Improvement of email and twitter classification accuracy based on pre-processing bayes naive classifier optimization in integrated digital assistant

Abstract

This research focuses on improving the accuracy of email and twitter classification. Spelling mistakes and lack of matches with bag of word causes the low accuracy in classifying. This research used naïve Bayes as a text classification algorithm. Text is divided into three categories: personal, work and family. To achieve maximum likelihood value for the category, a better preprocessing techniques is needed. It is necessary for the process to normalize the preprocessing and search for words that correspond to classes in the bag of word. So that the text can be classified by category or has a higher precision accuracy.