Research on Book Information Accurate Retrieval Recommendation System Based on Large Language Model

In this paper, we design and implement an innovative system that integrates the deep semantic understanding capabilities of BERT models, the powerful sequence modeling properties of Transformer networks, and the efficient decision-making strategies of deep Q-networks (DQN). The introduction of BERT greatly improves the understanding accuracy of user query intention, so that the system can match user needs and book content more accurately. The use of Transformer enables the system to update user profiles in real time, adapt to their changing preferences, and enhance the effect of personalized recommendation. The combination of the three improves the accuracy of recommendation, enhances the scalability and robustness of the system, and provides users with a more rich, diverse and personal preference book recommendation experience.