Everything EdB says in his note below is quite reasonable, and I agree with
most of it. The need for some demonstration of the utility of ontologies is
a theme that occurs frequently in discussions of ontologies. The only
thing Ed's note leaves out is the question of whether there are any economic
costs in approaching the problem by the incremental proof-of-concept in
increasingly large aggregates of systems. If one believes that the 100B/yr
estimate of costs for lack of semantic interoperability is grossly
exaggerated, or that a common ontology can only solve a microscopic fraction
of the problem, then the urgency is much diminished, and one might well want
to pursue a leisurely path with very small projects. Even if, as PatH
suggests, a common FO developed by a consortium would only solve 0.01% of
the problem, then the costs of delay in implementing this project amount to
$10 million per year, and the project would have an expected economic return
of 30% per year, well justified even on this very pessimistic guess, even if
one assumes implausibly that the common FO would not be good for *anything*
else. And then, the Feds have already spent over 200 million on CALO, and
more on other AI projects, so it is not the case that the Feds haven't
already invested a lot more than 30M in the problem. My sense of urgency,
which in my view justifies a project of 30M size, is base on the belief that
the lack of semantic interoperability is a real and very large economic
problem, and that a broadly acceptable Foundation Ontology is a tool that
would be an important part of the solution. From those premises, I have
concluded that funding not just one, but many 30M projects that could solve
the problem would be justified. (01)
Yes, I would also like to see proof-of-concept projects beyond the simple
examples of combining a few facts to generate a new inference, or the small
demos that one can cobble together in a year or so. One of the reasons I
went to work at MITRE is precisely to discover or develop examples of
ontology utility to answer such skepticism; but the projects I encountered
were either too tentative to reach the execution stage, or confidential.
The proposal for a common FO includes the need to develop example uses, as
part of the project itself. I have previously suggested that it would be
helpful to create a public demonstration of database integration via a
common ontology. But the DB and demo systems have to be publicly
inspectable, and the DB/ontology projects I know of are all confidential. I
have previously inquired on this forum if anyone knows of publicly available
existing databases that have some overlapping data and would be suitable as
a test case for at least the database integration function. Thus far I have
not heard of any candidates. Some federal DB's should be publicly
releasable, so perhaps this tactic can be tried; If anyone has a suggestion
for a specific database, I would be quite interested to hear about it. I
have seen the "movie database" used in some projects, but don't know of any
other DB that has enough overlap with movies to provide an interesting demo.
This doesn't mean that there are no DB's that could be usefully integrated -
I know of a few from my own very limited experience. But they are all
proprietary or otherwise confidential. A demo needs something public.
Perhaps a couple of made-up DBs could be used, but that might well be
unconvincing, or less convincing than a demo based on existing real-world
info. (02)
Another point that could be relevant is: that developing a *common* ontology
by a consortium process (not a proprietary one like Cyc) that has enough
functionality to support reasoning at close to the human level is something
that has never been done, and is also qualitatively different from other
kinds of standards development issues. Of course, one wants to be sure that
lessons from prior experience are taken into account. But my suggestion is
based on lessons I have gleaned from the prior experience of efforts to
develop a common FO (aka "upper ontology"). The main points that I have
concluded is that there needs to be a much wider direct participation, at a
level that can only be achieved by funding the participants; and that the
process and product needs to accommodate multiple viewpoints, with a clear
goal and timetable, and include development of utilities that make it easier
to use. SUMO was an instructive effort, but the level of funding and the
process chosen did not allow for all of the alternative views and desiderata
to be accommodated. In general, while staying aware of prior experience in
developing resources for use by many people, it is also important to stay
aware of how any new project differs from previous ones. (03)
So, ignoring the economic argument, Ed's suggestion that the hypothesis of
ontology-based interoperability be tested is eminently reasonable: (04)
> OK. Provide that functionality for two systems that currently have the
> problem. Then try to do it for two more. About 4 efforts in, you
> should begin to understand the problem well enough to describe the
> nature of the general solution.
> (05)
. . . and I would be happy to participate in such a demo. Any *specific*
suggestions, anyone? We do need publicly available systems to test. (06)
Pat (07)
Patrick Cassidy
MICRA, Inc.
908-561-3416
cell: 908-565-4053
cassidy@xxxxxxxxx (08)
> -----Original Message-----
> From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx [mailto:ontolog-forum-
> bounces@xxxxxxxxxxxxxxxx] On Behalf Of Ed Barkmeyer
> Sent: Wednesday, February 11, 2009 7:34 PM
> To: [ontolog-forum]
> Subject: Re: [ontolog-forum] standard ontology
>
> Pat Hayes wrote:
>
> >> Hardly anyone needs to be able to interoperate with everybody.
>
> Pat Cassidy wrote:
> > No, I don't either, but a lot of people want to interoperate with a
> lot of
> > other people,
>
> Well, yes and no. A lot of people (and more importantly their software)
> need to interoperate with others in the same domain (=
> company/industry/profession) and with others in joint activities that
> cross specific domains.
>
> People in the same domain develop some means of interoperability for
> their survival. Usually that involves "common semantics". In order to
> work with others in joint activities across domains, they also develop
> some means of interoperability. Standards, of one kind or another are
> a
> favorite. But inter-human negotiation has worked for far longer.
>
> The problem we are talking about is *much* smaller. It is systems that
> are using ontologies as part of the communication within a domain and
> are now required to deal with systems using different ontologies in the
> course of a joint activity. That is when we start talking about
> "semantic interoperability".
>
> Right now the number of commercial systems using ontologies as part of
> anything is small, and the likelihood of their encountering one another
> in joint activities is even smaller. We expect that to change, but
> that
> will only happen when systems using ontologies do more things or better
> things or new things than systems that don't, as Don Conklin observed.
>
> > and I think that those of us who are in a position to create
> > the technology that can make it happen should make an effort to
> provide that
> > functionality.
>
> OK. Provide that functionality for two systems that currently have the
> problem. Then try to do it for two more. About 4 efforts in, you
> should begin to understand the problem well enough to describe the
> nature of the general solution.
>
> At the moment all you have is a theory, and it is not yet clear to me
> what that theory really is. And it is even less clear that accepting
> it
> requires less than a leap of faith. The old adage applies:
> The difference between theory and practice in practice is greater
> than the difference between theory and practice in theory.
>
> So tell us about your practical experience with this approach.
> There is reported practical experience with other approaches (but I
> imagine some of them got their 30M$ to acquire it).
>
> -Ed
>
> --
> Edward J. Barkmeyer Email: edbark@xxxxxxxx
> National Institute of Standards & Technology
> Manufacturing Systems Integration Division
> 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528
> Gaithersburg, MD 20899-8263 FAX: +1 301-975-4694
>
> "The opinions expressed above do not reflect consensus of NIST,
> and have not been reviewed by any Government authority."
>
> _________________________________________________________________
> 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
> (09)
_________________________________________________________________
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 (010)
|