中华医学教育杂志 ›› 2026, Vol. 46 ›› Issue (7): 515-520.DOI: 10.3760/cma.j.cn115259-20251102-01390

• 教育技术 • 上一篇    下一篇

基于大语言模型与Neo4j的《中国医学史》教材知识图谱构建研究

刘思言1, 周雅倩2, 褚嘉琪1, 姜雨彤3, 孙灵芝4   

  1. 1北京中医药大学管理学院2023级公共事业管理专业学生,北京 102488;
    2北京中医药大学管理学院2023级工商管理专业学生,北京 102488;
    3北京中医药大学中医学院2023级中医学(卓越5+3一体化)专业学生,北京 102488;
    4北京中医药大学中医学院医史文献教研室 教育部哲学社会科学实验室北京中医药大学中医药典籍数智化与人文创新实验室 国家中医药典籍翻译与传播研究中心,北京 102488
  • 收稿日期:2025-11-02 出版日期:2026-07-01 发布日期:2026-07-01
  • 通讯作者: 孙灵芝, Email: sunlingzhi@bucm.edu.cn
  • 基金资助:
    北京中医药大学岐黄英才科研新锐人才培育专项(K2023D02);北京中医药大学教育科学研究课题(XJY24011);北京中医药大学基本科研业务费科研平台建设项目(2026-JYB-KYPT-12)

Construction of a knowledge graph for the textbook History of Chinese Medicine based on large language models and Neo4j

Liu Siyan1, Zhou Yaqian2, Chu Jiaqi1, Jiang Yutong3, Sun Lingzhi4   

  1. 1Public Administration Major, Enrolled in 2023, School of Management, Beijing University of Chinese Medicine, Beijing 102488, China;
    2Business Administration Major, Enrolled in 2023, School of Management, Beijing University of Chinese Medicine, Beijing 102488, China;
    3Chinese Medicine Major (Excellence 5+3 Integrated Program), Enrolled in 2023, School of Traditional Chinese Medicine, Beijing University of Chinese Medicine, Beijing 102488, China;
    4Department of Medical History and Classical Literature, School of Traditional Chinese Medicine & Philosophy and Social Sciences Laboratory of the Ministry of Education: Laboratory of Digitalization and Humanistic Innovation of Traditional Chinese Medicine Classics & National Research Center for the Translation and Communication of Traditional Chinese Medicine Classics, Beijing University of Chinese Medicine, Beijing 102488, China
  • Received:2025-11-02 Online:2026-07-01 Published:2026-07-01
  • Contact: Sun Lingzhi, Email: sunlingzhi@bucm.edu.cn
  • Supported by:
    Qihuang Elite Research Talent Cultivation Program, Beijing University of Chinese Medicine (K2023D02); Educational Science Research Project, Beijing University of Chinese Medicine (XJY24011); Fundamental Research Funds for the Central Universities, Research Platform Construction Project of Beijing University of Chinese Medicine (2026-JYB-KYPT-12)

摘要: 中国医学史是中医类院校必修课程,其知识体系繁杂,给教学带来一定挑战。现有中医药知识图谱研究未覆盖此门课程所涉及的系统性知识。本研究旨在借助现代信息技术实现《中国医学史》教材的知识结构化与可视化,助力中医药教育。通过大语言模型结合人工核验的方式自底向上构建图谱,并存储于Neo4j图数据库,最终成功构建含2 879个实体、2 487条语义关系的知识图谱。模型抽取三元组精确率达65.42%、召回率77.82%、F1值71.09%,人工核验与模型抽取结果具有较高的一致性。该图谱可提供教学可视化与关联检索支持,具备多元教育潜力,为医学人文教育智能化创新提供参考实例与框架。

关键词: 教育技术, 中国医学史, 知识图谱, 大语言模型, Neo4j, 中医药数字化教育

Abstract: History of Chinese medicine is a compulsory course in traditional Chinese medicine (TCM) universities and colleges. Its complex knowledge system poses certain challenges to teaching. Existing research on TCM knowledge graphs has not covered the systematic knowledge involved in this course. This study aims to leverage modern information technology to achieve knowledge structuring and visualization of the History of Chinese Medicine textbook, thereby supporting TCM education. By employing a large language model combined with manual verification, a bottom-up approach was adopted to construct the knowledge graph, which was then stored in the Neo4j graph database. A knowledge graph containing 2 879 entities and 2 487 semantic relationships was successfully built. The model achieved a triplet extraction precision of 65.42%, recall of 77.82%, and an F1 score of 71.09%, demonstrating high consistency between manual verification and model extraction results. This knowledge graph provides teaching visualization and associative retrieval support, possesses diverse educational potential, and offers a reference example and framework for intelligent innovation in medical humanistic education.

Key words: Educational technology, History of Chinese medicine, Knowledge graph, Large language model, Neo4j, Digital education in traditional Chinese medicine

中图分类号: