Re: shapes as classes

I find it easier to explain what I mean via specific examples (in RDF). 
 From that perspective, a Shape can be regarded as a Resource that has 
constraints attached to it. In OWL such shapes often correspond to 
anonymous classes, used via owl:allValuesFrom and owl:someValuesFrom.

# Every Person must have a social security number
ex:Person
     a owl:Class ;
     rdfs:subClassOf [
         a owl:Restriction ;
         owl:onProperty ex:ssn ;
         owl:cardinality 1 ;
     ] ...

However, the shapes such as the owl:Restriction above are technically 
still owl:Classes, only that are used in a specific context only, and 
not of interest to the outside world (therefore blank nodes). 
Furthermore, such blank nodes are usually not instantiated via rdf:type, 
but are only used to declare additional conditions that certain values 
must fulfill. Following this, a Class is a Shape that can be 
instantiated directly, usually via rdf:type.

Technically, the terms Class and Shape are very similar to me. To avoid 
confusion, we could try to use the term Class only, assuming that a 
class describes a set of instances with shared characteristics.

Holger


On 12/13/2014 10:09, Holger Knublauch wrote:
> Please feel free to edit the page and come up with a better wording. 
> It was just a proposal. We can fight about words forever, without 
> making progress.
>
> Holger
>
>
> On 12/13/14, 9:35 AM, Peter F. Patel-Schneider wrote:
>> So RDF classes are not shapes, and neither are OWL classes.
>>
>> But why then does 
>> https://www.w3.org/2014/data-shapes/wiki/Resource_Shape_Association 
>> say that classes are shapes?
>>
>> peter
>>
>>
>> On 12/12/2014 03:11 PM, Holger Knublauch wrote:
>>> This depends on how you define the term Shape. To me, a Shape is a 
>>> group of
>>> connected constraints, so if a class has no constraints then it is 
>>> not really
>>> a Shape (unless you include the unconstrained Shape to be a Shape too).
>>>
>>> Holger
>>>
>>>
>>> On 12/13/14, 9:02 AM, Peter F. Patel-Schneider wrote:
>>>> But are all classes shapes?
>>>>
>>>> peter
>>>>
>>>>
>>>> On 12/12/2014 02:52 PM, Holger Knublauch wrote:
>>>>>
>>>>> On 12/13/14, 3:40 AM, Peter F. Patel-Schneider wrote:
>>>>>> If "Shape" is more general than "Class", shouldn't all classes be 
>>>>>> shapes?
>>>>>
>>>>> Some classes are just named entities without any constraints.
>>>>>
>>>>> Holger
>>>>>
>>>>>
>>>
>>>
>

Received on Monday, 15 December 2014 04:18:24 UTC