Utilizando word embeddings para verificar o alinhamento político de deputados federais brasileiros através de seus tweets.

Abstract

Nowadays Twitter is widely used by politicians as a more direct channel of communication with their voters and to give their opinion on various issues. Due to the large amount of content generated on social media, measuring the alignment of politicians in relation to certain subjects through their tweets makes it possible to automatically observe how they behave and express themselves using these media. Here, we measure the political alignment of Brazilian federal deputies with regard to Social Security Reform by applying the Word Embedding Association Test (WEAT), which quantifies association between word sets. The results show that the method used can measure the alignment of deputies to the Social Security Reform, coinciding with the positions taken by these politicians on the subject.