Design of Intelligent Q&A System Based on Knowledge Graph Combined with Large Language Model

LLM can transform natural language questions into structured queries, and the KG-based Q&A system can provide accurate and reliable answers. The combination of LLM and KG is the key technology to support the modern Q&A model. Through the two-wheel drive of structured knowledge and semantic understanding, the processing ability of complex problems is significantly improved, and the Q&A system is jointly promoted from the primary to the advanced intelligence evolution. In this paper, based on LLM and KG, knowledge distillation based LLM fusion KG technology is studied, so that the target model can absorb the advantages of both, not only have the language processing capability of large models, but also use the structured knowledge of KG to improve performance and interpretability. On this basis, the multi-layer architecture of intelligent Q&A system is designed, which is easy for developers to work together. The ClaudeKG model constructed in this paper is compared with DeepSeek and Doubao baseline models, and the function and performance are analyzed.