Feature selection and text classification for Chinese Web documents

A great deal of methods for feature selection and text classification have been widely applied to English Web documents, while few studies have been done on Chinese Web documents. This paper gives a term weighting method based on inverse document frequency, HTML tags and length of Chinese phrase, reports our method to select Web text feature based on the messy genetic algorithm, provides an algorithm for Web text classification based on improvement on lattice machine approach. Our experiments show that these methods are valuable.