Sentiment Analyses on Movie Reviews using Machine Learning-Based Methods

Abstract

As a result of the rapid development of information and communication technologies, social networks have become an essential part of our daily life. The comments shared on social network platforms such as Twitter, Instagram, Facebook and Tumblr about watched movies and purchased products, advertisements and campaigns have a very important effect on the target users using these platforms. However, the users need a long time to read and understand huge amounts of textual data stacks on the relevant platforms. This is a major problem for the users. Nowadays, different types of automatic text processing methods and algorithms based on sentiment analysis and artificial intelligence are applied to solve the problem. In this study, in order to determine whether the movie reviews in the IMDB dataset obtained on Kaggle are positive or negative, 6 different models based on two-class machine learning (ML) are developed in the Microsoft Azure ML studio, and a series of analyses are performed to test the performance of the relevant models by confusion matrix metrics and ROC curves. The results obtained show that a single model could not reach the most successful result in terms of all confusion matrix metrics, while the two-class neural network (NN) model achieves this result in ROC analyses.