Stereotype and Skew: Quantifying Gender Bias in Pre-trained and Fine-tuned Language Models

Daniel de Vassimon Manela, David Errington, Thomas Fisher, Boris van Breugel, Pasquale Minervini


Abstract
This paper proposes two intuitive metrics, skew and stereotype, that quantify and analyse the gender bias present in contextual language models when tackling the WinoBias pronoun resolution task. We find evidence that gender stereotype correlates approximately negatively with gender skew in out-of-the-box models, suggesting that there is a trade-off between these two forms of bias. We investigate two methods to mitigate bias. The first approach is an online method which is effective at removing skew at the expense of stereotype. The second, inspired by previous work on ELMo, involves the fine-tuning of BERT using an augmented gender-balanced dataset. We show that this reduces both skew and stereotype relative to its unaugmented fine-tuned counterpart. However, we find that existing gender bias benchmarks do not fully probe professional bias as pronoun resolution may be obfuscated by cross-correlations from other manifestations of gender prejudice.
Anthology ID:
2021.eacl-main.190
Volume:
Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume
Month:
April
Year:
2021
Address:
Online
Editors:
Paola Merlo, Jorg Tiedemann, Reut Tsarfaty
Venue:
EACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
2232–2242
Language:
URL:
https://aclanthology.org/2021.eacl-main.190
DOI:
10.18653/v1/2021.eacl-main.190
Bibkey:
Cite (ACL):
Daniel de Vassimon Manela, David Errington, Thomas Fisher, Boris van Breugel, and Pasquale Minervini. 2021. Stereotype and Skew: Quantifying Gender Bias in Pre-trained and Fine-tuned Language Models. In Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume, pages 2232–2242, Online. Association for Computational Linguistics.
Cite (Informal):
Stereotype and Skew: Quantifying Gender Bias in Pre-trained and Fine-tuned Language Models (de Vassimon Manela et al., EACL 2021)
Copy Citation:
PDF:
https://aclanthology.org/2021.eacl-main.190.pdf
Code
 12kleingordon34/NLP_masters_project
Data
BookCorpusWinoBias