Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

G-reasoner: Foundation Models for Unified Reasoning over Graph-structured Knowledge

Linhao Luo, Zicheng Zhao, Junnan Liu, Zhangchi Qiu, Junnan Dong, Serge Panev, Chen Gong, Thuy-Trang Vu, Gholamreza Haffari, Dinh Phung, Alan Wee-Chung Liew, Shirui Pan

TL;DR

G-reasoner is presented, a unified framework that integrates graph and language foundation models for scalable reasoning over diverse graph-structured knowledge and introduces a 34M-parameter graph foundation model that jointly captures graph topology and textual semantics, and is integrated with LLMs to enhance reasoning in downstream applications.

Abstract

Large language models (LLMs) excel at complex reasoning but remain limited by static and incomplete parametric knowledge. Retrieval-augmented generation (RAG) mitigates this by incorporating external knowledge, yet existing RAGs struggle with knowledge-intensive tasks due to fragmented information and weak modeling of knowledge structure. Graphs offer a natural way to model relationships within knowledge, but LLMs are inherently unstructured and cannot effectively reason over graph-structured data. Recent graph-enhanced RAG (GraphRAG) attempts to bridge this gap by constructing tailored graphs and enabling LLMs to reason on them. However, these methods often depend on ad-hoc graph designs, heuristic search, or costly agent pipelines, which hinder scalability and generalization. To address these challenges, we present G-reasoner, a unified framework that integrates graph and language foundation models for scalable reasoning over diverse graph-structured knowledge. Central to our approach is QuadGraph, a standardized four-layer abstraction that unifies heterogeneous knowledge sources into a common graph representation. Building on this, we introduce a 34M-parameter graph foundation model (GFM) that jointly captures graph topology and textual semantics, and is integrated with LLMs to enhance reasoning in downstream applications. To ensure scalability and efficiency, mixed-precision training and distributed message-passing are implemented to scale GFM with more GPUs. Extensive experiments on six benchmarks show that G-reasoner consistently outperforms state-of-the-art baselines, significantly enhances LLM reasoning, and achieves strong efficiency and cross-graph generalization.

G-reasoner: Foundation Models for Unified Reasoning over Graph-structured Knowledge

TL;DR

G-reasoner is presented, a unified framework that integrates graph and language foundation models for scalable reasoning over diverse graph-structured knowledge and introduces a 34M-parameter graph foundation model that jointly captures graph topology and textual semantics, and is integrated with LLMs to enhance reasoning in downstream applications.

Abstract

Large language models (LLMs) excel at complex reasoning but remain limited by static and incomplete parametric knowledge. Retrieval-augmented generation (RAG) mitigates this by incorporating external knowledge, yet existing RAGs struggle with knowledge-intensive tasks due to fragmented information and weak modeling of knowledge structure. Graphs offer a natural way to model relationships within knowledge, but LLMs are inherently unstructured and cannot effectively reason over graph-structured data. Recent graph-enhanced RAG (GraphRAG) attempts to bridge this gap by constructing tailored graphs and enabling LLMs to reason on them. However, these methods often depend on ad-hoc graph designs, heuristic search, or costly agent pipelines, which hinder scalability and generalization. To address these challenges, we present G-reasoner, a unified framework that integrates graph and language foundation models for scalable reasoning over diverse graph-structured knowledge. Central to our approach is QuadGraph, a standardized four-layer abstraction that unifies heterogeneous knowledge sources into a common graph representation. Building on this, we introduce a 34M-parameter graph foundation model (GFM) that jointly captures graph topology and textual semantics, and is integrated with LLMs to enhance reasoning in downstream applications. To ensure scalability and efficiency, mixed-precision training and distributed message-passing are implemented to scale GFM with more GPUs. Extensive experiments on six benchmarks show that G-reasoner consistently outperforms state-of-the-art baselines, significantly enhances LLM reasoning, and achieves strong efficiency and cross-graph generalization.

Paper Structure

This paper contains 31 sections, 9 equations, 7 figures, 16 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Preliminary
  4. Approach
  5. Unified Graph Interface: QuadGraph
  6. Graph Foundation Model Reasoning
  7. Language Foundation Model Reasoning
  8. Experiment
  9. Experimental Setup
  10. Main Results (RQ1)
  11. Generalization Across Graph Structures (RQ2)
  12. Ablation Study (RQ3)
  13. Efficiency Analysis (RQ4)
  14. Conclusion
  15. Appendix
  16. ...and 16 more sections

Figures (7)

  • Figure 1: The overall framework of G-reasoner. First, G-reasoner provides a unified graph interface, QuadGraph, that integrates diverse graph-structured knowledge from different domains into a standard format. Then, it adopts a GNN-powered foundation model to jointly reason over the graph-structured knowledge and make versatile predictions. Last, we enhance the LLMs with the graph reasoning results to improve the performance on downstream applications.
  • Figure 2: Illustration of QuadGraph for unifying existing graph-structured knowledge.
  • Figure 3: Memory and throughput gain brought by mixed precision training.
  • Figure 4: Compute scaling of G-reasoner.
  • Figure 5: The illustration of distributed message passing in G-reasoner.
  • ...and 2 more figures