Vector-Combination-Applied KNN Method for Chinese Text Categorization
Xiao Zhang · Mini-micro Systems · 2004
On account of the fact that the traditional method lacks for the consideration of words association, this paper proposes an improved KNN (k-Nearest Neighbor) method for Chinese Text Categorization. This method applies vector-combination technology to extract the associated discriminating words according to the CHI statistic distribution, which indicates the relationship between words and classes. One of the merits of this method is to combine the associated discriminating words to be one feature and abandon the traditional method -- one word per dimension. It not only decreases the dimensions of the text vector, but also strengthens the contribution to categorization of each feature. The experiment shows that this improvement augments the categorization recall and precision obviously.