An overview of Brazilian researches in the Computer Science field in last years

Abstract

Recently, most traditional market services have joined online service platforms. Despite the practicality achieved, such services eventually bring a large amount of data to the Web. In this sense, data analysis, data engi- neering, and data science activities have become extremely necessary. In general, they can extract extra information about systems and users, allowing the owners to produce insights and analyze patterns. Then, we propose an evalua- tion methodology to be applied in the online scenario of registration of publications and scientific productions, such as ResearchGate and Lattes Platform of CNPq. This methodology is unsupervised and divided into three main stages: (i) obtaining and representing the data; (ii) application of topic modeling; and (iii) the labeling of topics. This proposal diverges from the literature’s proposes that are based on collaborative networks and supervised techniques. We applied this methodology to a Lattes database and were able to observe the evolution of Computer Science research in Brazil. Based on this analysis, it is possible to identify the most popular and least explored research lines in order to direct public investments according to a certain interest.