Personalized library knowledge services driven by chain-of-thought large language models

With the increasing demand for personalized and intelligent information services, traditional library systems face increasing limitations in meeting the complex and diverse needs of users. This paper proposes a novel framework that integrates large language models with chain-of-thought reasoning and retrieval-augmented generation to enable automated, context-aware knowledge services in libraries. The framework empowers large language models to interpret user queries, construct structured retrieval plans, and dynamically collect and process relevant data and knowledge for downstream tasks such as question answering and recommendation. By combining internal library databases with external knowledge sources, the system achieves a balance between semantic depth, response accuracy, and information relevance. In the evaluated setting, the results provide comparative evidence that the framework improves perceived relevance, personalization, and completeness over a zero-shot baseline. This study provides proof-of-concept evidence for integrating reasoning-capable language models with structured knowledge retrieval in library services; its applicability across institutions and service settings remains to be established through broader validation.