Análise multidimensional da produção científica em ciência da computação

Abstract

The amount of Computer Science graduate programs in Brazil has increased over 200\% in the past twelve years, from 17 to 50 programs. In light of this growth, it is important to assess how the quality of the scientific production of these programs compares to their equivalent in other countries. The objective of this dissertation is to perform an analysis of the Computer Science scientific production, using data collected from the eight major programs in the country according to official data from CAPES for the triennium 2004-2006, and 22 of the most important programs in North America and Europe. Therefore, the first step was to design and implement a data warehouse of scientific publications from the 30 programs to allow multidimensional analysis of bibliographic data. The data warehouse provides an opportunity for analysis using several dimensions, which allows the acquisition of statistical data regarding the graduate programs related in this study, such as: average of publications per professor per program, publications distribution among subareas of Computer Science, most popular subareas per program, average of citations per article per program, productivity measurements per professor, per program or per country, among others. Thus, the data warehouse proposed allowed the analysis of the Brazilian programs publication profile, compared to major programs in North America and Europe. Results demonstrate that national scientific production, represented by articles published in journals and international conferences, is comparable in volume, quantity and impact to some of the major programs in North America and Europe.