Re: Relationships in SKOS Schemes

On 28 Nov 2006, at 07:56, Antoine Isaac wrote:

> Another important point
>
>>
>> The SKOS Core Guide described the notion of concept schemes:
>>
>> http://www.w3.org/TR/swbp-skos-core-guide/#secscheme
>>
>> [[
>> Usually, concepts are defined in relation to other concepts, as part
>> of an internally coherent concept scheme. As mentioned in the
>> introduction, a 'concept scheme' is defined here as: a set of
>> concepts, optionally including statements about semantic  
>> relationships
>> between those concepts.
>> ]]
>>
>> How might I represent the fact that the semantic relationships  
>> between
>> concepts occurs within a particular scheme? If my graph contains more
>> than one vocabulary or concept scheme, how do you tell which scheme
>> the relationships belong to?  If I understand things correctly, we  
>> can
>> make assertions that a particular concept is in a scheme through the
>> inScheme property, but this doesn't cover the semantic relationships
>> themselves. In fact I don't see any way of doing this without
>> resorting to reification....
>
>
> And you are right. Relying on existing semantic web technologies,  
> SKOS cannot offer more than what these technologies enable (at  
> least if we want to keep relationships tractable for knowledge  
> engineers, and perhaps also tools)
> Note that a few patterns accounting for some kind of reification  
> were already proposed, as in
> http://www.w3.org/TR/2005/WD-swbp-skos-core-guide-20050510/ 
> #secdocnodestyle

Ok, so I see this as two issues.

First is making it clearer in the document what is expected w.r.t the  
containment relationship.

The second is then thinking what we might do about addressing these  
shortcomings. Issues like containment are very natural in this  
setting and if we don't think carefully about it, it'll come back and  
bite us. A question this raises to me is exactly how we intend to  
specify SKOS. The SKOS guide describes SKOS as an application of RDF  
that can be used to express a concept scheme. The presentation of the  
SKOS "model" is done through the presentation of the RDF vocabulary  
used.

An alternative could be to provide some more abstract model of what  
SKOS vocabularies are and then a mapping into RDF triples that  
provides the interchange format. This would be similar to
the way that OWL was defined, with an abstract syntax and then a  
mapping into the underlying triples representing the structure. Again  
referring to the OWL 1.1 effort [1], this is being done using UML  
diagrams to describe what's "in" the ontology. I believe doing things  
this way can be of benefit in providing a "clean" underlying model  
that isn't cluttered by, for example, things like the RDF collection  
properties, which are really just syntactic machinery.

	Sean

[1] http://owl1_1.cs.manchester.ac.uk/owl_specification.html

--
Sean Bechhofer
School of Computer Science
University of Manchester
sean.bechhofer@manchester.ac.uk
http://www.cs.manchester.ac.uk/people/bechhofer

Received on Tuesday, 28 November 2006 17:20:00 UTC