Principals

Our mapping language (R2RML) needs to support both ETL mapping style (the RDB content is mapped into RDF store and user query the RDF store), as well as on-demand mapping style (user in real time query the RDB and maps the returned result only to RDF.

With the on-demand approach and depending on the deployment environment, it might be possible to modify the RDB schema to do most of the work inside the SQL engine and leverage SQL optimizations or not. R2RML will support both paradigms for different deployment environments.

• Creating a SQL view to present the results of the RDB query results as RDF triples. The idea is to do most of the work in the SQL engine and leverage traditional SQL optimizations. This approach would require modifying the RDB schema.

• Use customized mapping, where the SPARQL query is translated into SQL query and is executed against the RDB. The returned results is mapped into RDF by mapping SQL tables into RDF/OWL classes and table attributes/columns into class properties outside the RDB engine. This approach does not required modifying the RDB schema.

In addition R2RML will support vendor specific data types by integrating a proprietary mapping functions provided by the vendor with the R2RML runtime.