A Semi-Supervised Approach to Sentiment Analysis of Tweets during the 2022 Philippine Presidential Election

With the increasing popularity of Twitter as both a social media platform and data source for companies, decision makers, advertisers, even researchers alike, have been so massive that manual labeling is no longer feasible. This research uses semi-supervised approach to sentiment analysis English Tagalog tweets using base classifier. In this study involving Philippines, where played central role in campaign candidates, during widely contested race between son Philippines’ former President Dictator, outgoing Vice Philippines were used. Using Natural Language Processing techniques, these annotated, processed, trained classify into three polarities: positive, neutral, negative. Through Self-Training with Multinomial Naïve Bayes classifier 30% unlabeled data, results yielded an accuracy 84.83%, which outweighs other studies from Philippines.