Graph-Anchored Knowledge Indexing for Retrieval-Augmented Generation
This addresses the problem of mitigating hallucinations in LLMs for users of RAG systems, though it is incremental as it builds on existing graph-based methods by making them active and evolving.
The paper tackles the challenge of integrating scattered evidence in noisy documents for Retrieval-Augmented Generation (RAG) by proposing GraphAnchor, a graph-anchored knowledge indexing approach that incrementally updates a graph during retrieval to guide LLMs, resulting in improved performance on four multi-hop question answering benchmarks.
Retrieval-Augmented Generation (RAG) has emerged as a dominant paradigm for mitigating hallucinations in Large Language Models (LLMs) by incorporating external knowledge. Nevertheless, effectively integrating and interpreting key evidence scattered across noisy documents remains a critical challenge for existing RAG systems. In this paper, we propose GraphAnchor, a novel Graph-Anchored Knowledge Indexing approach that reconceptualizes graph structures from static knowledge representations into active, evolving knowledge indices. GraphAnchor incrementally updates a graph during iterative retrieval to anchor salient entities and relations, yielding a structured index that guides the LLM in evaluating knowledge sufficiency and formulating subsequent subqueries. The final answer is generated by jointly leveraging all retrieved documents and the final evolved graph. Experiments on four multi-hop question answering benchmarks demonstrate the effectiveness of GraphAnchor, and reveal that GraphAnchor modulates the LLM's attention to more effectively associate key information distributed in retrieved documents. All code and data are available at https://github.com/NEUIR/GraphAnchor.