面向医学教育的多模态知识图谱构建方法及其在智能教学辅助平台中的应用  

Multi-modal Knowledge Graph Construction Method for Medical Education and its Application in Intelligent Teaching Assistance Platform

作　　者：何晓茜 王若瑶 张炯娜 HE Xiaoxi;WANG Ruoyao;ZHANG Jiongna(Shanghai Chest Hospital(School of Medicine,Shanghai Jiao Tong University),Shanghai 200030,China)

机构地区：[1]上海市胸科医院(上海交通大学医学院附属胸科医院),上海200030

出　　处：《微型电脑应用》2024年第10期42-45,共4页Microcomputer Applications

基　　金：上海交通大学医学院2022年度思政教育研究基金(SZ-2022-14)。

摘　　要：采用多模态知识图谱,利用YOLOv3-tiny模型进行目标检测训练,结合自顶向下和自底向上方法构建医学知识图谱。智能教学平台使用MVC 3层模型,MySQL和Neo4j数据库,Elasticsearch作为搜索引擎数据库。研究模型第五次训练显著提升效率,医学教学图片搜索中,模型识别精度达81.0%,运行时间2.7 s,损失函数下降至约0.25。实践证明构建的多模态知识图谱很成功,利用智能教学平台实现了对医学教学图片的高效搜索,为医学教育领域的智能化发展提供了基础。Multi-modal knowledge map is adopted,YOLOv3-tiny model is used for target detection training,and the medical knowledge graph is constructed by top-down and bottom-up methods.The intelligent teaching platform uses the MVC three-tier model,MySQL and Neo4j databases,and Elasticsearch is used as the search engine database.The experiment shows that the fifth training of the research model significantly improves efficiency.In medical teaching image search,the model recogni-tion accuracy reaches 81.0%,the running time is 2.7 seconds,and the loss function drops to about 0.25.The practice proves that the multi-modal knowledge graph constructed is very successful,and the intelligent teaching platform is used to realize the efficient search of medical teaching images,which provides the basis for the intelligent development of medical education.

关 键 词：医学教育 多模态知识图谱 智能教学 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]