Diretrizes para a escolha de técnicas de visualização aplicadas no processo de extração do conhecimento

Abstract

The search for new and useful information on data, which tends to increases due to technological advancement, requires more sophisticated methods as those developed for the knowledge discovery in database. Such methods may employ data mining and visualization techniques in order to extract information and enable the domain expert to have a better understanding about the extracted information for decision making. Thus, data mining and visualization techniques can be applied together. However, this integration is not mandatory. It is possible to extract information from data just using visualization techniques as an exploitation tool. Therefore, the definition of visualization techniques, which may better fulfill this role during the extraction of knowledge and also make it the best way, requires an understanding about what can influence this choice, including the knowledge about the domain. In this context, the main contribution of this work is an analysis of visualization techniques in order to establish guidelines for the best choice of these techniques. These guidelines were based on parameters named as: data type, task type, volume, dimensionality and position of the attributes on the display. The procedure adopted to identify these parameters was based in the Grounded Theory methodology, where parameters were described and analyzed within visualization techniques, and also using computational tools in the practical application of visualization techniques on real and fictitious databases.