Dual hypergraph indexing improves multi-hop reasoning in knowledge retrieval
Dual-Hypergraph Indexing: Bridging Knowledge Islands for Multi-Hop Reasoning in Retrieval-Augmented Generation
Information Retrieval
Summary
When computers try to answer complex questions, they need to connect many small facts together like building blocks. The authors found that current methods treat these facts as isolated, making it hard to link them logically over time or through cause and effect. They created a new approach called Dual-Hypergraph Indexing that organizes facts into two connected layers, helping to group related ideas and track how information changes over time. This method leads to better performance in reasoning tasks, especially in areas like medical diagnosis where understanding complex relationships is crucial.
What this means in practice
- •For clinical decision support teams: Improve complex medical diagnosis reasoning by better connecting related medical facts over time for more coherent reasoning processes.
- •For enterprise knowledge management teams: Enhance retrieval systems that need to integrate multiple connected pieces of business knowledge for in-depth causal and timeline analyses.
Authors
Qi Sun, Xingliang Hou, Caibo Li, Yijia Zhang, Qiang Li, Yu Guo
Abstract
While hypergraph-based Retrieval-Augmented Generation (RAG) effectively captures higher-order multi-entity correlations, existing paradigms treat extracted hyperedges as isolated factual assertions. This structural fragmentation engenders rigid "knowledge islands" that bottleneck multi-hop causal inference, temporal tracking, and narrative synthesis. To systematically address these challenges, we introduce Dual-Hypergraph Indexing (DHI), a hierarchical representation framework that elevates discrete facts into structured analytical insights. DHI couples a foundational entity-relation factual hypergraph ($H_K$) with an elevated deep-insight hypergraph ($H_D$) via a dual-pathway aggregation algorithm. Specifically, DHI employs: (1) importance-driven hub aggregation via 5-metric topological profiling and adaptive thresholding to capture spatial semantic clusters; and (2) temporal chunk-chain progressive aggregation via sliding-window greedy exploration to track chronological evolutions. Across five benchmarks, DHI achieves state-of-the-art performance, boosting logical coherence by +1.53 on the multidisciplinary Mix benchmark and scoring 85.78\% on complex medical pathology reasoning tasks. DHI provides a robust architecture for next-generation multi-hop RAG.