Whose Tweet is it Anyway? Classifying Tweets by Brazilian Users Through Supervised Machine Learning Techniques

Abstract

In two of the world’s largest democracies far-right sentiment is on the rise. Within the last four years the United States elected conservative, political outsider Donald Trump and Brazil elected former army captain and congressman Jair Bolsonaro. The similarities between the two leaders are many; the most jarring being their xenophobic, sexist and racist rhetoric. Many have inquired and researched the reasons as to why far right ideology has endured such recent growth in such developed democracies. In this thesis, I will examine the role of social media in today’s society. This will include its role in the proliferation of ideology – due to the publishing and sharing of tweets. In this work, I will pivot away from the United States and focus on Brazil – specifically Brazilian Twitter. The thesis will analyze how politicians and regular people use Twitter; it will also aim to examine the effects of politicians’ tweets on their constituents. Twitter has a plethora of data regarding how people feel. I could use this data to learn about how a country feels towards their president. This thesis falls under the field of textual analysis, but it will aim to do more. The research is threefold: first, I mined the necessary data using Twitter’s API, then I conducted sentiment analysis on these tweets, and lastly, I created a model that accurately predicts the political affiliation or stance of the creator of a tweet. I hope to provide an accurate depiction of how politicians are using the social media application as well as what that means for regular people like you and me.