- From: Matthew Perry <matthew.perry@oracle.com>
- Date: Wed, 02 Mar 2011 09:08:20 -0500
- To: Axel Polleres <axel.polleres@deri.org>
- CC: Birte Glimm <birte.glimm@comlab.ox.ac.uk>, SPARQL Working Group <public-rdf-dawg@w3.org>
Hi, To answer Axel's question about canonicalization, Oracle canonicalizes all xsd-typed literals. For example, if you insert: :a :p "001"^^xsd:decimal . :a :p "01"^^xsd:decimal . Only a single triple (:a :p "1"^^xsd:decimal) makes it into the triple store. We maintain information that allows us to recreate the original non-canonicalized triples, but SPARQL queries only match against the canonicalized triples. - Matt On 3/2/2011 4:36 AM, Axel Polleres wrote: > On 1 Mar 2011, at 19:42, Birte Glimm wrote: > >> >> On 1 March 2011 14:52, Axel Polleres<axel.polleres@deri.org> wrote: >> just looked quickly over those, manual inspection... >> >> >> On 22 Feb 2011, at 16:02, Birte Glimm wrote: >> >>> Hi all, >>> I ran the following OWL Direct Semantics tests cases and they pass: >>> :owlds01 -- Test: OWL DS bnodes are not existentials >> looks ok to me. >> >>> :owlds02 -- Test: OWL DS bnodes are not existentials with answer >> looks ok to me. >> >>> :plainLit -- Test: Plain literals with language tag are not the same >> looks ok to me (but why is this OWL/Entailment specific? It would be, potentially if you asked for >> "name"^^xsd:string under D-entailment?) >> >> Well, another disadvantage of D-entailmen is that the datatype map is not fixed, i.e., there is no guaranee that systems support the same datatypes and one does not have to support rdf:PlainLiteral or even xsd:string, which also makes testing relatively difficult. > The lists in http://www.w3.org/TR/rdf-mt/#DTYPEINTERP or http://www.w3.org/TR/rif-dtb/#Symbol_Spaces (or intersection thereof) could be a start? > >> As I said, I am for removing D-entailment alltogether ;-) > I would like this to be discussed at least once more, it seems there is use out there of datatypes - the fact that implementation do canonicalisation is IMO an indication that something should be done about datatypes at least. (We had some earlier discussion about a > D$^-$-Entailment a while back, but I think just nobody had time to spend on it. > > At least I would like to gather once more which implementation does *what* about Datatypes and see whether there's need to standardise that, before we decide to drop it alltogether... but, yes, it's a matter of time as well. > > Axel >> Birte >> >> >> didn't look into the bind0x tests yet... >> >> Axel >> >>> as the same literal without >>> :bind01 -- Test: bind01 - BIND fixed data for OWL DL >>> :bind02 -- Test: bind02 - BIND fixed data for OWL DL >>> :bind03 -- Test: bind03 - BIND fixed data for OWL DL >>> :bind04 -- Test: bind04 - BIND fixed data for OWL DL >>> :bind05 -- Test: bind05 - BIND fixed data for OWL DL >>> :bind06 -- Test: bind06 - BIND fixed data for OWL DL >>> :bind07 -- Test: bind07 - BIND fixed data for OWL DL >>> >>> The bind0x test cases are as for simple entailment, but the input data >>> is extended o make it an OWL 2 DL ontology. The test :plainLit is >>> applicable also under OWL 2 RDF Based semantics. >>> Birte >>> >>> -- >>> Dr. Birte Glimm, Room 309 >>> Computing Laboratory >>> Parks Road >>> Oxford >>> OX1 3QD >>> United Kingdom >>> +44 (0)1865 283520 >>> >> >> >> >> -- >> Dr. Birte Glimm, Room 309 >> Computing Laboratory >> Parks Road >> Oxford >> OX1 3QD >> United Kingdom >> +44 (0)1865 283520 >
Received on Wednesday, 2 March 2011 14:09:28 UTC