Re: "shape" as a relationship, not a class

On 2/9/2015 10:16, Michel Dumontier wrote:
> Holger,
>
>> there is zero practical difference between
>> the following options:
>>
>>      ex:Class
>>          a owl:Class ;
>>          rdfs:subClassOf [
>>              a owl:Restriction ;
>>              owl:onProperty ex:property ;
>>              owl:minCardinality 1 ;
>>          ] .
>>
>>      ex:Class
>>          a owl:Class ;
>>          ldom:property [
>>              a ldom:PropertyConstraint ;
>>              ldom:predicate ex:property ;
>>              ldom:minCount 1 ;
>>          ] .
>>
>> Holger
>    is it not the case that (1) *entails* that every instance of
> ex:Class has at least one relation ex:property to some unidentified
> object of any type, whereas (2) should be used to check those
> instances of ex:Class have at least one ex:property to one concrete
> object?  will the shape in (2) trivially validate on an OWL axiom in
> (1)?

Yes that is correct. The OWL restriction can be used for entailment, 
while the LDOM property can be used for constraints. The discussion here 
is about whether those ldom:properties can be attached to classes or 
not. And in my judgement it is perfectly valid to attach these 
closed-world constraints to classes just like the open-world OWL 
restrictions are. The class provides a natural way of organizing such 
constraints.

Hope this clarifies it.

Holger

Received on Monday, 9 February 2015 00:20:57 UTC