[Fwd: Re: [OWL] Annotations on subclass axioms etc.]

[Denny Vrandecic]

Just to keep you informed on our break-through results :) (I always 
forget to type the reply-to-all button, gosh)

denny

-------- Original Message --------

On 11/17/05, Denny Vrandecic <dvr at aifb.uni-karlsruhe.de> wrote:
> >>Well, you could kinda put the axioms x,y and z into the ontology o and
> >>then make statements about o, because o is just another URI as well, and
> >>if I remember correctly the spec doesn't state that ontologies and
> >>instances have to be disjoint.
> >
> >
> > hmm..but the problem still remains: there is no way to say that it is
> > the specific combination of those three axioms alone which achieves
> > some purpose. I mean you could use reification to talk about
> > individual statements in the ontology, but not about some particular
> > statements as a collection unless you use rdf:Bag or Xpointers or
> > something funky like that..(which doesnt have standard OWL-DL
> > semantics)
> >
> > For example, suppose I have three axioms:
> > Jack rdf:type PhdStudent
> > PhDStudent subClassOf someValuesFrom(hasAdvisor, Professor)
> > hasAdvisor subPropertyOf knows
> >
> > I want to add an annotation in my ontology that the above three axioms
> > taken together entail "Jack knows some Professor"...how can one do
> > this easily?
>
>
> I think it is pretty easy. You have the following ontologies:
> o:
> Jack type PhDStudent
> PhDStuden sco someValuesFrom(hasAdvisor, Professor)
> hasAdvisor spo knows
>
> p:
> Jack type someValuesFrom(knows, Professor)
>
> and finally we can say in the ontology q:
> o entails p
>
> It's just, your ontologies get much more finegrained then you're used
> to. And I don't like this too much, but it seems to solve the problem.

hmm..I see what you mean...its interesting

> >>But Sean keeps telling me that there is no real semantic connection
> >>betwen o and (x,y,z), which I am simply not clever enough to grasp.
> >
> > my understanding is that the ontology (o) gives the *context* for
> > those axioms (x,y,z)..i.e., in general an ontology is a set of axioms
> > taken to mean something together (in the context of the ontology) and
> > so I do think there is a semantic connection in that regard.
> >
> > For example, in your ontology, you may have the three axioms "Penguin
> > subClassOf Bird", "Bird subClassOf hasValue(canFly, "true"),
> > "FunctionalProperty(canFly)". But I can take those three exact axioms,
> > add it to my ontology which has a fourth axiom "Penguin subClassOf
> > hasValue(canFly, "false")" and assuming I use the same namespaces, I
> > now have a contradiction and the class "Penguin" is unsatisfiable.
> > Clearly, the context (ontology) has made all the difference for the
> > same 3 axioms.
>
> Let me paraphrase this, only to see if I understood it right: only
> axioms that belong to the same ontology may be regarded together. If I
> have ontology o with axiom a and ontology p with axiom b, I actually
> would have to create a new ontology q (which may be virtual) that
> imports o and p, thus has the axioms (a,b) and only then I can reason
> about q (and maybe find out its unsatisfiable or whatever).
>
> Well, OK, if that is the case I am happy, cause this is the way I
> understood it as well.
>
> Or did I get it wrong?

yep, thats right..after all, when we say reason over an ontology your
reasoning over all axioms together in that ontology..

> denny
>


--
Aditya Kalyanpur
PhD Candidate, Dept. of Computer Science
GRA, MINDSWAP Research Group,
University of Maryland, College Park
Web page: http://www.mindswap.org/~aditkal

-- 
Denny Vrandecic
Institute AIFB, University of Karlsruhe (TH)
http://www.aifb.uni-karlsruhe.de/WBS/
Blog: http://semantic.nodix.net