基于本体分类结构计算医疗领域语义相似度的方法

摘要

相似度计算能提高从医疗源数据进行信息检索的效率并使得异构临床数据的集成变得更加容易。不同学者基于单个医疗本体，将经典的相似度计算方法用于医疗术语的相似性评估。本文选定基于距离的LCH方法，依据Pederson基准，对比该算法在基于MeSH、SNOMEDCT、UMLS本体时的相关度值，并就计算结果进行分析和解释。%Semantic similarity computation can promote the efficiency of information retrieval of biomedical resources, and make the integration of heterogeneous clinical data more easier. Various experts devote themselves on the application of classic semantic similarity measures over single biomedical ontology and develop them. In this paper, we compare the results for LCH measure over various ontologies such as MeSH SNOMED CT and UMLS. Finally we analysis and explain the results.