一种基于粗糙集的文本分类规则抽取方法  

A Method of Rules Extraction of Text Categorization Based on Rough Set Theory

作　　者：孟庆春[1] 王汉萍[1] 魏天滨[1] 葛艳[1] 高云[1] 

机构地区：[1]中国海洋大学计算机系

出　　处：《青岛海洋大学学报（自然科学版）》2003年第6期943-949,共7页Journal of Ocean University of Qingdao

基　　金：山东省自然科学基金 (Y2 0 0 2 G1 8)

摘　　要：随着文本数据库的日益增大 ,寻找新的文本数据处理方法变得十分紧迫。本文将粗糙集理论应用于文本自动分类的规则提取 ,提出了基于粗糙集理论的文本分类方法。把文本特征项的权值进行离散化处理后 ,作为规则的条件属性 ,文本所属的类别用作决策属性 ,构造决策表 ,然后通过决策表的知识约简算法提取出文本的分类规则。实验结果表明 ,该方法提取规则的分类正确率较高 。With the increasing of text databases, it is urgent to find out a new method of text categorization. In this paper, a method of text categorization based on rough set theory is presented. RST is applied to rules' extraction of the text categorization. First, the weights of the key words are made discrete. Then, the decision table is made with the weights as the condition at tributes and the classes of texts as the decision attributes. Finally, the categorizing rules are extracted by reduction of attributes. It shows in the experiment that both the accuracy and the speed of categorization are high.

关 键 词：粗糙集 文本分类 规则抽取 决策表 属性约简 万维网 机器学习问题 

分 类 号：TP391.6[自动化与计算机技术—计算机应用技术]