基于模糊关系的文本分类方法  

Text Classification Approach Based on Fuzzy Relationship

作　　者：张玉芳[1] 娄娟[1] 李智星[1] 熊忠阳[1] 

机构地区：[1]重庆大学计算机学院,重庆400030

出　　处：《计算机工程》2011年第16期149-151,共3页Computer Engineering

基　　金：中国博士后科学基金资助项目(20070420711);重庆市科委基金资助项目(CSTC;2008BB2191)

摘　　要：为更好地对未标记文本进行分类,通过定义文本和类别的隶属函数,将测试文本和类别表示为特征的模糊集,计算模糊集之间的相关系数并用来度量测试文本到每个类别的隶属度,根据最大隶属度原则确定测试文本所属类别。实验结果表明,与k-NN算法相比,该方法有较好的准确率,分类速度有较大提高。To better classify the unlabeled text,this paper presents a text classification approach based on fuzzy relationship.Through defining membership function of test document relationship and test category relationship,the test document and category can be represented as fuzzy sets.Then evaluating the membership degree between test document and each category by computing the correlation coefficient of fuzzy sets,the test document is decided to category using maximum membership principle.Compared with k-NN,experimental result shows that the precision is increased and the classification process is speeded up to a considerable degree.

关 键 词：文本分类 隶属函数 模糊关系 相关系数 隶属度 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]