Re: Anyone able to help improve Dublin Core namespace XSLT to support RDFS RDFa?

On Fri, May 11, 2012 at 02:26:32PM -0400, Gregg Kellogg wrote:
> > I'm not sure what you mean by "reduced".  The Web document says:
> > 
> >    Term Name:    Agent
> >    URI:          http://purl.org/dc/terms/Agent
> >    Label:        Agent
> >    Definition:   A resource that acts or has the power to act.
> >    Comment:      Examples of Agent include person, organization, and software agent.
> >    Type of Term: Class
> >    Instance Of:  http://purl.org/dc/terms/AgentClass
> >    Version:      http://dublincore.org/usage/terms/history/#Agent-001
> > 
> > This matches the RDF/XML output except for the addition, in the RDF/XML, of
> > rdfs:isDefinedBy and of the date issued (of the individual property or class).
> 
> Sorry, I ment _reversed_. It looks like the <dcterms:description> is the
> comment and the <rdfs:comment> is the description in the RDF/XML output.

DCMI's use of rdfs:comment for the "Definition" goes back to the very first
experimental RDF schema of October 1997 [1].  The use of dcterms:description
for the "Comment" dates back to August 2002 [2].  What would you suggest as
best practice today?

[1] http://dublincore.org/1997/10/02-dces
[2] http://dublincore.org/2002/08/13/dces

> >> The HTML output ends up pulling in several definitions for
> >> http://purl.org/dc/dcam/VocabularyEncodingScheme. Perhaps this needs some
> >> additional filter to not define anything not in dcterms?
> > 
> > In the dcmi-terms/index.shtml output (RDFa-to-Turtle) I'm looking at, I only
> > see one entry for VocabularyEncodingScheme, as in [1]:
> > 
> >    <http://purl.org/dc/dcam/VocabularyEncodingScheme>
> >       dc:issued "2008-01-14";
> >       rdfs:identifier <http://purl.org/dc/dcam/VocabularyEncodingScheme>;
> >       rdfs:seeAlso <http://dublincore.org/documents/2007/06/04/abstract-model/>;
> >       rdf:type rdfs:Class;
> >       dc:hasVersion <http://dublincore.org/usage/terms/history/#VocabularyEncodingScheme-001> .
> > 
> > The rdfs:identifier is wrong, if only because rdfs:identifier does not exist...
> > 
> > Compared to the Web document [1], what is systematically missing in this
> > RDFa-to-Turtle output (and in the RDFa-to-Turtle output for all other
> > properties and classes) is:
> > 
> >    rdfs:label for the "Term Name"
> >    rdfs:comment for the "Definition"
> >    dcterms:description for the "Comment" (though not for "VocabularyEncodingScheme", which does not have one)
> > What is systematically missing compared to the RDF/XM
> 
> Corrected in my REPO. as is adding Date Issued and Date Modified to the
> tabluar definitions; it was explicitly removed before. If you'd like to keep
> it out of the term tables as hidden elements, I can go back to that.

Including the issued and modified dates for each property and class makes the
printout longer without adding information that will be of use to most readers
of the HTML.  Including the more fine-grained information such as dates, per
term, was always the function of the "history" document [3].  My impulse would
therefore be to keep those dates out of the term tables in the main DCMI
Metadata Terms document.  What do others think?

[3] http://dublincore.org/usage/terms/history/

> > Do I correctly understand that if the HTML/RDFa document completely expresses the 
> > contents of the current HTML and RDF/XML documents, and we direct the PURLs to that 
> > document, we could delete (or rather archive) the XSL scripts used to generate the
> > RDF/XML?  The other simplifications I would like to make are:
> > 
> > -- Delete (or archive) the XSL script for generating the (redundant) stand-alone
> >   DCMI Type Vocabulary document.
> > 
> > -- Revert to the pre-RDFa XSL script for generating the "history" document [2].
> > 
> > This means that only one XSL script for generating RDFa would need to be
> > maintained.
> 
> That's true, but there is value in providing alternate representations
> (perhaps using content negotiation), which should probably include Turtle.
> You could certainly go to auto-generating this from the HTML+RDFa, though.

Auto-generating the Turtle from HTML+RDFa seems like it would be less risky
because they would come from the same RDF source -- not from XML-markup source
via multiple scripts, all of which would have to be maintained and kept in
sync, as now...

Tom

-- 
Tom Baker <tom@tombaker.org>

Received on Friday, 11 May 2012 18:58:14 UTC