Identifying influential members of parliament using topological features in a co-votation network

• Ano de defesa: 2017
• Autor(a) principal: Nunes, Marcelo Granja
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Universidade Federal do Rio de Janeiro; Brasil; Instituto Alberto Luiz Coimbra de Pós-Graduação e Pesquisa de Engenharia; Programa de Pós-Graduação em Engenharia de Sistemas e Computação; UFRJ
• Palavras-chave em Português: Redes sociais; Modelos matemáticos; Aprendizado computacional; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO::MATEMATICA DA COMPUTACAO::MODELOS ANALITICOS E DE SIMULACAO
• Link de acesso: http://hdl.handle.net/11422/8653

This works proposes to investigate whether influent members of the parliament can be identified solely by using voting results. Data from voting sessions in the plenary of the Brazilian House of Representatives during the year of 2015 were used to create a co-voting network. In this network, vertices are congressmen and weighted edges represent pairwise similarity between congressmen regarding their voting behavior. Ground truth data about most influential congressmen were obtained from a report prepared by a political science think tank. Initially, congressmen were ranked according to different centrality metrics. Afterwards, those topological proprieties were combined by using them as input features to feed classification algorithms. Results indicate that, as measure by the average precision over the precision-recall curve, this method performed almost 3 times better than what would be expected if influent congress were selected by random chance. This suggests that information regarding congressmen influence is encoded into voting results.