基于句法分析的临床指南事件及事件关系提取  

An Approach for Event Extraction and Event Relation Extraction of Clinical Guideline Based on Parsing

作　　者：李井竹[1] 陆玉婷[2,3] 顾进广[2,3] 

机构地区：[1]河南牧业经济学院计算机应用系,河南郑州450002 [2]武汉科技大学计算机科学与技术学院,湖北武汉430065 [3]智能信息处理与实时工业系统湖北省重点实验室,湖北武汉430065

出　　处：《武汉大学学报（理学版）》2015年第2期156-162,共7页Journal of Wuhan University:Natural Science Edition

基　　金：国家自然科学基金(60803160;61100133;61272110);国家社会科学基金重大计划(11&ZD189);湖北省自然科学基金(2013CFB334);湖北省教育厅科研项目(Q20101110;D2009110);湖北省高等学校优秀中青年科技创新团队计划(T201202);武汉市科技攻关计划(201110821225);软件工程国家重点实验室(武汉大学)开放基金SKLSE2012-09-07

摘　　要：研究了临床指南事件及事件关系的提取,提出了基于句法分析的事件信息抽取方法.通过分析临床指南语料库,总结出临床指南事件及关系模型;将临床指南数据进行句法分析,根据事件触发词来匹配事件与事件关系模型;将抽取出的事件信息进行优化处理,得到最终结果.通过具体临床指南数据对方法进行评估,获得了较好的准确率与召回率,验证了该方法的可行性与有效性.As an important direction for the research in rational use,the information processing of clinical guidelines has attracted more and more attention,and the clinical guideline event is one of the core problems of it. The research area of this paper is event extraction and event relation extraction of clinical guideline. This paper proposes an approach for event extraction and event relation extraction of clinical guidelines based on parsing. Based on the clinical guideline corpus,the models of medicinal events and event relations are summarized. The parsing of clinical guidelines is proceeded to extract the candidate event trigger words which are used to match the event models with event relation models and to optimize the results of extracting. By the evaluation with specific data of guideline,this approach achieves a better precision and recall rate. Therefore,this approach is feasible and effective in extracting event and event relation of clinical guidelines.

关 键 词：合理用药 临床指南 事件抽取 事件关系抽取 句法分析 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术] R441[自动化与计算机技术—计算机科学与技术]