Work

Remedying Gender Bias in Open English Wordnet

Year: 2025

Type: conference-paper

Abstract: Open English Wordnet aims to improve and maintain a wordnet for English, based on the Princeton WordNet.In this context, we identify a number of gender biases in the existing wordnet and consider the challenges of remediating the biases in the resource.In particular, we look at structural, contextual and definitional biases in the resource and examine how changes to the structure of the wordnet and to the textual definitions can create a wordnet that more fairly represents reality.We propose a number of changes that introduce 317 new synsets as well as changing the definitions or relations of over 400 further synsets.We show that these changes reduce certain kinds of gender bias within the resource.

Source: Global WordNet Conference (GWC)

DOI: 10.18653/v1/2025.gwc-1.16

Authors: John P. McCrae, Haotian Zhu, Fei Xia, Al Waskow, Kexin Gao

Institutions: Ollscoil na Gaillimhe – University of Galway, Insight SFI Research Centre for Data Analytics, University of Washington

Language: English

Topic: Gender Studies in Language

Subfield: Gender Studies

Field: Social Sciences

Domain: Social Sciences

SDGs: Reduced inequalities, Gender equality

Open Access status: hybrid

Funders: Science Foundation Ireland, Insight SFI Research Centre for Data Analytics