SUO: RE: OMG, UML and Ontologies
I would tend to agree that UML graphical representation adds at least a
valuable dimension to the development work in ontology. UML software has
proven to be a valuable tool in developing software and even organizational
architectures.
However, does it have all of the 'bells and whistles' to qualify to be a
serious tool in modeling ontologies?
And
Would UML become a precursor for OML (Ontology Modeling language) tool?
Edward Dawidowicz
US Army, CECOM, RDEC
Command and Control Directorate
Tel: (732) 427-4122 DSN 987-4122
Fax (732) 427-3440
E-Mail edward.dawidowicz@mail1.monmouth.army.mil
-----Original Message-----
From: Francis G. McCabe [mailto:fgm@fla.fujitsu.com]
Sent: Wednesday, February 28, 2001 1:41 PM
To: standard-upper-ontology@ieee.org
Subject: SUO: OMG, UML and Ontologies
At the risk of opening up some wounds, I would like to bring people's
attention
to some recent developments at OMG and possible implications for the SUO.
OMG has traditionally been a `plumbing' oriented organization. However,
things
are changing.
1. UML has had a significantly greater take up in industry than CORBA; this
is
now being reflected in an increased emphasis on abstract modelling over hard
plumbing. For those that dont know, UML is a graphical language that has
been
effectively used for a variety of tasks. It does have a formal semantics,
which
is ultimately based on OCL which is a 1st order constraint language, not a
million miles away from FOL.
2. OMG has always had a very rich domain orientation. Fully half of the
activities are in areas such as medical, space science, real-time, business
etc.
etc.
3. The `O' word is cropping up more and more in OMG circles.
4. Earlier this week there was an Ontology workshop held as part of the
Agents
SIG. (Adam Pease gave an effective talk about the SUO)
Several things stood out here: the OMG needs Ontologies and is beginning to
know
it; and there is a rich source of raw material for developing Ontologies.
For
the OMG community, and my guess is for the business community at large also,
using UML as the primary vehicle for generating Ontologies would make a lot
more
sense than using KIF, or even an XML version of KIF.
This is where there may be some pain: if the SUO community were to adopt
UML/OCL
(with possible extensions) as a preferred Ontology representation language,
then
I believe that the `take up' of the SUO would be greater and faster.
Incidentally, UML has rich support for things like packages as well as
support
for representing dynamic processes; neither of which are currently well
handled
by KIF (in whatever flavour).
The bottom line is this: is the SUO effort a primarily academic exercise, or
is
there willingness to use the tools of Industry to reach Industry?