科技文献信息抽取方法浅析  

An Analysis on Methods of Information Extraction from Foreign Scientific Literature

作　　者：敖龙[1] 谢海先 Ao Long;Xie Haixian(Shenzhen Polytechnic,Shenzhen,Guangdong 518055,China)

机构地区：[1]深圳职业技术学院图书馆,广东深圳518055

出　　处：《高校图书馆工作》2022年第2期24-27,共4页Library Work in Colleges and Universities

基　　金：深圳市哲学社会科学“十三五”规划课题“深圳智慧图书馆联盟设计研究”(SZ2018B030)研究成果之一。

摘　　要：文章在Web of Science等影响力较大的国际数据库中检索内容与“科技文献”和“信息抽取”相关的文献,经设定条件筛选后获得63篇相关文献。回顾相关文献,从抽取的信息与抽取的方法两个角度进行分类与分析,总结该领域已有的研究成果和存在的不足。从科技文献中抽取的信息主要为结构化信息、显式信息和隐式信息,最新最先进的抽取方法主要集中在机器学习、自然语言处理和统计学中。语义信息的抽取有一定的进步空间及挑战性,灵活结合机器学习和自然语言处理方法是处理此领域问题的未来趋势。Using international databases with great influence such as Web of Science, studies relevant to scientific literature and information extraction were searched and 63 studies were included in this research. By reviewing relevant literature, this research classifies and analyzes the extracted information and the extraction methods and summarizes the contributions of existing research as well as their limitations. The information extracted from scientific literature mainly includes structured information, explicit information and implicit information. The latest and most advanced extraction methods mainly focus on the fields of machine learning, natural language processing and statistics. There is room for improvement as well as challenges in the extraction of semantic information. A flexible combination of methods in machine learning and natural language processing is a future trend for solving the problems in this area.

关 键 词：信息抽取 科技文献 语义信息 机器学习 自然语言处理 

分 类 号：G253[文化科学—图书馆学]