- From: Dave Reynolds <der@hplb.hpl.hp.com>
- Date: Fri, 01 Oct 2004 15:59:17 +0100
- To: "Miles, AJ (Alistair) " <A.J.Miles@rl.ac.uk>
- Cc: "'public-esw-thes@w3.org'" <public-esw-thes@w3.org>
Hi Al,
>>By the way, having written out this example I'd argue that if
>>you want to
>>build in some links between RDFS and SKOS then some official
>>replacement
>>for the "ex:classifiedAs" I invented above would be much higher up my
>>priority list than "denotesSameAs".
>
>
> What would be the semantics of such a property, and how would it be
> different from e.g. skos:broaderInstantive?
I'd define it something like:
Short: "Used to link a resource to a skos:Concept which that resource can
be classified under."
Longer: "We can think of a SKOS concept as standing for the set of
resources which can be classified under that concept. For example, in a
thesaurus of computer science topics used for document classification one
might have a 'Distributed systems' concept which stands for all the papers
which can be classified under 'Distributed systems'. When the things being
classified are represented by RDF resources then it is useful to have a
property which can link the resource to the SKOS Concept(s) under which it
is classified - skos:classifiedAs is such a property though applications
are free to define more specific sub-properties."
Perhaps it should be a sub-property of dc:subject with the added constraint
that the range is skos:Concept.
As to semantics I'm not sure. You could define its semantics to include
inheritance of classification:
if (r skos:classifiedAs s) holds and (s skos:broader sb) then
(r skos:classifedAs sb) holds
Or you could leave that open. In SWED we keep the transitively-closed
version of the classification relations separate because that's easier to
work with in engineering terms.
The difference from skos:broaderInstantive is that it has domain
rdfs:Resource rather than skos:Concept.
Cheers,
Dave
Received on Friday, 1 October 2004 14:59:27 UTC