W3C home > Mailing lists > Public > www-webont-wg@w3.org > April 2002

WOWG: report from test breakout group at f2f

From: Jeremy Carroll <jjc@hplb.hpl.hp.com>
Date: Thu, 18 Apr 2002 09:35:02 +0100
To: <www-webont-wg@w3.org>
Message-ID: <CEECKEAMDAJDDEDGJNBEMECECAAA.jjc@hpl.hp.com>
Summary: named classes or unnamed class expression, who cares?


During the f2f, I declined to report immediately from the TEST focus area
breakout, because of the time pressure.

I indicated I would send an e-mail report.

This is it.

===

We worked on producing test cases that would help clarify the difference
between a light syntax with only named classes, and a heavy syntax allowing
unnamed class expressions.

We concluded that there were no substantive differences, and that all the
test cases we could think of illustrated the lack of content to this
divisive issue.

Some of the tests were in the abstract syntax of the light syntax proposal,
others in a more familiar daml like triple syntax.

The tests all concerned orphans, sometimes we called them orphans, sometimes
the class of orphans was unnamed or named with a skolem function.

Here are some of the test we discussed:

A:

DefinedClass(GenSym1,Person1,slot(hasParent1,maxCardinality=0))

B:

DefinedClass(GenSym2,Person1,slot(hasParent1,maxCardinality=0))



Test 1:

A & B

entail

SameClassAs(GenSym1,GenSym2)

Comment:
Using a skolem function to generate names of otherwise unnamed classes is
harmless, because equivalent names will be provably equivalent.


C:

DefinedClass(Orphan2,Person2,slot(hasParent2,maxCardinality=0))


D:
SameClassAs(Person1,Person2)
SamePropertyAs(hasParent1,hasParent2)



Test 2:

A,C,D

entail

sameClassAs(GenSym1,Orphan2)

Comment:
Two independently developed ontologies (A and C) for the same concepts, one
of which (D) names the Orphan class the other of which (A) does not name it
can be aligned by aligning the named components (D). In this way the unnamed
class can be seen to be equivalent to the named class.


E:

_:orphan <rdf:type> _:intersect .
_:intersect <rdf:type> <owl:Class> .
_:intersect <owl:intersectionOf> _:b .
_:b <rdf:member> <Person2>
_:b <rdf:member> _:r
_:r <rdf:type> <owl:Restriction> .
_:r <owl:onProperty> <hasParent2> .
_:r <owl:maxCardinalty> "0" .


Test 3:

A,E,D

entail

_:orphan <owl:sameClassAs> <GenSym1> .

Comment:
Heavy and Light syntax interoperate. Whether the class expression is named
or not is wholly irrelevant.




Moral:
The WG had a fairly severe division that was content-free.
Test cases help elucidate.
Received on Thursday, 18 April 2002 04:36:36 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:57:49 GMT