Implementation of Web-Based Sentiment Analysis Application on Movie Reviews Using the Naive Bayes Algorithm

Abstract

Technological advances have brought significant changes to the film industry, including easy access to various content review and rating platforms such as IMDb. The increasing number of moviegoers has led to a surge in the number of movie reviews, making sentiment analysis important to understand audience reactions to a movie. Sentiment analysis uses technology to evaluate and understand emotional expressions in movie reviews, helping to identify positive, neutral, or negative sentiments. This study aims to eliminate the need for users to read long reviews that take a long time. With this system, users can immediately find out the sentiment of the review without having to read it all using the Naïve Bayes algorithm. The movie review data used in this study was taken from the Kaggle platform, consisting of 1000 data, where 800 data were used for training and 200 data for testing. The test results showed that the system was successfully implemented. The Naive Bayes model used showed adequate performance with an accuracy of 81%. This figure was obtained through an evaluation using a confusion matrix, which measures the model's performance in classifying reviews into positive, neutral, and negative sentiment categories. The confusion matrix shows that this model is quite effective in recognizing positive sentiment, but less accurate in identifying negative and neutral sentiments. This indicates that although the Naive Bayes model is able to provide a good overview of positive sentiment in reviews, there is room for improvement especially in the classification of negative and neutral sentiments. This study is expected to provide better insight to the audience regarding the sentiment of movie reviews, thus helping in making better decisions regarding the movies they want to watch.