Re: (Round 2) Proposed Extensions to OWL

Richard H. McCullough wrote:
> I think you're jumping to a wrong conclusion here.

I don't. :)

> Seems to me if cardinality of length is 1,
> an OWL inference would conclude that the input contains a contradiction,
> and one of the two lengths is wrong.

Remember that in RDF/OWL, two different URIs can refer to the same 
individual (resource).

That's pretty fundamental; I may want to create a name for a concept, 
and you may independently want to do so, and without central control we 
end up having two different names (URIs) for the same thing.

Therefore, if you have two different URIs or bnodes both of which are 
"length" values for the same resource, you cannot assume that the two 
represent different concepts. (Especially with two bnodes-- as in the 
example-- that would really not make sense!)

 From "length" having cardinality 1, we can conclude that the two 
resources would be the same-- i.e., A owl:sameIndividualAs B. Now if we 
could *also* somehow conclude that A owl:differentFrom B, then we could 
conclude that there's a contradiction; but from the above alone, we can't.

For more discussion, see:

   http://www.w3.org/TR/2003/WD-owl-guide-20030210/#owl_sameIndividualAs

This even includes an example very similar to ours (where it's concluded 
that <#Bancroft> and <#Beringer> are the same).

- Benja

Received on Thursday, 3 July 2003 19:08:04 UTC