Combining Multiple Classifiers for Automatic Text Categorization

Abstract

In this paper we study the development of mul- tiple classier systems in the Chinese text cate- gorization. Our objective is to develop ecient techniques to combine the strength of well-known classiers such as linear classiers, decision trees, Bayesian methods, neural networks, and support vector machines. We have experimented with Chi- nese documents from the Central News Agency and from the Web Opennd. Experiments show that our approaches signicantly improve the clas- sication accuracy of individual classiers for Chi- nese text categorization as well as for web page classication.