OPINION MINING UNTUK ULASAN PRODUK DENGAN KLASIFIKASI NAIVE BAYES

Abstract

Nowadays, micro blogs have become the most used tools for users to share many things: from just updating things to telling their conditions or thoughts. Some popular micro blogs mostly used to give comments and opinions are facebook, instagram, and twitter. Twitter has 259 million active users each month as for January until April 2017. This made twitter one of the best micro blogs to know the most updated opinions.

The system uses Naive Bayes Classification to classify opinions about smartphone and computer from twitter. The sentiments are divided to positive, neutral, and negative. After that, Confusion Matrix is used to evaluate the algorithm and count the accuracy. Naive Bayes Classification gives 77.7% accuracy for Unigram, 50.7% for Bigram, and 31.7% for Trigram