SiReRAG: Indexing Similar and Related Information for Multihop Reasoning

2024-12-09

arXiv.org

AI Summary

Existing RAG systems predominantly rely on single-dimensional indexing—either semantic similarity or relevance—limiting their capacity for multi-hop reasoning and knowledge composition. To address this, we propose a novel dual-tree indexing framework that jointly models semantic similarity and entity-driven relevance: a similarity tree and a relevance tree are constructed separately and then synergistically integrated via recursive summarization, proposition-entity extraction, shared-entity clustering, and dual-tree flattening. We further enhance retrieval with a re-ranking module. Evaluated on MuSiQue, 2WikiMultiHopQA, and HotpotQA, our method achieves an average F1 gain of +1.9%; when combined with state-of-the-art re-rankers, the improvement reaches up to +7.8%. Our core contribution is the first dual-tree indexing paradigm that explicitly unifies similarity and relevance modeling, significantly strengthening multi-hop knowledge association and compositional reasoning in RAG.

Technology Category

• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Knowledge Representation and Reasoning: Computational Complexity of Reasoning
• Search and Optimization: Metareasoning and Metaheuristics

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: Querying, indexing, and retrieval in Web-related graphs

Abstract

Indexing is an important step towards strong performance in retrieval-augmented generation (RAG) systems. However, existing methods organize data based on either semantic similarity (similarity) or related information (relatedness), but do not cover both perspectives comprehensively. Our analysis reveals that modeling only one perspective results in insufficient knowledge synthesis, leading to suboptimal performance on complex tasks requiring multihop reasoning. In this paper, we propose SiReRAG, a novel RAG indexing approach that explicitly considers both similar and related information. On the similarity side, we follow existing work and explore some variances to construct a similarity tree based on recursive summarization. On the relatedness side, SiReRAG extracts propositions and entities from texts, groups propositions via shared entities, and generates recursive summaries to construct a relatedness tree. We index and flatten both similarity and relatedness trees into a unified retrieval pool. Our experiments demonstrate that SiReRAG consistently outperforms state-of-the-art indexing methods on three multihop datasets (MuSiQue, 2WikiMultiHopQA, and HotpotQA), with an average 1.9% improvement in F1 scores. As a reasonably efficient solution, SiReRAG enhances existing reranking methods significantly, with up to 7.8% improvement in average F1 scores. Our code is available at https://github.com/SalesforceAIResearch/SiReRAG .

Problem

• Addresses insufficient knowledge synthesis in RAG indexing
• Combines both semantic similarity and related information perspectives
• Improves multihop reasoning performance in complex tasks

Innovation

• Dual indexing with similarity and relatedness trees
• Entity-based proposition grouping for relatedness
• Unified retrieval pool from flattened tree structures