[Top] [All Lists]

Re: [ontolog-forum] Semantic Web shortcomings [was Re: ANN: GoodRelation

To: "'[ontolog-forum] '" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: "Patrick Cassidy" <pat@xxxxxxxxx>
Date: Wed, 20 Aug 2008 14:20:32 -0400
Message-id: <02fa01c902f1$6f017470$4d045d50$@com>
   In order for a technology as complex as a foundation ontology to get
adopted in a commercial setting, I am assuming that those who have to pay
for the implementation will have to see some conspicuously successful
applications first.  I am not aware of any commercial implementation of Cyc
that could be described as successful, though Cyc people may have examples
in mind.   Until there are publicly observable examples of applications, I
wouldn't expect to see much investment in the technology.  As for the cases
where Cyc funders tried to use it and found it wanting, I can't learn
anything from their experience unless I have access to the details of what
was tried and what deficiency of Cyc made it unusable.    (01)

   There are several other issues raised by your comments:
   (1) is any ontology-based application likely to contribute to the bottom
line of a corporation in the near future?
   (2) if so, does a foundation ontology make that specialized ontology
easier to build?  Or add anything to its performance?
   (3) If I build an ontology is there anyone else with an ontology-based
application that I need to interoperate with?    (02)

 My best guesses:
   (1) I don't know the situation in the IT departments of large
corporations, but I have not seen any ontology-based applications, so I am
guessing that thus far few of them believe that the answer to #1 is "yes".
Small demos don't score many points.
   (2) Even if an ontology does seem useful, for a specialized
ontology-based application that doesn't need to interoperate with other
applications, incorporating a full foundation ontology may be more of a
distraction than a help.  I think that it's easier to build a specialized
ontology than to learn how to use Cyc, and then use that to build the
ontology.  No need for Cyc in many cases.
   (3) If there are no applications out there, there's no one to
interoperate with.    (03)

   I can well imagine why the adoption of a foundation ontology like Cyc
would be slow, even if the ontology were perfect as a logical encoding of
basic knowledge.    (04)

   So what is a foundation ontology good for anyway?    (05)

   Two things in the **near-term**, I think:
   (1) database interoperability.  To provide a common basic schema from
which different specialized databases can derive their elements; this will
allow true federated search over all the databases mapped to the common
ontology.  A public foundation ontology will allow separated groups to
design for interoperability with other users without every contacting those
other users.
   (2) As a common standard of meaning for use in research in AI.  I
believe, as I think that you do, that multiple agents or procedures will be
needed to solve problems that have a significant semantic content.  The
foundation ontology can provide the standard of meaning that will allow
diverse components to communicate among each other and contribute
efficiently to solving the problem.  I believe that research benefitting
from interaction with other projects, will allow much more rapid advance
than the relatively isolated project we have had over the past 50 years.    (06)

   The second point is addressed by my suggestion for a project to develop a
foundation ontology by a large group of developers and users.  Once the
first phase of such a project (building an agreed foundation ontology) nears
its goal, the next phase will be to demonstrate its usefulness by showing
how it can allow multiple components to interact effectively.  I don't
expect many outside the participating group to try using such a foundation
ontology until the group itself builds applications demonstrating its
utility.  A common foundation ontology can provide the ability to reuse
research results more efficiently among diverse semantic-based applications.
But as with other communication issues, a language is of little use unless
many people use it.  That is why we need to jump-start the use of a
foundation ontology as a communications medium for AI research, by funding a
sufficiently large group that will make a commitment to use it for a while
in their work.    (07)

   But why don't commercial organizations use an ontology for DB
   (a) some do.  There are companies that will use an ontology to integrate
your databases.  Unfortunately, the details seem to be proprietary.
   (b) there is enormous inertia in IT departments that just prefer to do
things the way they already know how.  They are using common schemas for the
same purpose - which are fine for a few databases, but do not support
interoperation of multiple DBs as well as a public ontology would.  Until
some good example of such use of an ontology can be made public, it is going
to be slow adoption.  I feel certain that it is inevitable, but it is too
new and complex and has too few examples at this point.
   (c) even when one can get IT managers and developers to pay attention for
a short time, they are simply gut-skeptical, and need to see an operating
example.    (08)

