Re: Classes and predicates as first class objects

On August 19, R.V.Guha writes:
> 
> Ian,
> 
> Tell me if you agree with the following ... At least one of the 
> following has to occur:

Not necessarily. E.g., I could imagine more than one kind of layering,
and I could imagine partial layering.

Ian

> 
> a) RDF provides for the notion of a reserved vocabulary. This reserved 
> vocabulary will include terms like first and last, which are crucial to 
> constructing paradoxes. They will not include general predicates and 
> hence it will be possible to use predicates as arguments to other 
> predicates.
> b) OWL is not layered on RDF/S, i.e., owl does not allow for classes and 
> predicates to be arguments to predicates.  
> c) OWL's semantics are very different from that used by DAML+OIL.  
> 
> guha
> 
> 
> Ian Horrocks wrote:
> 
> >This last point is the nub of the matter. Of course the problem can be
> >resolved by treating some RDF triples as a reserved vocabulary - this
> >much is obvious. It does, however, require a modification of RDF to
> >which some are strongly opposed.
> >
> >My point is that, without such a modification, extending RDF with the
> >expressive power of OWL would result in a language so seriously broken
> >that the question as to whether classes/predicates can be treated as
> >arguments to other predicates would become an irrelevance.
> >
> >Ian
> >  
> >
>

Received on Tuesday, 20 August 2002 13:10:02 UTC