RE: Question about value vocabularies and range

Dear Karen,

Ideally you would give your RDF "property meaning something like 'language of text'" a range.  If you wanted to use dcterms:language, then that range would logically be a dcterms:LinguisticSystem.  Individual languages (all of which would be Classes) would be sub-classes of the dcterms:LinguisticSystem.  In any event, your right about owl:allValuesFrom - it sounds promising, but isn't quite what you're looking for.  

Now, if you need URIs for languages, you have a few options [1,2,3,4].

Unfortunately, we haven't quite given you the hooks you need to integrate these value lists into an ontology.  But, I'm glad you brought this up because I've recognized this for sometime, but haven't had the opportunity to make the next step.  Partly this has been a lack of time to fully investigate our options here.  But, without promising anything, would defining a dcterms:LinguisticSystem per value list at ID [1,2,3,4], and then making each of the individuals in those lists sub-classes of the dcterms:LinguisticSystem be something you and/or the community in general would be interested in? Are there other options (other defined "LinguisticSystem"-type types, e.g.) open to us here?

Warmly,

Kevin

[1] http://id.loc.gov/vocabulary/languages
[2] http://id.loc.gov/vocabulary/iso639-1
[3] http://id.loc.gov/vocabulary/iso639-2
[4] http://id.loc.gov/vocabulary/iso639-5


________________________________________
From: public-lld-request@w3.org [public-lld-request@w3.org] On Behalf Of Young,Jeff (OR) [jyoung@oclc.org]
Sent: Thursday, May 05, 2011 21:09
To: Karen Coyle; public-lld
Subject: RE: Question about value vocabularies and range

Karen,

For the specific example of "language of text", I think that BCP-47 tags
<http://tools.ietf.org/html/rfc5646> would be ideal. This would account
for language, region, and script and a few other facets of language.

It would be nice if we could figure out how to specify BCP-47 in terms
of Dublin Core. One would hope that dc:language and/or dcterms:language
could be configured somehow to accept BCP-47 tokens, but I'm not sure
how. A dc:language solution might look like this:

abox:manifestation1 a frbr:Manifestation;
        dc:language "en-US"^^tbox:BCP-47.

The tbox:BCP-47 datatype would need to be defined using XML Schema
somewhere, but it would be surprising if this was left as an exercise
for the user. Maybe someone knows if such a datatype has already been
defined for BCP-47 somewhere. I suspect not because the permutations of
language facets is quite large.

Alternately, the range on dcterms:language is set to
dcterms:LinguisticSystem, which means the object must be a URI. This
implies a solution like this:

abox:manifestation a frbr:Manifestation;
        dcterms:language tbox:en-US.

tbox:en-US a dcterms:LingusticSystem;
        rdf:value "en-US";
        rdfs:label "US English".

Again, though, I don't know if anyone has already defined this or how
they dealt with the language facet permutation problem.

I know you are asking a more abstract question, but simpler cases than
this could be boiled down the SKOS ConceptSchemes. BCP-47 seems like an
unusual and difficult case, though.

Jeff

> -----Original Message-----
> From: public-lld-request@w3.org [mailto:public-lld-request@w3.org] On
> Behalf Of Karen Coyle
> Sent: Thursday, May 05, 2011 8:34 PM
> To: public-lld
> Subject: Question about value vocabularies and range
>
> I have a question that MAY be an OWL question, and it MAY be an
> Application Profile question.
>
> Presume I have an RDF-define property meaning something like "Language
> of text." I would like to say that the values for this property must
> be taken from an EXTERNAL (but URI-identified) list, like ISO 639-n.
> owl:allValuesFrom looks like it's heading in the right direction, but
> can't exactly do this (or if it can, PLEASE explain!).
>
> This is a pattern that I think we will want to use frequently in
> library data. When I was working on the DCMI Application Profile
> document I ran into exactly this same problem but didn't pursue it
> further.
>
> All ideas welcome.
> kc
>
> --
> Karen Coyle
> kcoyle@kcoyle.net http://kcoyle.net
> ph: 1-510-540-7596
> m: 1-510-435-8234
> skype: kcoylenet
>
>

Received on Friday, 6 May 2011 14:24:41 UTC