一种基于领域本体的药品研发信息抽取方法

摘要

以生物医药公司年度报告为载体，设计了一种基于领域本体的信息抽取方法获取药品研发信息。首先依据药品研发固有的基本流程构建了领域本体词典，在对样本文档进行预处理的基础上介绍了时间词和否定词的处理；然后基于映射原理并结合触发机制、继承机制和选择机制，抽取并规范化药品研发信息；最后对抽取结果进行了查准率和查全率分析，证明了信息抽取方法的有效性。%　　Taking annual report as a carrier, this paper proposes an information extraction method for biological companies' drugs R&D in-formation based on domain ontology. First, the domain ontology dictionaries according to the basic process of drugs R&D is constructed and the process of times words and negative words after preprocessing the sample PDF documents is introduced in detail. Then, extracting and normalizing drugs R&D information are done by using mapping principle as well as trigger, inheritance, and selection mechanisms. Fi-nally, computing precision ratio and recall ratio based on the results of information extraction proves the validity of the method.