Registration of Heterogeneous Data Sources in the Case of the Grid Semantic Data Access Middleware (G-SDAM)

G-SDAM is a semantic data mediation middleware for Grids which allows the integration of multiple data sources using different data structures and semantics. While traditional data integration requires all data sources to follow common rules concerning the structure of data (and even more, requires the users to have knowledge about the structure of each single data source to be accessed), the G-SDAM approach allows a seamless and transparent view on distributed data. Local data sources as well as users accessing those data can both be part of Virtual Organizations since G-SDAM is planned to build upon the Globus Toolkit. The enabling element of the middleware is ontologymanagement including mapping between different data models (e.g. relational models, hierarchical file systems, and also ontologies itself) to a global OWL-based ontology database residing on a Global Repository Node (GRN). The GRN acts as the gateway to the distributed data sources. Queries can be composed in SPARQL using global semantics which can be seen as a common language. Without an agreement on facts and terms, seamless data integration would not be possible. For each application domain there will be different global semantics defined by domain experts. Ontology versioning support and further requirements concerning security, copyright issues, data provenance, etc. are also addressed by the G-SDAM approach.