Design and implementation of a data mining grid-aware architecture

Current business processes often use data from several sources. Data is characterized to be heterogeneous, incomplete and usually involves a huge amount of records. This implies that data must be transformed in a set of patterns, rules or some kind of formalism, which helps to understand the underlying information. The participation of several organizations in this process makes the assimilation of data more difficult. Data mining is a widely used approach for the transformation of data to useful patterns, aiding the comprehensive knowledge of the concrete domain information. Nevertheless, traditional data mining techniques find difficulties in their application on current scenarios, due to the complexity previously mentioned. Data Mining Grid tries to fix these problems, allowing data mining process to be deployed in a grid environment, in which data and services resources are geographically distributed, belong to several virtual organizations and the security can be flexibly solved. We propose both a novel architecture for Data Mining Grid, named DMGA, and the implementation of this architecture, named WekaG.