高校图书馆AI虚拟馆员服务框架与实施路径研究

摘要/Abstract

摘要：

人工智能从传统机器学习( ML-AI) 迈向生成式人工智能( GenAI) 的变革正在重塑高校图书馆的服务模式,基于大模型的AI 虚拟馆员已成为智慧图书馆建设的重要突破口。本研究通过调研国内“双一流”高校及国际U. S. News 前100 高校图书馆的在线咨询服务,揭示了国内高校在AI虚拟馆员覆盖面与功能创新方面的突破,以及国际高校图书馆在服务精准性与人工协作模式中的实践经验。结合华东师范大学图书馆的应用探索,本研究创新性地提出了一种面向未来的AI 虚拟馆员设计框架,通过构建问答库、文档库、应用库3 层知识库结构,并融合语义向量化、操作编排、提示词工程和ChatECNU 大语言模型技术,打造智能化、个性化、全天候的精准AI 虚拟馆员。实践结果表明,AI 虚拟馆员不仅显著优化了服务效率和用户体验,还解放了馆员从事重复性工作的时间,为其专注于高价值任务创造了条件,为高校图书馆提供了一种技术创新的可复制范式。未来,AI 虚拟馆员将在多模态交互、智能学术支持及数据隐私保护等领域进一步推动图书馆服务的智能化转型。

Abstract:

The evolution of artificial intelligence AI from traditional machine learning ML-AI togenerative AI GenAI is reshaping the service models of university libraries. AI-powered virtual librarian based on large language models have become an important break through for the development of smart libraries. This study investigates the online reference services of libraries at China’s Double First-Classuniversities and at the top 100 global universities ranked by U. S. News. It reveals breakthroughs in theservice coverage and functional innovation of AI virtual librarians in Chinese universities as well aspractical experiences related to service accuracy and human-AI collaboration models in internationaluniversity libraries. Based on the application exploration of East China Normal University Library thispaper proposes an innovative future-oriented design framework for AI virtual librarian services. Byconstructing a three-layer knowledge base comprising question-answering document and applicationdatabases and integrating semantic vectorization operational orchestration prompt engineering and theChatECNU large language model technology it creates an intelligent personalized and 24 / 7 accurate AIvirtual librarian. Practical results show that the AI virtual librarian not only significantly optimizes serviceefficiency and user experience but also frees librarian from repetitive tasks enabling them to focus onhigher-value activities and fostering innovation in university library services. In the future AI virtuallibrarians will further promote the intelligent transformation of library services in areas such as multimodalinteraction intelligent academic support and data privacy protection.

Key words: AI virtual librarian, GenAI, Prompt engineering, Knowledge base, ChatECNU

张　毅. 高校图书馆AI虚拟馆员服务框架与实施路径研究[J]. 图书馆杂志, 2025, 45(417): 49-60.

Zhang Yi. Research on AI Virtual Librarian Service Framework and Implementation Path in University Libraries[J]. Libraly Journal, 2025, 45(417): 49-60.

参考文献

[ 1 ] Nobel Prize. Nobel prizes 2024 EB / OL . 2024-11-24 . https / / www. nobelprize. org / all-nobelprizes-2024.

[ 2 ] 王文清,陈凌,关涛. 融合发展的CALIS 新一代图书馆服务平台[J]. 数字图书馆论坛,2020(1):2 10.

[ 3 ] Ghali M K Farrag A Won D et al. Enhancingknowledge retrieval with in-context learning andsemantic search through generative AI J .Knowledge-Based Systems 2025 311 113047.

[ 4 ] Clarivate. Pulse of the library EB / OL . 2024-11-24 . https / / clarivate. com / academia-government /wp-content / uploads / sites / 3 / dlm_uploads / Pulse-ofthe-Library-Report-2024. pdf.

[ 5 ] Sharma S Khan Q A. Embracing the digital shifta revolution in library technology J . AcademicLibraries 2024 184 190.

[ 6 ] AI 馆员,您身边的学习伙伴[EB / OL]. [2025-02-27]. https / / www. lib. sjtu. edu. cn / f / content /detail. shtml id =8259&lang =zh-cn.

[ 7 ] 案例集萃:云瀚社区AI 馆员与服务运营场景案例实践分享[EB / OL]. [2025-02-27]. https / /www. calsp. cn / 2024 / 10 / 09 / bulletin-202409-04.

