Re: The Open world assumption shoe does not always fit - was: RE: [ontolog-forum] Fwd: Ontolog invited speaker session - Dr. Mark Greaves on the Halo Project - Thu 2008.06.19

At 10:35 PM -0400 6/26/08, Bob Futrelle wrote:
>If I have a database of *all* employees in a company and a query for a
>person returns nothing, then that failure allows me to assert that
>that person is not an employee.

As long as you know that the database has that all-encompassing 
quality, yes. But then, since that knowledge is crucial to the 
conclusion being correct, it - the knowledge of the completeness - 
should be made explicit as a premis in the argument. And then the 
reasoning is classical.

Heres a way to say it. Logic is about what entails what. Now, suppose that:
John is not listed in database D of employees. Does it follow -is it 
entailed - that John is not an employee? Well, no, actually. But NAF 
would say that it does.

Now, in your scenario, we also know that the database is a list of 
ALL the employees. So add this as another assumption, since this is 
an important fact. NOW indeed it does follow that John is not an 
employee. But we don't need NAF to conclude this: it follows by 
classical logic from the two assumptions. Either way, in order to 
express the reasoning correctly, classical logic is exactly what we 
want. Using NAF on just the failure is an enthymeme: it has a missing 
premis, like saying "Plato is a man, so Plato is mortal", forgetting 
to add the 'obvious' premis that all men are mortal. As this example 
might suggest, this is a VERY old mistake in reasoning.

>It's a matter of deciding what your
>universe of discourse is, is it not?

Well, if your entire universe of discourse is just the employees of 
one company, then sure. But by and large, most ontologies have a 
rather wider view of the universe. And if you plan to publish stuff 
on a Web, then expecting everyone who reads it to agree with you that 
the entire universe comprises nothing but the employees listed in 
your database is asking rather a lot.

Bear in mind that the start of this thread was John Sowa's suggestion 
that CL be adopted as a basic notation for ontologies in general, a 
kind of super-OWL, and Adrian's response suggesting that there was 
something inherently wrong with classical negation for such a role.

Pat

-- 
---------------------------------------------------------------------
IHMC		(850)434 8903 or (650)494 3973   home
40 South Alcaniz St.	(850)202 4416   office
Pensacola			(850)202 4440   fax
FL 32502			(850)291 0667    cell
http://www.ihmc.us/users/phayes      phayesAT-SIGNihmc.us
http://www.flickr.com/pathayes/collections

Received on Friday, 27 June 2008 04:14:56 UTC