John,
I agree with your three points, but have some additional comment: (01)
[JS] > 2. We do have billions, if not trillions, of dollars of software
> that supports the world's economy, and we see legacy software
> that is 40 years old or older still in daily use -- none of
> which is or will be converted to any "integrating" ontology.
>
The programs won't but the databases can be. Cyc and Ontology Works claim
to be able to integrate databases with an ontology. I am involved in a
project exploring the technology to do that. Each of these methods uses a
single upper ontology for the integration process.
But to my knowledge there is no public demonstration of that integration
technique on a database of significant size. Not because the tactic has
flaws, but because (1) Cyc and OWI methods are proprietary - so are client
databases; and (2) the process at this stage, without well-developed
utilities to make it easy, is very expensive and simply hasn't yet been done
for a publicly available set of databases. So of course no existing
ontology has caught on because there are no impressive public demonstrations
of its capability. Building such applications is costly; some projects are
underway (e.g. CALO) that may provide more evidence of the utility of an
ontology. Right now we have mostly projections of utility from very small
and limited demos, which have not convinced the big money people to open up
their wallets for more.
Perhaps if we could find some *public* relational databases that could
benefit from federated search it will be possible to use some (almost any)
public foundation ontology to provide a good demonstration of utility. Do
you know of any public RDB's that we might investigate as candidates for
such a demo? (02)
> 3. But we do see a history of thousands of years of civilization,
> in which people with very divergent views manage to cooperate
> on task-specific terminology to accomplish major goals.
>
> As I have said many times, I have not abandoned the search for
> suitable foundations for ontology. But given the above, the major
> question is how to get from where we are to where we want to be.
> My suggestion is that point #3 is a good place to start.
>
The point of the "Conceptual Defining Vocabulary" is precisely to enable
people to cooperate by sharing data in a form that will support accurate
automated inferencing. No one needs to modify the local application, just
to map the data to the common foundation ontology. If one or a very small
number of common foundation ontologies became widely used, it would make the
process of mapping much faster and cheaper by allowing database developers
to use the most widely used pre-existing ontology to provide the top-level
data model, and then add the local details needed. As I understand it, this
is the same point that Matthew was making.
To (try at least) to develop the foundation ontology as a CDV provides
what I believe to be the best *tactic* to achieve the widest possible use
for a foundation ontology, by including the largest number of views in the
CDV, which one may view as a single ontology or as a fully consistent
lattice of ontologies. It doesn't have to be *universally* accepted to save
lots of DB developers a lot of money and have great economic impact.
How do we get from here to there? It starts with some agreement among a
group of knowledge engineers that a common foundation ontology is worth the
cost of substantial effort, and then to get enough money for them to make
that effort. The SUO project didn't have enough funding for substantial
participation by anyone outside of Teknowledge. The successful project will
have to support a lot more participants. (03)
One additional point about retroactive fitting of existing applications.
I agree that it is important, but it is likely to be a short phase of
diminishing importance, once a common foundation ontology is shown by
several impressive applications to provide important data-sharing
capabilities not available otherwise. In 40 years, there may well be
applications still used that are in existence now, but the vast majority
will be developed between now and then. We are compelled to focus on
accommodating legacy systems for a while, but we need to remember that the
more important goal is to accommodate the vastly larger number of systems
that will be developed in the next few decades, which can have enhanced
capability from using an ontology. I think a foundation ontology viewed as
a CDV can do both very well. (04)
Pat (05)
Patrick Cassidy
MICRA, Inc.
908-561-3416
cell: 908-565-4053
cassidy@xxxxxxxxx (06)
> -----Original Message-----
> From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx [mailto:ontolog-forum-
> bounces@xxxxxxxxxxxxxxxx] On Behalf Of John F. Sowa
> Sent: Sunday, March 16, 2008 3:15 PM
> To: [ontolog-forum]
> Subject: Re: [ontolog-forum] Ontology similarity and accurate
> communication
>
> Dear Matthew and Pat,
>
> An integrating ontology is a goal that many of us have been pursuing
> for over 20 years:
>
> PC> You don't think you need to worry about the upper ontology for
> > integration?
> >
> > Then everyone will have their own (potentially incompatible)
> meanings
> > for person, place, organization, time, object, substance, artifact
> > (manufactured object), event (happening), group, path, properties
> > (attributes/relations), organism, reasoning, emotion, gravity,
> > agents, goals, obligations, exchange, money, liquid flow....
>
> I worry about it constantly. The only correction I would make
> to that point is the replacement of "will have" with "does have".
> If God handed down Avril's "perfect" ontology on tablets of stone,
> it would do as much to stop the proliferation of ontologies as
> His earlier tablets did to stop vice.
>
> Following is our current state:
>
> 1. We don't have a perfect ontology, we don't know whether any
> of our very many proposed ontologies are even adequate, and we
> don't see anybody who has a current proposal who is willing to
> drop it in favor of any other.
>
> 2. We do have billions, if not trillions, of dollars of software
> that supports the world's economy, and we see legacy software
> that is 40 years old or older still in daily use -- none of
> which is or will be converted to any "integrating" ontology.
>
> 3. But we do see a history of thousands of years of civilization,
> in which people with very divergent views manage to cooperate
> on task-specific terminology to accomplish major goals.
>
> As I have said many times, I have not abandoned the search for
> suitable foundations for ontology. But given the above, the major
> question is how to get from where we are to where we want to be.
> My suggestion is that point #3 is a good place to start.
>
> MW> When you develop your interfaces with a single ontology, then
> > for each system you have only one ontology to integrate with,
> > rather than one per system that you interface to. Also it is the
> > same ontology you integrate to for all systems, and that is
> > another benefit.
>
> Yes, that has been the goal for at least 20 years or more.
>
> JFS>> But the types at the upper levels may become very abstract
> >> and sophisticated.
>
> MW> I agree that there are some abstract concepts. But they are
> > relatively few (tens rather than hundreds in an ontology of
> > tens of thousands).
>
> Yes, but those very few concepts may have an enormous number of
> implications. Otherwise, they wouldn't be in the upper level.
>
> MW> Of course those that are dealing with things at a lower level
> > of abstraction do not even need to be aware of these. Most people
> > are happy when they can see that the next immediate supertype is
> > appropriate.
>
> I agree. And that is why task-specific interoperability has
> worked so successfully for people (and, I would claim, for most
> computer agents as well).
>
> JFS>> For example, consider some of the sophisticated notions in
> >> relativity and quantum mechanics, which must be accommodated
> >> by any truly global ontology...
>
> MW> I would not see this as upper ontology stuff at all. This
> > is the stuff of detail at a very low level of reality.
>
> I agree that we do *not* want to put relativity and QM into
> the upper levels. But by the term 'accommodate' I meant that
> the upper levels must be neutral with respect to any or all
> physical theories, including Newtonian, QM, string theory, etc.
> Whitehead's system achieves that, but without mentioning QM or
> relativity in any way.
>
> MW> So what you are saying is that when you are integrating
> > some external ontology with yours, you do not need to worry
> > about its upper ontology. That I would agree with.
>
> Great! That is the primary point I wanted to emphasize, and
> I'm happy that we agree.
>
> John
>
>
> _________________________________________________________________
> Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
> Subscribe/Config: 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 Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx
> (07)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Subscribe/Config: 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 Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx (08)
|