# correction RE: The Ugly Test

From: Jos De_Roo <jos.deroo@agfa.com>
Date: Sat, 1 Feb 2003 16:28:50 +0100

Cc: "Jos De_Roo" <jos.deroo@agfa.com>, www-webont-wg@w3.org, www-webont-wg-request@w3.org
Message-ID: <OFF0969E52.921EA962-ONC1256CC0.005436B2-C1256CC0.00550C23@agfa.be>

[wasn't completely awake when I wrote earlier reply; hope I'm now]

>> suppose you have a similar case
>>   the range of p is one of 1,2,3,4
>>   the range of p is one of 3,4,5,6
>>   i is in a restriction on property p
>>     with mincardinality of 2
>>
>> is it then the case that
>>   i p 3
>>   i p 4
>
>
>Yes that is correct.
>
>>
>> I wouldn't think so as I (still) think
>> that ranges can come in via RDF merges
>>   the range of p is one of 2,4,6,8
>>
>> (I think I can't live with the idea of
>> *closed* ranges...)
>>
>
>The ranges aren't closed the related case ls:
>
>>   the range of p is one of 1,2,3,4
>>   the range of p is one of 3,4,5,6
>>   the range of p is one of 2,4,6,8
>>   i is in a restriction on property p
>>     with mincardinality of 2
>
>This is inconsistent and thus it entails
>
>>   i p 3
>>   i p 4
>
>(but it is less interesting this time round!).

| that's a nice reasoning hmm...
| so you would make it impossible to further make
| the range of a property sharper (doing so would
| make the kb inconsistent)
| I believe that this is not following from AS&S
| at least I can't conclude it from such piece as
|   if E is rdfs:range
|   then for x element of IOP, y element of IOC U IDC
|   <x,y> element of EXTi(Si(E)) iff
|     <w,z> element of EXTi(x) -> z element of CEXTi(y)
| which we interpret as
|   {?x rdfs:range ?y. ?w ?x ?z} => {?z rdf:type ?y}.
|   {?z rdfs:subClassOf ?y. ?x rdfs:range ?z} => {?x rdfs:range ?y}.
|
| so I still think that we can't call the proposed testcase
| http://www.w3.org/2002/03owlt/oneOf/Manifest004#test a
| http://www.w3.org/2002/03owlt/testOntology#PositiveEntailmentTest

given that
the range of p is one of 1,2,3,4
the range of p is one of 3,4,5,6
the range of p is one of 2,4,6,8

then it is the case that
(or we could entail that)
the range of p is one of 1,2,3,4,10,20
the range of p is one of 3,4,5,6,10,20
the range of p is one of 2,4,6,8,10,20

(per {?z rdfs:subClassOf ?y. ?x rdfs:range ?z} => {?x rdfs:range ?y}.
and having that one of 1,2,3,4 is a subclass of one of 1,2,3,4,10,20)

which is consistent with
i is in a restriction on property p with mincardinality of 2
but we can't in general derive the specific elements of EXTi(Si(p))

so I still think that we can't call the proposed testcase
http://www.w3.org/2002/03owlt/oneOf/Manifest004#test a
http://www.w3.org/2002/03owlt/testOntology#PositiveEntailmentTest

-- ,
Jos De Roo, AGFA http://www.agfa.com/w3c/jdroo/
Received on Saturday, 1 February 2003 10:29:34 GMT

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