RE: [Proposal][SKOS-Core] Local (non-URI) identifiers for concept s

> From: Alistair Miles [mailto:a.j.miles@rl.ac.uk] 
> Sent: 06 September, 2004 11:31
> Subject: Re: [Proposal][SKOS-Core] Local (non-URI) 
> identifiers for concepts
> 
> dc:identifier has recently occurred to me here also.
> 
> What if we go ahead with the deprecation of skos:externalID, 
> and replace with dc:identifier?  (And drop the proposal for a 
> new property skos:localID).
> 
> The problem of how to cope with concepts that may have more 
> than one identifier (possibly deriving from different 
> schemes) still remains, however.
> 
> What if we recommend here that a sub-property of 
> dc:identifier is created by the scheme owners for each 
> identifier?  So e.g.
> 
> <skos:Concept rdf:about="http://example.org/topics/098">
> 	<ex:conceptNumber>098</ex:conceptNumber>
> 	<ex2:id>A.34.12</ex2:id>
> </skos:Concept>
> 
> Where ex:conceptNumber and ex2:id are both hypothetical 
> properties created by scheme owners, and both are declared to 
> be subPropertyOf dc:identifier by scheme owners.  (Scheme 
> owners can also declare these props as inversefunctional if 
> they choose).
> 
> Comments?

I believe that Dublin Core/RDF already accommodates differing schemes
for <dc:subject> and that idea could be extended into the <dc:identifier>
element.  In qualified Dublin Core you can specify a scheme for the
<dc:subject> element:

<?xml version="1.0"?>
<rdf:RDF
  xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
  xmlns:rdfs="http://www.w3.org/2000/01/rdf-schema#"
  xmlns:dc="http://purl.org/dc/elements/1.1/"
  xmlns:dcterms="http://purl.org/dc/terms/"
>
  <rdf:Description rdf:ID="MeSH.D08.586.682.075.400">
    <dc:subject>
      <dcterms:MESH>
        <rdf:value>D08.586.682.075.400</rdf:value>
        <rdfs:label>Formate Dehydrogenase</rdfs:label>
      </dcterms:MESH>
    </dc:subject>
  </rdf:Description>
</rdf:RDF>

In the future, DCMI may deprecate encoding schemes, since they are not
all inclusive and DCMI doesn't want to track other encoding scheme lists.
However, the above RDF really boils down to simple RDF and I'm working
on a proposal to DCMI on this issue.  If DCMI wanted to maintain backward
compatibility with <dcterms:MESH> and allow metadata authors the ability
to specify *any* encoding scheme, then the same RDF as the above example
could be specified as:

<?xml version="1.0"?>
<rdf:RDF
  xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
  xmlns:rdfs="http://www.w3.org/2000/01/rdf-schema#"
  xmlns:dc="http://purl.org/dc/elements/1.1/"
  xmlns:dcterms="http://purl.org/dc/terms/"
>
  <rdf:Description rdf:ID="MeSH.D08.586.682.075.400">
    <dc:subject>
      <rdf:Description>
        <rdf:type>http://purl.org/dc/terms/MESH</rdf:type>
        <rdf:value>D08.586.682.075.400</rdf:value>
        <rdfs:label>Formate Dehydrogenase</rdfs:label>
      </rdf:Description>
    </dc:subject>
  </rdf:Description>
</rdf:RDF>

This produces the same RDF graph as if you had used <dcterms:MESH> and can
easily be verified when graphing both examples in the RDF validator [1].

The above concepts that work for <dc:subject> could be applied to
<dc:identifier>.  This would allow you to have multiple identifiers with
author specific schemes:

<?xml version="1.0"?>
<rdf:RDF
  xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
  xmlns:rdfs="http://www.w3.org/2000/01/rdf-schema#"
  xmlns:dc="http://purl.org/dc/elements/1.1/"
  xmlns:dcterms="http://purl.org/dc/terms/"
>
  <rdf:Description rdf:ID="MeSH.D08.586.682.075.400">
    <dc:identifier>
      <rdf:Description>
        <rdf:type>http://www.nlm.nih.gov/mesh/</rdf:type>
        <rdf:value
rdf:datatype="http://www.w3.org/2001/XMLSchema#string">D08.586.682.075.400</
rdf:value>
        <rdfs:label>MeSH Identifier</rdfs:label>
      </rdf:Description>
    </dc:identifier>
  </rdf:Description>
</rdf:RDF>


Andy.

[1] http://www.w3.org/RDF/Validator/

Received on Tuesday, 7 September 2004 14:13:48 UTC