Uma análise da produção científica em Ciência da Computação na América Latina

Abstract

Different studies have been conducted aiming to compare the scientific production in Computer Science (CS) of Brazil and other Latin American countries with that of different countries in North America, Europe and Asia. However, little is still known about the scientific production of researcher groups from these countries in the area. In this MSc dissertation, we study the profile of the most important research groups from Argentina, Brazil, Chile, Colombia, Costa Rica, Cuba, Mexico, Paraguay, Peru, Uruguay and Venezuela in a period of 20 years, from 1994 to 2013. To do so, our study addresses researchers of 48 graduate programs in CS from academic institutions in Latin American. The data was collected from DBLP (Digital Bibliography & Library Project), the largest CS digital library. Our work includes a statistical analysis of the scientific production of each country, as well as analyses of the coauthorship networks based on metrics and properties of complex networks. Our results show a stable growth of the scientific production in the last decade in Argentina, Brazil, Chile and Mexico, and a significant growth in countries with less tradition in CS such as Colombia, Cuba, Costa Rica, Paraguay and Peru. Particulary, we show that the scientific collaboration among prestigious research groups in the region has resulted in better quality publications and an increase in terms of publication rate of each research group. Thus, we confirm that collaborations are an important mechanism to increase the rate of publications. In addition, we show that exists a correlation between the degree and closeness centrality metrics with the total number of publications of the researchers. We also confirm that scientific collaboration networks follow the power law, where there is evidence of few researchers with a large degree within the network structure that at act as hubs on the network.