一种新型英语基本名词短语识别方法——基于边界概率与N_Gram词性串规则相结合  

Combining Boundary Probability and N_Gram String of Part of Speech:A New Method of Identifying English Base Noun Phrases

作　　者：韩朝阳[1] 刘国兵[2] 王跃武[1] 

机构地区：[1]上海杉达学院计算机科学与技术学院,上海211209 [2]河南师范大学外国语学院,河南新乡453007

出　　处：《软件导刊》2015年第8期14-18,共5页Software Guide

基　　金：2014年国家社科基金项目(14BYY084)

摘　　要：英语基本名词短语识别是一种重要的基础性自然语言处理活动,其识别准确率与召回率直接影响其它相关自然语言处理活动效果。在分析、总结几种具有代表性基本名词短语识别方法的基础上,提出了一种新型识别方法,其核心是:把边界概率与N_Gram词性串规则相结合作为识别判断条件,以"假拟中心词"为起点,分别向左、向右识别出当前"假拟中心词"所在基本名词短语的左、右边界。实验证明,该方法的识别准确率为97.13%、召回率为98.75%,FB=1为97.93%。The identification of English base noun phrases is an important and basic task in natural language processing since the precision and recall rates of such identification have a direct impact on other related tasks. After a summary and analysis of three representative methods of base noun phrase identification, this paper proposes a new method in which boundary probability and NGram string of part of speech are combined as a basic recognition procedure and a pseudo cen- ter word is used to identify the left and right boundaries of the base noun phrase where the pseudo center word is located. Experiments show that the method reaches a precision of 97.13% and recall of 98.75% in English base noun phrase iden- tification, with FB=1 of 97.93%.

关 键 词：英语基本名词短语识别 边界概率 N_Gram词性串规则 假拟中心词 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]