中文实体关系抽取研究综述  

Review of Chinese Entity Relation Extraction

作　　者：武文雅 陈钰枫[1] 徐金安[1] 张玉洁[1] WU Wen-ya;CHEN Yu-feng;XU Jin-an;ZHANG Yu-jie(School of Computer and Information Technology,Beijing Jiaotong University,Beijing 100044,China)

机构地区：[1]北京交通大学计算机与信息技术学院,北京100044

出　　处：《计算机与现代化》2018年第8期21-27,34,共8页Computer and Modernization

基　　金：国家自然科学基金资助项目(61473294;61370130);北京市自然科学基金资助项目(4172047);中央高校基本科研业务费专项资金资助项目(2015JBM033)

摘　　要：作为信息抽取任务中极为关键的一项子任务,实体关系抽取对于语义知识库的构建和知识图谱的发展都有着重要的意义。对于中文而言,语义关系更加复杂,实体关系抽取的作用也就愈加显著,因此,对中文实体关系抽取的研究方法进行详细考察极为必要。本文从实体关系抽取的产生和发展开始,对目前基于中文的实体关系抽取技术现状作了阐述;按照关系抽取方法对语料的依赖程度分为4类:有监督的实体关系抽取、无监督的实体关系抽取、半监督的实体关系抽取和开放域的实体关系抽取,并对这4类抽取方法进行具体的分析和比较;最后介绍深度学习在中文实体关系抽取上的应用成果和发展前景。Entity relation extraction is an important sub-task of information extraction. It is of great significance for the construction of semantic knowledge base and the development of knowledge graph. For Chinese,semantic relations are more complex,and the effect of entity relation extraction is more significant. So discussing the details of Chinese entity relation extraction methods is very necessary. From the beginning of the emergence and development of entity relation extraction,the current status of Chinese entity relation extraction technology is discussed. Relation extraction methods can be divided into four categories according to the degree of dependence on the corpus： entity supervised relation extraction,unsupervised relation extraction,semi-supervised relation extraction and open domain relation extraction. This paper analyzes and compares these four methods. Finally,the application results and development prospects of deep learning in Chinese entity relation extraction are introduced.

关 键 词：中文实体关系抽取 有监督方法 无监督方法 半监督方法 开放域实体关系抽取方法 深度学习 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]