Brazilians Divided: Political Protests as Told by Twitter

摘要

After a fierce presidential election campaign in 2014, the re-elected president Dilma Rousseff became a target of protests in 2015 asking for her impeachment. This sentiment of dissatisfaction was fomented by the tight results between the two favorite runners-up and the accusations of corruption in the media. Two main protests in March were organized and largely reported with the use of Social Networks like Twitter: one pro-government and other against it, separated by two days. In this work, we apply two supervised learning algorithms to automatically classify tweets during the protests and to perform an exploratory analysis to acquire insights of their inner divisions and their dynamics. Furthermore, we can identify a slightly different behavior from both parts: while the pro-government users criticized the opposing arguments prior the event, the group against the government generated attacked during different times, as a response to supporters of government.