Sentiment Analysis on English and Greek Twitter Data Regarding Vaccinations

COVID-19 is one of the most discussed diseases worldwide. To control its spread, several vaccines have been developed and approved. However, low acceptance and worries about the efficacy of vaccines exist due to insufficient information, and social media influence. This study aims to classify people who are in favor or against vaccination, as well as people's preferences for three main types of available vaccines, Pfizer, Moderna and AstraZeneca. Twitter data were retrieved over seven months. This work proposes the adoption of an on-the-fly approach involving a sentiment analyzer that includes Machine Learning (ML). Moreover, this paper compares sentiment analysis techniques by applying the following Supervised ML algorithms: Logistic Regression (LR), Decision Trees (DT), Random Forest (RF), XGBoost Classifier (XGB) and Support Vector Machines (SVM). The dataset was divided into two periods in 2022: summer, and autumn. The main contribution is the achievement of high accuracy and F1-Score, when combining these models. The best results were obtained when tweets were analyzed using TextBlob and DT. Regarding the English tweets, the top accuracy was 97.99%, while the top F1-Score was 97.98% in the summer period. With Greek tweets, the performance was better in the autumn period reaching 99.92% accuracy and F1-Score.