	substantially	ADV
alusur-6996	107	25	(	(	PUNCT
alusur-6996	107	26	2	2	NUM
alusur-6996	107	27	-	-	SYM
alusur-6996	107	28	5%).19	5%).19	NUM
alusur-6996	107	29	thus	thus	ADV
alusur-6996	107	30	new	new	ADJ
alusur-6996	107	31	training	training	NOUN
alusur-6996	107	32	data	datum	NOUN
alusur-6996	107	33	for	for	ADP
alusur-6996	107	34	an	an	DET
alusur-6996	107	35	unknown	unknown	ADJ
alusur-6996	107	36	typeface	typeface	NOUN
alusur-6996	107	37	can	can	AUX
alusur-6996	107	38	be	be	AUX
alusur-6996	107	39	produced	produce	VERB
alusur-6996	107	40	by	by	ADP
alusur-6996	107	41	correcting	correct	VERB
alusur-6996	107	42	the	the	DET
alusur-6996	107	43	output	output	NOUN
alusur-6996	107	44	from	from	ADP
alusur-6996	107	45	a	a	DET
alusur-6996	107	46	model	model	NOUN
alusur-6996	107	47	for	for	ADP
alusur-6996	107	48	a	a	DET
alusur-6996	107	49	similar	similar	ADJ
alusur-6996	107	50	font	font	NOUN
alusur-6996	107	51	—	—	PUNCT
alusur-6996	107	52	in	in	ADP
alusur-6996	107	53	other	other	ADJ
alusur-6996	107	54	words	word	NOUN
alusur-6996	107	55	,	,	PUNCT
alusur-6996	107	56	generating	generate	VERB
alusur-6996	107	57	training	training	NOUN
alusur-6996	107	58	data	datum	NOUN
alusur-6996	107	59	for	for	ADP
alusur-6996	107	60	every	every	DET
alusur-6996	107	61	subsequent	subsequent	ADJ
alusur-6996	107	62	model	model	NOUN
alusur-6996	107	63	will	will	AUX
alusur-6996	107	64	require	require	VERB
alusur-6996	107	65	less	less	ADJ
alusur-6996	107	66	and	and	CCONJ
alusur-6996	107	67	less	less	ADJ
alusur-6996	107	68	time	time	NOUN
alusur-6996	107	69	.	.	PUNCT
alusur-6996	108	1	last	last	ADJ
alusur-6996	108	2	but	but	CCONJ
alusur-6996	108	3	not	not	PART
alusur-6996	108	4	least	least	ADJ
alusur-6996	108	5	,	,	PUNCT
alusur-6996	108	6	it	it	PRON
alusur-6996	108	7	is	be	AUX
alusur-6996	108	8	also	also	ADV
alusur-6996	108	9	possible	possible	ADJ
alusur-6996	108	10	to	to	PART
alusur-6996	108	11	train	train	VERB
alusur-6996	108	12	multi	multi	ADJ
alusur-6996	108	13	-	-	NOUN
alusur-6996	108	14	typeface	typeface	ADJ
alusur-6996	108	15	(	(	PUNCT
alusur-6996	108	16	so	so	ADV
alusur-6996	108	17	-	-	PUNCT
alusur-6996	108	18	called	call	VERB
alusur-6996	108	19	,	,	PUNCT
alusur-6996	108	20	“	"	PUNCT
alusur-6996	108	21	generalized	generalize	VERB
alusur-6996	108	22	”	"	PUNCT
alusur-6996	108	23	)	)	PUNCT
alusur-6996	108	24	models	model	NOUN
alusur-6996	108	25	by	by	ADP
alusur-6996	108	26	simply	simply	ADV
alusur-6996	108	27	combining	combine	VERB
alusur-6996	108	28	training	training	NOUN
alusur-6996	108	29	data	datum	NOUN
alusur-6996	108	30	,	,	PUNCT
alusur-6996	108	31	albeit	albeit	SCONJ
alusur-6996	108	32	some	some	DET
alusur-6996	108	33	parameter	parameter	NOUN
alusur-6996	108	34	tuning	tuning	NOUN
alusur-6996	108	35	is	be	AUX
alusur-6996	108	36	required	require	VERB
alusur-6996	108	37	to	to	PART
alusur-6996	108	38	account	account	VERB
alusur-6996	108	39	for	for	ADP
alusur-6996	108	40	the	the	DET
alusur-6996	108	41	richer	rich	ADJ
alusur-6996	108	42	typographic	typographic	ADJ
alusur-6996	108	43	morphology	morphology	NOUN
alusur-6996	108	44	that	that	SCONJ
alusur-6996	108	45	the	the	DET
alusur-6996	108	46	neural	neural	ADJ
alusur-6996	108	47	network	network	NOUN
alusur-6996	108	48	must	must	AUX
alusur-6996	108	49	learn	learn	VERB
alusur-6996	108	50	.	.	PUNCT
alusur-6996	109	1	5.1	5.1	NUM
alusur-6996	109	2	acknowledgements	acknowledgement	NOUN
alusur-6996	109	3	we	we	PRON
alusur-6996	109	4	would	would	AUX
alusur-6996	109	5	never	never	ADV
alusur-6996	109	6	have	have	AUX
alusur-6996	109	7	been	be	AUX
alusur-6996	109	8	able	able	ADJ
alusur-6996	109	9	to	to	PART
alusur-6996	109	10	complete	complete	VERB
alusur-6996	109	11	this	this	DET
alusur-6996	109	12	work	work	NOUN
alusur-6996	109	13	without	without	ADP
alusur-6996	109	14	the	the	DET
