高校图书馆工作

用户画像的构建是分析用户特征、了解用户需求的基础。本研究提出了一种基于大语言模型和数据挖掘技术,实现高校图书馆用户画像构建与更新的方法。首先,构建高校图书馆用户画像标签体系,包括用户属性、用户行为等5个一级指标,个人信息、在校背景等11个二级指标,以及姓名、院系、在馆时长、借阅频率等49个三级指标;其次,提出用户画像构建、更新和基于大语言模型的画像生成方法;最后,以北京大学图书馆的读者为例,通过实证研究验证了所提方法的有效性,设计了基于大语言模型生成的个人和群体用户动态画像,实现了用户画像的可视化展示和动态更新。

The construction of user profiles is fundamental to analyzing user characteristics and understanding user needs. This study proposes a method based on large language models(LLMs) and data mining techniques to construct and update user profiles in academic libraries. First, a hierarchical tagging system for academic library user profiles is developed. This system consists of 5 first-level indicators(including user attributes and behaviors), 11 second-level indicators( such as personal information and academic background), and 49 third-level indicators(such as name, department, duration of library visits, and borrowing frequency). Second, the study introduces methods for profile construction and updating, as well as an LLM-based approach for building and updating these profiles. Finally, an empirical study using readers from the Peking University Library is conducted to verify the effectiveness of the proposed methods. Based on this, dynamic profiles for individuals and groups based on LLMs are designed, achieving the visual presentation and dynamic updating of user profiles.