Structure-Aware RAG: Structured Retrieval Augmented Generation from Noisy Data for Conversational Agents

2026-05-22

AI Summary

This work addresses the limitations of existing retrieval-augmented generation (RAG) approaches in effectively integrating dynamic and domain-specific information when confronted with noisy or irrelevant context. To mitigate this issue, the authors propose a structure-aware RAG framework (SA-RAG) that leverages tables as structured intermediate representations to reduce noise interference while preserving essential knowledge. SA-RAG incorporates a quality-aware table metadata generation mechanism, combining both training-free and trainable table construction strategies, and employs generation verification alongside direct preference optimization to ensure semantic and structural consistency. Experimental results demonstrate that SA-RAG significantly outperforms current RAG baselines on two real-world noisy datasets, and the implementation has been made publicly available.

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 text-based and graph-based RAG methods often struggle with noisy or irrelevant contexts. In this work, we propose Structure-aware Retrieval Augmented Generation (SA-RAG), which uses tables as an intermediate structured representation to provide a compact and controllable interface that reduces noise while preserving essential information. We introduce a quality-aware table metadata generation framework that models metadata normalization and effectiveness, improving metadata quality and downstream performance. Furthermore, we explore both training-free and training-based table generation methods. Generation validation and direct preference optimization further improve table quality while maintaining semantic and structural consistency. Experiments on two noisy real-world datasets show that SA-RAG significantly outperforms existing RAG baselines. Our code is publicly available at a public repository.

Problem

• Retrieval-Augmented Generation
• noisy data
• conversational agents
• structured representation
• external knowledge

Innovation

• Structure-aware RAG
• structured representation
• table-based retrieval
• metadata normalization
• noise reduction