We would like to present a prototype system enabling a rather empirical than a formal approach to the problem of posing queries to a semantically rich (quality aspects, semantic distance, etc.) data integration system {G,S,M} (Global schema, Sources, Mediation) through integration not only of intensional but also of extensional semantics. While the first is provided by an alphabet Ag, as given by an ontology based global schema G, and a high level query language (conjunction/disjunction + inequalities + statistical operations), the latter enables synthesizing of data source specific and previously transformed query results according to well-defined set operations for heterogeneous, distributed data sources. Our approach contrasts with other GAV (Global-As-View) related architectures for mediation of integrated read-only views, in that it simplifies query processing while preserving flexibility when adding new data sources, despite the inherited complexity of mappings due to enhanced semantic description of data (semantic distance, quality parameters, etc.) such that statistical results and comparisons become more meaningful. |