About

BirdDog is a project founded by Roney Fraga Souza, a professor at Faculty of Economics, Federal University of Mato Grosso in Cuiabá, and, editor of Revista de Estudos Sociais (Journal of Social Studies).
Our main line of research is a technology forecasting project which maps fields of knowledge by different dimensions. It is used for reading of curriculum of the platform lattes to find which professionals work with a certain area of knowledge, considering the weight of the publications and the relations of co-authorship. With articles it is possible to use data obtained from databases such as Scopus and WoS Web of Science to build networks of scientific publications, group articles by similarity of connections, calculate topological means of importance of each article, extract the content of groups with NLP Natural Language Processing, among other procedures. The same procedures applied to scientific articles are finally applied to patents, where it is possible to find the frontier of knowledge in the world of patents (USPTO). All the procedures used are composed by unsupervised methods, which allow the applications of thousands of authors, articles and patents, and spend little processing time.
We also investigate economics issues using quantitative methods, computation algorithms, network analysis and big data.

Main Skills

Big Data Analysis

Many of the data that we work on is obtained via scraping data from internet pages. We also usually follow and manipulate some databases the Brazilian government makes available. These are:

Censo Demográfico
Pesquisa Nacional por Amostra de Domicílios - PNAD
Pesquisa de Orçamento Familiar - POF
Pesquisa Mensal de Emprego - PME
Censo da Educação Superior
Censo Escolar
ENEM
ENADE

Text Mining

We use NLP Natural Language Processing to analyze the contents of articles and patents. The idea is to obtain the content of a set of documents, via language filters to return candidate terms, without the need to read those documents. The analysis of the importance of these candidate terms is done by metrics such as tf-idf.

Web Scraping

Automated processes implemented using a bot. It is a form of copying, in which specific data is gathered and copied from the web, typically into a central local database or spreadsheet, for later retrieval or analysis.

Main Work Tools

R

Statistical Software•Programming language

allows you to analyze data in a proficient way with a few lines of code. The great availability of packages, free software, community and flexibility are the strongest points to choose R as a work tool.

Python

Programming Language

Python supports multiple programming paradigms, including object-oriented, imperative, functional and procedural, and has a large and comprehensive standard library.