摘要
针对民生的医疗健康大数据,运用知识图谱构建医疗知识图谱,通过F1值比较,在知识图谱构建的知识抽取阶段采用Bert模型自然语言处理算法,而关系抽取阶段采用基于注意机制的卷积神经网络(Att-RCNN),通过上述两种算法将建立好的知识图谱存储于图存储引擎Neo4j中;其次,为了系统能够识别用户的意图,通过F1值比较,最终采用BERT-TextCNN算法模型处理用户意图识别和槽位匹配;最后,利用Django框架搭建后端,前端采用微信接口实现用户与该系统的交互。
Aiming at the big data of medical and health care of the people's livelihood,the knowledge graph is used to construct the medical knowledge graph,and through the F1 value comparison,the Bert model natural language processing algorithm is used in the knowledge extraction stage of the knowledge graph construction,and the convolutional neural network(Att-RCNN)based on the attention mechanism is used in the relation extraction stage.Through the above two algorithms,the established knowledge graph is stored in the graph storage engine Neo4j.Secondly,in order for the system to recognize the user's intention,through the F1 value comparison,the BERTTextCNN algorithm model is finally used to process the user's intention recognition and slot matching.Finally,the Django framework is used to build the back-end,and the front-end uses the WeChat interface to realize the user's interaction with the system.
作者
谢崇波
XIE Chongbo(Sichuan Vocational College of Information Technology,Guangyuan 628040,China)
出处
《现代信息科技》
2023年第12期1-5,9,共6页
Modern Information Technology
基金
四川信息职业技术学院校级课题(2022C18)。
关键词
知识图谱
自然语言处理
智能问答系统
深度学习
knowledge graph
natural language processing
intelligent question answering system
Deep Learning
