Machine Reading Comprehension-Enabled Public Service Information System: A Large-Scale Dataset and Neural Network Models

Table 4

Analysis of each MRC dataset.

Dataset	Question style	Answer style	Corpora source	Dataset generation method
Facebook CBT	Cloze	Free-form	Children’s books	Automated
MovieQA	Natural	Free-form	Movie plot	Crowdsourcing
DREAM	Natural	Free-form	Exam	Crowdsourcing
CliCR	Cloze	Free-form	BMJ case reports	Automated
MCScript	Natural	Free-form	Narrative texts	Crowdsourcing
ARC-C	Natural	Free-form	Science questions	Experts
ARC-E	Natural	Free-form	Science questions	Experts
WikiQA	Natural	Free-form	Wikipedia	Crowdsourcing
TQA	Natural	Free-form	Science curricula	Experts
ProPara	Natural	Spans	Process paragraph	Crowdsourcing
SQUAD2.0	Natural	Spans	Wikipedia	Crowdsourcing
SQuAD1.1	Natural	Spans	Wikipedia	Crowdsourcing
Qangaroo-W	Synthesis	Spans	Paper abstracts	Crowdsourcing
Qangaroo-M	Synthesis	Spans	Wikipedia	Crowdsourcing
OpenBookQA	Natural	Free-form	Elementary level science facts	Crowdsourcing
C-Pulse	Natural	Spans	Public service information	All three methods