面向医药信息的知识图谱构建

Construction of Knowledge Graph for Medical Data

随着在线售药的普及,开源药品数据量激增,各类“寻医问药”在线平台不断完善,患者在线寻药的需求不断扩增,智慧医疗已然成为新的发展方向。但医药数据信息冗杂,各平台数据互不相通,患者依据疾病寻找相关药品的难度增大。为了解决医药数据分布庞杂,数据间缺乏良好管理与组织问题,通过网上开源的医药数据构建医药知识图谱并进行可视化展示。该文以疾病为中心,进行数据获取,设计相应数据模型,利用预先定义的医学词典以规则匹配的方式进行知识抽取,利用双向最大向前匹配算法对其结果进行知识验证,以综合考察实体名称相似度与属性相似度方式进行实体链接,从知识关联的角度出发构建医药知识图谱。通过Neo4J图数据库实现知识可视化,实现医药知识图谱构建的预期目标。

With the popularity of online medicine sales,the volume of open source medicine data has increased dramatically,various online platforms for"medicine searching"have been improved,and the demand for patients to find medicine online has been expanding and wise medical service has become a new development direction.However,the medical data information is redundant and the data on various platforms are not interconnected.It is more difficult for patients to find relevant drugs based on diseases.In order to solve the problem of the complex distribution of medical data and the lack of good management and organization among the data,the medical knowledge graph is constructed through online open source medical data and visualized.We conduct disease-centric data acquisition,design corresponding data models,use pre-defined medical dictionaries to extract knowledge in a rule-matching manner,and use two-way maximum forward matching algorithm to verify the results of knowledge to comprehensively investigate entity names similarity and attribute similarity methods for link entities.A medical knowledge graph is constructed from the perspective of knowledge association.Knowledge visualization is realized through Neo4J graph database,and the expected goal of medical knowledge graph construction is achieved.