Twitter sentiment analysis of the 2016 US Presidential Election using an emoji training heuristic

Twitter Sentiment Analysis can be a useful vehicle to provide deep insight into how citizens feel, thus it enables Governments across the world to track the public’s political views. Various Sentiment Analysis techniques have been proposed in recent years, and automatic annotation of training sets using an emoticon heuristic has been proved useful. As emojis are becoming increasingly popular in online written communication, this research investigates the feasibility of an emoji training heuristic for Twitter Sentiment Analysis of the 2016 U.S. presidential election. Multinomial Naïve Bayes classifier is used to build a sentiment classifier, which employs a variety of features. The results demonstrate the emoji heuristic together with our methodological framework achieve satisfying performance. We also apply our model to real-world presidential election tweets and present how the public views the top election candidates.