RE: DAML+OIL (March 2001) released: a correction

Some comments on how I think RDF could be changed to make it better.  This
is my personal opinion, so far.

From: Peter Crowther <Peter.Crowther@melandra.com>
Subject: RE: DAML+OIL (March 2001) released: a correction
Date: Mon, 2 Apr 2001 08:37:42 +0100 

> > From: Jim Hendler [mailto:jhendler@darpa.mil]
> > DAML+OIL did not have authority to change anything in RDF or to 
> > otherwise impact the RDF spec except by example.
> 
> Jim has (as usual) got right to the heart of the problem.  Feedback between
> designers of different architecture layers is essential if you want to
> create a coherent architecture.  So far, the sequence has been that RDF has
> taken XML as a given, and DAML has taken RDF as a given --- feed-forward
> with no feedback.  In fact, both XML and RDF are subject to periodic
> revision, but there hadn't been the time and experience of using them at the
> time the initial versions were formalised (obviously!).  That experience is
> now starting to accrue.
> 
> I think two interesting questions can be posed here:
> 
> 1) How could RDF be changed/augmented/better documented to make it a firmer
> base on which to build DAML+OIL?

1/ Provide a firm semantic foundation.

2/ Remove the parts that cannot be given a firm semantic foundation.

3/ Remove the insistence that everything can be represented as
   RDF-interpreted triples.

4/ Include facilities for datatypes.

5/ Remove the non-monotonic portions of RDFS (e.g., domain) and the
   non-compositional portions of RDFS where possible (e.g., range and
   acyclicity). 

> 2) How could XML be changed/augmented/better documented to make it a firmer
> base on which to build RDF?

1/ Extend the use of namespaces so that they can be used throughout RDF
   wherever a resource name appears.

2/ Make a lightweight version of XML Schema datatypes for those who are not
   interested in, say, regular expression-restricted datatypes.

> ... and, I guess, (3) is there any chance of these changes happening?  (3),
> at least, would be helped by keeping the dialogue constructive; but (1) and
> (2) require constructive criticism of those existing standards in the light
> of practical experience.  For example, could the appropriate parts of the
> Fikes & McGuinness paper be used within the next revision of RDF to provide
> that clear semantics that many within DAML+OIL wish to see?

I am hopeful that at least some good things will happen.  There is at least
some concern that under-specified standards are not a good thing.

I think that the KIF axiomatization is an excellent start for an
axiomatization of RDF(S).  There may be other methods for providing
semantics for RDF(S).  I am much more partial to model-theoretic
semantics.  It would be useful to have several potential specifications for
RDF and RDF(S).  (No, I am not volunteering to provide one.)

> 		- Peter
> 
> [For "RDF" read "RDF and/or RDFS"; I tried to construct  similar statement
> for XML but was defeated by the plethora of extensions out there.  X*?]


Peter F. Patel-Schneider
Bell Labs Research

Received on Monday, 2 April 2001 13:52:08 UTC