基于医疗本体的语义相似度评估方法

Semantic similarity estimation method based on medical ontology

为了能够正确地理解医疗概念和精确地分析临床记录,提出了一种基于概念信息量的方法来衡量概念之间的语义相似度。引进了计算概念信息量的算法,从医疗本体的分类知识中来计算概念的信息量。介绍和分析了常用的语义相似度算法,根据概念的信息量来重定义这些语义相似度算法,产生新的基于概念信息量的语义相似度算法。通过使用一个医疗术语的评估标准和一个标准的医疗本体来评估和比较这些算法。实验结果表明,相比常用的语义相似度算法,重定义后的算法有效地改善了概念相似性评估的准确性。

To properly understanding medical concept and precisely analysing clinical records, a method based on information content （IC） of concept is proposed to measure the semantic similarity between concepts. Firstly, the algorithm of computing IC of concept is introduced, which computes the IC of concept from the taxonomical knowledge modelled in medical ontologies. Then, well-known semantic similarity algorithms is introduced and analyzed. After that, new semantic similarity algorithms based on IC of concept are expressed by redefining these well-known semantic similarity measures in terms of IC of concept. Fi- nally, these algorithms are evaluated and compared by using a benchmark of medical terms and a standard medical ontology. Ex- periment shows that these redefined algorithms results in noticeable improvements in accuracy comparing with well-known se- mantic similarity algorithms.