【摘要】 聚焦AI赋能的高校图书馆馆藏资源服务效能评估，剖析传统评估模式的局限，结合人工智能技术在数据处理、智能分析、需求预测等方面的优势，基于高校图书馆资源服务一体化理念，文章构建了“目标—过程—结果”闭环三层评估框架，通过对哈尔滨工业大学图书馆馆藏资源服务效能的AI评估实践探索，为高校图书馆优化馆藏资源服务、提升服务效能提供了理论范式与方法论支撑。更多还原

【Abstract】 Traditional evaluation of academic library collections has largely relied on manual statistics and empirical judgment, leading to limitations such as reliance on singular evaluation methods, incomplete indicator systems, difficulty in assessing the cultural value of special collections, insufficient metrics for assessing the “in-depth utilization” of digital resources, and the absence of cross-institutional benchmarks. These constraints hinder the ability to meet the diverse needs of faculty and students. Addressing these challenges, this study explores an AI-empowered evaluation model for assessing the academic library collection service effectiveness and proposes a scientific evaluation framework and methods to provide theoretical paradigms and methodological support for optimizing resource services and improving service effectiveness. Based on the integrated concept of resources and services, this study introduces a closed-loop, three-layer evaluation framework of “Objective-Process-Outcome” which incorporates AI technologies, statistical analysis, bibliometric methods, and survey results to implement multi-index quantitative and qualitative evaluations. Taking Harbin Institute of Technology library as a case study, the evaluation of paper resources applies NeuralProphet tool to predict circulation trends, employs Principal Component Analysis and K-means clustering to categorize materials by circulation patterns, and integrates text semantic clustering to identify characteristics of high-demand and low-demand titles. In the evaluation of electronic resources, database coverage rates are assessed through citation matching, and citation themes are mined through text semantic clustering. Empirical findings reveal that circulation trend prediction for paper resources enables proactive strategies, such as adjusting internal operations during off-season and optimizing resource allocation prior to peak demand. By clustering analysis, books are divided into four types of circulation, informing differentiated management strategies. The identification of high-demand book themes has optimized acquisitions and collection layout. The evaluation of electronic resources shows that the Harbin Institute of Technology Library achieved a 95.59% coverage rate of journal citations from top-tier articles, with databases such as ScienceDirect playing a pivotal role. Citation theme clustering further provides precise evidence to guide resource development. Looking ahead, academic library evaluation is expected to evolve toward multi-dimensional, multi-agent, and multi-technology integration, facilitating the transition of libraries from “manual evaluation” to “intelligent governance”. The proposed AI-based intelligent evaluation model, driven by data, supports a closed-loop optimization mechanism that enables precise resource development and personalized services. It enhances the library’s contribution to teaching, research, and talent cultivation, reinforcing its position as a core pillar of the academic ecosystem.更多还原