Complex networks applied to political analysis: Group voting behavior in the Brazilian congress

The Senate and the Chamber of Deputies constitute the Brazilian Congress and are responsible for the Brazilian legislative management. Complex networks were shown to be a suitable tool to analyze this type of system. Several researches explored party dynamics in the Chamber of Deputies, however, no attention has been given to the Senate. Previous works that have stated the necessity of a backbone extraction methodology to be used in these types of networks also failed to define an automatic backbone extraction methodology to uncover group structure in legislative networks, reverting to heuristics or subjective approaches. In this work, we explore both legislative houses and compare them to see their differences and similarities. We also systematize an automatic backbone extraction methodology. Further, we expand on previous analyses by bringing spectrum and government x opposition analysis based on voting data. Our results show that the Senate and the Chamber of Deputies have behaved differently during major events in Brazil over the second decade of the century. From the obtained results it is fair to say that the dynamics for both houses are different and that the best backbone extraction algorithm varies over time and is different for each house.