W3C home > Mailing lists > Public > public-semweb-lifesci@w3.org > March 2010

Re: Fw: SKOS/Datasets

From: Bernard Vatant <bernard.vatant@mondeca.com>
Date: Mon, 1 Mar 2010 10:40:51 +0100
Message-ID: <9d93ef961003010140rf88faep7b1eb4bdf6baf29c@mail.gmail.com>
To: hans.cools@agfa.com
Cc: SKOS <public-esw-thes@w3.org>, public-semweb-lifesci@w3.org, Rebecca S Guenther <rgue@loc.gov>
Hello Hans

Skipping all the technical part ... (re. Protégé processing of your
ontology, where we agree something is going wrong, and again my guess it has
something to do with the user-defined dataypes. But it is not the first time
I meet a case were Protégé can't re-open a file it has generated. I thought
Protégé 4.0 has overcome this, but apparently you found an exception. Maybe
you shoud forward this exchange to the Protégé list?)

... and going directly to your last question which many followers of this
thread might have missed, lost at the bottom of lines of code :)

> About SKOS:
> Making the class fos:Language a subclass of skos:Concept sounds like a use
> mention bug, since the things in the class skos:Concept are within a scheme
> of an iso standard of codes, describing a language, not being the language,
> not?
This is a recurrent issue in various lists around SKOS, e.g., recent debates
on "can a Person be represented as a skos:Concept". Several options :

In current published version of lingvoj ontology I define the "Language"
class http://www.lingvoj.org/ontology#Lingvo as a subclass of
dcterms:LinguisticSystem http://purl.org/dc/terms/LinguisticSystem, and
there is no use of skos elements altogether in this ontology, but on the
other hand nothing prevents to declare a language as a skos:Concept in an
open world.

If I remember well, Rebecca Guenther in a private exchange suggested that in
the SKOS publication at LoC the languages will indeed be represented as
instances of skos:Concept.

In your ontology, the very use of skos:exactMatch such as in ...

        a fos:Language;
        rdfs:isDefinedBy language:;
        skos:exactMatch [
                a skos:Concept; skos:inScheme language:iso639-1;
skos:notation "fr"^^language:iso639-1DT], [
                a skos:Concept; skos:inScheme language:iso639-2;
skos:notation "fre"^^language:iso639-2DT];
         foaf:name """French"""@en.

... entails that fre:language is indeed a skos:Concept, based on axioms
defined at http://www.w3.org/TR/skos-reference/#L2055

skos:exactMatch  is a subproperty of  skos:semanticRelation, of which domain
is skos:Concept

So maybe you do not explicitly define your fos:language class as a subclass
of skos:Concept in intension, but in extension, every instance of it you
define is indeed a skos:Concept by entailment.

At this point it would be simpler to define fos:Language as a subclass of
skos:Concept, with each instance being skos:inScheme each of the relevant
ISO-639 flavours.


Bernard Vatant
Senior Consultant
Vocabulary & Data Engineering
Tel:       +33 (0) 971 488 459
Mail:     bernard.vatant@mondeca.com
3, cité Nollez 75018 Paris France
Web:    http://www.mondeca.com
Blog:    http://mondeca.wordpress.com
Received on Monday, 1 March 2010 09:41:28 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 14:52:42 UTC