Mitigating Gender Bias for Neural Dialogue Generation with Adversarial Learning

TLDR

Dialogue systems are increasingly integral to daily life, yet existing debiasing methods for NLP are not directly applicable to them and risk reducing response diversity. This study aims to train dialogue models free from gender bias while preserving performance. We introduce Debiased‑Chat, an adversarial learning framework that encourages gender‑neutral responses. Experiments on two real‑world datasets demonstrate that Debiased‑Chat substantially reduces gender bias without compromising response quality.

Abstract

Dialogue systems play an increasingly important role in various aspects of our daily life. It is evident from recent research that dialogue systems trained on human conversation data are biased. In particular, they can produce responses that reflect people’s gender prejudice. Many debiasing methods have been developed for various NLP tasks, such as word embedding. However, they are not directly applicable to dialogue systems because they are likely to force dialogue models to generate similar responses for different genders. This greatly degrades the diversity of the generated responses and immensely hurts the performance of the dialogue models. In this paper, we propose a novel adversarial learning framework Debiased-Chat to train dialogue models free from gender bias while keeping their performance. Extensive experiments on two real-world conversation datasets show that our framework significantly reduces gender bias in dialogue models while maintaining the response quality.