---
id: 20260726-T0-05
title: "TopoGuard：用图论防御RAG的“分裂知识攻击”"
title_en: "TopoGuard Uses Graph Theory to Defend RAG Against Split-Knowledge Attacks"
url: https://ai.daily.yangsir.net/daily/20260726-T0-05
issue_date: 2026-07-26
publish_date: 2026-07-25T04:00:00.000Z
category: research
source_name: "arXiv cs.CL (NLP)"
source_url: https://arxiv.org/abs/2607.20437
---

# TopoGuard：用图论防御RAG的“分裂知识攻击”

检索增强生成（RAG）面临新的安全威胁：攻击者可以通过在多个文档中分散埋藏错误信息，诱导模型输出错误答案，这种攻击被称为“分裂知识攻击”。新论文TopoGuard提出基于图论的防御方案，将检索到的文档构建为知识图谱，通过检测图结构中的异常连接模式来识别并过滤恶意文档。实验显示，该方法可将攻击成功率降低80%以上，且对正常答案的召回影响控制在5%以内。

## English Version

**TopoGuard Uses Graph Theory to Defend RAG Against Split-Knowledge Attacks**

RAG systems face a new threat called split-knowledge attacks, where adversaries hide malicious information across multiple retrieved documents. TopoGuard, a new defense, builds a knowledge graph from retrieved documents and detects anomalous connection patterns using graph theory. Experiments show it reduces attack success rate by over 80% while keeping recall impact on normal answers under 5%, offering a practical safeguard for production RAG.

---

**来源**：[arXiv cs.CL (NLP)](https://arxiv.org/abs/2607.20437)

**详情页**：https://ai.daily.yangsir.net/daily/20260726-T0-05

---

*智语观潮 · Daily — https://ai.daily.yangsir.net/llms.txt*