服装网页自动分类技术研究  

Research on Automatic Classification of Clothing Webpage

作　　者：周毅灵[1] 耿增民[1] 

机构地区：[1]北京服装学院计算机信息中心,北京100029

出　　处：《北京服装学院学报（自然科学版）》2011年第1期55-59,共5页Journal of Beijing Institute of Fashion Technology：Natural Science Edition

基　　金：中国纺织经济研究中心资助项目(H2009-39)

摘　　要：如何对Internet上的服装网页进行自动识别是服装资源挖掘研究的重要内容.本文研究了KNN算法和Bayes算法对服装网页测试集进行自动分类的效果.实验表明KNN和Bayes的分类准确率相当;特征提取的维数可以控制在一定的范围;过高的特征数目不一定有较高的分类准确率;针对领域特点进行算法或特征提取调整才能进一步提高分类的准确率.It＇s an important part of clothing mining technologies that how to automatically identify the clothing category to webpages in the Internet.In this paper,the KNN algorithm and the Bayes algorithm were tested to clothing pages for classification result.Experiments showed that the classification accuracy rate of KNN and Bayes algorithm was similar;the dimension of feature extraction could be controlled in a certain range;high number of features didn＇t not necessarily have higher classification accuracy.It＇s necessary to adjust algorithm or feature selection method according to domain characteristics to improve the classification accuracy.

关 键 词：文本分类 WEB挖掘 BAYES算法 KNN算法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]