alusur-6996	109	15	help	help	NOUN
alusur-6996	109	16	of	of	ADP
alusur-6996	109	17	our	our	PRON
alusur-6996	109	18	team	team	NOUN
alusur-6996	109	19	members	member	NOUN
alusur-6996	109	20	at	at	ADP
alusur-6996	109	21	leipzig	leipzig	PROPN
alusur-6996	109	22	university	university	PROPN
alusur-6996	109	23	,	,	PUNCT
alusur-6996	109	24	university	university	PROPN
alusur-6996	109	25	of	of	ADP
alusur-6996	109	26	maryland	maryland	PROPN
alusur-6996	109	27	(	(	PUNCT
alusur-6996	109	28	college	college	NOUN
alusur-6996	109	29	park	park	NOUN
alusur-6996	109	30	)	)	PUNCT
alusur-6996	109	31	,	,	PUNCT
alusur-6996	109	32	and	and	CCONJ
alusur-6996	109	33	aga	aga	PROPN
alusur-6996	109	34	khan	khan	PROPN
alusur-6996	109	35	university	university	PROPN
alusur-6996	109	36	,	,	PUNCT
alusur-6996	109	37	london	london	PROPN
alusur-6996	109	38	.	.	PUNCT
alusur-6996	110	1	we	we	PRON
alusur-6996	110	2	would	would	AUX
alusur-6996	110	3	also	also	ADV
alusur-6996	110	4	like	like	VERB
alusur-6996	110	5	to	to	PART
alusur-6996	110	6	thank	thank	VERB
alusur-6996	110	7	elijah	elijah	PROPN
alusur-6996	110	8	cooke	cooke	PROPN
alusur-6996	110	9	(	(	PUNCT
alusur-6996	110	10	roshan	roshan	PROPN
alusur-6996	110	11	institute	institute	PROPN
alusur-6996	110	12	,	,	PUNCT
alusur-6996	110	13	umd	umd	PROPN
alusur-6996	110	14	)	)	PUNCT
alusur-6996	110	15	for	for	ADP
alusur-6996	110	16	helping	help	VERB
alusur-6996	110	17	us	we	PRON
alusur-6996	110	18	to	to	PART
alusur-6996	110	19	process	process	VERB
alusur-6996	110	20	the	the	DET
alusur-6996	110	21	data	datum	NOUN
alusur-6996	110	22	,	,	PUNCT
alusur-6996	110	23	samar	samar	PROPN
alusur-6996	110	24	ata	ata	PROPN
alusur-6996	110	25	(	(	PUNCT
alusur-6996	110	26	roshan	roshan	PROPN
alusur-6996	110	27	institute	institute	PROPN
alusur-6996	110	28	,	,	PUNCT
alusur-6996	110	29	umd	umd	PROPN
alusur-6996	110	30	)	)	PUNCT
alusur-6996	110	31	for	for	ADP
alusur-6996	110	32	generating	generate	VERB
alusur-6996	110	33	several	several	ADJ
alusur-6996	110	34	sets	set	NOUN
alusur-6996	110	35	of	of	ADP
alusur-6996	110	36	high	high	ADJ
alusur-6996	110	37	-	-	PUNCT
alusur-6996	110	38	quality	quality	NOUN
alusur-6996	110	39	scans	scan	NOUN
alusur-6996	110	40	for	for	ADP
alusur-6996	110	41	us	we	PRON
alusur-6996	110	42	,	,	PUNCT
alusur-6996	110	43	and	and	CCONJ
alusur-6996	110	44	layal	layal	X
alusur-6996	110	45	mohammad	mohammad	PROPN
alusur-6996	110	46	(	(	PUNCT
alusur-6996	110	47	aku	aku	PROPN
alusur-6996	110	48	,	,	PUNCT
alusur-6996	110	49	ismc	ismc	PROPN
alusur-6996	110	50	)	)	PUNCT
alusur-6996	110	51	,	,	PUNCT
alusur-6996	110	52	mohammad	mohammad	PROPN
alusur-6996	110	53	meqdad	meqdad	PROPN
alusur-6996	110	54	(	(	PUNCT
alusur-6996	110	55	aku	aku	PROPN
alusur-6996	110	56	,	,	PUNCT
alusur-6996	110	57	ismc	ismc	PROPN
alusur-6996	110	58	)	)	PUNCT
alusur-6996	110	59	,	,	PUNCT
alusur-6996	110	60	and	and	CCONJ
alusur-6996	110	61	fatemeh	fatemeh	VERB
alusur-6996	110	62	shams	sham	NOUN
alusur-6996	110	63	(	(	PUNCT
alusur-6996	110	64	aku	aku	PROPN
alusur-6996	110	65	,	,	PUNCT
alusur-6996	110	66	ismc	ismc	NOUN
alusur-6996	110	67	)	)	PUNCT
alusur-6996	110	68	for	for	ADP
alusur-6996	110	69	helping	help	VERB
alusur-6996	110	70	us	we	PRON
alusur-6996	110	71	to	to	PART
alusur-6996	110	72	generate	generate	VERB
alusur-6996	110	73	and	and	CCONJ
alusur-6996	110	74	double	double	ADJ
alusur-6996	110	75	check	check	VERB
alusur-6996	110	76	the	the	DET
alusur-6996	110	77	training	training	NOUN
alusur-6996	110	78	data	datum	NOUN
alusur-6996	110	79	.	.	PUNCT
alusur-6996	111	1	lastly	lastly	ADV
alusur-6996	111	2	,	,	PUNCT
alusur-6996	111	3	we	we	PRON
alusur-6996	111	4	would	would	AUX
alusur-6996	111	5	like	like	VERB
alusur-6996	111	6	to	to	PART
alusur-6996	111	7	express	express	VERB
alusur-6996	111	8	our	our	PRON
