An empirical evaluation of lazy learning classifiers for text categorization

Abstract

With the rapid growth of online documents available on the World Wide Web necessitate the task of classifying those documents into semantic categories. Text categorization is the task of automatically classifying the textual documents into a set of predefined categories. In this paper, we report the empirical evaluation of lazy learning classifier such as kNN and its variant like distance weighted kNN and our newly proposed evident theoretic kNN for text categorization task over two benchmark datasets. We observed the superiority of evident theoretic kNN method over others in all experiments we conducted.