arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2606.12469cs.CR
收藏

RAG投毒的影响因素

Influence Factors on RAG Poisoning

Polytechnic of Porto(波尔图理工学院);THALES SIX GTS(泰雷兹六世GTS)

Pedro Pereira, Eva Maia, Isabel Praça, Adrien Bécue

AI总结:

通过432种配置的全因子实验,研究数据集、检索器类型、检索深度、数据库组成、分块策略和生成模型对RAG系统投毒鲁棒性的影响,发现检索器架构、数据集和检索深度是主要因素,且投毒脆弱性源于多组件交互。

摘要:

检索增强生成(RAG)系统通过在推理时从外部知识源检索文档来增强大型语言模型,使响应基于检索内容。然而,这种对检索内容的依赖引入了投毒攻击的脆弱性,对抗性文档可以操纵检索过程和生成输出。本文通过涵盖432种配置的全因子实验研究,分析了数据集、检索器类型、检索深度、数据库组成、分块策略和生成模型对检索级和生成级指标的影响。结果表明,检索器架构、数据集和检索深度是影响投毒暴露的最强因素,而生成器选择和数据库组成对下游攻击成功率有重大影响。与BM25相比,密集和基于图的检索器通常提高鲁棒性,而更大的检索深度增加了检索到投毒段落的可能性。我们进一步表明,在多个数据库中复制投毒内容会放大对抗性影响,而额外的干净来源可以缓解这种影响。这些发现强调,RAG中的投毒脆弱性并非归因于单一组件,而是源于检索、生成和知识库配置的交互。

英文摘要:

Retrieval-Augmented Generation (RAG) systems enhance large language models by grounding responses in retrieved documents from external knowledge sources at inference time. However, this reliance on retrieved content introduces vulnerabilities to poisoning attacks, in which adversarial documents can manipulate both the retrieval process and the generated outputs. This paper investigates poisoning robustness in RAG through a full factorial experimental study covering 432 configurations. We analyze the impacts of dataset, retriever type, retrieval depth, database composition, chunking strategy, and generator model on retrieval-level and generation-level metrics. The results show that retriever architecture, dataset, and retrieval depth are the strongest factors affecting poisoning exposure, while generator choice and database composition have a major impact on downstream attack success. Dense and graph-based retrievers generally improve robustness relative to BM25, whereas larger retrieval depth increases the likelihood of retrieving poisoned passages. We further show that replicating poisoned content across multiple databases amplifies adversarial influence, while additional clean sources can mitigate it. These findings highlight that poisoning vulnerability in RAG is not attributable to a single component, but instead arises from the interaction of retrieval, generation, and knowledge-base configuration.

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。

↑