- Automatically discover SPARQL endpoints and their query ontologies.
- Offer up a SPARQL endpoint, which takes SPARQL queries and evaluates them by running sub-queries over the discovered endpoints, collects a covering of instance data which satisfies the original query, then runs the original query over the retrieved instance data.
- Run SPARQL-DL queries for introspecting the available ontologies.
- Should we be able to run queries over explicitly referenced ontologies, e.g. for managing ontologies in Plone we'd want to run an ontology-lint process which checks various attributes of a given ontology — could this be done with named graphs?
- Need a way to record provenance — again could use named graphs.
- Need to investigate the requirements on the query ontologies for query optimization, e.g. disjointedness of top level classes.
Design & Implementation
The description above suggests two new content types:
- Source: containing metadata about a SPARQL end point
- Ontology: containing metadata about an OWL ontology