TY - GEN
T1 - Removal of Hallucination on Hallucination
T2 - 63rd Annual Meeting of the Association for Computational Linguistics, ACL 2025
AU - Hu, Wentao
AU - Zhang, Wengyu
AU - Jiang, Yiyang
AU - Zhang, Chen Jason
AU - Wei, Xiaoyong
AU - Li, Qing
N1 - Publisher Copyright:
© 2025 Association for Computational Linguistics.
PY - 2025/7
Y1 - 2025/7
N2 - Retrieval-Augmented Generation (RAG) enhances factual accuracy by integrating external knowledge, yet it introduces a critical issue: erroneous or biased retrieval can mislead generation, compounding hallucinations, a phenomenon we term Hallucination on Hallucination. To address this, we propose Debate-Augmented RAG (DRAG), a training-free framework that integrates Multi-Agent Debate (MAD) mechanisms into both retrieval and generation stages. In retrieval, DRAG employs structured debates among proponents, opponents, and judges to refine retrieval quality and ensure factual reliability. In generation, DRAG introduces asymmetric information roles and adversarial debates, enhancing reasoning robustness and mitigating factual inconsistencies. Evaluations across multiple tasks demonstrate that DRAG improves retrieval reliability, reduces RAG-induced hallucinations, and significantly enhances overall factual accuracy. Our code is available at https://github.com/Huenao/Debate-Augmented-RAG.
AB - Retrieval-Augmented Generation (RAG) enhances factual accuracy by integrating external knowledge, yet it introduces a critical issue: erroneous or biased retrieval can mislead generation, compounding hallucinations, a phenomenon we term Hallucination on Hallucination. To address this, we propose Debate-Augmented RAG (DRAG), a training-free framework that integrates Multi-Agent Debate (MAD) mechanisms into both retrieval and generation stages. In retrieval, DRAG employs structured debates among proponents, opponents, and judges to refine retrieval quality and ensure factual reliability. In generation, DRAG introduces asymmetric information roles and adversarial debates, enhancing reasoning robustness and mitigating factual inconsistencies. Evaluations across multiple tasks demonstrate that DRAG improves retrieval reliability, reduces RAG-induced hallucinations, and significantly enhances overall factual accuracy. Our code is available at https://github.com/Huenao/Debate-Augmented-RAG.
UR - https://www.scopus.com/pages/publications/105021009081
U2 - 10.18653/v1/2025.acl-long.770
DO - 10.18653/v1/2025.acl-long.770
M3 - Conference article published in proceeding or book
AN - SCOPUS:105021009081
T3 - Proceedings of the Annual Meeting of the Association for Computational Linguistics
SP - 15839
EP - 15853
BT - Long Papers
A2 - Che, Wanxiang
A2 - Nabende, Joyce
A2 - Shutova, Ekaterina
A2 - Pilehvar, Mohammad Taher
PB - Association for Computational Linguistics (ACL)
Y2 - 27 July 2025 through 1 August 2025
ER -