- From: Jeremy Carroll <jjc@hplb.hpl.hp.com>
- Date: Mon, 01 Sep 2003 13:48:09 +0100
- To: Ian Horrocks <horrocks@cs.man.ac.uk>
- Cc: Dan Connolly <connolly@w3.org>, Steven Gollery <sgollery@cadrc.calpoly.edu>, www-rdf-logic@w3.org
If I have understood this thread the question is: eg:p1 owl:equivalentProperty eg:p2 . eg:p1 rdfs:domain eg:c . entails eg:p2 rdfs:domain eg:c . This is true in OWL, by the OWL CR Semantics. However, the related RDFS entailment, eg:p2 rdfs:subPropertyOf eg:p1 . eg:p1 rdfs:subPropertyOf eg:p2 . eg:p1 rdfs:domain eg:c . entails eg:p2 rdfs:domain eg:c . while true in OWL is not true in RDFS (by the drafts to be published at the end of the week). i.e. the OWL semantics is an extension of RDFS semantics in a way that is visible in this entailment. Jeremy Ian Horrocks wrote: > On August 27, Dan Connolly writes: > >>On Wed, 2003-08-27 at 12:39, Steven Gollery wrote: >> >>>In DAML: >>> >>>if there is a property P1 whose domain is class C1 and range is class C2 >>> >>I presume you mean to use domain/range in the sense >>of rdfs:domain and rdfs:range; i.e. C1 is *a* range >>of P1 and C2 is *a* range of P1. i.e. >> forall X, Y, if P1(X, Y) then X in C1 and Y in C2. >> >>and *not* the other fairly common usage of range, i.e. >> for all Y, if Y in C2 then there is some X with P1(X, Y). >> >> >>>and there is a property P2 whose domain is class C3 and range is class >>>C4 >>>and there is a statement that P1 is the samePropertyAs P2, >>> >>>does it follow that C1 is the sameClassAs C3 and C2 is the sameClassAs >>>C4, >>> >>no. Counterexample: >> eats rdfs:domain Animal; rdfs:range Food.# P1/C1/C2 >> consumes rdfs:domain rdf:Resource; rdfs:range rdfs:Resource.# P2/C3/C4 >> >>Now we can have >> eats daml:samePropertyAs consumes. #i.e. same extension >>but lots of things can be in rdf:Resource without >>begin in Animal nor Food. >> >> >> >> >>> or does it just add a new class to the domain and range of each of >>>the properties? >>> >>Umm... no, I don't think so; I'm not even sure what that means. >> >> >>> Or neither one? >>> >>It just means that P1 and P2 have the same property extension. >>i.e. for every X and Y, P1(X,Y) iff P2(X,Y). >> > > Yes, and this means that for every x, if there exists some y > s.t. P1(x,y) (or P2(x,y)), then x has type (intersectionOf C1 C3), and > y has type (intersectionOf C2 C4). The semantics of DAML+OIL do not > make it completely clear if (domain P1 (intersectionOf C1 C3)) and > (range P1 (intersectionOf C2 C4)) (and the same for P2) are > entailed. However, this *is* the case in OWL, i.e., the intersection > domain and range restrictions are entailed. > > As Dan points out, the fact that the domain of P1 is (intersectionOf > C1 C3) does *not* entail that C1 and C3 are equivalent. > > Ian > > > >>fyi, the OWL equivalent of daml:samePropertyAs >>is owl:equivalentProperty. >> >>cf >> >>4.2.1 owl:equivalentProperty >>http://www.w3.org/TR/2003/CR-owl-ref-20030818/#equivalentProperty-def >> >>and the links to the other parts of the specification >>that discuss equivalentProperty in >> >>Appendix C. OWL Quick Reference >>http://www.w3.org/TR/2003/CR-owl-ref-20030818/#appC >> >> >>-- >>Dan Connolly, W3C http://www.w3.org/People/Connolly/ >> >> >> >
Received on Monday, 1 September 2003 08:54:31 UTC