Skip to content

Evidence Sentence Extraction for Machine Reading Comprehension

Notifications You must be signed in to change notification settings

nlpdata/evidence

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 

Repository files navigation

Annotated evidence sentences for 500 questions from the dev set of the RACE-Middle dataset described in the paper Evidence Sentence Extraction for Machine Reading Comprehension. We provide the annotations of two annotators.

If you find this resource useful, please cite the following paper.

@inproceedings{wang2019evidence,
  title={Evidence Sentence Extraction for Machine Reading Comprehension},
  author={Wang, Hai and Yu, Dian and Sun, Kai and Chen, Jianshu and Yu, Dong and McAllester, David and Roth, Dan},
  booktitle={Proceedings of the CoNLL},
  address={Hong Kong, China},
  url={https://arxiv.org/abs/1902.08852v2},
  year={2019}
}

The format of these files are as follows.

{
 docid: [
  evidence sentences for the 1st question,
  evidence sentences for the 2nd question,
  ...
 ], 
 ...
}

Note that for some questions, annotators cannot find appropriate evidence sentences. For these cases, we leave the evidence sentence list empty.

About

Evidence Sentence Extraction for Machine Reading Comprehension

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published