Integration of Heterogeneous Semistructured Data Models in the Canonical One

To provide for interoperability of heterogeneous information objects it is required to establish a global, uniform view of the underlying digital collections and services. An information model is needed which is able to express uniformly the structure and semantics of heterogeneous data collections as well as the available services. Usually the mediator's layer is introduced to provide the users with the metainformation uniformly characterizing content of the underlying collections and with the canonical information model applied for deenition of such metainformation and for querying integrated world of digital collections. The paper focuses on the canonical model intended for homogeneous representation of various semistructured and hybrid data models that have been developed recently with orientation on the data contained in Web sites. The paper provides a short overview of several representative semistruc-tured and hybrid data models. The canonical information model (based on the SYNTHESIS language) intended for uniform representation of heterogeneous information in mediators is introduced. The paper shows how diierent categories of semistructured models can be equivalently and homegeneously represented in the canonical model.