Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: An, Seonho, Hyun, Chaejeong, Kim, Min-Soo
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2601.18579
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866917223787921408
author An, Seonho
Hyun, Chaejeong
Kim, Min-Soo
author_facet An, Seonho
Hyun, Chaejeong
Kim, Min-Soo
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2601_18579
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle FastInsight: Fast and Insightful Retrieval via Fusion Operators for Graph RAG
An, Seonho
Hyun, Chaejeong
Kim, Min-Soo
Information Retrieval
Artificial Intelligence
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.
title FastInsight: Fast and Insightful Retrieval via Fusion Operators for Graph RAG
topic Information Retrieval
Artificial Intelligence
url https://arxiv.org/abs/2601.18579