Emotion Analysis and Classification of Movie Reviews Using Data Mining

Abstract

This paper proposes a model for classification of movie reviews by using Data Mining. The paper also proposes the method of creating word cloud from word frequency in movie reviews, for the purpose of partially helping in analyzing for interested topic and opinion of reviewer. The research uses movie review data from Metacritic website. The review data consists of reviews from 21 movies, separated into two parts to be used as training set and test set. Training set have 462 reviews and test set have 238 reviews. The data preparation process started collecting review data by removing special symbols case and preprocessing into Weka program. Change the review text into structured data by using StringToWordVector filter. This process includes removing stop words with Rainbow stop words list, change word that have the same root origin into word stem by using Snowball Stemmer algorithm and then given weight value by using TF-IDF technique. After that, Naïve bayes, Random Forest and J48 algorithms were used to classify the review data into positive and negative groups. The experimental result given is 80.25%, 79.83% and 68.06%, respectively.