Know3-RAG: A Knowledge-aware RAG Framework with Adaptive Retrieval, Generation, and Filtering
This addresses the critical concern of factual unreliability in large language models for applications like open-domain question answering, representing an incremental improvement over existing RAG systems.
The paper tackles the problem of hallucinations and unreliable adaptive control in Retrieval-Augmented Generation (RAG) systems by proposing Know3-RAG, a framework that integrates structured knowledge from knowledge graphs to guide retrieval, generation, and filtering, resulting in significant reductions in hallucinations and enhanced answer reliability on multiple open-domain QA benchmarks.
Recent advances in large language models (LLMs) have led to impressive progress in natural language generation, yet their tendency to produce hallucinated or unsubstantiated content remains a critical concern. To improve factual reliability, Retrieval-Augmented Generation (RAG) integrates external knowledge during inference. However, existing RAG systems face two major limitations: (1) unreliable adaptive control due to limited external knowledge supervision, and (2) hallucinations caused by inaccurate or irrelevant references. To address these issues, we propose Know3-RAG, a knowledge-aware RAG framework that leverages structured knowledge from knowledge graphs (KGs) to guide three core stages of the RAG process, including retrieval, generation, and filtering. Specifically, we introduce a knowledge-aware adaptive retrieval module that employs KG embedding to assess the confidence of the generated answer and determine retrieval necessity, a knowledge-enhanced reference generation strategy that enriches queries with KG-derived entities to improve generated reference relevance, and a knowledge-driven reference filtering mechanism that ensures semantic alignment and factual accuracy of references. Experiments on multiple open-domain QA benchmarks demonstrate that Know3-RAG consistently outperforms strong baselines, significantly reducing hallucinations and enhancing answer reliability.