- From: Jeremy Carroll <jjc@hpl.hp.com>
- Date: Fri, 29 Feb 2008 15:45:36 +0000
- To: "Web Ontology Language ((OWL)) Working Group WG" <public-owl-wg@w3.org>
- CC: "Seaborne, Andy" <andy.seaborne@hp.com>
A further query I have had from colleagues concerns the relationship with the semantics of SPARQL queries. http://www.w3.org/TR/2008/REC-rdf-sparql-query-20080115/#BGPsparql talks about pattern matching, and blank nodes, complete with a link into RDF Semantics, which appears to be pertinent. This allows a blank node in a basic graph pattern to match a named node or a blank node in the rdf graph being matched. Boris at one point called the skolem semantics as 'SPARQL based' or something like that, and I am having increasing need for clarification as to quite what this means. To what extent would this change impact compatibility between OWL DL and SPARQL, as we migrate from OWL 1.0 to OWL 1.1? To make this concrete, here is a test case SELECT ?name FROM <http://example.org/A> WHERE { _:x eg:dp ?name } where A is as before A: _:a rdf:type owl:Thing. eg:dp rdf:type owl:DatatypeProperty . _:a eg:dp "foo". If we read A with an RDF tool the answer is { ?name="foo" }. If we read A with an OWL 1.0 DL toolset the answer is that {?name="foo"} (I am not completely clear how we use SPARQL above OWL DL - is this an intended point of compatibility?) If we read A with an OWL 1.1 DL toolset with the proposed resolution to ISSUE-3, then the answer is the empty set. I am unclear as to what interop is expected between OWL DL and SPARQL, so maybe this example is unfair. However, clarification of this would be helpful. a) what interoperability is expected between OWL DL and SPARQL and b) what changes between 1.0 and 1.1 Jeremy
Received on Friday, 29 February 2008 15:46:34 UTC