arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2602.09552cs.CLcs.AIcs.IR
收藏

跨多领域对话问答中RAG方法的全面比较

Comprehensive Comparison of RAG Methods Across Multi-Domain Conversational QA

Hub of Computing and Data Science (HCDS) University of Hamburg（计算与数据科学中心（HCDS）乌姆斯大学）

Klejda Alushi, Jan Strich, Chris Biemann, Martin Semmann

备注：
Accepted to EACL SRW 26
阅读 PDF
arXiv 页面 ↗HTML 原文 ↗

AI总结：

本文通过跨多领域对话问答数据集的全面比较，发现简单策略在多轮对话中表现更优，而复杂方法未必有效，强调检索策略与数据集结构的匹配至关重要。

摘要：

对话问答越来越多地依赖检索增强生成（RAG）来使大型语言模型（LLMs）基于外部知识。然而，大多数现有研究对RAG方法进行孤立评估，主要集中在单轮设置上。本文针对多轮对话问答中RAG方法系统比较的缺乏问题，其中对话历史、核心ference和用户意图的转移显著增加了检索的难度。我们对八个多样化的对话问答数据集进行了全面的实证研究，这些数据集覆盖多个领域。使用统一的实验设置，我们评估了检索质量及答案生成，使用生成器和检索指标进行评估，并分析了性能在对话轮次中的演变。我们的结果表明，稳健且简单的策略，如重排序、混合BM25和HyDE，始终优于基础RAG。相比之下，几种高级技术未能带来提升，甚至可能使性能低于无RAG基线。我们进一步证明，数据集特征和对话长度强烈影响检索效果，解释了为什么没有单一RAG策略在所有设置中占据主导。总体而言，我们的发现表明，有效的对话RAG取决于检索策略与数据集结构之间的匹配，而非方法的复杂性。我们发布了所用代码。\footnote{\href{https://github.com/Klejda-A/exp-rag.git}{GitHub Repository}}

英文摘要：

Conversational question answering increasingly relies on retrieval-augmented generation (RAG) to ground large language models (LLMs) in external knowledge. Yet, most existing studies evaluate RAG methods in isolation and primarily focus on single-turn settings. This paper addresses the lack of a systematic comparison of RAG methods for multi-turn conversational QA, where dialogue history, coreference, and shifting user intent substantially complicate retrieval. We present a comprehensive empirical study of vanilla and advanced RAG methods across eight diverse conversational QA datasets spanning multiple domains. Using a unified experimental setup, we evaluate retrieval quality and answer generation using generator and retrieval metrics, and analyze how performance evolves across conversation turns. Our results show that robust yet straightforward methods, such as reranking, hybrid BM25, and HyDE, consistently outperform vanilla RAG. In contrast, several advanced techniques fail to yield gains and can even degrade performance below the No-RAG baseline. We further demonstrate that dataset characteristics and dialogue length strongly influence retrieval effectiveness, explaining why no single RAG strategy dominates across settings. Overall, our findings indicate that effective conversational RAG depends less on method complexity than on alignment between the retrieval strategy and the dataset structure. We publish the code used.\footnote{\href{https://github.com/Klejda-A/exp-rag.git}{GitHub Repository}}

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。

↑