Author(s)

F Angiulli, T Catarci, P Ciaccia, G Ianni, S Kimani, S Lodi, M Patella, G Santucci & C Sartori

Abstract

The process of data mining includes many steps, starting with the choice and preparation of data sources and ending with the presentation of the data mining results. In addition, it is generally accepted that the data mining is not a \“one shot” process, but rather the result is obtained through iterative refinement steps of algorithm choice, parameters settings and intermediate results presentation. An effective architecture for a data mining tool should therefore allow easy integration of three components: acquisition of data sources, data mining algorithms and results presentation. In this paper we present the architecture of a data mining tool which is under development in the framework of the project D2I (Data to Information), supported by the Italian MIUR (Ministry of Instruction, University and Research). The architecture is based on the concept of \“metadata repository”: it is a specification of the data exchanged by the various modules which guarantees flexibility and extensibility: new algorithm and presentation method can be added, provided that the metadata specification is available. As a guideline and a testbed for the architecture, we present the specification of some data mining methods and we sketch how their results can be presented.

Keywords