alusur-6996	111	9	gratitude	gratitude	NOUN
alusur-6996	111	10	to	to	ADP
alusur-6996	111	11	gregory	gregory	PROPN
alusur-6996	111	12	crane	crane	PROPN
alusur-6996	111	13	(	(	PUNCT
alusur-6996	111	14	alexander	alexander	PROPN
alusur-6996	111	15	von	von	PROPN
alusur-6996	111	16	humboldt	humboldt	PROPN
alusur-6996	111	17	chair	chair	PROPN
alusur-6996	111	18	for	for	ADP
alusur-6996	111	19	digital	digital	ADJ
alusur-6996	111	20	humanities	humanity	NOUN
alusur-6996	111	21	,	,	PUNCT
alusur-6996	111	22	lu	lu	PROPN
alusur-6996	111	23	)	)	PUNCT
alusur-6996	111	24	,	,	PUNCT
alusur-6996	111	25	fatemeh	fatemeh	PROPN
alusur-6996	111	26	keshavarz	keshavarz	PROPN
alusur-6996	111	27	(	(	PUNCT
alusur-6996	111	28	roshan	roshan	PROPN
alusur-6996	111	29	institute	institute	PROPN
alusur-6996	111	30	for	for	ADP
alusur-6996	111	31	persian	persian	PROPN
alusur-6996	111	32	studies	study	NOUN
alusur-6996	111	33	,	,	PUNCT
alusur-6996	111	34	umd	umd	NOUN
alusur-6996	111	35	)	)	PUNCT
alusur-6996	111	36	,	,	PUNCT
alusur-6996	111	37	and	and	CCONJ
alusur-6996	111	38	david	david	PROPN
alusur-6996	111	39	taylor	taylor	PROPN
alusur-6996	111	40	(	(	PUNCT
alusur-6996	111	41	aku	aku	PROPN
alusur-6996	111	42	,	,	PUNCT
alusur-6996	111	43	ismc	ismc	NOUN
alusur-6996	111	44	)	)	PUNCT
alusur-6996	111	45	for	for	ADP
alusur-6996	111	46	their	their	PRON
alusur-6996	111	47	guidance	guidance	NOUN
alusur-6996	111	48	and	and	CCONJ
alusur-6996	111	49	support	support	NOUN
alusur-6996	111	50	of	of	ADP
alusur-6996	111	51	our	our	PRON
alusur-6996	111	52	work	work	NOUN
alusur-6996	111	53	.	.	PUNCT
alusur-6996	112	1	17	17	NUM
alusur-6996	112	2	.	.	PUNCT
alusur-6996	113	1	graves	grave	NOUN
alusur-6996	113	2	et	et	PROPN
alusur-6996	113	3	al	al	PROPN
alusur-6996	113	4	.	.	PROPN
alusur-6996	113	5	,	,	PUNCT
alusur-6996	113	6	2006	2006	NUM
alusur-6996	113	7	,	,	PUNCT
alusur-6996	113	8	as	as	SCONJ
alusur-6996	113	9	elaborated	elaborate	VERB
alusur-6996	113	10	in	in	ADP
alusur-6996	113	11	breuel	breuel	NOUN
alusur-6996	113	12	et	et	PROPN
alusur-6996	113	13	al	al	PROPN
alusur-6996	113	14	.	.	PROPN
alusur-6996	113	15	,	,	PUNCT
alusur-6996	113	16	2013	2013	NUM
alusur-6996	113	17	.	.	PUNCT
alusur-6996	114	1	18	18	NUM
alusur-6996	114	2	.	.	X
alusur-6996	115	1	see	see	VERB
alusur-6996	115	2	:	:	PUNCT
alusur-6996	115	3	https://github.com/tesseract-ocr	https://github.com/tesseract-ocr	ADJ
alusur-6996	115	4	and	and	CCONJ
alusur-6996	115	5	https://en.wikipedia.org/wiki/tesseract_(software	https://en.wikipedia.org/wiki/tesseract_(software	ADJ
alusur-6996	115	6	)	)	PUNCT
alusur-6996	115	7	19	19	NUM
alusur-6996	115	8	.	.	PUNCT
alusur-6996	116	1	for	for	ADP
alusur-6996	116	2	example	example	NOUN
alusur-6996	116	3	,	,	PUNCT
alusur-6996	116	4	if	if	SCONJ
alusur-6996	116	5	a	a	DET
alusur-6996	116	6	glyph	glyph	NOUN
alusur-6996	116	7	is	be	AUX
alusur-6996	116	8	in	in	ADP
alusur-6996	116	9	a	a	DET
alusur-6996	116	10	slightly	slightly	ADV
alusur-6996	116	11	different	different	ADJ
alusur-6996	116	12	font	font	NOUN
alusur-6996	116	13	than	than	ADP
alusur-6996	116	14	the	the	DET
alusur-6996	116	15	one	one	NOUN
alusur-6996	116	16	that	that	PRON
alusur-6996	116	17	the	the	DET
alusur-6996	116	18	model	model	NOUN
alusur-6996	116	19	was	be	AUX
alusur-6996	116	20	trained	train	VERB
alusur-6996	116	21	on	on	ADP
alusur-6996	116	22	,	,	PUNCT
alusur-6996	116	23	it	it	PRON
alusur-6996	116	24	may	may	AUX
alusur-6996	116	25	sometimes	sometimes	ADV
alusur-6996	116	26	be	be	AUX
alusur-6996	116	27	misrecognized	misrecognize	VERB
alusur-6996	116	28	as	as	ADP
alusur-6996	116	29	another	another	DET
alusur-6996	116	30	one	one	NOUN
alusur-6996	116	31	(	(	PUNCT
alusur-6996	116	32	or	or	CCONJ
alusur-6996	116	33	not	not	PART
alusur-6996	116	34	at	at	ADV
alusur-6996	116	35	all	all	ADV
alusur-6996	116	36	)	)	PUNCT
