Re: [sio-dev] Fwd: [ontolog-forum] Sharing and IntegratingOntologies

[Cameron Ross <cross@xxxxxxxxxxxx>]

On Wed, Apr 14, 2010 at 12:25 PM, John F. Sowa <sowa@xxxxxxxxxxx> wrote:

Ron,

There are some kinds of vetting that can be done fairly easily,
and the major part of the burden should be placed on the people
who design the ontology and contribute it to the OOR.

RW> It is difficult to see how any group can vet the entire range
 > of human activities.

Since the full range is unlimited, that is obviously impossible.
But the metadata of the OOR can and should include evaluations,
pro and con, by people who actually used the ontology.

One of the most important tests is consistency.  In general,
a proof of consistency can be undecidable or take an exponential
amount of time.  For example, any theory that has at least one
model is guaranteed to be consistent, but finding a model for
an arbitrary theory is as difficult as using a theorem prover.

However, people don't develop ontologies in a vacuum.  They
usually start with at least one application, and they begin
by describing the entities that actually occur in that
application.  That application is indeed a model, and its
existence demonstrates consistency.

See slide 6 of the following talk for that point:

   http://www.jfsowa.com/talks/sionto.pdf

The minimal vetting that I would recommend for the OOR is to
require at least one model for each contributed ontology.
Then the consistency test for that ontology can be done by
pushing a button to check the ontology against the model.

RW> What are you going to do if the DoD adopts an ontology for
 > contracting that has not been vetted? Or Federal Reserve for
 > interbank reporting? Or Walmart for provisioning, payments and
 > warranty claims?

I would push the button and check whether it was consistent.
If they didn't include a model, I'd ask them to send the one
that they used in designing and testing the ontology.

If they didn't have such a model, I'd notify my congressman
that the gov't is screwing up.

This is reminiscent of software engineering best-practices.  Specifically,

the idea of automated testing.  I believe "models" would be analogous

to "functional tests" in agile software development.  You'd be hard pressed 

to find a professional software developer that would argue against using

automated unit and functional tests as a vetting process.  This holds 

true in both the open source community and the proprietary software

companies.

The SIO tool suite should consider tooling for automated model testing

a la agile development.  I wonder how for the Agile development analogy 

can be taken?

Cameron.

 

 

John


_________________________________________________________________
Msg Archives: http://ontolog.cim3.net/forum/sio-dev/
Join Community: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/sio-dev/
Unsubscribe: mailto:sio-dev-leave@xxxxxxxxxxxxxxxx
Community Shared Files: http://ontolog.cim3.net/file/work/SIO/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?SharingIntegratingOntologies

-- 
Kojeware Corporation
_________________________________________________________________ 
Msg Archives: http://ontolog.cim3.net/forum/sio-dev/   
Join Community: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J 
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/sio-dev/  
Unsubscribe: mailto:sio-dev-leave@xxxxxxxxxxxxxxxx 
Community Shared Files: http://ontolog.cim3.net/file/work/SIO/ 
Community Wiki: 
http://ontolog.cim3.net/cgi-bin/wiki.pl?SharingIntegratingOntologies     (01)