Contrastive Self-Supervised Learning for Commonsense Reasoning

Tassilo Klein, Moin Nabi


Abstract
We propose a self-supervised method to solve Pronoun Disambiguation and Winograd Schema Challenge problems. Our approach exploits the characteristic structure of training corpora related to so-called “trigger” words, which are responsible for flipping the answer in pronoun disambiguation. We achieve such commonsense reasoning by constructing pair-wise contrastive auxiliary predictions. To this end, we leverage a mutual exclusive loss regularized by a contrastive margin. Our architecture is based on the recently introduced transformer networks, BERT, that exhibits strong performance on many NLP benchmarks. Empirical results show that our method alleviates the limitation of current supervised approaches for commonsense reasoning. This study opens up avenues for exploiting inexpensive self-supervision to achieve performance gain in commonsense reasoning tasks.
Anthology ID:
2020.acl-main.671
Volume:
Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics
Month:
July
Year:
2020
Address:
Online
Venue:
ACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
7517–7523
Language:
URL:
https://aclanthology.org/2020.acl-main.671
DOI:
10.18653/v1/2020.acl-main.671
Bibkey:
Cite (ACL):
Tassilo Klein and Moin Nabi. 2020. Contrastive Self-Supervised Learning for Commonsense Reasoning. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 7517–7523, Online. Association for Computational Linguistics.
Cite (Informal):
Contrastive Self-Supervised Learning for Commonsense Reasoning (Klein & Nabi, ACL 2020)
Copy Citation:
PDF:
https://aclanthology.org/2020.acl-main.671.pdf
Video:
 http://slideslive.com/38929108
Code
 SAP-samples/acl2020-commonsense +  additional community code
Data
WSC