Characterizing Semantic Service Parameters with Role Concepts to Infer Domain-specific Knowledge at Runtime

E-Science platforms leverage Service Oriented Architecture (SOA) principles to deliver large catalogs of data processing services and experiments description workflows. In spite of their growing success, the usability of these platforms is hampered by their catalogs size and the domain-specific knowledge needed to manipulate the services provided. Relying on domain ontologies and semantic services to enhance the understanding and usability of e-Science platforms, our contribution is twofold. First, we propose to delineate role concepts from natural concepts at domain ontology design time which leads to a neuroimaging role taxonomy, making explicit how neuroimaging datasets are related to the data analysis services. Then we propose to exploit, at workflow runtime, provenance information extended with these domain roles, to infer new meaningful semantic annotations. Platform semantic repositories are thus transparently populated, with newly inferred annotations, through the execution of e-Science workflows. A concrete example in the area of neurosciences illustrates the use of role concepts to create reusable inference rules.