W3C home > Mailing lists > Public > public-esw-thes@w3.org > October 2004

RE: Resources, Classes, and Concepts

From: Miles, AJ (Alistair) <A.J.Miles@rl.ac.uk>
Date: Mon, 4 Oct 2004 14:07:57 +0100
Message-ID: <350DC7048372D31197F200902773DF4C05E50C67@exchange11.rl.ac.uk>
To: 'David Menendez' <zednenem@psualum.com>, public-esw-thes@w3.org

Hi David,

> The problem is that most of the annotation
> properties SKOS provides only make sense in the context of a single
> concept scheme.

Yes.  This is the key argument for why skos concepts should not be merged in
the graph, even if they look like they carry identical meaning.

However, there is a further issue here, which revolves around the
relationship between skos concepts, rdfs/owl classes and rdfs/owl
individuals ... (here is a different slant on the problem from me):

The requirement is: we want to be able to say that some skos concept X has a
relationship of meaning to some rdfs/owl class/individual.

Danbri has argued that a relationship of meaning between a skos concept and
an rdfs/owl class/individual **should always be asymmetric** (i.e. 'skos
concept X *denotes* rdfs/owl class/individual Y', see e.g. [1]).

Dave R. has argued that a relationship of meaning between a skos concept and
an rdfs/owl class/individual **can be symmetric** (i.e. 'skos concept X
*denotesSameAs* rdfs/owl class/individual Y', see e.g. [2]).

At this moment, both arguments are compelling to me, am quietly waiting for
an epiphany :)

Al.

[1] http://lists.w3.org/Archives/Public/public-esw-thes/2004Sep/0067.html
[2] http://lists.w3.org/Archives/Public/public-esw-thes/2004Oct/0005.html


> 
> 
> I missed most of the skos:denotes thread, so I'll try to summarize my
> position here. Let me restate Alistair's example:
> 
>     a:ajm a skos:Concept
>         ; skos:prefLabel "Alistair Miles"
>         ; skos:scopeNote "The man from SKOS"
>         .
>     
>     b:ajm a foaf:Person
>         ; foaf:name "Alistair Miles"
>         ; foaf:mbox <mailto:A.J.Miles@rl.ac.uk>
>         .
> 
> My feeling is that a:ajm and b:ajm should not be considered to denote
> the same resource.
> 
> Let's say they did. That would mean that we can replace a:ajm 
> with b:ajm
> and vice versa anywhere in the graph without changing its
> interpretation. Now let's say that another concept scheme 
> also includes
> a concept for Alistair:
> 
>     c:ajm a skos:Concept
>         ; skos:prefLabel "Alistair Miles"
>         .
> 
> In that case, we'd have to say that c:ajm also denotes the 
> same resource
> as a:ajm and b:ajm. The problem is that most of the annotation
> properties SKOS provides only make sense in the context of a single
> concept scheme. My feeling is that we do not want a:ajm and c:ajm to
> denote the same resource.
> 
> What we want to say is:
> 
>     a:ajm skosmap:exactMatch c:ajm.
> 
> An analogous situation applies to classes in RDFS and OWL. rdfs:Class
> and owl:Class are intensional, which means that two classes 
> can have the
> same extension but still be distinct. For example, here are 
> two distinct
> classes with the same extension:
> 
>     d:ajm a owl:Class
>         ; rdfs:label "A class whose only instance is Alistair"
>         ; owl:oneOf ( b:ajm )
>         .
>     
>     e:ajm a owl:Class
>         ; rdfs:label "A different class whose only instance 
> is Alistair"
>         ; owl:oneOf ( b:ajm )
>         .
> 
> Additionally, if a:ajm and b:ajm are distinct, we can make a useful
> distinction between dc:subject and foaf:topic:
> 
>     ex:something_about_alistair
>         ; dc:subject a:ajm
>         ; foaf:topic b:ajm
>         .
> 
>     ex:something_about_concept_ajm
>         ; foaf:topic a:ajm
>         .
> -- 
> David Menendez <zednenem@psualum.com> 
<http://www.eyrie.org/~zednenem/>
Received on Monday, 4 October 2004 13:08:29 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 2 March 2016 13:32:04 UTC