[ 8 ] Balnaves E Bultrini L Cox A et al. Newhorizons in artificial intelligence in libraries M .Berlin Boston De Gruyter Saur 2025. [ 9 ] Manjunath A Chaithra B C Nagaraju B et al.AI-powered virtual library assistant enhancing userexperience with intelligent features C / /International Conference on Digital Landscape inLibrary 4. 0 DLL 4. 0 . Bengaluru 2024.

[10] IFLA FAIFE. IFLA statement on libraries andartificial intelligence R . 2020-10 . https / /repository. ifla. org / handle / 20. 500. 14598 / 1646.

[11] Li L L Coates K. Academic library online chatservices under the impact of artificial intelligence J .Information Discovery and Delivery 2025 53 2192 205.

[12] 郭亚军,寇旭颍,冯思倩,等. 大语言模型赋能图书馆参考咨询服务:逻辑、场景与体系[J]. 图书馆论坛,2025, 45(1):118 127.

[13] 刘泽,徐潇洁,邵波. 基于多策略混合问答系统模型的图书馆咨询机器人的设计与应用[J]. 新世纪图书馆,2022(5):43 49.

[14] 吴育良. 大语言模型对数字参考咨询服务的影响与应对策略研究[ J]. 山西档案,2024(5): 134 137.

[15] Wheatley A Hervieux S. Comparing generativeartificial intelligence tools to voice assistants usingreference interactions J . The Journal of AcademicLibrarianship 2024 50 5 102942.

[16] 郭利敏,付雅明. 以大语言模型构建智慧图书馆:框架和未来[J]. 图书馆杂志,2023, 42(11): 22 30;133.

[17] 王翼虎,白海燕,孟旭阳. 大语言模型在图书馆参考咨询服务中的智能化实践探索[J]. 情报理论与实践,2023, 46(8):96 103.

[18] Leger-Rousseau M. Performance of large models ofthe pre-trained generative transformer family insupport of automated virtual reference services inlibrary a narrative review J . Documentation etBibliothèques 2024 70 3 28 37.

[19] 余望枝,刘芳. 基于智能问答的图书馆参考咨询服务设计及优化[J]. 图书馆理论与实践,2024(1):128 136.

[20] 王晰巍,罗然,刘宇桐,等. 智慧图书馆在线聊天机器人使用行为影响因素及实证研究[J]. 情报学报,2023, 42(2):217 230.

[21] Chase M. Academic libraries can develop AIchatbots for virtual reference services with minimaltechnical knowledge and limited resources J .Evidence Based Library and Information Practice2024 19 2 136 138.

[22] 徐芳. 智慧图书馆生成式人工智能应用场景及其法律问题[J]. 情报资料工作,2024, 45(2): 24 29. [23] Moiseeva N A. Chatbots as a trend in libraryservice digitalization J . Scientific and TechnicalLibraries 2020 12 120 138.

[24] 顾伟,郑丽娜,张静怡. 我国省级公共图书馆智能咨询机器人应用现状调查与分析[J]. 图书馆学研究,2024(10):87 95.

[25] 张晓林. Library-Inside: AI 赋能图书馆新质生产力的一种基础模型[J]. 中国图书馆学报,2024,50(3):4 16.

[26] DeLone W H McLean E R. Information systemssuccess the quest for the dependent variable J .Information Systems Research 1992 3 1 60 95.

[27] Davis F D. Perceived usefulness perceived ease ofuse and user acceptance of information technology J .MIS Quarterly 1989 319 340.

[28] Alsafari B Atwell E Walker A et al. Towardseffective teaching assistants from intent-basedchatbots to LLM-powered teaching assistants J .Natural Language Processing Journal 2024 8100101. [29] Vidivelli S Ramachandran M Dharunbalaji A.Efficiency-driven custom chatbot developmentunleashing langchain RAG and performanceoptimizedLLM fusion J . Computers Materials &Continua 2024 80 2 2423 2442.

[30] Rangan K Yin Y. A fine-tuning enhanced RAGsystem with quantized influence measure as AIjudge J . Scientific Reports 2024 141 27446.

[31] Panda S Kaur N. Exploring the viability ofChatGPT as an alternative to traditional chatbotsystems in library and information centers J .Library Hi Tech News 2023 40 3 22 25.

[32] Lo L S. AI policies across the globe implicationsand recommendations for libraries J . IFLA Journal2023 49 4 645 649.