analiza aiqbench
This research introduces HyGRL, a novel framework addressing the limitations of current retrieval-augmented generation by integrating unstructured text with structured knowledge graphs for complex multi-entity reasoning. By employing imitation learning followed by reinforcement learning, it achieves high accuracy and low token costs while overcoming structural sparsity issues inherent in traditional graph-based approaches.
HyGRL: Adaptive Hybrid Graph Reasoning for Multi-Entity Questions
arXiv:2607.19398v1 Announce Type: new 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) .
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 this https URL) .
Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
arXiv:2607.19398 [cs.AI]
(or arXiv:2607.19398v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2607.19398 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 222 cuvinte
Subiecte conexe
From RAG to Agentic AI: Building the Next Generation of Intelligent Enterprise Systems
KDnuggets · 8 sept. 2026
Presentation: Engineering AI for Creativity and Curiosity on Mobile
InfoQ · 21 iul. 2026
Vector RAG vs. GraphRAG: Which retrieval do you need?
Neo4j · 23 sept. 2026
Weaving Multi-Source Evidence for Biomedical Reasoning: The BioMedHop Benchmark and BioWeave Framework
arXiv · 17 iun. 2026