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