W3C home > Mailing lists > Public > public-swbp-wg@w3.org > January 2005

Re: OMG Ontology Metamodel Definition Review

From: Phil Tetlow <philip.tetlow@uk.ibm.com>
Date: Thu, 27 Jan 2005 11:45:42 -0500
To: Jeremy Carroll <jjc@hplb.hpl.hp.com>
Cc: public-swbp-wg@w3.org, ewallace@cme.nist.gov, Grady Booch <gbooch@us.ibm.com>, Cliff.jones@newcastle.ac.uk
Message-ID: <OF857BE86C.053D9029-ON80256F96.004BADD4-85256F96.005AC04B@uk.ibm.com>


Purely for the record. I find your colleague's points valid and
interesting, but I genuinely don't think that they relate to the intent
behind my review.

The 'U' in the name 'UML' does not relate to low level syntactical
implementation - it never could. It,in fact, historically relates to the
unification of a number other 'graphical' languages, similar to the
scenario that the ODM paints.  Implementation languages such as EMOF, MOF
and XMI where created long after the concepts behind the UML were born and
the language was christened. Indeed the UML is, actually, an abstract
representation, developed to be deliberately technology and implementation
independent. The ODM doc, does however strongly refer to specific UML
implementations such as that found in the Rational suite of products and I
thank you for pointing out that fact. I, nevertheless, used the term
'unification' to convey the original architectural and design intent behind
the language's name and, IMHO, its use to characterise Superstructure
aspects is indeed correct. Unification of implementation or translational
syntax  such as MOF or XMI is another matter as the ODM document accurately
outlines in section 8.4.1 and further acknowledges in section 8.4.3. I
hence deliberately did not address my review comments at such low levels of
abstraction. I also used the specific words 'a number of acknowledged
weaknesses within the language (UML)' and did not go on to say that this
statement was incorrect as you suggest. I actually agree with this
statement (and your point below), and merely commented that the spirit of
the UML was unfairly represented. My comments were, hence, more directed at
the content in section 9 of the ODM document (ODM Overview) from an
architectural standpoint.

So, I accept your colleague's comment that  'sections 8 and 9 of the
document are more or less in line with that level of unification (i.e. at
meta modelling level using common core MOF/XMI constructs)' - ie mappings,
but do not see any supporting evidence in your mail to suggest why an
equalised collection of metamodels within the ODM should be more
architecturally favourable than one based on a hub and spoke pattern. I
hence again look forward to any relevant thoughts you might have on this
matter. If you wish to oppose my review then that's fine. My points were
made simply to avoid the potential isolation of individual metamodels in
the future - surely a valid concern? To reject such a position without
appropriate forethought would be a shame.


Phil Tetlow
Senior Consultant
IBM Business Consulting Services
Mobile. (+44) 7740 923328

             Jeremy Carroll                                                
             com>                                                       To 
                                       Phil Tetlow/UK/IBM@IBMGB            
             27/01/2005 08:17                                           cc 
                                       ewallace@cme.nist.gov, Grady Booch  
                                       Re: OMG Ontology Metamodel          
                                       Definition Review                   

After initial feedback from colleagues I am intending to oppose the WG
endorsing Phil's review.

I agree with the positive statements, but the paragraph suggesting a
hub-and-spoke model does not receive my support.

A colleague articulated his concerns thus:

I agree with your position on opposing Phil's position in the specific
areas that you articulate in your message. The "U" in UML for the
purposes of unification is not in the Superstructure aspects of UML but
in the UML Core and that too specifically in the MOF Core (EMOF as it
appears in Eclipse.org for example). I think the proposal in sections 8
and 9 of the document are more or less in line with that level of
unification (i.e. at meta modeling level using common core MOF/XMI

Received on Thursday, 27 January 2005 16:42:08 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:31:04 UTC