Emoji-based Opinion Mining using Deep Learning Techniques

Abstract

Abstract Informal media, such as Twitter, are more relevant today than ever before. Twitter is still a valuable tool for friends to communicate, but it has evolved into a public bulletin board where ordinary people, companies, and even big personalities like politicians and sports routinely publish their thoughts and participate in conversations. Since, Twitter is so extensively utilized throughout the world, the ability to do reliable opinion mining and gauge public opinion and perception on a variety of issues is more vital than ever. Emojis are frequently used to communicate feelings or sentiments that are difficult to express succinctly in language. For emoji-based opinion mining, a deep learning framework is presented to model the influence of emojis on text sentiment polarity. The emojis and words in microblog posts are combined to create emoji representations that include contextual information.