arXiv 2506.225182025-06-26cs.CLcs.AI

Weak-to-Strong GraphRAG：将弱检索器与大语言模型对齐以用于基于图的检索增强生成

Weak-to-Strong GraphRAG: Aligning Weak Retrievers with Large Language Models for Graph-based Retrieval Augmented Generation

The Chinese University of Hong Kong（香港中文大学）；Georgia Institute of Technology（佐治亚理工学院）；Hong Kong Baptist University（香港 Baptist大学）

Deyu Zou, Yongqiang Chen, Mufei Li, Siqi Miao, Chenxi Liu, Bo Han, James Cheng, Pan Li

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

本文提出ReG框架，通过引入LLM反馈消除虚假信号和结构感知重组模块构建证据链，对齐弱检索器与LLMs以改进基于图的RAG，显著提升性能并降低推理成本。

摘要：

基于图的检索增强生成（RAG）使大语言模型（LLMs）能够利用来自最新知识图谱（KGs）的结构化外部知识来为响应提供依据，并减少幻觉。然而，LLMs在基于图的RAG中通常依赖弱检索器：I）由于缺乏真实标签，检索器通常在弱监督下训练，这往往会向LLMs引入虚假信号。II）由于图数据的抽象性，检索到的知识通常以无组织的形式呈现。为缓解此问题，我们提出了精炼的基于图的RAG（Refined Graph-based RAG, ReG），以在基于图的RAG中将弱检索器与LLMs对齐。具体而言，ReG结合LLM反馈来消除虚假信号并提高监督质量。同时，ReG引入了结构感知重组模块，将检索结果重构为逻辑连贯的证据链。在主要基准上的实验表明，ReG在不同LLM骨干网络上显著且一致地带来了高达10%的提升。提升的监督质量使ReG能够仅用5%的训练数据就达到最先进的性能，并能迁移到分布外的KGs。值得注意的是，当应用于基于推理的LLMs时，ReG将推理token成本降低了高达30%，并将性能提升了高达4%。

英文摘要：

Graph-based retrieval-augmented generation (RAG) enables large language models (LLMs) to ground responses with structured external knowledge from up-to-date knowledge graphs (KGs) and reduce hallucinations. However, LLMs often rely on a weak retriever in graph-based RAG: I) Due to the lack of ground truth, the retriever is often trained on weak supervision, which often introduces spurious signals to the LLMs. II) Due to the abstraction of graph data, the retrieved knowledge is often presented in unorganized forms. To mitigate the issue, we present Refined Graph-based RAG (ReG) to align weak retrievers to LLMs for graph-based RAG. Specifically, ReG incorporates LLM feedback to get rid of spurious signals and improve the quality of the supervision. Meanwhile, ReG introduces a structure-aware reorganization module to refactor the retrieval results into logically coherent evidence chains. Experiments on prominent benchmarks demonstrate that ReG significantly and consistently brings improvements across different LLM backbones by up to 10%. The improved supervision quality enables ReG to match the state-of-the-art performance with 5% training data and to transfer to out-of-distribution KGs. Notably, when adopted to reasoning-based LLMs, ReG reduces the reasoning token cost by up to 30% and improves the performance by up to 4%.

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。