HyGRL: Adaptive Hybrid Graph Reasoning for Multi-Entity Questions

A new framework called HyGRL, detailed in a preprint on arXiv, outperforms state-of-the-art baselines in answer accuracy and reasoning fidelity for multi-entity compositional questions while maintaining low token costs and near real-time inference. The method embeds unstructured text into structured knowledge graphs and uses imitation and reinforcement learning to refine reasoning. The code is available on GitHub.

arXiv:2607.19398v1 Announce Type: cross Abstract: Multi-entity compositional questions pose significant challenges to existing retrieval-augmented language models. Conventional methods fall into a dilemma: standard RAG lacks dynamic reasoning, traditional Graph-RAG is limited by structural sparsity, and LLM-constructed Graph-RAG incurs prohibitive costs. We propose \textbf{\fwa}, a unified framework that embeds unstructured text into structured knowledge graphs, creating a heterogeneous network for flexible evidence retrieval. Reasoning is formulated as adaptive structure induction, learned via a robust two-stage process: 1 imitation learning distills heuristic expert signals, and 2 reinforcement learning refines the policy using LLM-driven preference rewards. Experiments demonstrate that {\fwa} effectively merges textual richness with structural knowledge, outperforming SOTA baselines in answer accuracy and reasoning fidelity while maintaining extremely low token costs and near real-time inference code available at https://github.com/wjywjy123/HyGRL .