   Hardly anyone wants to be first - "bleeding edge" is something to be
avoided.    (09)

   And in any case, for DB operability, the complexity of Cyc is not needed.
Ontology Works has a lot simpler basic ontology that they provide for
customers to customize.       (010)

  The difference between the Cyc and SUMO project and the one I am
suggesting is not just the **scale**, it is a totally different orientation.
The great value of a common foundation ontology is in its ability to enable
*communication* among diverse applications.  Therefore to demonstrate that
ability, it is essential that multiple *separate* groups participate in its
construction and testing.  Multiple groups will bring the diverging
perspectives that a common foundation ontology will encounter in the real
world.  What we have seen is that development of a foundation ontology by a
single group leads to a single-perspective ontology that is uncongenial to
many other users.  That is true even in the case of the SUMO, where input
from people in the IEEE-SUO group, outside of Teknowledge, was actively
solicited.   They just didn't have the time or inclination to include
alternative perspectives or viewpoints in one ontology.    (011)

  Note that in the week we have taken for discussing this thread, by that
100B/yr estimate, the country has wasted 2 billion dollars in lost
productivity.  I really think that risking 10-20 million to solve the
problem has a very high expected return on investment.  But only for the
government or some other public-spirited agent.    (012)

Pat    (013)

Patrick Cassidy
cell: 908-565-4053
cassidy@xxxxxxxxx    (014)

> -----Original Message-----
> From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx [mailto:ontolog-forum-
> bounces@xxxxxxxxxxxxxxxx] On Behalf Of John F. Sowa
> Sent: Wednesday, August 20, 2008 9:21 AM
> To: [ontolog-forum]
> Subject: Re: [ontolog-forum] Semantic Web shortcomings [was Re: ANN:
> GoodRelations - The Web Ontology for E-Commerce]
> Pat,
> That inference doesn't seem to be convincing:
> PC> One thing we learned from the Cyc project (and also from SUMO) is
>  > that an ontology developed by one group will have great difficulty
>  > gaining adoption by those who did not participate in its creation.
>  > Cyc had the additional disadvantage of being proprietary for its
>  > first 15 years (i.e. **not open**) and is still partly proprietary.
> For many years, Cyc was being supported by sponsors who received
> copies of all the code and who had the opportunity to send some
> of their personnel to Austin to take courses and work with the Cyc
> developers.  The companies who supported it received a license to
> use any or all of that material, royalty free, in any of their
> products.
> Lenat gave a talk to Bill Gates at one time and convinced him to
> sign up as a sponsor.  But after a couple of years, they dropped
> their license, even though they would have had the right to package
> any or all of Cyc in any or all MSFT products -- royalty free.
> Around that time I also spoke with somebody in MSFT research,
> and I asked him what people there thought about Cyc.  He said
> that the were not impressed with its potential.  (Of course,
> that might be a biased view by somebody who was working on
> a different approach.)
> I also spoke with the manager of the AI department at a large
> corporation that had been a sponsor of Cyc from the beginning.
> While I was visiting there, I also got a demo from a person
> who was using Cyc for his project.  After that demo, I asked
> the manager what they thought of their investment in Cyc.
> He replied "Every person who had ever spent any time working
> with Cyc has been fired, and I don't think it's a coincidence."
> A few weeks after that visit, I sent an email note to the
> person who showed me the demo, and the note bounced back
> with an "address unknown" response.
> The companies who signed up for Cyc paid a lot of money for it,
> and they had employees who learned how to use the system.  But
> apparently, they weren't able to build any useful applications
> that the companies considered profitable or even promising.
> That experience does not give me any confidence that another
> Cyc-like or SUMO-like project on a bigger scale will produce
> a high return on investment.  For years, I had been telling
> Lenat that he should devote more time to work with customers
> to develop useful applications.  But he kept saying that he
> didn't want to "dilute" his research by working on applications.
> I still believe that Cyc is a valuable resource, from which
> we have all learned a great deal.  But one thing we learned is
> that an unfocused research effort is not likely to discover a
> pot of magic applications at the end of the rainbow.
> 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
>     (015)

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    (016)

<Prev in Thread] Current Thread [Next in Thread>