USE Case - Pres. PFPS (was Re: An intuition pump)

At 10:23 AM -0400 9/24/03, Peter F. Patel-Schneider wrote:

>
>Well, suppose I wrote a formal system that works by the strong-Tim hypothesis.
>This system would be in a contradictory state if given the following
>(single) premise
>
> 
>	<http://www-db.bell-labs.com/user/pfps/#Peter_Frederick_Patel-Schneider>
>	owl:differentFrom <http://www.whitehouse.gov/president/> .
>


I have been thinking hard (ouch, it hurts) about Peter's use case 
above.  I think it could shed a lot of light on the conversation, at 
least in terms of defining our own vocabulary for this dicussion (Tim 
is right, a big part of the work of a WG is defining its own 
agreements on terms)

Lets assume the above is on a web document and is the only thing on 
that document.

  That document, if the only RDF document in the universe, would 
clearly be consistent according to the RDF Model theory, and the 
graph would be straightforward.

  The document currently at 
http://www-db.bell-labs.com/user/pfps/index.html contains a number of 
RDF statements including one that includes:

<foaf:Person rdf:about="#Peter_Frederick_Patel-Schneider">
   <owl:sameAs rdf:resource="http://www.whitehouse.gov/president/" />
</foaf:Person>

which, again, if the only thing on the entire SW, would be consistent.

If, however, these documents were joined into a single document, that 
document would be inconsistent according to the OWL model theory

In fact, one can create "document sets" using OWL such that any 
proper subset of the documents joined onto a single document would be 
consistent, but the entire set joined wouldn't (I have a back of an 
envelope proof of this if anyone cares)


so we have two possibilities:
  1 - we can say the entire set of documents is inconsistent if, 
merged into a single graph the entire graph is inconsistent
  2 - we can say that the entire set of documents is consistent if 
every document in the set is consistent (i.e. would be consistent if 
it was alone the only document on the entire Semantic Web)

  In a certain sense, I wonder if this is somehow the heart of our debate?

  -JH

p.s. Note that OWL has a way to differentiate the two cases -- when 
imports is used, the entire "document set" of the imports closure 
becomes a single document, and thus definition 2 is used.  WHen there 
is no imports, each document essentially stands along, and definition 
1 is used.     I begin to wonder if that is as bad an idea as I 
thought at first -- looked at this way, maybe I should consider 
removing my objection to owl:imports...
-- 
Professor James Hendler				  hendler@cs.umd.edu
Director, Semantic Web and Agent Technologies	  301-405-2696
Maryland Information and Network Dynamics Lab.	  301-405-6707 (Fax)
Univ of Maryland, College Park, MD 20742	  *** 240-277-3388 (Cell)
http://www.cs.umd.edu/users/hendler      *** NOTE CHANGED CELL NUMBER ***