e-Government, open government data, data science, social network analysis, economics, clustering, classification model
A graph-based method is proposed for inferring similarities among companies from their affiliations in the context of expenditure financial transactions in the Brazilian Federal Government. There are trusted and untrusted companies. We performed a basic cluster analysis in the companies network to verify whether clusters (connected components) are discriminative concerning companies trustworthiness. Results show evidences that this is true, reinforcing the following hypotheses: (1) there are suppliers associations, which evidences the formation of cartels; and (2) public agencies and agents play an important role in the legality of financial transactions.
Pita, Marcelo and Torres, Gustavo da Gama, "Inferring Companies Similarities from Brazilian Government Expenditure Data" (2015). Proceedings of the XI Brazilian Symposium on Information Systems (SBSI 2015). 7.