Re: Layering on the Semantic Web

From: "Jeremy Carroll" <jjc@hplb.hpl.hp.com>
Subject: Re: Layering on the Semantic Web
Date: Fri, 4 Jan 2002 12:25:45 +0100

[...]

> We should only part company with RDF if OWL has requirements that are
> [...] not met by RDF.

This is indeed the situation, assuming that OWL is going to go
significantly beyond the expressive power of RDF(S).  The only question is
how and how far to part company with RDF.

> > - RDF is not well suited as syntax carrier:
> >   - the problems that Peter has identified with "additional" tuples
> >     (which specify only syntax) ending up in the RDF model and breaking
> >     [] inference
> 
> My reading of
>   http://lists.w3.org/Archives/Public/www-webont-wg/2001Dec/0156.html
> is that Peter is proposing dropping the problematic daml:collection
> construct.

The issues here have nothing to do with daml:collection.  They have to do
with the impossibility of creating an ontological *extension* of RDFS that
uses *only* RDF syntax.

> Jos is currently arguing that RDF allows unasserted triples, and his Euler
> system shows how that can work in practice. I think both these paths address
> this issue.

Unasserted triples are not part of RDF.  All information in the RDF graph
is currently asserted in RDF.  The RDF Core WG may change this, but that
would be a cataclysmic change.

> >  - problems with scoping
> 
> There are inevitably issues with ontological scoping, since in the semantic
> web I may well want to annotate your descriptions in your ontology with my
> descriptions in my ontology. There are issues as to how to merge information
> from multiple sources. I am far from convinced though that scoping problems
> are specific to OWL.

There are several scoping issues, and RDF syntax is unable to express any
of them.  Sure they are not specific to OWL but this only serves to point
out the fact that others will have to go beyond RDF.

[...]

> I don't think this group has any greater ability in syntax than either of
> the RDF working groups (the current one or the former one). I see no reason
> that we should believe that we can do a better job in syntax than the mess
> we inherit. Do we have people in the group who are world class at language
> syntax design?

I think that it would be extremely hard to have a lesser ability in syntax
than that exhibited by the initial RDF working group.  We certainly have in
the group people who have created syntaxes that are vastly better than the
syntax for RDF.  

> 
> Summary:
>   The problems identified with RDF are not *our* problems.

They are if we have to use RDF.

> Jeremy


Summary:
   Contrary to many such claims, RDF cannot be the semantic base of the
   semantic web, at least not if the semantic web has to stick to RDF
   syntax.  


Peter F. Patel-Schneider
Bell Labs Research

Received on Friday, 11 January 2002 10:01:55 UTC