中文中医本体定量评测研究  

Study on Quantitative Evaluation of Chinese TCM Ontology

作　　者：王大禹[1] 杨硕[1] 崔蒙[1] 

机构地区：[1]中国中医科学院中医药信息研究所

出　　处：《中国数字医学》2017年第2期16-18,44,共4页China Digital Medicine

摘　　要：目的:研究使用来源于实际病案中的二元关系评测中医本体的可行性。方法:利用中医医疗术语集及语义网构建被评测本体,利用中医药学术论文中提及的病例作为测试集来评测本体。结果:包含41652个实例的本体在评测实例名称覆盖度时使用来源于1000-3000个诊次的测试集时,评测结果趋于稳定。在评测本体的关系覆盖度时,使用1000或2000个诊次的测试集评测包含10000-30000个实例的本体时评测结果稳定。使用来源于3000个诊次的评测集时,测试结果在10000-40000个实例的本体上都略有波动。当本体包含40000个实例时,使用1000-3000个诊次的测试集不足以得到稳定结果。结论:利用来自于实际病案的小规模二元关系测试集测试本体的实例覆盖度是可行的,而测试本体的关系覆盖度需要更大规模的测试集。Objective： To study the feasibility of using binary relationships extracted from clinic records to evaluate Chinese TCM ontology. Methods： Built a Chinese TCM ontology with subset of term sets and sematic webs and build training set from the TCM clinical cases extracted in the papers appearing in modern academic journals. Results： When evaluating an ontology containing 41,652 instances, the evaluation results in terms of instance coverage are stable when the number of the TMC clinical cases ranges from 1000 to 3000. In terms of relationship coverage, the evaluate results are stable when the number of cases ranges from 1000 to 2000, but instable when increase at 3000. When evaluating an ontology with 40000 instances, 3000 clinical cases are inadequate to support. Conclusion： To use a small scale of TCM clinical cases extracted from academic journals to evaluate Chinese TCM ontology＇s instance coverage is feasible. But evaluating relationship coverage needs bigger scale of testing set.

关 键 词：本体评价 本体评测 中医药本体 中医药术语集 

分 类 号：R2-5[医药卫生—中医学] TP391[自动化与计算机技术—计算机应用技术]