Bean C A
Department of Medical Informatics, Columbia University College of Physicians and Surgeons, USA.
Proc AMIA Annu Fall Symp. 1997:625-9.
The verb structure of narrative text in a gross anatomy textbook was analyzed to identify locative relationships. The 169 locative indicators were organized semantically to construct a frame-based model. The validity and coverage of the model was assessed and compared with the UMLS Semantic Net Relations using a novel test set of 71 indicators. All mapped directly to the frame model, while 60% mapped directly to UMLS.
对一本大体解剖学教科书中叙述性文本的动词结构进行分析,以识别位置关系。对169个位置指示词进行语义组织,构建基于框架的模型。使用一组由71个指示词组成的新测试集,评估该模型的有效性和覆盖范围,并与统一医学语言系统(UMLS)语义网络关系进行比较。所有词都直接映射到框架模型,而60%直接映射到UMLS。