Comparative Study of Algorithms for Sentiment Analysis on IMDB Movie Reviews

Sentiment Analysis is one of the major NLP tasks, which mainly focuses on finding polarity(sentiment) and intention for a given piece of textual data. The data can range from a sentence level to a document level. When we communicate as human to human, it is much more easier to interpret our emotions. But when it comes to machines, it becomes difficult to identify the sentiments. Sentiment Analysis makes use of techniques like Text Analytics and other NLP methods. Movie reviews play a significant role in telling whether it's a success or failure. These reviews will have a great influence on larger set of viewers. So, its important to build a good model which classifies movie reviews accurately. In this paper, we have adopted six machine learning models to classify movie reviews and a comparative study is made among them which helps to decide the best classifier based on several evaluation metrics.

I. Introduction

With the increased use of Internet services, people always tend to share their feedback, feelings over a variety of platforms like social media apps, company survey forms, through mails, etc. They generally discuss elements like major happenings in the society, movie and product reviews etc which contains highly polarised text. [3]