面向中文信息处理的复句关系词提取算法研究  

Research on the Extraction of Relation Markers in Compound Sentences Oriented to Chinese Information Processing

作　　者：胡金柱[1,2] 舒江波[1,2] 姚双云[1] 周星[2] 吴锋文[1] 肖升[1] 

机构地区：[1]华中师范大学语言与语言教育研究中心,湖北武汉430079 [2]华中师范大学计算机科学系,湖北武汉430079

出　　处：《计算机工程与科学》2009年第10期90-93,共4页Computer Engineering & Science

基　　金：教育部人文社会科学重点研究基地重大项目(07JJD740063);湖北省科技攻关项目(2007AA101C49)

摘　　要：关系词语对于标明复句关系有重要的作用。在用计算机来实现汉语多重关系复句的关系层次分析的过程中,关系词语的提取和标引是首要的任务。本文针对利用计算机处理汉语复句的研究需求,结合词性标记和关系词搭配理论,提出了一种关系词提取算法——正向选择算法。通过测试可知,关系词提取的正确率达到89.88%,这表明了算法的有效性以及用于利用计算机处理汉语复句的可行性。Relation markers play an important role in indicating the relation of compound sentences. In Chinese information processing, the extraction and marking up of relation markers is the first step in the process of using a computer to accomplish an analysis of the relation hierarchy of Chinese multi-hierarchy compound sentences. According to the requirements of using a computer to process Chinese compound sentenees, and combining with the theory of tag marker and collocation, a relation marker extraction algorithm called the Forward Extraction Algorithm, is proposed in this paper. Through the o- pening test, the correct rate reaches 89.88%, and this shows the effectiveness of the algorithm.

关 键 词：复句 关系词提取 正向选择算法 关系词搭配理论 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]