Fakhri Karray
2023
Can a Prediction’s Rank Offer a More Accurate Quantification of Bias? A Case Study Measuring Sexism in Debiased Language Models
Jad Doughman
|
Shady Shehata
|
Leen Al Qadi
|
Youssef Nafea
|
Fakhri Karray
Proceedings of the 4th Workshop on Evaluation and Comparison of NLP Systems
Pre-trained language models are known to inherit a plethora of contextual biases from their training data. These biases have proven to be projected onto a variety of downstream applications, making their detection and mitigation imminent. Limited research has been conducted to quantify specific bias types, such as benevolent sexism, which may be subtly present within the inferred connotations of a sentence. To this extent, our work aims to: (1) provide a benchmark of sexism sentences; (2) adapt two bias metrics: mean probability score and mean normalized rank; (3) conduct a case study to quantify and analyze sexism in base and de-biased masked language models. We find that debiasing, even in its most effective form (Auto-Debias), solely nullifies the probability score of biasing tokens, while retaining them in high ranks. Auto-Debias illustrates a 90%-96% reduction in mean probability scores from base to debiased models, while only a 3%-16% reduction in mean normalized ranks. Similar to the application of non-parametric statistical tests for data that does not follow a normal distribution, operating on the ranks of predictions rather than their probability scores offers a more representative bias measure.
Search