Towards Coordination Patterns for Complex Experimentations in Data Mining

In order to support the management of experimental activities in a networked scientic community, the exploitation of serviceoriented paradigm and technologies is a hot research topic in E-science. In particular, scientic workows can be modeled by resorting to the notion of process. In this paper we present a preliminary study about the exploitation of the formal coordination language Reo for representation and verication of complex coordination patterns typically appearing in distributed experimental processes for the extraction of potentially useful knowledge from data (i.e., Data Mining). The study demonstrates the feasibility of the approach, its advantages and potential for the Data Mining eld and for E-science in general. It also highlights the peculiarities of Data Mining processes that solicit an extension of the existing process modeling/checking approach.