alusur-6996	116	37	,	,	PUNCT
alusur-6996	116	38	thus	thus	ADV
alusur-6996	116	39	leading	lead	VERB
alusur-6996	116	40	the	the	DET
alusur-6996	116	41	overall	overall	ADJ
alusur-6996	116	42	accuracy	accuracy	NOUN
alusur-6996	116	43	rate	rate	NOUN
alusur-6996	116	44	to	to	PART
alusur-6996	116	45	be	be	AUX
alusur-6996	116	46	slightly	slightly	ADV
alusur-6996	116	47	lower	low	ADJ
alusur-6996	116	48	despite	despite	SCONJ
alusur-6996	116	49	the	the	DET
alusur-6996	116	50	fact	fact	NOUN
alusur-6996	116	51	that	that	SCONJ
alusur-6996	116	52	most	most	ADJ
alusur-6996	116	53	of	of	ADP
alusur-6996	116	54	the	the	DET
alusur-6996	116	55	other	other	ADJ
alusur-6996	116	56	text	text	NOUN
alusur-6996	116	57	is	be	AUX
alusur-6996	116	58	recognized	recognize	VERB
alusur-6996	116	59	correctly	correctly	ADV
alusur-6996	116	60	.	.	PUNCT
alusur-6996	117	1	https://github.com/tmbdev/clstm	https://github.com/tmbdev/clstm	NOUN
alusur-6996	117	2	https://en.wikipedia.org/wiki/tesseract_%28software%29	https://en.wikipedia.org/wiki/tesseract_%28software%29	PROPN
alusur-6996	117	3	al-ʿuṣūr	al-ʿuṣūr	PROPN
alusur-6996	117	4	al	al	PROPN
alusur-6996	117	5	-	-	PUNCT
alusur-6996	117	6	wusṭā	wusṭā	NOUN
alusur-6996	117	7	25	25	NUM
alusur-6996	117	8	(	(	PUNCT
alusur-6996	117	9	2017	2017	NUM
alusur-6996	117	10	)	)	PUNCT
alusur-6996	117	11	new	new	ADJ
alusur-6996	117	12	developments	development	NOUN
alusur-6996	117	13	in	in	ADP
alusur-6996	117	14	arabographic	arabographic	ADJ
alusur-6996	117	15	optical	optical	ADJ
alusur-6996	117	16	character	character	NOUN
alusur-6996	117	17	recognition	recognition	NOUN
alusur-6996	117	18	•	•	NOUN
alusur-6996	117	19	11	11	NUM
alusur-6996	117	20	bibliography	bibliography	NOUN
alusur-6996	117	21	and	and	CCONJ
alusur-6996	117	22	links	link	VERB
alusur-6996	117	23	computer	computer	NOUN
alusur-6996	117	24	-	-	PUNCT
alusur-6996	117	25	science	science	NOUN
alusur-6996	117	26	bibliography	bibliography	NOUN
alusur-6996	117	27	hochreiter	hochreiter	PROPN
alusur-6996	117	28	,	,	PUNCT
alusur-6996	117	29	sepp	sepp	PROPN
alusur-6996	117	30	,	,	PUNCT
alusur-6996	117	31	and	and	CCONJ
alusur-6996	117	32	jürgen	jürgen	PROPN
alusur-6996	117	33	schmidhuber	schmidhuber	PROPN
alusur-6996	117	34	.	.	PUNCT
alusur-6996	118	1	“	"	PUNCT
alusur-6996	118	2	long	long	ADJ
alusur-6996	118	3	short	short	ADJ
alusur-6996	118	4	-	-	PUNCT
alusur-6996	118	5	term	term	NOUN
alusur-6996	118	6	memory	memory	NOUN
alusur-6996	118	7	.	.	PUNCT
alusur-6996	118	8	”	"	PUNCT
alusur-6996	119	1	neural	neural	ADJ
alusur-6996	119	2	computation	computation	NOUN
alusur-6996	119	3	9.8	9.8	NUM
alusur-6996	119	4	(	(	PUNCT
alusur-6996	119	5	1997	1997	NUM
alusur-6996	119	6	):	):	PUNCT
alusur-6996	119	7	1735	1735	NUM
alusur-6996	119	8	-	-	SYM
alusur-6996	119	9	1780	1780	NUM
alusur-6996	119	10	.	.	PUNCT
alusur-6996	120	1	graves	grave	NOUN
alusur-6996	120	2	,	,	PUNCT
alusur-6996	120	3	alex	alex	PROPN
alusur-6996	120	4	,	,	PUNCT
alusur-6996	120	5	et	et	PROPN
alusur-6996	120	6	al	al	PROPN
alusur-6996	120	7	.	.	PUNCT
alusur-6996	121	1	“	"	PUNCT
alusur-6996	121	2	connectionist	connectionist	ADJ
alusur-6996	121	3	temporal	temporal	ADJ
alusur-6996	121	4	classification	classification	NOUN
alusur-6996	121	5	:	:	PUNCT
alusur-6996	121	6	labelling	label	VERB
alusur-6996	121	7	unsegmented	unsegmente	VERB
alusur-6996	121	8	sequence	sequence	NOUN
alusur-6996	121	9	data	datum	NOUN
alusur-6996	121	10	with	with	ADP
alusur-6996	121	11	recurrent	recurrent	ADJ
alusur-6996	121	12	neural	neural	ADJ
alusur-6996	121	13	networks	network	NOUN
alusur-6996	121	14	,	,	PUNCT
alusur-6996	121	15	”	"	PUNCT
alusur-6996	121	16	proceedings	proceeding	NOUN
alusur-6996	121	17	of	of	ADP
alusur-6996	121	18	the	the	DET
alusur-6996	121	19	23rd	23rd	ADJ
alusur-6996	121	20	international	international	ADJ
