- From: Graham Klyne <GK@dial.pipex.com>
- Date: Mon, 02 Oct 2000 14:16:29 +0100
- To: Ian Horrocks <horrocks@cs.man.ac.uk>
- Cc: rdf interest <www-rdf-interest@w3.org>
Ian, I think what you describe has exactly the same effect as Jan Grant's recent suggestion: >and loosely: > >P has a range of (a member of the union of A and B) > > A --[rdfs:subclassOf]-> anon:C > B --[rdfs:subclassOf]-> anon:C > P --[rdfs:range]-> anon:C > >(give anon:C a real URI if you prefer). except that Jan's approach doesn't depend on additional application-specific awareness. Or am I missing something? #g -- At 02:17 PM 9/30/00 +0100, Ian Horrocks wrote: >In the OIL language (see http://www.ontoknowledge.org/oil/) we extend >RDF schema with (amongst other things) logical constructs that allow >you to say things like: > > P --[rdfs:range]-> (A or B) > >This approach has many advantages: it allows disjunctive semantics to >be exactly captured and it saves cluttering up the class hierarchy with >unwanted classes. Of course the meaning would only be accessible to >OIL-aware agents. > >In this setting, given > > S --P--> O > >we can infer > > O --rdf:type--> (A or B) > >Note that this is NOT the same as being able to infer either > > (O --rdf:type--> A) >or > (O --rdf:type--> B) > >As for validity checking, without being able to express e.g., negation >or disjointness, then "validation" is not very meaningful as we can >never infer invalidity - we can only infer tighter constraints. > >Regards, Ian ------------ Graham Klyne (GK@ACM.ORG)
Received on Monday, 2 October 2000 09:40:58 UTC