Sentiment Analysis of IMDB Movie Reviews

Sentiment analysis is very important and feedback from the customers or the people plays a big role in many fields. If done manually for a huge amount of data, it will waste lots of time and effort. The purpose of our study is to make the sentiment analysis of large datasets as easy and precise as possible. We used Machine Learning and Natural Language Processing to conduct sentiment classification on IMDb movie reviews. Sentiment analysis of movie reviews is important as it not only guides the production team on how to improve the quality of their content but also tells the people if they should spend their precious time as well as money on the film. It gives them a fair idea of the type and quality of the film beforehand. Our findings are specific to IMDb movie reviews, but their applications are not just limited to movie reviews. We have used the following classification algorithms to classify the reviews as either negative or positive: K Nearest Neighbors, Support Vector Machine, Naive Bayes, Decision Tree, Random Forest and Linear Discriminant Analysis. Most accurate of these was Support Vector Machine which gave us an accuracy of 88%. We also used the famous BERT Model which was recently published by Google researchers. It is very rightly said that the BERT Model is a game changer in the field of NLP. We achieved an accuracy of 93.97% using the BERT Model. We conclude that the BERT Model gave us the highest accuracy as compared to the six classifiers that we implemented.