ArchRAG: Attributed Community-based Hierarchical Retrieval-Augmented Generation
Association for Artificial Intelligence 2026, Yingli Zhou, Shu Wang, Mayuchi Mayuchi, Xilin Liu
Open MIND
Problems Identified (4)
Graph RAG relevance identification: Existing graph-based RAG approaches cannot accurately identify relevant information from graph data for questions.
Graph RAG token cost: Existing graph-based RAG approaches consume many tokens during online retrieval.
Graph RAG relevance identification: Existing graph-based RAG approaches cannot accurately identify relevant information from graph data for questions.
Graph RAG token cost: Existing graph-based RAG approaches consume many tokens during online retrieval.
Proposed Solutions (5)
ArchRAG: ArchRAG is a graph-based RAG approach that uses attributed communities and hierarchical retrieval.
Attributed community question augmentation: The approach augments questions using attributed communities.
LLM hierarchical clustering: The approach introduces an LLM-based hierarchical clustering method.
Hierarchical community index retrieval: The approach builds a hierarchical index over attributed communities and uses an online retrieval method to retrieve relevant graph information.
ArchRAG: ArchRAG is a graph-based RAG approach that uses attributed communities and hierarchical retrieval.
Results (2)
Improved accuracy and token cost:
Improved accuracy and token cost:
Research Domain
Graph-based retrieval-augmented generation for LLM question answering