Desinformação na pandemia de COVID-19: modelagem de tópicos e análise de discurso de um corpus de mídia social

Abstract

The growing expansion of social media platforms has potentiated two important processes, usually related, and already considered a social concern at the beginning of the 21st century: The massive dissemination of disinformation and the polarization of political discourse. In the context of the COVID-19 pandemic, in Brazil and in the world, we cannot exclude the impact of these platforms and political narratives as potentiators of social movements against vaccination, some already widespread in some countries like the antivax. Considering this scenario, this project aims to analyze and classify the most relevant topics in each group, classified by the political spectrum, during the COVID-19 pandemic in Brazil. Using a dataset with more than 220,000 posts related to this subject and extracted from the social network Facebook, between February and July 2021 in Brazil, through a web scraping process, a COVID-19 pandemic and vaccination news texts Corpus was consolidated. By applying topic modeling methodologies and discourse analysis, this Corpus is analyzed in more detail to evaluate what was the discourse focus and the topics temporal evolution for each group according to their political view. Sources and news are also evaluated in relation to their reliability, misinformation and whether there are positions contrary to the vaccination measures in this period. Although there is not much evidence of this process having taken place in Brazil, there is a clear difference in the objectives, narratives and in the reliability of the sources, some sources listed in the COVID CPI as disinformation disseminators during the same period. Even when applied to a specific, somewhat limited period of the pandemic, the results are very promising for future related works, whether complementary, in other periods, or even in similar analyzes applied to other subjects.