Twitter Sentiment Analysis of Movie Reviews Using Information Gain and Naïve Bayes Classifier

The opinion of Twitter users related with movie is a valuable asset to be studied. This is because business and organization casts always want to know the public and consumers' opinions about their movies and consumers also want to know others' opinion before purchasing movie products. Sentiment analysis was applied in this research to gain information whether a tweet is positive opinion, negative opinion, or neutral opinion. Naive Bayes Classifier was chosen as the algorithm due to its strength on accuracy. In order to increase the run time efficiency, it was combined with Information Gain as feature selection method. By collecting tweets from 12 popular movie titles as dataset, the experimental evaluations show that the proposed techniques have the accuracy 82.19% with 0.006 as the optimal threshold of gain.