Análise das emoções em redes sociais como ferramenta para estudar manifestações populares

Abstract

This work aims to use computational linguistics tools to analyze the emotions expressed in texts posted by users on social media in response to the impactful event of George Floyd’s murder and the subsequent anti-racism protests worldwide, as well as addressing the arrival of COVID-19 in the United States and users’ reactions during commemorative dates. To achieve this goal, methods such as word embeddings, recurrent neural networks, and transformers are applied based on reference works, in order to capture the nuances of emotions expressed in user texts. By studying these events, it is observed that they trigger different types of public outcry, revealing variations in the predominant emotions at specific moments. The obtained results demonstrate a strong relationship between the event dates and the changes in emotions expressed by users. Analyzing emotions in response to impactful events provides valuable insights into how people react and cope with such situations, contributing to a deeper understanding of the emotional impact of these events on society. Furthermore, these findings can guide future actions and policies, aiming to better address such situations.