ArchRAG: Attributed Community-based Hierarchical Retrieval-Augmented Generation
Shu Wang, Yuchi Ma, Xilin Liu, Yixiang Fang, Yingli Zhou
Proceedings of the AAAI Conference on Artificial Intelligence
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 (4)
Attributed community hierarchical RAG: The paper proposes ArchRAG, a graph-based RAG approach that augments questions using attributed communities and uses hierarchical clustering.
Attributed community hierarchical index retrieval: The paper builds a hierarchical index structure for attributed communities and an online retrieval method to retrieve graph information relevant to a question.
Attributed community hierarchical RAG: The paper proposes ArchRAG, a graph-based RAG approach that augments questions using attributed communities and uses hierarchical clustering.
Attributed community hierarchical index retrieval: The paper builds a hierarchical index structure for attributed communities and an online retrieval method to retrieve graph information relevant to a question.
Results (2)
Higher accuracy and lower token cost:
Higher accuracy and lower token cost:
Research Domain
Graph-based retrieval-augmented generation for question answering