John (01)
I'm not sure that I agree with you (though whether I do may
depend on how we ground terms like "experience"). (02)
The STEP Product Life Cycle Support standard was developed to
fill a gap between a large number of logistics products, none of which
cover the whole scope of PLCS. One of its starting point was the
existing STEP standards, particularly for product structure, which
have
been in use for some ten years in industrial strength applications. It
also pulled in a lot of other standards, but still required a high
level
of innovation. (03)
PLCS has worked because it combined two groups - some 40 domain
experts and a dozen data modellers (who also had years of experience
in
the aerospace/ship/vehicle industries). The modellers' job has been
not
only to represent the user requirements, but also to keep the domain
experts honest, not biasing the model too much to their own company's
way of working. Conversely, the modellers had to be able to explain
back
to the domain experts what the models actually said, and how they
could
be used to cover what the experts want to say. (04)
The spine of STEP's model of a product is the six element
sequence of product, version, view, property, property-representation
and presentation. These form an ontological vocabulary used to apply
STEP to parts, interfaces, documents, slots etc. Although STEP uses
an
Entity-Relation modelling language, models are written by combining
*modules*, rather than down at the entity level (this also allows
relationship subtyping as well as entity subtyping). That is, the
vocabulary relates directly to the things modelled. Using a simpler
model than STEP would not have worked, since then the modellers would
have to spend a great deal of ingenuity "interpreting" the model to
make
it work - most of the interpretation of the PLCS model is about
finding
honest routes through it. The biggest area of difficulty is
maintaining
consistency with the STEP standards already in use, where the concepts
turn out to been modelled in a form that is too specific to the
original
applications. (05)
It should be possible to translate PLCS directly into some
ontology language, such as RDF-S. As a starting point for an ontology,
it would make sense, since the cost of formalising domain knowledge is
by far the major cost in any model. However, the problem is not simply
a
translation of the model to an ontology, but also of the ontology back
to language which is understood by our non-geek domain experts. (06)
That is to say, I think the problem of creating standards is not
about being pro-active or reactive, or about testing what has been
produced, or starting with existing technologies. It is about two way
communication between modellers and domain experts. I have found it
hard
enough to work out what the ontology community is banging on about,
and
almost as hard explain it to my more sophisticated users. Creating
ontologies based on existing standards sounds like a good idea - a
chance to clear out the Augean stables. However, the real challenge
will
be getting the ontology community to understand the experts in the
field, and make sense in return. (07)
Also, since I've been writing this, Pat Cassidy and Ed's postings have
arrived. Pat - based on PLCS, I'd say your costs and timescales are
too
low. And, if it fills a big enough hole (which PLCS does) marketing is
less time critical. Ed - getting the CAD implementers to agree a
standard is not the problem - they agree lots of them; its getting
them
to implement them that is the hard problem, since if you can get
information out of their system, they see it as a threat to their
market
share. In practice, a lot of work on the CAD standards was and is done
users and other-than-CAD-system vendors. (08)
Sean Barker
Bristol, UK (09)
> -----Original Message-----
> From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx
> [mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx] On Behalf Of
> John F. Sowa
> Sent: 06 January 2009 14:16
> To: [ontolog-forum]
> Cc: janez.potocnik@xxxxxxxxxxxx
> Subject: Re: [ontolog-forum] Next steps in using ontologies
> as standards
>
>
> *** WARNING ***
>
> This mail has originated outside your organization, either
> from an external partner or the Global Internet.
> Keep this in mind if you answer this message.
>
> Azamat,
>
> I have some concerns about such pronouncements, which sound
> good on the surface:
>
> > I hold with Ravi that it is a great undertaking. Though to
> > become such, the initiative needs deliberate planning...
> >
> > the Forum has time to debate and decide on a principal matter:
> > which general world model is most fitting to science,
> arts, > technology, commerce and industry, to conclude if
> "Standard > Ontology: a single malt or blended".
>
> The standards that have proved to be the most valuable in
> practice have been based on successful technologies that many
> independent groups have adopted, used, developed, and
> extended on major applications. In most cases, those
> standards started with a successful implementation (e.g., SQL
> or HTML), polished up the rough edges, made it more
> systematic, and added new features.
>
> About 20 years ago, some people working on standards got the
> idea that it would be good for the standards organizations to
> take a "proactive" stance in developing and promoting
> cleaner, more elegant systems that take advantage of the
> latest theories and practices. But the results have been
> decidedly "mixed".
>
> I once thought that "proactive standards" seemed promising,
> but after observing many attempts, I have very serious doubts.
> Among the problems with proactive standards is that they are
> inevitably designed by committees. The basic strength *and*
> weakness of a committee is the diversity of people with
> different backgrounds, views, and requirements. That gives
> them great strength in *evaluating* proposals from many
> different points of view. But it also means that committees
> inevitably have "too many cooks" who "spoil the broth" when
> they try to do the design.
>
> I don't believe that any proposed system should be adopted as
> a standard until *after* there has been a considerable amount
> of experience in using and testing it on a wide range of
> practical applications. Instead of "deliberate planning", we
> need extensive testing, comparison, and evaluation of
> proposed alternatives on major applications.
>
> John (010)
50% off Norton Security 2009 - http://www.tiscali.co.uk/security (011)
________________________________________________ (012)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx (013)
|