Likelihood calculation classification for Indonesian language news documents

Abstract

Text categorization has been an important research area that seeks to classify textual documents into a group of predetermined categories. Unfortunately, the interest towards Indonesian news classification has been very little. In this paper, we propose a text categorization algorithm based on Bracewell method that uses the likelihood calculation between the article and the category's keywords. Through experiments, the algorithm succeeded in classifying Indonesian news corpus with accuracy as high as 93,84% in offline environment, 93,82% in online environment, and 80% benchmarking against human evaluation.