[SKOS] proposal for ISSUE-83 SemanticsOfSchemeContainmentProperties

Dear all,

I'd like to propose the following resolution to ISSUE-83 
SemanticsOfSchemeContainmentProperties [1]

PROPOSAL: to add to the SKOS reference that the graph:
ex:cs skos:hasTopConcept ex:c.
entails the graph:
ex:c skos:inScheme ex:cs
But *not* to add extra semantics regarding the combination of 
skos:hasTopConcept, skos:broader/skos:narrower and skos:inScheme properties

--- Explanation of the issue:
This issue was raised by a comment from Margherita saying that she was 
using the combination of skos:hasTopConcept, skos:narrower to infer 
scheme containment for concepts [2]
The initial (assumed) position is that if we have the following statements
ex:cs skos:hasTopConcept ex:c1
ex:c1 skos:narrower ex:c2
then we can infer that both c1 and c2 belong to the concept scheme ex:cs

--- Motivation for resolution:
I think there in sense in having, for the above example, the inference 
that ex:c1 belong to ex:cs. Not saying so would intuitively amount to 
support cases where concept schemes have as top concepts concepts that 
do not belong to them, which is weird.
But I'm sure we cannot allow the second entailment (ex:cs2 belong to 
ex:cs). ex:c2 might not indeed belong to the scheme ex:cs, since we have 
resolved that skos:narrower and skos:broader can be asserted for 
concepts that belong to different schemes (among others, as a side 
effect of having skos:broadMatch a sub-property of skos:broader [3])

Note: there is a note in the current version of the Reference about this 
[4], but I think it could be made stronger: hence my proposing this 



[1] http://www.w3.org/2006/07/SWD/track/issues/83
[2] http://lists.w3.org/Archives/Public/public-swd-wg/2008Jan/0150.html
[3] http://www.w3.org/2006/07/SWD/SKOS/reference/master.html#L4160
[4] http://www.w3.org/2006/07/SWD/SKOS/reference/master.html#L2446

Received on Tuesday, 27 May 2008 16:09:52 UTC