alusur-6996	121	21	conference	conference	NOUN
alusur-6996	121	22	on	on	ADP
alusur-6996	121	23	machine	machine	NOUN
alusur-6996	121	24	learning	learning	NOUN
alusur-6996	121	25	.	.	PUNCT
alusur-6996	122	1	acm	acm	PROPN
alusur-6996	122	2	,	,	PUNCT
alusur-6996	122	3	2006	2006	NUM
alusur-6996	122	4	.	.	PUNCT
alusur-6996	123	1	breuel	breuel	PROPN
alusur-6996	123	2	,	,	PUNCT
alusur-6996	123	3	thomas	thomas	PROPN
alusur-6996	123	4	m.	m.	PROPN
alusur-6996	123	5	,	,	PUNCT
alusur-6996	123	6	et	et	PROPN
alusur-6996	123	7	al	al	PROPN
alusur-6996	123	8	.	.	PUNCT
alusur-6996	124	1	“	"	PUNCT
alusur-6996	124	2	high	high	ADJ
alusur-6996	124	3	-	-	PUNCT
alusur-6996	124	4	performance	performance	NOUN
alusur-6996	124	5	ocr	ocr	NOUN
alusur-6996	124	6	for	for	ADP
alusur-6996	124	7	printed	print	VERB
alusur-6996	124	8	english	english	PROPN
alusur-6996	124	9	and	and	CCONJ
alusur-6996	124	10	fraktur	fraktur	NOUN
alusur-6996	124	11	using	use	VERB
alusur-6996	124	12	lstm	lstm	ADJ
alusur-6996	124	13	networks	network	NOUN
alusur-6996	124	14	,	,	PUNCT
alusur-6996	124	15	”	"	PUNCT
alusur-6996	124	16	12th	12th	ADJ
alusur-6996	124	17	international	international	ADJ
alusur-6996	124	18	conference	conference	NOUN
alusur-6996	124	19	on	on	ADP
alusur-6996	124	20	document	document	NOUN
alusur-6996	124	21	analysis	analysis	NOUN
alusur-6996	124	22	and	and	CCONJ
alusur-6996	124	23	recognition	recognition	NOUN
alusur-6996	124	24	.	.	PUNCT
alusur-6996	125	1	ieee	ieee	PROPN
alusur-6996	125	2	,	,	PUNCT
alusur-6996	125	3	2013	2013	NUM
alusur-6996	125	4	.	.	PUNCT
alusur-6996	126	1	clstm	clstm	NOUN
alusur-6996	126	2	neural	neural	ADJ
alusur-6996	126	3	network	network	NOUN
alusur-6996	126	4	library	library	NOUN
alusur-6996	126	5	:	:	PUNCT
alusur-6996	126	6	https://github.com/tmbdev/clstm	https://github.com/tmbdev/clstm	NOUN
alusur-6996	126	7	links	link	NOUN
alusur-6996	126	8	to	to	PART
alusur-6996	126	9	open	open	VERB
alusur-6996	126	10	source	source	NOUN
alusur-6996	126	11	software	software	NOUN
alusur-6996	126	12	:	:	PUNCT
alusur-6996	126	13	nidaba	nidaba	NOUN
alusur-6996	126	14	:	:	PUNCT
alusur-6996	126	15	https://openphilology.github.io/nidaba/	https://openphilology.github.io/nidaba/	PRON
alusur-6996	126	16	kraken	kraken	NOUN
alusur-6996	126	17	:	:	PUNCT
alusur-6996	126	18	https://github.com/mittagessen/kraken	https://github.com/mittagessen/kraken	VERB
alusur-6996	126	19	ocr	ocr	PROPN
alusur-6996	126	20	-	-	PUNCT
alusur-6996	126	21	evaluation	evaluation	NOUN
alusur-6996	126	22	tools	tool	NOUN
alusur-6996	126	23	:	:	PUNCT
alusur-6996	126	24	https://github.com/ryanfb/ancientgreekocr-ocr-evaluation-tools	https://github.com/ryanfb/ancientgreekocr-ocr-evaluation-tools	PROPN
alusur-6996	126	25	openiti	openiti	PROPN
alusur-6996	126	26	gold	gold	NOUN
alusur-6996	126	27	-	-	PUNCT
alusur-6996	126	28	standard	standard	NOUN
alusur-6996	126	29	data	datum	NOUN
alusur-6996	126	30	for	for	ADP
alusur-6996	126	31	arabic	arabic	ADJ
alusur-6996	126	32	ocr	ocr	PROPN
alusur-6996	126	33	:	:	PUNCT
alusur-6996	126	34	link	link	NOUN
alusur-6996	126	35	:	:	PUNCT
alusur-6996	126	36	https://github.com/openarabic/ocr_gs_data	https://github.com/openarabic/ocr_gs_data	NOUN
alusur-6996	126	37	editions	edition	NOUN
alusur-6996	126	38	of	of	ADP
alusur-6996	126	39	printed	print	VERB
alusur-6996	126	40	texts	text	NOUN
alusur-6996	126	41	(	(	PUNCT
alusur-6996	126	42	when	when	SCONJ
alusur-6996	126	43	two	two	NUM
alusur-6996	126	44	dates	date	NOUN
alusur-6996	126	45	are	be	AUX
alusur-6996	126	46	given	give	VERB
alusur-6996	126	47	,	,	PUNCT
alusur-6996	126	48	the	the	DET
alusur-6996	126	49	second	second	ADJ
alusur-6996	126	50	one	one	NOUN
alusur-6996	126	51	is	be	AUX
alusur-6996	126	52	ce	ce	PROPN
alusur-6996	126	53	)	)	PUNCT
alusur-6996	127	1	[	[	X
