摘要
临床科学研究往往以科技文献的形式储存。文章对医学领域科技文献表示模型进行概述和分析,以PIBOSO模型为基础,采用支持向量机对科技文献的摘要句子进行分类,实现了科技文献摘要信息的自动化抽取及关键句子的识别,从而将科技文献的摘要内容进行语义关系的量化和结构化表示,为临床医师和相关研究人员在寻找证据资源时提供有效借鉴和帮助。实验结果表明,该方法的F值在大多数类别上高于其他方法,表明研究方案具有可行性和有效性。
It is well known that clinical scientific research is often stored in scientific and technical(S&T)literature.The knowledge hidden in S&T literature can provide clinicians and researchers with the clinical decision-making evidence in the practice of evidence-based medicine.After the representation models for S&T literature are summarized and analyzed in the medical field,are fined PIBOSO model is used in this study.For purpose of the automatic extraction of the abstract and the identification of key sentences,Support Vector Machine(SVM)is utilized here to classify abstract sentences.The classification results with SVM help quantify the semantic relations and structure the abstracts,thus providing effective reference for clinicians and researchers to find evidence.From experimental results,one can see that F-score in this work is higher than the counterparts in most categories,which indicates that our research framework is feasible and effective in the sentence classification task from the biomedical field.
作者
杜圣梅
朱礼军
徐硕
DU Shengmei;ZHU Lijun;XU Shuo(Institute of Scientific and Technical Information of China,Beijing 100038;Beijing University of Technology,Beijing 100124)
出处
《中国科技资源导刊》
2018年第6期94-100,共7页
China Science & Technology Resources Review
基金
北京市社会科学基金项目"大数据驱动的可制造性知识挖掘与管理方法研究"(17GLB074)
北京市优秀人才培养资助青年骨干个人项目"(2015000020124G052)
关键词
循证医学
SVM
句子分类
知识挖掘
机器学习
evidence based medicine
SVM
sentence classification
knowledge mining
machine learning
