Re: [SKOS] Resolutions on concept semantics

OK, going back to the subject, the minutes of today's telecon will say:
> We make no statement either way about whether skos:Concept is disjoint 
> or not disjoint with owl:Class

I can live with this formulation, which is actually different from the 
one in the F2F minutes which was "we will not say anything about the 
disjointness. we should make clear that the omission is explicit"

But I would like to have things put clearly here. *If there is something 
about this issue in one of the SKOS document, then it cannot be 
something like "skos:Concept may be disjoint with owl:Class or not."*
We must be coherent he: if for us there is a possibility for someone 
somewhere somewhen to create something that is both a skos:Concept and 
an owl:Class, then we cannot honnestly say that "owl:Class and 
skos:Concept may be disjoint".

And I cannot imagine something like a "local interpretation of SKOS" for 
which skos:Concept and owl:Class are disjoint. This would be 
overcommitting to what was defined in the SKOS *standard* and, since the 
two alternatives are contradictory, could raise serious interoperability 
issues, as Daniel mentioned.

Cheers,

Antoine



>
> It seems there are several interpretations of "saying nothing" here...
> To me saying nothing on the disjointness amounts to say that there are 
> not disjoint. The idea is that people can do what they want with their 
> concepts, including declaring some as classes or keeping them 
> distinct. If we have to find a common denominator for these different 
> situations, then it is "non-disjointess".
>
> As Daniel I also don't like the idea of letting people decide whether 
> skos:concept is disjoint or not. SKOS is defined by us, and we cannot 
> have people making different interpretations of the shared SKOS 
> construct, especially when they can be conflicting.
> What users can do of course is to say that *their* SKOS concepts are 
> disjoint from owl:Class (e.g. by creating a subclass of skos:Concept 
> that is disjoint with owl:Class). But this is another story...
>
> Cheers,
>
> Antoine
>>
>> I am opposed to *saying nothing* about disjointness between 
>> skos:Concept and owl:Class (actually, in general I'm opposed to 
>> leaving any semantics undefined in SKOS)--if you leave any semantics 
>> up to the user to decide, then you ensure people will have different 
>> meanings for the same entities and relations, making interoperability 
>> impossible.
>> Daniel
>>
>> Quoting "Miles, AJ (Alistair)" <A.J.Miles@rl.ac.uk>:
>>
>>>
>>> Hi Antoine,
>>>
>>>> >
>>>> > [following discussion on the OWL/SKOS patterns] ... we are not
>>>> > discussing the introduction of new properties, but the semantics of
>>>> > skos:Concept, in particular its disjointness with owl:Class
>>>> > aliman: we will not say anything about the disjointness
>>>> > sean: we should make clear that the omission is explicit
>>>>
>>>>  3. RESOLUTION: skos:Concept is not disjoint with owl:Class .
>>>> Some instances of SKOS concept may be also declared (and
>>>> treated) as OWL classes, and vice versa.
>>>
>>> I thought our resolution was to *say nothing* about disjointness  
>>> between skos:Concept and owl:Class. That would give people the  
>>> freedom to interpret them as disjoint, if they want to do that, or  
>>> not, if they don't.
>>>
>>> That's what I tried to capture in:
>>>
>>> [1]  
>>> <http://www.w3.org/2006/07/SWD/wiki/SKOS/Reference/Concepts?action=recall&rev=10> 
>>>
>>>
>>>> From [1] ...
>>>
>>> "The decision to leave the formal semantics of skos:Concept  
>>> undefined has been made to allow different design patterns for using 
>>>  SKOS in combination with more formal languages such as OWL to be  
>>> explored.
>>>
>>> For example, interpreting skos:Concept and owl:Class as disjoint  
>>> classes would be consistent with the semantics of SKOS.  
>>> Alternatively, interpreting skos:Concept as a super-class of  
>>> owl:Class would also be consistent with the semantics of SKOS."
>>>
>>> Cheers,
>>>
>>> Al.
>>>
>>>
>>
>>
>>
>>
>
>
>

Received on Tuesday, 30 October 2007 17:14:51 UTC