Sentiment Analysis of Covid-19 Vaccination on Twitter Using Classification Algorithms based on PSO

Abstract

Twitter social media is widely used by internet users to provide opinions on an event. In 2021, the opinion about the Covid-19 vaccination was widely discussed by the Indonesian people on social media. Togetherness of opinion can be categorized as positive sentiment and negative sentiment. To categorize negative and positive sentiments, data mining processes can be used. This study discusses the sentiment of Covid-19 vaccination using classification. Data from Twitter is collected in a crawling process, then labeled into two classes, namely positive and negative sentiments. After labeling (polarity) the next stage is data preprocessing which consists of transform case, annotation removal, tokenizing and stemming. The classification algorithm used is Naïve Bayes (NB) and Support Vector Machine (SVM) and then compared with the classification algorithm using Particle Swarm Optimization (PSO). Tests were carried out using k-Fold Cross Validation to obtain accuracy values, Confusion Matrix tables and Area Under Curve. The test results on the classification using PSO are better than those without using PSO. The results of the accuracy of the NB and SVM algorithms are 64.04% and 72.55%, while the accuracy results after PSO on the SVM and NB algorithms are 70.43% and 76.38%, respectively.