We believe that the possibility to use SPARQL as a front end to heterogeneous data without significant cost in performance or expressive power is key to RDF taking its rightful place as the lingua franca of data integration. To this effect, we demonstrate how RDF and SPARQL can tackle a mix of standard relational workload and data mining in public data sources. We discuss extending SPARQL for business intelligence (BI) workloads and relate experiences on running SPARQL against relational and native RDF databases. We use the well known TPC H benchmark as our reference schema and workload. We define a mapping of the TPC H schema to RDF and restate the queries as BI extended SPARQL. To this effect, we define aggregation and nested queries for SPARQL. We demonstrate that it is possible to perform the TPC H workload restated in SPARQL against an existing RDBMS without loss of performance or expressivity and without changes to the RDBMS. Finally, we demonstrate how to combine TPC-H or XBRL financial reports with RDF data from CIA factbook and DBpedia.
展开▼
