SentGraph: Hierarchical Sentence Graph for Multi-hop Retrieval-Augmented Question Answering

2026-01-06

arXiv.org

AI Summary

This work addresses the limitations of traditional retrieval-augmented generation (RAG) approaches in multi-hop question answering, which often fail to construct coherent evidence chains and are prone to reasoning errors due to irrelevant or logically inconsistent retrieved contexts. To overcome this, the authors propose SentGraph, the first framework to integrate Rhetorical Structure Theory into multi-hop RAG. SentGraph constructs a hierarchical sentence graph composed of nucleus and satellite sentences, and links cross-document entities to form topical subgraphs, thereby explicitly modeling logical dependencies among sentences. By combining graph neural networks with a graph-guided retrieval strategy, SentGraph enables fine-grained evidence selection and dynamic expansion of reasoning paths. Extensive experiments on four multi-hop QA benchmarks demonstrate that SentGraph significantly outperforms existing methods, validating the effectiveness of sentence-level logical modeling for enhancing multi-hop reasoning capabilities.

Technology Category

• Knowledge Representation and Reasoning: Computational Complexity of Reasoning
• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.
• Reasoning under Uncertainty: Graphical Models

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Traditional Retrieval-Augmented Generation (RAG) effectively supports single-hop question answering with large language models but faces significant limitations in multi-hop question answering tasks, which require combining evidence from multiple documents. Existing chunk-based retrieval often provides irrelevant and logically incoherent context, leading to incomplete evidence chains and incorrect reasoning during answer generation. To address these challenges, we propose SentGraph, a sentence-level graph-based RAG framework that explicitly models fine-grained logical relationships between sentences for multi-hop question answering. Specifically, we construct a hierarchical sentence graph offline by first adapting Rhetorical Structure Theory to distinguish nucleus and satellite sentences, and then organizing them into topic-level subgraphs with cross-document entity bridges. During online retrieval, SentGraph performs graph-guided evidence selection and path expansion to retrieve fine-grained sentence-level evidence. Extensive experiments on four multi-hop question answering benchmarks demonstrate the effectiveness of SentGraph, validating the importance of explicitly modeling sentence-level logical dependencies for multi-hop reasoning.

Problem

• multi-hop question answering
• retrieval-augmented generation
• evidence integration
• logical coherence
• sentence-level reasoning

Innovation

• sentence-level graph
• multi-hop reasoning
• Rhetorical Structure Theory
• graph-guided retrieval
• Retrieval-Augmented Generation