Investigation of Movie Review Data Set by the Use of Opinion Mining

Abstract

With the growth of the internet and its users over the past decade, there is an extremely large magnitude of data online. There are various websites and forums which provide options to review a particular product or place. These reviews are helpful to all the customers, who are thinking of buying that product, as well as the organisation building that product, so they can improve its flaws. The opinions and sentiments gathered through online forums and review sites help in a more accurate way of understanding the customer’s preferences and tastes. In this study, our objective is to determine the underlying opinion in the movie reviews which could be positive, neutral or negative, through various machine learning algorithms like Support Vector Machine (SVM), Decision trees and Naive Bayes classifier and to compare the results given by each algorithm. Further, we also appraise the role of pre-processing online reviews before polarising them. The outcome of this study shows the different accuracies given by the various classifiers applied to the same movie review dataset.