Framework to Study Optimizing Sentiment Analysis Using Hybrid Machine Learning Techniques

Abstract: The increasing demands on government organizations and private businesses motivate researchers to finish their work in sentiment analysis. People's perspectives on different goods, services, and events are reflected in the way they express themselves on social media. Sentiment analysis is a branch of natural language processing that seeks to identify positive or negative polarities in content from social media networks. In order to maximize sentiment analysis, this paper presents three state-of-the-art machine learning classifiers: Naïve Bayes, SVM, and OneR. The investigations make use of two benchmark datasets, one from IMDB movie reviews and the other from Amazon. The outcomes of different classification techniques are compared and examined. While the Naïve Bayes learned very rapidly, OneR exhibits greater promise with a 93.4% correctly classified occurrence rate, an F-measure of 96%, and a precision of 92.6%. Keywords: Sentiment analysis, Naïve Bayes, SVM, OneR.