W3C home > Mailing lists > Public > public-owl-wg@w3.org > November 2007

Re: UFDTF Metamodeling Document

From: Peter F. Patel-Schneider <pfps@research.bell-labs.com>
Date: Thu, 29 Nov 2007 10:22:20 -0500 (EST)
Message-Id: <20071129.102220.112919105.pfps@research.bell-labs.com>
To: conrad.bock@nist.gov
Cc: public-owl-wg@w3.org

From: "Conrad Bock" <conrad.bock@nist.gov>
Subject: RE: UFDTF Metamodeling Document
Date: Thu, 29 Nov 2007 10:06:06 -0500

> 
> Peter,
> 
>  >  I'm assuming that you mean the OWL Metamodel in:
>  >  
>  >  Ontology Definition Metamodel
>  >  OMG Adopted Specification
>  >  OMG Document Number: ptc/2007-09-09
> 
> Yes, that's the latest, see Chapters 10 and 11 at
> http://doc.omg/ptc/2007-09-09.
> 
>  >  I just took a look at that document and, frankly, it is so full of
>  >  inaccuracies that there is no reason whatsoever to align with it.
> 
> Could you give some examples from the OWL and RDF metamodels?  

For RDF:

p35: "URIReferenceNode, BlankNode, and RDFSLiteral form a complete
      covering of RDFSResource and are pairwise disjoint."  Among other
      problems, there are unnamed resources in RDF.

p35: RDFGraphs are unordered sets of triples, not ordered collections of
     statements.

p35: RDFGraphs don't have names.

p35: The treatment of reification implies that there is only one triple
     with a given subject, predicate, and object over all RDF graphs,
     and that its reification status is the same in all RDF graphs.

p37: A typed literal does not have two names.

p37: Literals are not assigned a meaning by interpretations.

p38: Literals have special status, and are not replacable by URIs.

I gave up listing problems after page 36.  The RDF section has a number
of fundamental misconceptions, including a severe case of mixing syntax
and semantics.

For OWL:

I started looking more closely at the start of the OWL section:

p63: "Not all RDF graphs are valid OWL graphs, however".  WRONG!

p67: Naming non-atomic descriptions is not allowed in OWL DL.

p69: "OWLAllDifferent ... links an individual to an idiomatic class."
     WRONG! 

p70: "... individuals in OWL have a 'default type' (i.e., owl:Thing)".
     WRONG!

p70: "Multiple inheritance is also supported." WRONG.

But I soon gave up.  

I have no idea what possible use the RDF and OWL metamodels could be put
to.

> If you'd
> like to file issues on them, see
> http://www.omg.org/technology/agreement.htm.
> They must be filed by February 22, 2008 to require the task force to
> address them before issuing the finalized specification.  You can also
> become involved in the task force, and since Alcatel-Lucent is an OMG
> member, you can have voting privileges with a simple motion made at the
> next meeting (December 14th, I can arrange for the motion if you like).

Well, I'm not sure that a comment along the lines of:

	This has too many inaccuracies to be accepted.

would be appreciated. 

> Just as background, several of the authors of the OWL and RDF/S
> metamodels in ODM are members of the OWL working group and were members
> of WebOnt (you might recall being invited a number of times to review
> them).  The OWL metamodel in particular underwent significant peer
> review, within OMG as well as by the broader user community (academic
> and industrial), and is successfully implemented in a number of open
> source and commercial applications.  I'm sure they would welcome
> specific suggestions for improvement.

Well, again, I'm not sure that a suggestion like "Start over" would be
welcome, but that's the best advice I can give.

> Conrad

peter
Received on Thursday, 29 November 2007 15:48:59 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 8 January 2008 14:13:27 GMT