Gender Bias in Coreference Resolution: Evaluation and Debiasing Methods

TL;DR

This paper introduces WinoBias, a benchmark for evaluating gender bias in coreference resolution systems, and proposes a debiasing method that reduces bias without harming overall performance.

Contribution

The paper presents WinoBias, a new dataset for gender bias evaluation, and combines data augmentation with embedding debiasing to mitigate bias in coreference models.

Findings

Coreference systems favor stereotypical gender-entity links by 21.1 F1 points.

Debiasing techniques reduce gender bias in models.

Performance on existing benchmarks remains unaffected by debiasing.

Abstract

We introduce a new benchmark, WinoBias, for coreference resolution focused on gender bias. Our corpus contains Winograd-schema style sentences with entities corresponding to people referred by their occupation (e.g. the nurse, the doctor, the carpenter). We demonstrate that a rule-based, a feature-rich, and a neural coreference system all link gendered pronouns to pro-stereotypical entities with higher accuracy than anti-stereotypical entities, by an average difference of 21.1 in F1 score. Finally, we demonstrate a data-augmentation approach that, in combination with existing word-embedding debiasing techniques, removes the bias demonstrated by these systems in WinoBias without significantly affecting their performance on existing coreference benchmark datasets. Our dataset and code are available at http://winobias.org.