RE: Proposed response to Hugh WInkler - allDisjoint

Jim,

I did try to write a description of the test case.  It seemed to be 
getting too long, so I backed off.  Perhaps it could be explained 
by example:

--------------------------------------------------
When n is large, alternate approaches can be used to avoid exponential
assertion explosion.  One such method is illustrated in the OWL test
suite.

The illustrated method works as follows.  Given the following:

- The class Reptile is a subclass of things with exactly one
  family-name.

- The class Amphibian is a subclass of Reptile whose members have
  family-name 'Amphisbaenidae'.

- The class Crocodilian is a subclass of Reptile whose members have
  family-name 'Crocodylidae'.

Any member of the class Amphibian cannot also be a member
of the class Crocodialian since their literal family-names are not
equal.  By repeating the family-name assertion for each subclass of
Reptile, where each class uses a distinct literal name, we can ensure
that all the subclasses of Reptile are disjoint using only order n 
assertions.
--------------------------------------------------

Any help appreciated.

- Mike

Michael K. Smith, Ph.D., P.E.
EDS - Austin Innovation Centre
98 San Jacinto, #500
Austin, TX  78701

phone: +01-512-404-6683
email: michael.smith@eds.com
  
-----Original Message-----
From: Jim Hendler [mailto:hendler@cs.umd.edu] 
Sent: Monday, October 27, 2003 5:07 PM
To: Jeremy Carroll; Smith, Michael K; www-webont-wg@w3.org
Subject: RE: Proposed response to Hugh WInkler - allDisjoint



At 15:03 +0100 10/27/03, Jeremy Carroll wrote:
>Looks fine to me, except the URL will change once the test is approved ...
>
>Jeremy

OK, I've been thinking about this for a bit, and I'm still not happy. 
We had 3 LC comments about all Disjoint, a long discussion about it, 
aand it arose again during CR -- and all we ended up doing is not 
much.  The problem is this -- we refer people to a test case -- but 
can anyone reading that test case see how to apply the solution to 
their own problem?    let me make that a real, instead of rhetorical, 
comments -- Guide Editors -- take a look at the proposed test case 
and see if you feel comfortable that someone reading the Guide, who 
gets directed to this as the solution to allDisjoint would see how to 
apply it to some other problem (i.e. would be able to cut and paste 
and get it right)??   If you feel in good conscience that this test 
is enough, than I'm okay sending this response --- otherwise, I'd 
sure like someone to write up a description of how to do this (a 
paragraph or two would be fine) and include that with a pointer to 
the example -- would strengthen Guide on what is clearly an important 
issue (since it has come up 4 times).
  thanks
  JH


>
>
>>
>>   [EXISTING] As the number of mutually disjoint classes
>>   grows, the number of disjointness assertions grows proportionally to
>>   n<sup>2</sup>.  However, in the use cases we have seen, n is typically
>>   small.
>>
>>   [ADD] When n is large, alternate approaches can be used to avoid
>>   an exponential growth in the number of assertions.
>>   One such method is illustrated in the
>>   <a
>>  href="http://www.w3.org/TR/2003/CR-owl-test-20030818/proposedByIss
>>  ue#I5.21-0
>>  02">
>>   OWL test suite</a>.
>>
>>  - Mike
>>
>>  Michael K. Smith, Ph.D., P.E.
>>  EDS - Austin Innovation Centre
>>  98 San Jacinto, #500
>>  Austin, TX  78701
>>
>>  phone: +01-512-404-6683
>>  email: michael.smith@eds.com
>>
>>
>>  -----Original Message-----
>>  From: Smith, Michael K
>>  Sent: Monday, September 22, 2003 4:47 PM
>>  To: 'Jim Hendler'; Jeremy Carroll; www-webont-wg@w3.org
>>  Subject: RE: Proposed response to Hugh WInkler - allDisjoint
>>
>>
>>  Will do.  But not for a few days.
>>
>>  ACTION: Add text to Guide re disjoint classes and pointer to test.
>>
>>  - Mike
>>
>>  Michael K. Smith, Ph.D., P.E.
>>  EDS - Austin Innovation Centre
>>  98 San Jacinto, #500
>>  Austin, TX  78701
>>
>>  phone: +01-512-404-6683
>>  email: michael.smith@eds.com
>>
>>
>>  -----Original Message-----
>>  From: Jim Hendler [mailto:hendler@cs.umd.edu]
>>  Sent: Monday, September 22, 2003 6:35 AM
>>  To: Jeremy Carroll; www-webont-wg@w3.org
>>  Subject: Re: Proposed response to Hugh WInkler - allDisjoint
>>
>>
>>
>>  At 12:40 PM +0300 9/22/03, Jeremy Carroll wrote:
>>  >Mike Smith
>>  >>  1. There is problem with the Guide text as it is.  It notes the most
>>  >>  naïve approach to creating disjoint classes, which is O(n^2).
>>  >
>>  >How about simply pointing to the test case. e.g.
>>  >
>>  >After:
>>  >  "However, in the use cases we have seen, n is typically small. "
>>  >
>>  >Add:
>>  >[[
>>  >When n is large,
>>  >an <a href="??">example</a> in [OWL Test Cases] shows a
>>  >possible more efficient encoding.
>>  >]]
>>  >
>>  >The current URI for the example is:
>>
>http://www.w3.org/TR/2003/CR-owl-test-20030818/proposedByIssue#I5.21-002
>>  >
>>  >on approval this will become
>>  >
>>  >http://www.w3.org/TR/owl-test/byIssue#I5.21-002
>>  >
>>  >Jeremy
>>
>>
>>  I still think we need to actually describe the solution, not just
>>  point at the test -- Mike, how about putting a paragraph of
>>  description (take it from Ian's email or work with him to get it
>>  written) describing what to do - something like
>>
>>  "When n is large, an alternate encoding can be used.  In this case, ..."
>>  etc.
>>
>>  For some reason I thought we had agreed to do this when I withdrew my
>>  objection to not adding allDisjoint, but I cannot find the record of
>>  that - so now I ask again.  Also, we should have something in Ref
>>  that points back to this section of Guide or has a discussion there.
>>    -JH
>>
>>
>>
>>  --
>>  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
***
>>

-- 
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 ***

Received on Monday, 27 October 2003 18:49:26 UTC