PerAnSel:  A  Novel Deep Neural Network-Based System for Persian Question Answering

Table 1

Review of answer selection datasets for various languages.

Dataset	Language	Type	Domain	Train	Dev	Test
TrecQA (raw) [23]	English	Native	Open	1229/94	82	100
TrecQA (clean) [24]	English	Native	Open	1229/94	65	68
WikiQA [18]	English	Native	Open	2118	396	633
InsuranceQA [25]	English	Native	Close	12889	2000	2000
SelQA [26]	English	Native	Open	5529	785	1590
cMedQA v1 [53]	Chinese	Native	Close	50000	2000	2000
cMedQA v2 [54]	Chinese	Native	Close	100000	4000	4000
cEpilepsyQA [55]	Chinese	Native	Close	3920	490	490
DBQA [56]	Chinese	Native	Open	8772	4779	2500
MilkQA [57]	Portuguese	Native	Close	2307	50	300
WikiQAar [58]	Arabic	Translation	Open	2118	396	633
CQA-MD [59]	Arabic	Native	Close	1031	250	250
PerCQA [60]	Persian	Native	Open	692	99	198
PASD	Persian	Native	Open	17567	1000	1000

The bold row indicates our new dataset.