基于特征词匹配的政策文本分类算法研究与实现  

Research and implementation of text categorization algorithm based on feature word matching

作　　者：王丽鹏 张鹏云[1] 和志强[1] 

机构地区：[1]河北经贸大学信息技术学院,河北石家庄050061

出　　处：《河北省科学院学报》2017年第3期1-6,共6页Journal of The Hebei Academy of Sciences

基　　金：基金项目:河北省科技创新政策法规需求库建设研究(17960119D)

摘　　要：在基于特征词遍历匹配的文本分类算法中,字符串匹配算法的选取及相似度阈值控制对文本分类结果起着决定性的作用。针对三种常用的字符串匹配算法做了分析及对比实验,选取了最适合政策文本分类的一种字符串匹配算法。并通过研究政策文本具有的特征提出了一种基于特征词加权的相似度阈值计算方法,经实验证明相似度阈值符合分类要求。In the text categorization algorithm based on the feature word traversal matching,the selection of the string matching algorithm and the similarity threshold control play a decisive role in the text classification result.In this paper,three commonly used string matching algorithms are analyzed and compared experimentally,then,a string matching algorithm which is most suitable for policy text classification is selected.Finally,a similarity threshold calculation method based on feature word weighting is proposed by studying the characteristics of policy text.It is proved that the similarity threshold can meet the classification requirement.

关 键 词：字符串匹配算法 阈值计算 文本分类 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]