Mineração de Textos: Análise de sentimentos de tweets sobre o Projeto de Lei n° 2630/2020

Abstract

The internet is embedded in various aspects of human life, facilitating communica tion and information exchange. In the political realm, social media platforms are used to disseminate governmental decisions and information, enhancing individuals’ access to politics. Both lawmakers and citizens turn to these platforms to express opinions and influence social and political issues. However, the misuse of social media can escalate political polarization and spread misinformation, posing a threat to democratic regime. This work aims to analyze the sentiments expressed by users on the Twitter/X so cial network regarding Brazilian Senate Bill No. 2630/2020. To achieve this, data from this network were collected, and Sentiment Analysis steps were performed. Multiple text classification algorithms were applied to the dataset, both before and after text preproces sing. The performance evaluation of classifiers revealed a significant improvement when text preprocessing was applied. Additionally, Linear SVM was chosen as the classifier algorithm with the highest accuracy. The results indicated that everyday events involving the dissemination of fake news on social media contribute to increased tweets about Bill 2630/2020, which were predo minantly negative. Furthermore, it was observed that right-leaning legislators exhibit greater engagement on social media.