WinoPron: Revisiting English Winogender Schemas for Consistency, Coverage, and Grammatical Case
This work addresses data quality problems for researchers evaluating gender bias in coreference resolution, though it is incremental as it builds on an existing dataset.
The authors identified and fixed issues in the Winogender Schemas dataset, creating WinoPron to improve evaluation of gender bias in coreference resolution, and found that accusative pronouns are harder to resolve for models, with bias varying across pronoun forms.
While measuring bias and robustness in coreference resolution are important goals, such measurements are only as good as the tools we use to measure them. Winogender Schemas (Rudinger et al., 2018) are an influential dataset proposed to evaluate gender bias in coreference resolution, but a closer look reveals issues with the data that compromise its use for reliable evaluation, including treating different pronominal forms as equivalent, violations of template constraints, and typographical errors. We identify these issues and fix them, contributing a new dataset: WinoPron. Using WinoPron, we evaluate two state-of-the-art supervised coreference resolution systems, SpanBERT, and five sizes of FLAN-T5, and demonstrate that accusative pronouns are harder to resolve for all models. We also propose a new method to evaluate pronominal bias in coreference resolution that goes beyond the binary. With this method, we also show that bias characteristics vary not just across pronoun sets (e.g., he vs. she), but also across surface forms of those sets (e.g., him vs. his).