[Top] [All Lists]

Re: [ontology-summit] Track on Cyc?

To: ontology-summit@xxxxxxxxxxxxxxxx
From: "John F. Sowa" <sowa@xxxxxxxxxxx>
Date: Sat, 28 Jan 2012 17:49:22 -0500
Message-id: <4F247B72.5060903@xxxxxxxxxxx>
Mike, Leo, Amanda, Nicola, and Eric,    (01)

I want to emphasize that I was *not* suggesting that we address Cyc's
ontology.  In fact, Lenat has said many times (and I agree with him)
that the upper level ontology is the *least important* part of Cyc.    (02)

Having an upper level ontology is useful, but the most important work
is in the methodologies for using an ontology and integrating it with
large systems.  I wouldn't want a talk about Cyc, but about the issues
-- good and bad -- about connecting Cyc to large IT systems.    (03)

Furthermore, my suggestion is to have no more than *one* speaker who
currently works at Cyc.  The most important speakers are ones who
formerly worked with or at Cyc and who can compare their experiences
before *and* after.  The *comparison* is vital, not a sales pitch.    (04)

> It would also be valuable to unpack the stated disconnect between
> the ontology and IT in terms of ontology quality / QA related issues.
> Dealing with the connection or otherwise between ontologies and IT
> is something I think is relevant to a lot of people.    (05)

This is the reason why AI technology in general and ontologies
in particular are still outside the mainstream of IT.  Preaching
to the choir about how wonderful ontology can be is useless if we
don't address the "disconnect" at the interfaces.    (06)

That is the major reason why I would like to hear from Bill Anderson.
He worked with Cyc and developed some very interesting software for
*translating* Cyc ontologies to software that can be directly used
with commercial IT.  That project was so promising that he and his
colleagues founded their own company, but they quickly learned that
a lot more is needed. But they did learn, and they've been in business
for over a dozen years.    (07)

> I think some of the Cyc folks are participating in this Summit.    (08)

That's fine.  But my suggestion was to emphasize the "disconnect"
between large AI systems and large commercial systems.  We don't
need a talk about Cyc, but about the problems of connecting it
to mainstream IT systems.    (09)

> One thing I am often struck by is that many of the issues that appear
> as new or emergent in semantic technologies now were already being
> addressed in thoughtful and sophisticated ways in the Cyc system then.
> Cyc still has challenges, of course, but there is much to learn from
> the many person years of experience, and multiple rounds of implementation,
> learning, and revision, done on many of these fronts.    (010)

Yes.  There has been a lot of reinventing the wheel without making
the necessary improvements.    (011)

The SemWeb failed to address the interfaces and methodologies.
They ignored mainstream IT, and now they're surprised that the
mainstream is ignoring them.    (012)

> the Ontology Summit this year is not mainly focusing on ontologies as
> such, but on how ontologies can be used to better understand and realize
> large scale systems.    (013)

Yes, that's important.  I definitely do *not* recommend a talk about the
Cyc ontology.  The most important issue is how to use any ontology to
"better understand and realize large scale systems."  Nobody has as
much experience as the people who have used Cyc on actual applications.    (014)

> Picking one ontology over others seems off base.    (015)

I totally agree with that point.  The Cyc ontology is not of interest.    (016)

> I personally have never seen Cyc successfully used in fielded govt
> programs or in industry (if so please correct that statement).    (017)

I'll let the Cyc people address that issue.    (018)

Unfortunately, most of the things that are called ontologies are just
terminologies.  In last year's talk, Dennis W. said that he stopped
talking about ontology at DoD and shifted to the word 'terminology'.    (019)

Dennis put the word 'OWL' on his charts, but he didn't seem to have
any clue about the difference between a terminology and an ontology.
If you look at so-called OWL ontologies on the WWW, the overwhelming
majority are just terminologies.  Their definitions are in English
comments, and all they specify is a hierarchy of undefined labels.    (020)

> Having a track on Cyc would contribute to shift the focus inside
> the community  so to say  while I wish we look more "outside"...    (021)

Dennis W. is an example of an outsider who goes around the world
giving talks about terminology and calling it ontology.  He's a
typical outsider, and his time slot was wasted.    (022)

The critical issue is at the interface between AI-style reasoning
and mainstream IT.  That requires people who understand both sides.
That's why I recommended Bill A.  He worked with Cyc.  But when he
was forced to earn enough money to sustain a company, he had to
adapt his technology to what the outsiders do.  He and his group
learned a lot in the process.    (023)

John    (024)

Msg Archives: http://ontolog.cim3.net/forum/ontology-summit/   
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontology-summit/  
Unsubscribe: mailto:ontology-summit-leave@xxxxxxxxxxxxxxxx
Community Files: http://ontolog.cim3.net/file/work/OntologySummit2012/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2012  
Community Portal: http://ontolog.cim3.net/wiki/     (025)
<Prev in Thread] Current Thread [Next in Thread>