GraphRunner: A Multi-Stage Framework for Efficient and Accurate Graph-Based Retrieval

TL;DR

GraphRunner is a three-stage graph-based retrieval framework that enhances accuracy and efficiency by planning, verifying, and executing multi-hop traversals, significantly outperforming existing methods in robustness and speed.

Contribution

It introduces a novel multi-stage framework with high-level traversal planning and verification, reducing reasoning errors and hallucinations in graph-based retrieval.

Findings

Achieves 10-50% performance improvements over baselines

Reduces inference cost by 3.0-12.9x

Speeds up response generation by 2.5-7.1x

Abstract

Conventional Retrieval Augmented Generation (RAG) approaches are common in text-based applications. However, they struggle with structured, interconnected datasets like knowledge graphs, where understanding underlying relationships is crucial for accurate retrieval. A common direction in graph-based retrieval employs iterative, rule-based traversal guided by Large Language Models (LLMs). Such existing iterative methods typically combine reasoning with single hop traversal at each step, making them vulnerable to LLM reasoning errors and hallucinations that ultimately hinder the retrieval of relevant information. To address these limitations, we propose GraphRunner, a novel graph-based retrieval framework that operates in three distinct stages: planning, verification, and execution. This introduces high-level traversal actions that enable multi-hop exploration in a single step. It also generates a holistic traversal plan, which is verified against the graph structure and pre-defined traversal actions, reducing reasoning errors and detecting hallucinations before execution. GraphRunner significantly reduces LLM reasoning errors and detects hallucinations through validation. Our evaluation using the GRBench dataset shows that GraphRunner consistently outperforms existing approaches, achieving 10-50% performance improvements over the strongest baseline while reducing inference cost by 3.0-12.9x and response generation time by 2.5-7.1x, making it significantly more robust and efficient for graph-based retrieval tasks.