- From: Jeen Broekstra <jeen.broekstra@aidministrator.nl>
- Date: Tue, 19 Nov 2002 20:08:57 +0100
- To: Sandro Hawke <sandro@w3.org>
- CC: Manne Miettinen <manne.miettinen@csc.fi>, www-rdf-interest@w3.org
Sandro Hawke wrote: [snip] >> The first one produces these two triples: >> >> <_:MyClass> <rdf:type> <rdfs:Class> <_:MyClass> <rdfs:subClassOf> >> <rdfs:Class> >> >> The second one produces just this one triple: >> >> <_:MyClass> <rdf:type> <rdfs:Class> [snip] > While that's all true, the inputs have an identical RDFS closure [1]. > That means they'll be treated the same by software which implements > RDFS inference. When I use a term (such as rdfs:Class) in authoring > RDF documents, I generally assume my readers will "understand" it, > and that means they'll be doing RDFS inference. Its true there may > be useful RDF processors which do not understand RDFS, but it's hard > to author thinking about all the possible partial-understandings. > > -- sandro > > [1] http://www.w3.org/TR/rdf-mt/#rdfs_entail (although I can't find > my way through the rules to get the desired triple, it should be > there.) I'm sorry, but I do not think that that is true. I have an implementation of an RDFS statement inferencer running and it does not produce this result. And even though that's not definite proof, I think it is right in this case. It would also be very strange if it were, from a modeling perspective. Just to be sure, I have taken another look at the rdfs entailment rules, and can find no possible combination of productions that will arrive at this conclusion. The only possible entailment rules that could produce the subClassOf statement are rdfs6 and rdfs8. However, rdfs6 requires as input a subPropertyOf statement, and rdfs8 only encodes transitivity of the subClass relation itself. Neither premise is being fulfilled by the example input. Regards, Jeen
Received on Tuesday, 19 November 2002 14:09:02 UTC