The vision of the Semantic Web is to make Web content machine-readable. To describe data, the Resource Description Framework has been extended with a schema-level and description logics (e.g. RDF-S and OWL DL). To successfully retrieve data described by RDF-based ontologies, the query language SPARQL is currently being developed at W3C. However, although SPARQL also provides a client/server protocol, currently queries are targeted to single sites only. For a large-scale data integration, which

more »

... s the scope of a middleware developed within the Austrian Grid project, it is necessary to execute queries on distributed data nodes. Because of the large amount of research undertaken in the field of relational database systems, it is reasonable to apply relational algebra to SPARQL query processors. In fact, most RDF triple stores are based on relational database systems. In this paper, several concepts are proposed to enable the integration of heterogeneous, distributed data sources with SPARQL.