ISSN: 2577-4379
Authors: Kızılyer M and Çakıt E
The objectives of the study were two-fold: (1) To group mindsets related to COVID-19 vaccinations and examine their distribution by country. Then, based on this distribution, the study aimed to compare the number of vaccinations, deaths, and cases and analyze the relationship between these numbers and the mindset of the society. (2) To analyze people's tweets about the vaccine and compare them with the number of people vaccinated, in order to determine if there was a significant result. The study analyzed data from 17 countries among the top 20 countries with the highest gross national product in 2020. Machine learning methods such as multinomial logistic regression, random forest, naive Bayes, and ridge classification were used to evaluate the performance of predictive models. The accuracy achieved by these models were as follows: naive Bayes (76%), random forest (85.03%), ridge classification (85.72%), and multinomial logistic regression (86.67%). In conclusion, the study found that with increasing vaccination rates, positive interpretations of vaccines differed more than other moods. The study contributes to advancing awareness of the public's perception of COVID-19 vaccinations and supports the goal of eliminating coronavirus from the planet.
Keywords: Sentiment Analysis; Twitter®; COVID-19; Natural Language Processing; Machine Learning
