Opinion Mining on Twitter Data Using Machine Learning

Abstract

Twitter is a social media platform, where we can express our opinions about various events happening worldwide. Researchers are concentrating on opinion mining due to the exponential rise of electronic media-based society and online social networks. A method for figuring out how individuals feel about an event is called opinion mining, often known as sentiment analysis. This is done by using various machine learning algorithms. With the help of these algorithms, we can do sentiment analysis without reading tweets manually. The outcome of these models could help government and industry experts socialize their policies and products. This study investigates how emoji characters are used on social media and how that affects opinion mining.We used three machine learning models to determine the polarity of tweets, such as whether they were positive, negative, or neutral. After building the models, this method combines voting-based models, and additional comparison metrics are determined for each model. We applied the proposed ensemble model to a few tweets to show the emoji's usefulness.