Sentiment Analysis with Various Deep Learning Models on Movie Reviews

Abstract

Social media have led to the development of artificial intelligence tasks such as sentiment analysis to see whether people’s posts have a positive or negative effect on other people. Ideas that affect society directly or indirectly about various domains, such as a movie or a meal, are very important for many business operations. This paper presents a sentiment analysis study which was carried out with 7 models based on various methods of deep learning algorithms on IMDB dataset. The best result was obtained with the model consisting of 2 Bi-LSTM and 2 dropout layers with 80%–20% train-test separation and an accuracy value of 88.21%.