基于多属性规则的生物医学语义关系研究  

Study on Biomedical Semantic Relation Based on Multi-attribute Rules

作　　者：范少萍[1] 安新颖[1] FAN ShaoPing;AN XinYing(Institute of Medical Information,CAMS&PUMC,Beijing 100020,China)

机构地区：[1]中国医学科学院医学信息研究所,北京100020

出　　处：《数字图书馆论坛》2021年第1期18-23,共6页Digital Library Forum

基　　金：国家自然科学基金项目“面向精准医学的基因-疾病-药物语义关系抽取研究”(编号:71704188)资助。

摘　　要：生物医学领域文献指数级增长亟需快速识别出领域内核心且关键的语义关系,开展领域知识发现研究。本文借鉴DisGeNET基于规则的关系得分方法设计思想,在专家咨询基础上,提出生物医学领域语义关系具有可靠性、重要性和新颖性3种属性,设定每种属性对应的指标及定量计算方法。通过分析结直肠癌领域关系计算结果,验证本文所提关系得分方法在关系重要性排序方面的有效性。该方法考虑维度更加全面,为生物医学领域知识图谱构建、知识发现等提供有益参考。With the rapid growth of biomedical literature,it is urgent to identify the key semantic relations in the field quickly,and carry out domain knowledge discovery research.Based on the design idea of DisGeNET and expert consultation,this paper proposes that biomedical semantic relation has three attributes:reliability,importance and novelty,and sets the index and quantitative calculation method for each attribute.Through the results and analysis of colorectal cancer relation,the effectiveness of the proposed relation score method in the ranking of relations and knowledge discovery is verified.The dimension of the proposed method is more comprehensive,which can provide a useful reference for biomedical knowledge graph and knowledge discovery.

关 键 词：语义关系 关系得分 关系排序 知识图谱 知识发现 

分 类 号：G350[文化科学—情报学]