- From: <jos.deroo@agfa.com>
- Date: Thu, 8 Sep 2005 01:26:14 +0200
- To: franconi@inf.unibz.it
- Cc: bparsia@isr.umd.edu, connolly@w3.org, public-rdf-dawg@w3.org, public-rdf-dawg-request@w3.org
[...] > However, we spotted a case where the SPARQL document > fails to find such redundant triples in the answer, > if you want to take RDF-MT seriously. Consider the > following RDF graph: > > <http://example.org/book/book1> dc:title "SPARQL" . > _:b dc:title "SPARQL" . > > According the the RDF-MT semantics, this graph contains > a redundant triple (the second triple with the bnode), > since the original graph with that triple and the > graph without that triple are equivalent, i.e., they > RDF-entails each other. I.e., the above graph is > RDF-equivalent according to RDF-MT to the graph: > > <http://example.org/book/book1> dc:title "SPARQL" . > > In other words, the deductive closures according to > RDF-MT of the two graphs above *are the same*. And I > have to get the *same answer* when I query these two > graphs! > So, when I query each of the above graphs as follows: > > SELECT ?x > WHERE { ?x dc:title "SPARQL" } > > I should *not* get the bnode coming from the redundant > triple, but simply {<http://example.org/book/book1>}. Otoh, for CONSTRUCT { ?x dc:title "SPARQL" } WHERE { ?x dc:title "SPARQL" } I actually get <http://example.org/book/book1> dc:title "SPARQL". _:b_0_ dc:title "SPARQL". which I assume to be fine, no? -- Jos De Roo, AGFA http://www.agfa.com/w3c/jdroo/
Received on Wednesday, 7 September 2005 23:26:39 UTC