Abstract
Objectivity is a goal for Wikipedia and many news sites, as well as a guiding principle of many large language models. Indeed, several methods have recently been developed for automatic subjective bias neutralization. These methods, however, typically rely on parallel text for training (i.e. a biased sentence coupled with a non-biased sentence), demonstrate poor transfer to new domains, and can lose important bias-independent context. Toward expanding the reach of bias neutralization, we propose in this paper a new approach called FairBalance. Three of its unique features are: i) a cycle consistent adversarial network enables bias neutralization without the need for parallel text; ii) the model design preserves bias-independent content; and iii) through auxiliary guidance, the model highlights sequences of bias-inducing words, yielding strong results in terms of bias neutralization quality. Extensive experiments demonstrate how FairBalance significantly improves subjective bias neutralization compared to other methods.- Anthology ID:
- 2023.emnlp-main.882
- Volume:
- Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing
- Month:
- December
- Year:
- 2023
- Address:
- Singapore
- Editors:
- Houda Bouamor, Juan Pino, Kalika Bali
- Venue:
- EMNLP
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 14265–14278
- Language:
- URL:
- https://aclanthology.org/2023.emnlp-main.882
- DOI:
- 10.18653/v1/2023.emnlp-main.882
- Cite (ACL):
- Karthic Madanagopal and James Caverlee. 2023. Bias Neutralization in Non-Parallel Texts: A Cyclic Approach with Auxiliary Guidance. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 14265–14278, Singapore. Association for Computational Linguistics.
- Cite (Informal):
- Bias Neutralization in Non-Parallel Texts: A Cyclic Approach with Auxiliary Guidance (Madanagopal & Caverlee, EMNLP 2023)
- PDF:
- https://preview.aclanthology.org/corrections-2024-05/2023.emnlp-main.882.pdf