GraphReAct: Reasoning and Acting for Multi-step Graph Inference

2026-05-08

AI Summary

This work addresses the challenge of performing multi-hop reasoning on graph-structured data while effectively integrating both topological and semantic information. The authors propose GraphReAct, a novel framework that adapts the Reasoning–Acting (ReAct) paradigm to graph learning for the first time. GraphReAct introduces a graph-specific action space comprising three operations—topological retrieval, semantic retrieval, and context refinement—to dynamically expand and compress the reasoning context. By synergistically combining large language models, graph neural networks, and a dual-path retrieval mechanism, the method achieves substantial performance gains over state-of-the-art approaches across six benchmark datasets, demonstrating the efficacy and superiority of the ReAct framework in graph-based reasoning tasks.

Abstract

Reasoning-acting frameworks enhance large language models (LLMs) by interleaving reasoning with actions for dynamic information acquisition. However, extending this paradigm to graph learning remains underexplored. Graph data is inherently structured, with information distributed across nodes and edges and encoded through both topology and latent representations. As a result, effective reasoning over graphs requires not only retrieving informative evidence from the graph, but also progressively refining the accumulated context during multi-step inference. In this work, we propose GraphReAct, a graph reasoning-acting framework that enables step-by-step inference over graph-structured data. Specifically, we design a graph-based action space with two complementary retrieval actions: topological retrieval, which captures local structural dependencies, and semantic retrieval, which accesses non-local but relevant evidence in the representation space. These actions dynamically expand the reasoning context. To further support multi-step reasoning, we introduce another type of action, context refinement, which distills and reorganizes accumulated information into a compact representation. By interleaving reasoning with both retrieval and refinement actions, our framework enables a progressive transition from context expansion to compression. Extensive experiments on six benchmark datasets demonstrate that GraphReAct consistently outperforms state-of-the-art methods, validating the effectiveness of reasoning-acting for graph learning.

Problem

• graph reasoning
• multi-step inference
• reasoning-acting framework
• graph learning
• context refinement

Innovation

• graph reasoning
• reasoning-acting framework
• topological retrieval
• semantic retrieval
• context refinement