Re: [SKOS] on ISSUE-71 and ISSUE-74

Hello Guus

Does this mail from Guus actually acknowledge that none of my arguments 
in [1] is proper?

Shortly:
- there is a authoritative difference between broader and broadMatch 
that makes them different, especially every broadMatch is not a broader
- assuming this, there may be cases of inter-scheme broader links and 
cases of intra-scheme broadMatch links

These seems at first glance to go against some of your assumptions in [2]...

Antoine

[1] http://lists.w3.org/Archives/Public/public-swd-wg/2008Feb/0097.html
[2] http://lists.w3.org/Archives/Public/public-swd-wg/2008Feb/0113.html
>
> A this week's telecon I promised to make a proposal about the semantic 
> relation between skos:broader and skos:broaderMatch (and mutatis 
> mutandis for skos:narrower and skos:related).  This should be seen as 
> a possible addendum to Antoine's proposed resolutions [2, 3].
> My suggestion is that the following semantic relationship holds:
>
>  skos:broaderMatch rdfs:subPropertyOf skos:broader .
>
> The usage pattern would be the following:
>
> - an schema owner who wants to make a distinction between broader and 
> broaderMatch relations should use the subproperty typing
> - for such schema's  it would be reasonable to assume that instances 
> of skos:broader that are not instances of skos:broaderMatch, are 
> intra-scheme relations.
> A few remarks:
> - Through the inScheme property of concepts you can in principle 
> derive whether a skos:broader relation is inter-scheme or 
> intra-scheme. In this sense skos:broaderMatch is just a shorthand c.q. 
> syntactic sugar. This is actually the main argument against having the 
> Match relations at all.
> Turning this around, we could also say that if a broaderMatch relation 
> exists between two concepts, then they must be in different schemes 
> (useful if the inScheme property is not defined by the schema owner).
> - One could also say that we it would be neater to have two 
> subproperties of skos:broader, for both intra-scheme and inter-scheme 
> relations. In my opinion this creates too much new vocabulary and is 
> an unwanted complexity.
>
> - Another argument against this proposal is that some desired logical 
> meta-properties (e.g. symmetry of skos:related) do *not* automatically 
> inherit over the subproperty relation.
> Guus
>
> [1] http://lists.w3.org/Archives/Public/public-swd-wg/2008Feb/0095.html
> [2] http://lists.w3.org/Archives/Public/public-swd-wg/2008Feb/0076.html
> [3] http://lists.w3.org/Archives/Public/public-swd-wg/2008Feb/0077.html
>
>

Received on Wednesday, 27 February 2008 11:10:13 UTC