面向学术文献的知识挖掘方法研究  

Research on Method of Knowledge Minning in Academic Documents

作　　者：王凯[1] 孙济庆[1] 李楠[1] Wang Kai Sun Jiqing Li Nan(Institute of Science and Technology Information, East China University of Science and Technology, Shanghai 200237, Chin)

机构地区：[1]华东理工大学科技信息研究所,上海200237

出　　处：《现代情报》2017年第5期47-51,110,共6页Journal of Modern Information

基　　金：2015年度教育部人文社会科学研究青年项目"面向语义出版的富语义模型构建与应用研究"(项目编号:15YJC870014);中央高校基本科研业务费专项资金资助项目"信息环境下的数字文献内涵语义聚合研究"(项目编号:WR1522001)

摘　　要：学术文献中包含的大量有价值的知识往往无法在摘要中体现出来。本文提出一种基于位置加权的核心知识挖掘方法,旨在以句为知识处理粒度,抽取正文中的核心句子作为独立的知识单元。该方法通过量化句子间的关联,将正文表示成一个以句子为节点,句子间关联为边的文本关系网络,提出基于章节的位置加权算法,结合社会网络分析方法,挖掘出文本中核心知识单元部分的句子。实验结果表明,该方法可以实现对文章核心章节中重要句子的抽取,达到初步预期效果。There is abundant valuable knowledge inside academic documents that is not revealed in abstracts. This paper promoted a method of core knowledge discovery based on position weights, aiming to extract the core sentences as separated knowl- edge units in the main text with the processing size of sentence. By measuring the connection between sentences, the paper trans- formed main text into a text network that considers sentences as dots and connection between sentences as sides. An algorithm to compute position weights based on chapters was promoted in this paper. With the help of social network analysis, the paper could find sentences that revealed the core knowledge of the text. The result of the experiment showed that this method could realize the extraction of key sentences in the core chapter from the text, which is primarily expected.

关 键 词：学术文献 知识挖掘 方法 位置加权 知识抽取 文本网络 社会网络分析 

分 类 号：G203[文化科学—传播学]