Automatic Tweet Classification Based on News Category in Indonesian Language

Abstract

Tweet is being informative as well as news articles, so that the automatic tweet classifier based on news category could be useful to make ease in searching tweet based on certain interesting category. We identified those are 11 categories: religion, business, entertainment, law and crime, health, motivation, sport, government, education, politics and technology. In the learning process, we use ZeroR, Naive Bayes Multinomial (NBM), Support Vector Machine (SVM), Random Forest (RF) and Sequential Minimal Optimization (SMO) algorithm based on previous work that has similar topic with this paper. In experiments, we experiment classifier using all tweet and various maximum number of tweets and terms in each category. In evaluating performance system, we used 10-fold cross validation and use accuracy (correctly classified instances) as performance paramater. In the experiments result, NBM performs the highest performance with 77,47% accuracy with maximum number of tweets and terms in every category is 500 tweets and 1000 terms. At the last, we built automatic tweet classifier with NBM due to this classifier and experiment result perform the best performances using web-based programming.