Mitigating Implicit Bias in Chinese Toxic Speech Detection via Unbiased Contrastive Learning
ACM Transactions on Asian and Low-Resource Language Information Processing · 30.06.2026 · 25 (6) · 1-23
DOI: 10.1145/3816029 ↗Kurz gefasst
Fachartikel in der Zeitschrift ACM Transactions on Asian and Low-Resource Language Information Processing, online erschienen am 08.06.2026. Verfasst von Xuan Feng und 4 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 62 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Natural Science Foundation of China.
Abstract (Originalsprache)
Mitigating human-like biases and social stereotypes in pre-trained language models (PLMs) has become a crucial task in Chinese toxic speech detection. While PLMs have achieved state-of-the-art results in mitigating explicit bias (e.g., bias on sensitive attribute words), the study of implicit bias (e.g., bias against specific demographic groups) is still under-explored. Besides, existing debiasing methods focus on the tradeoff between debiasing efficiency and model performance, while ignoring robustness against noisy data. Therefore, a debiasing method that effectively reduces biases while maintaining robustness against noisy data is needed. In this article, we propose Unbiased Contrastive Learning (UCL), which can mitigate explicit and implicit bias while maintaining robustness to noisy data. Specifically, we first analyze the bias representation problem constrained by contrastive learning objective and implement unbiased contrastive objective for learning unbiased text representations to mitigate explicit and implicit biases in Chinese toxic speech detection tasks. UCL inherits the idea of supervised contrastive learning, which encourages representations of the same sensitive attribute to be closer than those of different sensitive attributes and ensures unbiasedness by penalizing the sensitive attribute information contained in the representations. Furthermore, we design conditional normalization to reduce biased classification caused by the imbalanced distribution of demographic groups in the data. Experimental results on Chinese and English datasets show that the proposed method outperforms the state-of-the-art methods and achieves the competitive performance.
Förderung
- Natural Science Foundation of China
Zitieren
APA
Feng, X., Yao, Y., Gu, T., Chang, L., & Liu, X. (2026). Mitigating Implicit Bias in Chinese Toxic Speech Detection via Unbiased Contrastive Learning. ACM Transactions on Asian and Low-Resource Language Information Processing, 25(6), 1-23. https://doi.org/10.1145/3816029
Vancouver
Feng X, Yao Y, Gu T, Chang L, Liu X. Mitigating Implicit Bias in Chinese Toxic Speech Detection via Unbiased Contrastive Learning. ACM Transactions on Asian and Low-Resource Language Information Processing. 2026;25(6):1-23. doi:10.1145/3816029
BibTeX
@article{Feng2026,
author = {Feng, Xuan and Yao, Yuan and Gu, Tianlong and Chang, Liang and Liu, Xiaoli},
title = {Mitigating Implicit Bias in Chinese Toxic Speech Detection via Unbiased Contrastive Learning},
journal = {ACM Transactions on Asian and Low-Resource Language Information Processing},
year = {2026},
volume = {25},
number = {6},
pages = {1--23},
doi = {10.1145/3816029}
}
EUVIMED