FastInsight: Fast and Insightful Retrieval via Fusion Operators for Graph RAG

2026-01-26

AI Summary

This work addresses the limitations of existing graph-based RAG methods, which rely on time-consuming LLM inference for insightful retrieval and suffer from semantic blind spots in graph search and topological blind spots in model reasoning. To overcome these challenges, we propose FastInsight, a novel framework that introduces the first triadic operator taxonomy for graph retrieval and designs two new fusion operators: the Graph Re-ranker (GRanker) and Semantic-Topological eXpansion (STeX). These components synergistically integrate vector search, graph structure, and model reasoning. Extensive experiments demonstrate that FastInsight significantly outperforms current state-of-the-art approaches across multiple retrieval and generation benchmarks, achieving a Pareto-optimal improvement that balances both effectiveness and efficiency.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Data Mining & Knowledge Management: Graph Mining, Social Network Analysis & Community
• Knowledge Representation and Reasoning: Computational Complexity of Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Existing Graph RAG methods aiming for insightful retrieval on corpus graphs typically rely on time-intensive processes that interleave Large Language Model (LLM) reasoning. To enable time-efficient insightful retrieval, we propose FastInsight. We first introduce a graph retrieval taxonomy that categorizes existing methods into three fundamental operations: vector search, graph search, and model-based search. Through this taxonomy, we identify two critical limitations in current approaches: the topology-blindness of model-based search and the semantics-blindness of graph search. FastInsight overcomes these limitations by interleaving two novel fusion operators: the Graph-based Reranker (GRanker), which functions as a graph model-based search, and Semantic-Topological eXpansion (STeX), which operates as a vector-graph search. Extensive experiments on broad retrieval and generation datasets demonstrate that FastInsight significantly improves both retrieval accuracy and generation quality compared to state-of-the-art baselines, achieving a substantial Pareto improvement in the trade-off between effectiveness and efficiency.

Problem

• Graph RAG
• insightful retrieval
• topology-blindness
• semantics-blindness
• retrieval efficiency

Innovation

• Graph RAG
• Fusion Operators
• Graph-based Reranker
• Semantic-Topological Expansion
• Efficient Retrieval