- From: Jonathan Rees <jar@creativecommons.org>
- Date: Sun, 4 Oct 2009 18:01:29 -0400
- To: Kevin Tyson <kevin.tyson@gmail.com>
- Cc: Bernard Vatant <bernard.vatant@mondeca.com>, semantic-web@w3c.org
OWL has no "instance of" in the OO sense, much less "direct instance of". It only has rdf:type or "member of" in the sense of logical class membership (which often, confusingly, is called "instance of"). "Class" is not used in the OO sense, it's used in the mathematical sense. So if Primate is the union of Human and Monkey, you're saying that everything that is a member of the Primate class is either a member of the Human class or a member of the Monkey class. The consistency question is then: does there exist a model in which all the axioms and their consequences are true. If you're missing the information of whether X is in Monkey vs. in Human (neither membership is entailed), then there may very well exist two distinct models of your ontology. If you stop using "instance of" and just say "member of" you'll be much less likely to get confused, I think. The terminological confusions (OO perversion of "class" and OWL-related perversion of "instance") are hateful, IMO, but we have to live with them. In theory one could probably roll an OO-like "direct instance of" relationship, but it would have to live outside of DL as an annotation property (as it would depend on the existence of distinct classes having the same members), and would therefore be inferentially weak. On behalf of all my bonobo and lemur friends, I take great offense to your biology, by the way. Jonathan On Thu, Oct 1, 2009 at 7:16 PM, Kevin Tyson <kevin.tyson@gmail.com> wrote: > Thanks Bernard. > If I understand correctly, this does not alter the semantics of the covering > axiom that can be implemented today using disjoint classes which in union > comprise the equivalent class or super class of the covered class. > What I don't understand is why an instance of the covered class does not > create an inconsistency. > Is this an artifact of the tools, in my case pellet and Protege 4, or is an > instance of a covered class which is not also an instance of a covering > class consistent? > TIA, > Kevin > > On Thu, Oct 1, 2009 at 3:46 AM, Bernard Vatant <bernard.vatant@mondeca.com> > wrote: >> >> Kevin >> >> I would say DisjointUnion in OWL 2 is exactly what you are looking for >> http://www.w3.org/TR/2009/PR-owl2-new-features-20090922/#F1:_DisjointUnion >> >> Best >> >> Bernard >> >> 2009/9/30 Kevin Tyson <kevin.tyson@gmail.com> >>> >>> Greetings, >>> Is it possible to create a covering axiom such that any instances of the >>> covered class must be a direct instance of one of the covering classes? >>> Such a structure would be analogous to the "abstract super class" pattern >>> popular in some O-O programming and modeling languages. >>> TIA, >>> Kevin >>> >>> -- >>> Kevin P. Tyson >>> Kevin.Tyson@gmail.com >> >> >> >> -- >> Bernard Vatant >> Senior Consultant >> Vocabulary & Data Engineering >> Tel: +33 (0) 971 488 459 >> Mail: bernard.vatant@mondeca.com >> ---------------------------------------------------- >> Mondeca >> 3, cité Nollez 75018 Paris France >> Web: http://www.mondeca.com >> Blog: http://mondeca.wordpress.com >> ---------------------------------------------------- > > > > -- > Kevin P. Tyson > Kevin.Tyson@gmail.com >
Received on Sunday, 4 October 2009 22:02:09 UTC