alusur-6996	127	2	#	#	SYM
alusur-6996	127	3	0	0	NUM
alusur-6996	127	4	]	]	X
alusur-6996	127	5	ibn	ibn	PROPN
alusur-6996	127	6	al	al	PROPN
alusur-6996	127	7	-	-	PUNCT
alusur-6996	127	8	faqīh	faqīh	PROPN
alusur-6996	127	9	(	(	PUNCT
alusur-6996	127	10	fl	fl	NOUN
alusur-6996	127	11	.	.	PUNCT
alusur-6996	128	1	late	late	ADJ
alusur-6996	128	2	3rd/9th	3rd/9th	NUM
alusur-6996	128	3	century	century	NOUN
alusur-6996	128	4	)	)	PUNCT
alusur-6996	128	5	.	.	PUNCT
alusur-6996	129	1	kitāb	kitāb	PROPN
alusur-6996	129	2	al	al	PROPN
alusur-6996	129	3	-	-	PUNCT
alusur-6996	129	4	buldān	buldān	NOUN
alusur-6996	129	5	.	.	PUNCT
alusur-6996	130	1	ed	ed	NOUN
alusur-6996	130	2	.	.	PUNCT
alusur-6996	131	1	yūsuf	yūsuf	PROPN
alusur-6996	131	2	al	al	PROPN
alusur-6996	131	3	-	-	PUNCT
alusur-6996	131	4	hādī	hādī	NOUN
alusur-6996	131	5	.	.	PUNCT
alusur-6996	132	1	beirut	beirut	PROPN
alusur-6996	132	2	:	:	PUNCT
alusur-6996	132	3	ʿālam	ʿālam	PROPN
alusur-6996	132	4	al	al	PROPN
alusur-6996	132	5	-	-	PUNCT
alusur-6996	132	6	kutub	kutub	PROPN
alusur-6996	132	7	,	,	PUNCT
alusur-6996	132	8	1996	1996	NUM
alusur-6996	132	9	ce	ce	X
alusur-6996	132	10	.	.	PUNCT
alusur-6996	133	1	[	[	X
alusur-6996	133	2	#	#	SYM
alusur-6996	133	3	1	1	NUM
alusur-6996	133	4	]	]	PUNCT
alusur-6996	133	5	ibn	ibn	PROPN
alusur-6996	133	6	al	al	PROPN
alusur-6996	133	7	-	-	PUNCT
alusur-6996	133	8	athīr	athīr	PROPN
alusur-6996	133	9	(	(	PUNCT
alusur-6996	133	10	d.	d.	PROPN
alusur-6996	133	11	630/1233	630/1233	NUM
alusur-6996	133	12	)	)	PUNCT
alusur-6996	133	13	.	.	PUNCT
alusur-6996	134	1	al	al	PROPN
alusur-6996	134	2	-	-	PUNCT
alusur-6996	134	3	kāmil	kāmil	PROPN
alusur-6996	134	4	fī	fī	PROPN
alusur-6996	134	5	al	al	PROPN
alusur-6996	134	6	-	-	PUNCT
alusur-6996	134	7	taʾrīkh	taʾrīkh	PROPN
alusur-6996	134	8	.	.	PUNCT
alusur-6996	135	1	ed	ed	NOUN
alusur-6996	135	2	.	.	PUNCT
alusur-6996	135	3	ʿabd	ʿabd	NOUN
alusur-6996	135	4	allāh	allāh	PROPN
alusur-6996	135	5	al	al	PROPN
alusur-6996	135	6	-	-	PUNCT
alusur-6996	135	7	qāḍī	qāḍī	NOUN
alusur-6996	135	8	.	.	PUNCT
alusur-6996	136	1	beirut	beirut	PROPN
alusur-6996	136	2	:	:	PUNCT
alusur-6996	136	3	dār	dār	PROPN
alusur-6996	136	4	al	al	PROPN
alusur-6996	136	5	-	-	PUNCT
alusur-6996	136	6	kutub	kutub	PROPN
alusur-6996	136	7	al-ʿilmiyya	al-ʿilmiyya	NOUN
alusur-6996	136	8	,	,	PUNCT
alusur-6996	136	9	1415/1994	1415/1994	NUM
alusur-6996	136	10	.	.	PUNCT
alusur-6996	137	1	[	[	X
alusur-6996	137	2	#	#	SYM
alusur-6996	137	3	2	2	NUM
alusur-6996	137	4	]	]	PUNCT
alusur-6996	137	5	ibn	ibn	NOUN
alusur-6996	137	6	qutayba	qutayba	NOUN
alusur-6996	137	7	(	(	PUNCT
alusur-6996	137	8	d.	d.	NOUN
alusur-6996	137	9	276/889	276/889	NUM
alusur-6996	137	10	)	)	PUNCT
alusur-6996	137	11	.	.	PUNCT
alusur-6996	138	1	kitāb	kitāb	PROPN
alusur-6996	138	2	adab	adab	PROPN
alusur-6996	138	3	al	al	PROPN
alusur-6996	138	4	-	-	PUNCT
alusur-6996	138	5	kātib	kātib	PROPN
alusur-6996	138	6	.	.	PUNCT
alusur-6996	139	1	ed	ed	NOUN
alusur-6996	139	2	.	.	PUNCT
alusur-6996	140	1	muḥammad	muḥammad	PROPN
alusur-6996	140	2	al	al	PROPN
alusur-6996	140	3	-	-	PUNCT
alusur-6996	140	4	dālī	dālī	PROPN
alusur-6996	140	5	.	.	PUNCT
alusur-6996	141	1	muʾassasat	muʾassasat	PROPN
alusur-6996	141	2	al	al	PROPN
alusur-6996	141	3	-	-	PUNCT
alusur-6996	141	4	risāla	risāla	PROPN
alusur-6996	141	5	,	,	PUNCT
alusur-6996	141	6	n.d	n.d	PROPN
alusur-6996	141	7	.	.	PUNCT
alusur-6996	142	1	[	[	X
alusur-6996	142	2	#	#	SYM
alusur-6996	142	3	3	3	NUM
alusur-6996	142	4	]	]	X
alusur-6996	142	5	al	al	PROPN
alusur-6996	142	6	-	-	PUNCT
alusur-6996	142	7	jāḥiẓ	jāḥiẓ	PROPN
alusur-6996	142	8	(	(	PUNCT
alusur-6996	142	9	d.	d.	PROPN
alusur-6996	142	10	255/868	255/868	PROPN
alusur-6996	142	11	-	-	PUNCT
alusur-6996	142	12	9	9	NUM
alusur-6996	142	13	)	)	PUNCT
alusur-6996	142	14	.	.	PUNCT
alusur-6996	143	1	kitāb	kitāb	PROPN
