用动态图结构提升多实体问题的推理能力，兼顾准确与效率。
HyGRL: Adaptive Hybrid Graph Reasoning for Multi-Entity Questions
- 将文本转为异构知识图，自适应构建推理结构
- 两阶段学习：模仿专家信号+强化学习优化策略
- 比现有方法更准更快，推理成本极低
多实体组合问题对现有检索增强型语言模型构成挑战。传统方法面临困境：标准RAG缺乏动态推理能力，传统Graph-RAG受限于结构稀疏性，而由大模型构建的Graph-RAG开销过大。本文提出HyGRL，一种统一框架，将非结构化文本嵌入结构化知识图谱，构建异构网络以实现灵活证据检索。推理被建模为自适应结构生成，通过稳健的两阶段训练过程学习：(1) 通过模仿学习提炼启发式专家信号；(2) 利用大模型驱动的偏好奖励进行强化学习优化策略。实验表明，HyGRL有效融合文本丰富性与结构化知识，在答案准确率和推理一致性上优于当前最优基线，同时保持极低的令牌消耗和近实时推理速度（代码见https://github.com/wjywjy123/HyGRL）。
原文摘要 · Abstract (English)
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) .
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。