Re: Side question : mapping OWL classes to SKOS concepts Re: SKOS Guide and owl:sameAs

Mark
> This is a real conversion issue instead of a simple mapping issue, 
> because you want to go from one way of viewing of the hotel-world (nr. 
> of stars) to another (in terms of cheap/standard/luxury). You would 
> also have this problem if the front-end was using an OWL ontology 
> instead of a SKOS representation.
Sure. But my point is that some applications are happy with SKOS, don't 
need OWL at all, but need to interface with applications using OWL.
> Yep. But it could also have been an OWL hierarchy that is a 
> "simplified" view of the back-office OWL hierarhcy.
Indeed. This is yet another issue : mapping of OWL ontologies, a complex 
issue which is not in this forum scope.
>> Seems to me this will be a frequent use case if both OWL ontologies 
>> and SKOS concept schemes are used in integrated environments.  
> Yes, but not necessarily for the use case you cite. Only if you have 
> to work together with other SKOS thesauri do you have this problem.
Actually, sort of, yes. The same ontology will potentially map to 
different SKOS schemes.
>> Since we need that right now for Mondeca applications, we are 
>> developing our own vocabulary for it, but of course would be happy to 
>> see this issue put on the standardisation track.
> The problem probably is that you have to go from a class hierarchy 
> (subClassOf between instances of rdfs/owl:Class) to an instance 
> hierarchy (skos:broader/narrower between instances of skos:Concept). 
That is not the issue actually : the hierarchy in OWL and the hierarchy 
in SKOS are defined independently. A OWL subsumption is not necessarily 
mapped to a SKOS broader-narrower, and in fact I would not like to lead 
people into thinking that such a mapping makes any sense.
But there is more, for example a concept may match e.g. all instances of 
a class which have a certain value for a certain attribute, IOW some 
owl:hasValue restriction which is not necessarily explicit in the 
ontology. In the Hotel example, the number of stars can be modeled as a 
property, not a subclass of Hotel.
> So what vocabulariy are you developing for it?
Guess what? An RDF extension of SKOS, allowing each skos:Concept to 
declare which ontology element(s) are relevant to the concept : which 
class, which instances etc ...
> It sounds like an off-line conversion of the ontology to SKOS would 
> solve the problem, removing the need for a mapping vocabulary.
Well I'm not sure what you mean exactly by an "off-line conversion of 
the ontology to SKOS", and as said above, I'm not even sure if such an 
expression makes sense at all. First it sounds like you're thinking 
there is a single and systematic way to convert OWL to SKOS, and I hope 
it's not the case. Second, it would lead folks to think that SKOS and 
OWL are just two languages to model the same things differently, which 
IMO is plainly wrong. Actually I hope the ongoing SKOS standardization 
will eventually clarify what SKOS is about vs ontology languages.

Bernard

<http://mondeca.wordpress.com/>

Received on Thursday, 2 November 2006 13:59:06 UTC