Knowledge discovery in public databases: a methodological structure proposal

Abstract

The contemporary world experiences a considerable growth of two phenomena that motivate this article. The first is the diffusion of digital technologies and the second is the growth of the share of its population that lives in poverty. Modern society generates and stores data and information in a very large scale. This article analyzes the transformation of public data in socialy valuable knowledge through the use of KDD (knowledge discovery in databases) with two main objectives: to propose a model for use of KDD in public databases and to demonstrate that the use of this methodology can generate useful knowledge for policies related to poverty relief. A complete model of KDD was elaborated and applied in the mining of Salvador¿s domiciliary data collected in the Brazilian census for the year 2000. In this process, the phenomenon of poverty was treated through a heuristical and statistical approach, resulting in a multidimensional representation based upon the characteristics of the domiciles and their inhabitants. This process allowed the creation of a tipology of poverty for the city of Salvador.