Re: OWL reasoning in rules

> Hmmm, I would like to see a small ontology which is necessarily infinite.

As the movie 300 has made the Greek phalanx popular again, here is a small
ontology that contains a class that has only infinite models:

Guard subclassOf ( (some shields.Guard) and ( <2 inverse(shields)) )
FirstGuard subclassOf ( Guard and ( <1 inverse(shields) )

"Every Guard shields some Guard, and is in turn shielded by at most 1
Guard."

"A FirstGuard is a Guard, and is not shielded by another."

FirstGuard has no finite model, because it generates an
infinite number of Guards (every Guard needs to shield someone, and it
can't be the FirstGuard).

The above example is found in this paper (Page 4, Sec 3):

http://citeseer.ist.psu.edu/calvanese96finite.html

other refs on topics of computing finite models in DL:

http://citeseer.ist.psu.edu/710125.html
http://www.inf.unibz.it/~franconi/dl/course/articles/reasoning-survey.ps.gz

I am sure others would chime in with newer papers, but these were helpful
to me.

Cheers,

Dave
----


> I've just being looking with google, and found my own
> http://www.w3.org/TR/owl-test/dl-900-arith#description-logic-908
>
> which I believe hinges on
>     2*3*n = 5*n & n>0
>      implies n >= aleph0,
> but I am still trying to understand it.
>
> thanks for a pointer
>
> Jeremy
>
>
>
> --
> Hewlett-Packard Limited
> registered Office: Cain Road, Bracknell, Berks RG12 1HN
> Registered No: 690597 England
>
>

Received on Tuesday, 29 May 2007 17:56:15 UTC