Mining Public Opinions on COVID-19 Vaccination: A Temporal Analysis to Support Combating Misinformation

Abstract

This article presents a study that applied opinion analysis about COVID-19 immunization in Brazil. An initial set of 143,615 tweets was collected containing 49,477 pro- and 44,643 anti-vaccination and 49,495 neutral posts. Supervised classifiers (multinomial naïve Bayes, logistic regression, linear support vector machines, random forests, adaptative boosting, and multilayer perceptron) were tested, and multinomial naïve Bayes, which had the best trade-off between overfitting and correctness, was selected to classify a second set containing 221,884 unclassified tweets. A timeline with the classified tweets was constructed, helping to identify dates with peaks in each polarity and search for events that may have caused the peaks, providing methodological assistance in combating sources of misinformation linked to the spread of anti-vaccination opinion.

Benchmarking non-pharmacological policies from an efficient administration perspective: a panel DEA approach with strategic insights for the post-pandemic

“…As discussed by Nepomuceno et al. (2022a, b) and de Carvalho et al. (2022a,b) assessments on how public opinion about COVID-19 preventive measures and vaccinations affect the spreading in Brazil, monitoring media rumors on news sites and social networks and disseminating suitable preventive measures for the Coronavirus, reorganization of health professional visits and integrating a unified front with all public administrations (at municipal, state and federal levels) in order to support the effort to combat misinformation can be the most prominent reactions to efficiently cope with future waves of the pandemic.…”

The COVID-19 Infodemic on Twitter: A Space and Time Topic Analysis of the Brazilian Immunization Program and Public Trust

“…The present article reports a new study expanding the findings of de Carvalho et al [33], who applied temporal opinion analysis to tweets about COVID-19 vaccination in Brazil. The main objective is to present the results of a spatio-temporal analysis of a set of geocoded tweets, demonstrating which topics were discussed between 2020 and 2021 through an interface that considers the distribution of posts over time and the Brazilian geographic space.…”

“…The texts were collected using Twitter API 2.0 through a scraping script developed in Python, applying functions from the "requests" library. Larger corpora were constructed, as reported in de Carvalho et al [33], but a subset containing only geocoded tweets was extracted, composing a georeferenced corpus. Table S1 in the Supplementary Materials presents the parameters used in the scraping script.…”

“…Table S1 in the Supplementary Materials presents the parameters used in the scraping script. The blocks of keywords applied for scraping are based on the studies by de Oliveira et al [35] and de Carvalho et al [33,36].…”

“…Considering the geocoded tweets extracted from one of the corpora presented by de Carvalho et al [33], the initial corpus composition for the present study contained 62,873 tweets. We applied a cleaning process to eliminate tweets that escaped location and language filters, resulting in 55,758 tweets.…”