W3C home > Mailing lists > Public > www-rdf-logic@w3.org > November 2000

Re: A Model Theoretic Semantics for DAML-ONT

From: Deborah McGuinness <dlm@ksl.stanford.edu>
Date: Tue, 14 Nov 2000 10:33:43 -0800
Message-ID: <3A118587.3D9F6736@ksl.stanford.edu>
To: Richard Fikes <fikes@ksl.stanford.edu>
CC: Jeff Heflin <heflin@cs.umd.edu>, www-rdf-logic@w3.org


Richard Fikes wrote:

> > 1) The "imports" statement is lacking semantics. The intent of "imports"
> > is so that you can say the current ontology or document includes the
> > rules and definitions from some other ontology. I imagine in order to
> > define this formally in KIF, you would need to add something like the
> > "ist" predicate (meaning "is true in") from context logic (see
> > McCarthy's and Guha's work). Assuming you had such a predicate, where
> > the first argument was an ontology, the second a predicate name, and the
> > rest arguments of that predicate, you might add this rule:
> >
> > (=> (imports ?o1 ?o2)(ist ?o2 ?p ?a1 ?a2)
> >     (ist ?o1 ?p ?a1 ?a2))
>
> Yes, we did not provide an axiomatization for "imports".  Your
> suggestion to define and use an "is true in" relation would seem to
> work.  Do you have an axiomization to suggest that defines "is true in"?

in the axiomatization, we should consider the input from alex borgida:
he suggests that using ist  allows circular imports.
(i am thinking that this is probably something that one would want to warn
people about since it could be problematic, but also probably something that
we would want to allow for broad web usage).

he also suggests that we might want to consider ist as a binary relation
rather than having it take an arbitrary number of arguments.

> > 2) The following is used for the definition of "domain":
> >
> > %% The object in an RDF statement of the form "(object property
> > resource)"
> > %% must be an instance of the domain of property.
> > (=> (Property ?p)(domain ?p ?d)(holds ?p ?x ?y)(type ?x ?d))
> >
> > This implies that if there is more than one domain stated for a
> > property, that the "object" must be an instance of the intersection of
> > these domains.
> > However, the RDF Schema spec states "If there is more than one domain
> > property, the constrained property can be used with instances of ANY of
> > the classes..." That is, the "object" must be an instance of the union
> > of the domains.
> >
> > Personally, I like your definition better. If you use the RDF schema
> > definition, then it becomes possible to extend an ontology in such a way
> > that content marked up with respect to new ontology is incompatible with
> > the extended ontology. Also, the RDF definition is meaningless (i.e.,
> > you cannot deduce anything from it or perform type checking) unless you
> > can close the world and guarantee that there do not exist any more
> > domain statements for the property in question elsewhere on the Web.
> >
> > I think this is a case where it is worth deviating from RDF, but would
> > like to hear the opinions of others.
>
> Yes, I would recommend adopting the axiomization of "domain" that we
> specified for at least the compelling reasons you mention.
>
> Thanks for your feedback.
>
> Richard

I agree that we want to highlight that this is a difference and I suggest
keeping our definition although presumably ora or dan might be able to shed
more light on why rdfs chose their route.


--
 Deborah L. McGuinness
 Knowledge Systems Laboratory
 Gates Computer Science Building, 2A Room 241
 Stanford University, Stanford, CA 94305-9020
 email: dlm@ksl.stanford.edu
 URL: http://ksl.stanford.edu/people/dlm/index.html
 (voice) 650 723 9770    (stanford fax) 650 725 5850   (computer fax)  801 705
0941
Received on Tuesday, 14 November 2000 13:33:46 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 2 March 2016 11:10:32 UTC