Table based KNN for categorizing words

Abstract

In this research, we propose the table based KNN as the approach to the text categorization. In previous works, we discovered that encoding texts into tables improved the performance in the text categorization, so in this research, become to consider the possibility of encoding words into tables as well as texts. In this research, we encode words into tables where entries are texts and their weights, and apply the table based version of the KNN to the task of word categorization. As the benefits from this research, we expect the better performance and more stability than the traditional version of the KNN, by doing so. Therefore, the goal of this research is to provide the improved approach to the word categorization task.