GUIDE: Report on OMG activities related to Presentation Syntax for UML

		Meeting report for WOWG (long)

Preface:

At the GUIDE portion of ftf4, I reported on the results of an Ontology
session that I had held the week before as part of the Orlando
Technical Committee meeting of the Object Management Group (OMG).  This
session had been motivated by discussions of formalizing (by which I
mean - producing a normative specification for) a mapping between
OWL/DAML+OIL and UML as an OMG "technology".  Because of the close
relationship of this work to OWL and in particular to the proposed UML
Presentation Syntax for OWL, Jim asked me to summarize the session in
an email message to WOWG.  The following is that summary.

Background:

As I mentioned in my review of Guus's strawman for a UML Presentation
Syntax for OWL (see
http://lists.w3.org/Archives/Public/www-webont-wg/2002May/0164.html),
interest arose some time ago at OMG in KR technologies in general and
the use of UML for ontology definition in particular.  This led to
work exploring mappings between DAML+OIL and UML, documenting tools
using such a mapping, and exploring changes that could be incorporated
into the next major revision (2.0) of UML and its underlayment the
Meta Object Facility (MOF).  This work resulted in the two documents
referenced in the previous email.

OMG Ontology working group meeting:

Discussing the UML Presentation Syntax for OWL with those who had been
involved in this work at OMG, rekindled interest in adapting OMG
modelware technologies for ontology development.  To kick off work in
this area, I held a session on Monday of the Orlando meeting week
after the normal sessions had ended for the day.  Thirty six people
attended this hastily arranged session, representing a cross section
of OMG attendees including: 
- those developing/implementing modelware such as MOF and the XML
Metadata Interchange specification (XMI);
- those involved in Web service standards, domain standards (such as
manufacturing application interfaces), and Business Rules standards;
- and finally those who had previously enhanced/created tools for
ontology development with UML.

A number of alternative actions for moving forward were considered.
These included:
1) define a UML (1.x or 2.x ?) Profile for OWL/DAML+OIL,
2) define a MOF (1.x or 2.x ?) Metamodel for OWL,
3) define a UML Profile or Metamodel for Frame Based Ontology 
  with mappings to languages such as OWL,
4) formalize UML,
5) define a MOF Metamodel for RDF.

A UML profile exploits a capability in UML to extend its metamodel by
creating new specializations of base constructs to support modeling for
a particular domain (e.g. network management).  This essentially
supports the definition of a new lexicon of modeling constructs
adapted from and in many ways similar to UML's base constructs.

A MOF metamodel defines new constructs corresponding to elements of a
modeling language or schema through assemblies of MOF constructs.
This is more flexible then the UML Profiling mechanism, but it doesn't
easily adapt a UML tool for some particular modeling usage.  It does,
however; lead to a clean XMI encoding.  Therefore what is typically
done is to develop a UML profile and a MOF metamodel together.

Note that the objective motivating all of the alternatives above is to
standardize the way that UML based modeling tools handle ontology
development and mapping to web ontology languages. In addition most of
these alternatives lead to the ability to export ontology models in
XMI form which could be used to interchange ontologies among tools or
to feed automated generation of OWL.

While the major revision currently underway for UML and MOF is an
opportunity for positive change in these specifications, it also
creates a quandry when making plans to develop mappings: One can
create a profile and/or metamodel against 1.x modelware to support
tools currently available OR aim for 2.x, anticipating a cleaner
mapping between ontology languages and UML/MOF.  However, 2.0 has not
yet reached final revision (see example RFP schedule below for what
that means), and will not for some months yet.  Furthermore, 2.0 may
somewhat change the meaning and form of future UML Profiles,
Metamodels, and even the relationship between UML and MOF.

The group decided to pursue the following:

A) Create a website accessible via ontology.omg.org which gathers
together material and references related to OMG technologies and
ontology development (Jeff Smith);

B) Agree on a definition for "ontology" (i.e. choose one ;);

C) Find or create a test ontology which demonstrates/excercises the
features of an ontology language such as DAML+OIL or OWL;

D) Draft an RFP for UML 1.x Profile for DAML+OIL and OWL 
(Lewis Hart - AT&T, Jeff Smith - Mercury Computer Systems, Manfred 
Koethe - 88solutions, Evan Wallace - NIST);

E) Draft an RFP for a MOF 2 metamodel for Ontology Definition
(Barbara Price - IBM, Pete Rivett - Adaptive, Lewis Hart - AT&T, Jim
Odell - Intellicorp, Manfred Koethe - 88Solutions);

Note: At OMG, a Request For Proposal (RFP) starts a technology
development process which leads to an adopted technology specification
and then a finalized specification.

Group schedule: 

A and D and candidates for B will all be ready for review prior to
Helsinki OMG meeting at end of September.  At the Helsinki meeting, B
should be decided and sufficient discussion of D should occur such
that the RFP could be issued at Washington meeting in November.  E may
begin at any time, but will not be issued until UML 2.0 and MOF 2.0
development stabilizes.

Given the above, an optimistic schedule for the UML Profile for
DAML/OWL could be as follows (note that this schedule is just a
strawman and does not reflect a consensus of any group at OMG):

		    		       Issue RFP    Nov 22,	 2002

  Deadline for Letters of Intent (LOI) to submit    Jan 22,	 2003
     
		     Initial submission deadline    March 3,	 2003

                Initial submission presentations    March 25-26, 2003

		     Revised submission deadline    May 12,	 2003

	        Revised submission presentations    June 3-4,	 2003

     Architecture Board (AB) and Task Force (TF)    September	 2003
     recommendation

     Technical Committee (TC) Vote to  recommend    December	 2003
     adoption completes

     Board of Directors (BoD) Adopts  technology    March	 2004


The milestones in this schedule which relate most strongly to WebOnt
and the lifecycle for OWL 1.0 development are: 

  RFP issuance - an official request to the OMG membership (but made 
  publicly available) for specifications which meet the requirements 
  in the RFP.

  Initial submission deadline - the time by which first draft proposal
  specifications must be submitted to OMG.  This should be scheduled
  so that OWL 1 will have been officially released with sufficient
  time for the submissions to be consistent with the specification.

  Revised submission (aka final revision) deadline - the time by which
  technical work on a specification must be completed and the
  resulting specification submitted to OMG.  Only minor
  changes/corrections can occur subsequent to this deadline, although
  the deadline itself can move under certain circumstances.

  BoD adopts technology - this is the point at which a specification
  becomes OMG Adopted Technology.  This last step is largely a
  formality, although evidence of intent to implement must be 
  available for a technology to make this transition.
  

That is where we are at the moment.  Future emails will update this
information with less background information and pointers into an
operational webpage for details.

-Evan


Evan K. Wallace
Manufacturing Systems Integration Division
NIST
ewallace@nist.gov

Received on Monday, 5 August 2002 14:02:08 UTC