Application Research on Intelligent Reference Services in Libraries Based on Large Language Models

Abstract

With the development of the digital information environment and the high demand for knowledge, the old model of library reference service can no longer handle open-ended, multi-turn and semantic-rich questions. Large Language Models are based on the Transformer architecture, have achieved good results in deep semantic understanding and generation, and are now being used in all sorts of reference services. First, we will present the theoretical basis for integrating large language models with library reference services; next, we will construct a technically layered and decoupled architecture of an intelligent reference system that includes domain adaptation and knowledge enhancement; finally, we will analyze the working mode of this system. Researchers have found all kinds of ways to add domain knowledge in Retrieval-Augmented Generation and fine-tuning of large models, as well as context-understanding mechanisms that support multi-turn dialogue and the combination of information from various sources. Finally, some directions for the development of intelligent reference systems are also put forward in this paper, such as personalised services, reconstruction of human-computer collaboration and trustworthy assurance systems. Present the theoretical basis and technical design ideas for building a new generation of intelligent knowledge service systems in libraries based on large language models in this paper.