|
|
|
|
LEADER |
01991 am a22002533u 4500 |
001 |
137401.2 |
042 |
|
|
|a dc
|
100 |
1 |
0 |
|a Schuster, Tal
|e author
|
100 |
1 |
0 |
|a Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory
|e contributor
|
100 |
1 |
0 |
|a Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science
|e contributor
|
700 |
1 |
0 |
|a Shah, Darsh J
|e author
|
700 |
1 |
0 |
|a Yeo, Yun Jie Serene
|e author
|
700 |
1 |
0 |
|a Filizzola, Daniel
|e author
|
700 |
1 |
0 |
|a Santus, Enrico
|e author
|
700 |
1 |
0 |
|a Barzilay, Regina
|e author
|
245 |
0 |
0 |
|a Towards Debiasing Fact Verification Models
|
260 |
|
|
|b Association for Computational Linguistics,
|c 2021-11-15T15:59:15Z.
|
856 |
|
|
|z Get fulltext
|u https://hdl.handle.net/1721.1/137401.2
|
520 |
|
|
|a © 2019 Association for Computational Linguistics Fact verification requires validating a claim in the context of evidence. We show, however, that in the popular FEVER dataset this might not necessarily be the case. Claim-only classifiers perform competitively with top evidence-aware models. In this paper, we investigate the cause of this phenomenon, identifying strong cues for predicting labels solely based on the claim, without considering any evidence. We create an evaluation set that avoids those idiosyncrasies. The performance of FEVER-trained models significantly drops when evaluated on this test set. Therefore, we introduce a regularization method which alleviates the effect of bias in the training data, obtaining improvements on the newly created test set. This work is a step towards a more sound evaluation of reasoning capabilities in fact verification models.
|
520 |
|
|
|a DSO (Grant DSOCL18002)
|
546 |
|
|
|a en
|
655 |
7 |
|
|a Article
|
773 |
|
|
|t 10.18653/V1/D19-1341
|
773 |
|
|
|t EMNLP-IJCNLP 2019 - 2019 Conference on Empirical Methods in Natural Language Processing and 9th International Joint Conference on Natural Language Processing, Proceedings of the Conference
|