arXiv cs.CL· Kaiqiao Han, LuAn Tang, Renliang Sun, Peng Yuan, Wei Cheng, Haoyu Wang, Wei Wang, Yizhou Sun, Haifeng Chen·· 2 天前AI 评分26
SA-RAG:从噪声对话中构建结构化表示以增强 RAG
SA-RAG: Structuring Noisy Conversations for Retrieval-Augmented Generation
AI 导读
SA-RAG 提出一种结构感知的检索增强生成框架,可直接从噪声对话数据中归纳出以表格为载体的可靠中间结构,并联合建模语义与结构一致性。该框架同时探索免训练与基于训练的结构生成策略,在两个真实噪声数据集上显著优于现有 RAG 基线。分析表明高质量结构表示对稳健的 RAG 应用至关重要。
正文 · 原文
Abstract:Large Language Models (LLMs) have been widely adopted in conversational applications. However, their reliance on parametric knowledge limits reliability in real-world scenarios that require dynamic or domain-specific information. Retrieval-Augmented Generation (RAG) addresses this limitation by incorporating external knowledge during generation, but existing RAG methods often struggle with noisy or irrelevant contexts, as they typically assume that high-quality structures are already available. In this work, we propose Structure-aware Retrieval-Augmented Generation (SA-RAG), a principled structure induction and generation framework that constructs reliable intermediate structures directly from noisy conversational data. SA-RAG uses tables as compact and controllable structured representations. We introduce a quality-aware structure induction framework that jointly models semantic and structural consistency. Furthermore, we explore both training-free and training-based structure generation strategies to further improve structure quality while maintaining fidelity. Experiments on two noisy real-world datasets show that SA-RAG significantly outperforms existing RAG baselines. Further analysis highlights that high-quality structural representations are crucial for robust RAG applications.
| Subjects: | Computation and Language (cs.CL); Machine Learning (cs.LG) |
| Cite as: | arXiv:2605.24366 [cs.CL] |
| (or arXiv:2605.24366v2 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2605.24366 arXiv-issued DOI via DataCite |
Submission history
From: Kaiqiao Han [view email]
[v1]
Sat, 23 May 2026 03:07:33 UTC (958 KB)
[v2]
Fri, 13 Nov 2026 03:34:05 UTC (981 KB)
来源:arXiv cs.CL · arxiv.org