Reducing Sentiment Bias in Language Models via Counterfactual Evaluation
This work addresses bias in language models for applications like text generation, though it is incremental as it builds on existing fairness and regularization techniques.
The paper tackled the problem of sentiment bias in language models by quantifying it using counterfactual evaluation and fairness metrics, and then reduced this bias through regularization methods, achieving improved fairness metrics while maintaining comparable perplexity and semantic similarity.
Advances in language modeling architectures and the availability of large text corpora have driven progress in automatic text generation. While this results in models capable of generating coherent texts, it also prompts models to internalize social biases present in the training corpus. This paper aims to quantify and reduce a particular type of bias exhibited by language models: bias in the sentiment of generated text. Given a conditioning context (e.g., a writing prompt) and a language model, we analyze if (and how) the sentiment of the generated text is affected by changes in values of sensitive attributes (e.g., country names, occupations, genders) in the conditioning context using a form of counterfactual evaluation. We quantify sentiment bias by adopting individual and group fairness metrics from the fair machine learning literature, and demonstrate that large-scale models trained on two different corpora (news articles, and Wikipedia) exhibit considerable levels of bias. We then propose embedding and sentiment prediction-derived regularization on the language model's latent representations. The regularizations improve fairness metrics while retaining comparable levels of perplexity and semantic similarity.