Análise de sentimentos expressos por políticos brasileiros na plataforma Twitter

Abstract

This study investigates how the dynamics of political polarization and the circulation of disinformation are structured and manifested on Twitter, through a sentiment analysis of posts related to Brazilian politicians. Initially, a systematic literature review is carried out to identify approaches and techniques used in international studies that explore the relationship between social networks, elections and the dissemination of fake news. Next, the research presents an overview of the most relevant investigations on entropy and polarization, highlighting models that employ the SPIN index to measure cycles of lexical homogenization. An experiment is conducted based on data collected from Twitter, involving millions of tweets mentioning national political leaders and parties, processed through cleaning, text normalization, and automated sentiment analysis using the Syuzhet package in R software. The polarity is calculated and predominance of positive, negative, or neutral sentiments, as well as the associated emotions, is identified by comparing different ideological spectrums. Is is observed that right-aligned groups exhibit a higher average positivity in their posts, while left-aligned groups show a lower positive balance, indicating important nuances in digital discursive construction. Retweet networks are also built to map the formation of clusters and the role of key influencers in spreading content, revealing structures that reinforce echo chambers and highly polarized interaction patterns. Finally, the implications of the findings for the democratic environment are discussed, pointing out how the architecture of platforms and informational consumption habits favor the spread of homogeneous narratives and reduce exposure to opposing views, creating fertile ground for radicalism and disinformation strategies. It is concluded that the combined analysis of sentiments, network, and entropy metrics provides valuable insights to understand the complexity of the political debate on Brazilian social networks.