Movie Recommendation System Based on Sentiment Analysis on Movie Reviews

Abstract

A movie review plays a significant role in determining whether the movie is recommended to them. Nowadays, movie reviews are filled with paid, sarcasm, and fake reviews that give users mixed feelings about the movie. With a movie recommendation system based on sentiment, the movie review is analyzed with specific keywords that give the user an absolute result on whether it is recommended or not recommended. The system uses three classifiers, Naïve Bayes, Support Vector Machines (SVM), and Deep Learning to determine the best classifier that gives better accuracy for user purposes. The core approach of this project is to provide the user with a simplified view that analyses all accumulated reviews into one single view. The project results show that SVM produces the best results with 81.17% accuracy. That result is because of the nature of the classification that works best in categorization. This project includes future works, adding more lists of movies and user input for better interactivity between users and machines.