Do Neural Language Models Overcome Reporting Bias?

Vered Shwartz, Yejin Choi

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

44 Scopus citations

Abstract

Mining commonsense knowledge from corpora suffers from reporting bias, over-representing the rare at the expense of the trivial (Gordon and Van Durme, 2013). We study to what extent pre-trained language models overcome this issue. We find that while their generalization capacity allows them to better estimate the plausibility of frequent but unspoken of actions, outcomes, and properties, they also tend to overestimate that of the very rare, amplifying the bias that already exists in their training corpus.

Original languageEnglish
Title of host publicationCOLING 2020 - 28th International Conference on Computational Linguistics, Proceedings of the Conference
EditorsDonia Scott, Nuria Bel, Chengqing Zong
PublisherAssociation for Computational Linguistics (ACL)
Pages6863-6870
Number of pages8
ISBN (Electronic)9781952148279
DOIs
StatePublished - 2020
Externally publishedYes
Event28th International Conference on Computational Linguistics, COLING 2020 - Virtual, Online, Spain
Duration: 8 Dec 202013 Dec 2020

Publication series

NameCOLING 2020 - 28th International Conference on Computational Linguistics, Proceedings of the Conference

Conference

Conference28th International Conference on Computational Linguistics, COLING 2020
Country/TerritorySpain
CityVirtual, Online
Period8/12/2013/12/20

Bibliographical note

Publisher Copyright:
© 2020 COLING 2020 - 28th International Conference on Computational Linguistics, Proceedings of the Conference. All rights reserved.

Funding

This research was supported in part by NSF (IIS-1524371, IIS-1714566), DARPA under the CwC program through the ARO (W911NF-15-1-0543), and DARPA under the MCS program through NIWC Pacific (N66001-19-2-4031).

FundersFunder number
National Science FoundationIIS-1714566, IIS-1524371
Army Research OfficeW911NF-15-1-0543
Defense Advanced Research Projects Agency
Naval Information Warfare Center PacificN66001-19-2-4031

    Fingerprint

    Dive into the research topics of 'Do Neural Language Models Overcome Reporting Bias?'. Together they form a unique fingerprint.

    Cite this