RE: UPDATE: why RDF syntax is not suitable for OWL

I will display my ignorance with a couple of comments relevant to this
topic.

1. If we are to take seriously the goal of providing a formal semantics for
OWL that completely includes RDF(S), we need to recognize that it is going
to be a difficult task.  The formal semantics of RDF(S) is incomplete
(containers, reification).

2. One of the things I have been puzzled by has been the apparent enthusiasm
for triples as a syntax.  Its like saying "I really prefer reading binary".
I thought Peter's examples were great.  They focus on the difficultly of
providing a sensible semantic interpretation for arbitrary sets of syntactic
triples.

Of course I am exaggerating the enthusiasm for triples. If we look at the
'RDF Model and Syntax Specification' and the DAML+OIL Reference Description,
the presentation of examples depends on XML, not triples.

We could certainly translate a clean OWL syntax to triples (just as the RDF
Model and Syntax Specification document does for RDF).  Going the other way,
in all possible cases, seems to me problematic.  

It does not seem to be the case that the official syntax for RDF is defined
as triples.  When I look at 'RDF Model and Syntax Specification' all I see
are the XML serializations.  Then there is a description of their
translation into triples.  But no syntax restricting the set of triples that
are permitted.

3. It wasn't immediately clear to me whether the bag example Peter presented
is ill-formed RDF (I've spent more time looking at the Model Theory than the
Specification). Note that the XML syntax used in
http://www.w3.org/TR/1999/REC-rdf-syntax-19990222/, where bag elements are
referenced using 'rdf:li', eliminates this problem.  Which illustrates why a
syntax for containers that includes a notion of scope is much simpler than
one that does not.  

Reading the RDF Model and Syntax Specification we find:

  Note: The definitions of Bag and Sequence explicitly permit duplicate 
  values. RDF does not define a core concept of Set, which would be a 
  Bag with no duplicates, because the RDF core does not mandate an 
  enforcement mechanism in the event of violations of such constraints. 

The document also states in the description of containers that 

  there may be at most one triple whose predicate is any given element 
  of Ord and the elements of Ord must be used in sequence starting with
RDF:_1

I don't understand what the consequences of violating this requirement would
be.  We have an open world.  The fact that we've got RDF:_1 and RDF:_3
explicitly asserted, but not RDF:_2, could just be an accident of the
current state.

- Mike

Michael K. Smith
EDS Austin Innovation Lab
98 San Jacinto, Suite 500
Austin, TX 78701
Work: 512 404-6683


-----Original Message-----
From: Dan Connolly [mailto:connolly@w3.org]
Sent: Wednesday, February 13, 2002 10:44 PM
To: Peter F. "Patel-Schneider
Cc: www-webont-wg@w3.org
Subject: Re: UPDATE: why RDF syntax is not suitable for OWL

On Wed, 2002-02-13 at 17:56, Peter F. Patel-Schneider wrote:
[...]
> 		Why Triples are Terrible for Syntax
[...the suspence builds...]
> None of the above problems are really solvable using RDF triples.  A much
> better syntax carrier would allow syntax to be represented as tuples or,
> even better, as trees.

Bummer. I was all set to say "yes, that's a much better alternative!"

But I don't see an alternative.

Oh well... there's always tomorrow...

-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/

Received on Thursday, 14 February 2002 15:13:14 UTC