WeKnow-RAG: An Adaptive Approach for Retrieval-Augmented Generation Integrating Web Search and Knowledge Graphs

TL;DR

WeKnow-RAG addresses the reliability gaps of large language models by integrating domain-specific Knowledge Graphs with Web-based Retrieval-Augmented Generation in a multi-stage retrieval framework. The system uses token-level chunking, BM25-based sparse retrieval, dense embedding-based retrieval with reranking, and a self-assessment mechanism to filter low-confidence answers, while KG querying provides precise, structured facts. An adaptive framework balances KG and Web RAG depending on domain dynamics, enabling robust performance in static and evolving information contexts. On the CRAG benchmark, WeKnow-RAG achieves significant gains in accuracy and reductions in hallucinations, demonstrating practical impact for high-stakes factual QA across diverse domains.

Abstract

Large Language Models (LLMs) have greatly contributed to the development of adaptive intelligent agents and are positioned as an important way to achieve Artificial General Intelligence (AGI). However, LLMs are prone to produce factually incorrect information and often produce "phantom" content that undermines their reliability, which poses a serious challenge for their deployment in real-world scenarios. Enhancing LLMs by combining external databases and information retrieval mechanisms is an effective path. To address the above challenges, we propose a new approach called WeKnow-RAG, which integrates Web search and Knowledge Graphs into a "Retrieval-Augmented Generation (RAG)" system. First, the accuracy and reliability of LLM responses are improved by combining the structured representation of Knowledge Graphs with the flexibility of dense vector retrieval. WeKnow-RAG then utilizes domain-specific knowledge graphs to satisfy a variety of queries and domains, thereby improving performance on factual information and complex reasoning tasks by employing multi-stage web page retrieval techniques using both sparse and dense retrieval methods. Our approach effectively balances the efficiency and accuracy of information retrieval, thus improving the overall retrieval process. Finally, we also integrate a self-assessment mechanism for the LLM to evaluate the trustworthiness of the answers it generates. Our approach proves its outstanding effectiveness in a wide range of offline experiments and online submissions.

Figures (2)

• Figure 1: WeKnow-RAG pipeline for End-to-End Retrieval-Augmented Generation.
• Figure 2: Multi-stage Retrieval methods.