alusur-6996	143	2	al	al	PROPN
alusur-6996	143	3	-	-	PUNCT
alusur-6996	143	4	ḥayawān	ḥayawān	PROPN
alusur-6996	143	5	.	.	PUNCT
alusur-6996	144	1	beirut	beirut	PROPN
alusur-6996	144	2	:	:	PUNCT
alusur-6996	144	3	dār	dār	PROPN
alusur-6996	144	4	al	al	PROPN
alusur-6996	144	5	-	-	PUNCT
alusur-6996	144	6	kutub	kutub	PROPN
alusur-6996	144	7	al-ʿilmiyya	al-ʿilmiyya	NOUN
alusur-6996	144	8	,	,	PUNCT
alusur-6996	144	9	1424/2003	1424/2003	NUM
alusur-6996	144	10	.	.	PUNCT
alusur-6996	145	1	[	[	X
alusur-6996	145	2	#	#	SYM
alusur-6996	145	3	4	4	NUM
alusur-6996	145	4	]	]	PUNCT
alusur-6996	145	5	al	al	PROPN
alusur-6996	145	6	-	-	PUNCT
alusur-6996	145	7	yaʿqūbī	yaʿqūbī	PROPN
alusur-6996	145	8	(	(	PUNCT
alusur-6996	145	9	fl	fl	NOUN
alusur-6996	145	10	.	.	NOUN
alusur-6996	145	11	second	second	ADJ
alusur-6996	145	12	half	half	NOUN
alusur-6996	145	13	of	of	ADP
alusur-6996	145	14	the	the	DET
alusur-6996	145	15	3rd/9th	3rd/9th	NUM
alusur-6996	145	16	century	century	NOUN
alusur-6996	145	17	)	)	PUNCT
alusur-6996	145	18	.	.	PUNCT
alusur-6996	146	1	al	al	PROPN
alusur-6996	146	2	-	-	PUNCT
alusur-6996	146	3	taʾrīkh	taʾrīkh	PROPN
alusur-6996	146	4	.	.	PUNCT
alusur-6996	147	1	beirut	beirut	PROPN
alusur-6996	147	2	:	:	PUNCT
alusur-6996	147	3	dār	dār	PROPN
alusur-6996	147	4	ṣādir	ṣādir	PROPN
alusur-6996	147	5	,	,	PUNCT
alusur-6996	147	6	n.d	n.d	PROPN
alusur-6996	147	7	.	.	PUNCT
alusur-6996	148	1	[	[	X
alusur-6996	148	2	#	#	SYM
alusur-6996	148	3	5	5	NUM
alusur-6996	148	4	]	]	PUNCT
alusur-6996	148	5	al	al	PROPN
alusur-6996	148	6	-	-	PUNCT
alusur-6996	148	7	dhahabī	dhahabī	PROPN
alusur-6996	148	8	(	(	PUNCT
alusur-6996	148	9	d.	d.	PROPN
alusur-6996	148	10	748/1347	748/1347	NUM
alusur-6996	148	11	)	)	PUNCT
alusur-6996	148	12	.	.	PUNCT
alusur-6996	149	1	taʾrīkh	taʾrīkh	PROPN
alusur-6996	149	2	al	al	PROPN
alusur-6996	149	3	-	-	PUNCT
alusur-6996	149	4	islām	islām	PROPN
alusur-6996	149	5	.	.	PUNCT
alusur-6996	150	1	al	al	PROPN
alusur-6996	150	2	-	-	PUNCT
alusur-6996	150	3	maktaba	maktaba	PROPN
alusur-6996	150	4	al	al	PROPN
alusur-6996	150	5	-	-	PUNCT
alusur-6996	150	6	tawfīqiyya	tawfīqiyya	PROPN
alusur-6996	150	7	,	,	PUNCT
alusur-6996	150	8	n.d	n.d	PROPN
alusur-6996	150	9	.	.	PUNCT
alusur-6996	151	1	[	[	X
alusur-6996	151	2	#	#	SYM
alusur-6996	151	3	6	6	NUM
alusur-6996	151	4	]	]	PUNCT
alusur-6996	151	5	ibn	ibn	PROPN
alusur-6996	151	6	al	al	PROPN
alusur-6996	151	7	-	-	PUNCT
alusur-6996	151	8	jawzī	jawzī	PROPN
alusur-6996	151	9	(	(	PUNCT
alusur-6996	151	10	d.	d.	NOUN
alusur-6996	151	11	597/1200	597/1200	NUM
alusur-6996	151	12	)	)	PUNCT
alusur-6996	151	13	.	.	PUNCT
alusur-6996	152	1	al	al	PROPN
alusur-6996	152	2	-	-	PUNCT
alusur-6996	152	3	muntaẓam	muntaẓam	NOUN
alusur-6996	152	4	.	.	PUNCT
alusur-6996	153	1	ed	ed	NOUN
alusur-6996	153	2	.	.	PUNCT
alusur-6996	154	1	muḥammad	muḥammad	PROPN
alusur-6996	154	2	al	al	PROPN
alusur-6996	154	3	-	-	PUNCT
alusur-6996	154	4	qādir	qādir	PROPN
alusur-6996	154	5	ʿaṭā	ʿaṭā	ADJ
alusur-6996	154	6	and	and	CCONJ
alusur-6996	154	7	muṣṭafá	muṣṭafá	PROPN
alusur-6996	154	8	al	al	PROPN
alusur-6996	154	9	-	-	PUNCT
alusur-6996	154	10	qādir	qādir	PROPN
alusur-6996	154	11	ʿaṭā	ʿaṭā	ADJ
alusur-6996	154	12	.	.	PUNCT
alusur-6996	155	1	beirut	beirut	PROPN
alusur-6996	155	2	:	:	PUNCT
alusur-6996	155	3	dār	dār	PROPN
alusur-6996	155	4	al	al	PROPN
alusur-6996	155	5	-	-	PUNCT
alusur-6996	155	6	kutub	kutub	PROPN
alusur-6996	155	7	al-ʿilmiyya	al-ʿilmiyya	NOUN
alusur-6996	155	8	,	,	PUNCT
alusur-6996	155	9	1412/1992	1412/1992	NUM
alusur-6996	155	10	.	.	PUNCT
alusur-6996	155	11	https://github.com/tmbdev/clstm	https://github.com/tmbdev/clstm	NOUN
alusur-6996	156	1	https://openphilology.github.io/nidaba/	https://openphilology.github.io/nidaba/	PROPN
alusur-6996	156	2	https://github.com/mittagessen/kraken	https://github.com/mittagessen/kraken	VERB
