Re: SUO: Re: OpenCyc Motion Open for Discussions
OpenCyc is an open source ontology. I have proposed its use
by the IEEE SUO, but anyone could have proposed OpenCyc for
this. The only requirement for the SUO group, as for any other
OpenCyc user, would be to make public any changes they implement in
their version of the ontology. (This fits with SUO's intentions, since
they intend to produce an open standard.) The reason for this requirement
(a requirement that comes from the Gnu LGPL license we use)
is precisely so that OpenCyc.org could incorporate useful
additions/modifications into future versions of the OpenCyc ontology.
It is certainly possible for OpenCyc to not incorporate some of
SUO's ontology. That would not be a case of disowning the
SUO ontology, since OpenCyc does not (and will never) own the
SUO ontology. So, there could be divergence, but it is more likely
that most of the SUO ontology would be subsumed by OpenCyc, since
OpenCyc will be capable of containing multiple "competing"
theories. (Depending on how SUO is implemented, it might be
capable of subsuming OpenCyc, as well.)
As for applications, if you're looking to make an objective
evaluation, I believe you'd be better to start from scratch.
Existing applications have been written with a particular
ontology in mind. If you come up with your own application,
you can try it against various ontologies. This could also
avoid the problem of confusing the OpenCyc ontology
with other things (e.g. inference engine) that come with
the OpenCyc distribution. (Besides, we have yet to include
much in the way of applications in the current beta release
of OpenCyc.)
- John De Oliveira
At 11:54 PM 4/7/2003 -0400, Patrick Cassidy wrote:
> I have a question for the proposer of the
>OpenCyc as an IEEE standard:
> If there are substantial changes in the OpenCyc
>proposed and accepted by a vote of the voting
>participants in this list, how would the resulting
>ontology be treated by Cycorp? If there are substantial
>changes to the OpenCyc, presumably some of the
>existing applications that Cycorp has built using
>the OpenCyc would not work without modification.
>Would Cycorp then disown the resulting modified
>ontology, or try to adopt it as their own
>ontology, or at least try to create some form
>of mapping of the original OpenCyc with the
>modified version? Is the OpenCyc an all-or-nothing
>proposition?
> My view of the present OpenCyc is that there is
>a great deal of useful structure in it, and much
>obvious careful thought has gone into the elements
>of the ontology. But the result appears far too
>complicated to serve as a starting ontology that
>will permit multiple participants to carefully
>consider the various choices that need to be made
>in building an ontology. Many of the choices of
>hierarchy and predicates appear to be tuned to a
>particular method of reasoning, the rationale for
>which is not obvious from the OpenCyc itself.
>Perhaps the OpenCyc would be logically consistent
>with a simpler representation of the same
>concepts, and could be mapped to a simpler
>ontology.
> In order to have an objective basis for making
>choices, I think we really need some kind of application,
>however simple, to start to evaluate alternatives.
> I haven't had the time recently to look carefully
>into the demo programs provided with the OpenCyc
>release. Has anyone examined any one of the applications
>and begun to figure out what elements of the
>OpenCyc are essential for that application?
>
> Pat
>
>
>=============================================
>Patrick Cassidy
>
>MICRA, Inc. || (908) 561-3416
>735 Belvidere Ave. || (908) 668-5252 (if no answer)
>Plainfield, NJ 07062-2054 || (908) 668-5904 (fax)
>
>internet: cassidy@micra.com
>=============================================