arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2604.20199cs.CL
收藏

所有语言都重要:理解并缓解多语言RAG中的语言偏差

All Languages Matter: Understanding and Mitigating Language Bias in Multilingual RAG

Chinese Information Processing Laboratory, Institute of Software, Chinese Academy of Sciences(中国科学院软件研究所信息处理实验室);University of Chinese Academy of Sciences(中国科学院大学);MYbank, AntGroup(蚂蚁集团MYbank)

Dan Wang, Guozhao Mo, Yafei Shi, Cheng Zhang, Bo Zheng, Boxi Cao, Xuanang Chen, Yaojie Lu, Hongyu Lin, Ben He, Xianpei Han, Le Sun

AI总结:

本文研究多语言RAG中的语言偏差问题,提出LAURA方法,通过多语言证据排名与生成效用对齐,有效缓解语言偏差并提升性能。

摘要:

多语言检索增强生成(mRAG)利用跨语言证据来使大型语言模型(LLMs)扎根于全球知识。然而,我们显示当前mRAG系统在重排序过程中存在语言偏差,系统性地偏向英语和查询的母语。通过引入估计的 oracle 证据分析,我们量化了现有重排序器与可实现上界之间的显著性能差距。进一步分析揭示了关键的分布不匹配:虽然最优预测需要证据分散在多种语言中,当前系统系统性地压制此类『答案关键』文档,从而限制下游生成性能。为弥合这一差距,我们提出 Language-Agnostic Utility-driven Reranker Alignment(LAURA),该方法将多语言证据排名与下游生成效用对齐。在多种语言和生成模型上的实验表明,LAURA有效缓解了语言偏差,并一致提升了mRAG性能。

英文摘要:

Multilingual Retrieval-Augmented Generation (mRAG) leverages cross-lingual evidence to ground Large Language Models (LLMs) in global knowledge. However, we show that current mRAG systems suffer from a language bias during reranking, systematically favoring English and the query's native language. By introducing an estimated oracle evidence analysis, we quantify a substantial performance gap between existing rerankers and the achievable upper bound. Further analysis reveals a critical distributional mismatch: while optimal predictions require evidence scattered across multiple languages, current systems systematically suppress such ``answer-critical'' documents, thereby limiting downstream generation performance. To bridge this gap, we propose \textit{\textbf{L}anguage-\textbf{A}gnostic \textbf{U}tility-driven \textbf{R}eranker \textbf{A}lignment (LAURA)}, which aligns multilingual evidence ranking with downstream generative utility. Experiments across diverse languages and generation models show that LAURA effectively mitigates language bias and consistently improves mRAG performance.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