alusur-6996	156	3	https://github.com/ryanfb/ancientgreekocr-ocr-evaluation-tools	https://github.com/ryanfb/ancientgreekocr-ocr-evaluation-tool	NOUN
alusur-6996	156	4	https://github.com/openarabic/ocr_gs_data	https://github.com/openarabic/ocr_gs_data	NOUN
alusur-6996	156	5	12	12	NUM
alusur-6996	156	6	•	•	NOUN
alusur-6996	156	7	kiessling	kiessle	VERB
alusur-6996	156	8	,	,	PUNCT
alusur-6996	156	9	miller	miller	PROPN
alusur-6996	156	10	,	,	PUNCT
alusur-6996	156	11	romanov	romanov	PROPN
alusur-6996	156	12	,	,	PUNCT
alusur-6996	156	13	&	&	CCONJ
alusur-6996	156	14	savant	savant	PROPN
alusur-6996	156	15	al-ʿuṣūr	al-ʿuṣūr	PROPN
alusur-6996	156	16	al	al	PROPN
alusur-6996	156	17	-	-	PUNCT
alusur-6996	156	18	wusṭā	wusṭā	NOUN
alusur-6996	156	19	25	25	NUM
alusur-6996	156	20	(	(	PUNCT
alusur-6996	156	21	2017	2017	NUM
alusur-6996	156	22	)	)	PUNCT
alusur-6996	156	23	appendix	appendix	NOUN
alusur-6996	156	24	:	:	PUNCT
alusur-6996	156	25	performance	performance	NOUN
alusur-6996	156	26	of	of	ADP
alusur-6996	156	27	text	text	NOUN
alusur-6996	156	28	-	-	PUNCT
alusur-6996	156	29	specific	specific	ADJ
alusur-6996	156	30	models	model	NOUN
alusur-6996	156	31	table	table	VERB
alusur-6996	156	32	a	a	DET
alusur-6996	156	33	:	:	PUNCT
alusur-6996	156	34	performance	performance	NOUN
alusur-6996	156	35	of	of	ADP
alusur-6996	156	36	#	#	SYM
alusur-6996	156	37	1	1	NUM
alusur-6996	156	38	-	-	PUNCT
alusur-6996	156	39	based	base	VERB
alusur-6996	156	40	model	model	NOUN
alusur-6996	156	41	on	on	ADP
alusur-6996	156	42	other	other	ADJ
alusur-6996	156	43	texts	text	NOUN
alusur-6996	156	44	table	table	NOUN
alusur-6996	156	45	b	b	NOUN
alusur-6996	156	46	:	:	PUNCT
alusur-6996	156	47	performance	performance	NOUN
alusur-6996	156	48	of	of	ADP
alusur-6996	156	49	#	#	SYM
alusur-6996	156	50	2	2	NUM
alusur-6996	156	51	-	-	PUNCT
alusur-6996	156	52	based	base	VERB
alusur-6996	156	53	model	model	NOUN
alusur-6996	156	54	on	on	ADP
alusur-6996	156	55	other	other	ADJ
alusur-6996	156	56	texts	text	NOUN
alusur-6996	156	57	al-ʿuṣūr	al-ʿuṣūr	PROPN
alusur-6996	156	58	al	al	PROPN
alusur-6996	156	59	-	-	PUNCT
alusur-6996	156	60	wusṭā	wusṭā	NOUN
alusur-6996	156	61	25	25	NUM
alusur-6996	156	62	(	(	PUNCT
alusur-6996	156	63	2017	2017	NUM
alusur-6996	156	64	)	)	PUNCT
alusur-6996	156	65	new	new	ADJ
alusur-6996	156	66	developments	development	NOUN
alusur-6996	156	67	in	in	ADP
alusur-6996	156	68	arabographic	arabographic	ADJ
alusur-6996	156	69	optical	optical	ADJ
alusur-6996	156	70	character	character	NOUN
alusur-6996	156	71	recognition	recognition	NOUN
alusur-6996	156	72	•	•	NOUN
alusur-6996	156	73	13	13	NUM
alusur-6996	156	74	appendix	appendix	NOUN
alusur-6996	156	75	:	:	PUNCT
alusur-6996	156	76	performance	performance	NOUN
alusur-6996	156	77	of	of	ADP
alusur-6996	156	78	text	text	NOUN
alusur-6996	156	79	-	-	PUNCT
alusur-6996	156	80	specific	specific	ADJ
alusur-6996	156	81	models	model	NOUN
alusur-6996	156	82	table	table	VERB
alusur-6996	156	83	d	d	NOUN
alusur-6996	156	84	:	:	PUNCT
alusur-6996	156	85	performance	performance	NOUN
alusur-6996	156	86	of	of	ADP
alusur-6996	156	87	#	#	SYM
alusur-6996	156	88	4	4	NUM
alusur-6996	156	89	-	-	PUNCT
alusur-6996	156	90	based	base	VERB
alusur-6996	156	91	model	model	NOUN
alusur-6996	156	92	on	on	ADP
alusur-6996	156	93	other	other	ADJ
alusur-6996	156	94	texts	text	NOUN
alusur-6996	156	95	table	table	NOUN
alusur-6996	156	96	c	c	NOUN
alusur-6996	156	97	:	:	PUNCT
alusur-6996	156	98	performance	performance	NOUN
alusur-6996	156	99	of	of	ADP
alusur-6996	156	100	#	#	SYM
alusur-6996	156	101	3	3	NUM
alusur-6996	156	102	-	-	PUNCT
alusur-6996	156	103	based	base	VERB
alusur-6996	156	104	model	model	NOUN
alusur-6996	156	105	on	on	ADP
alusur-6996	156	106	other	other	ADJ
alusur-6996	156	107	texts	text	NOUN
