On Fri, Sep 27, 2002 at 04:05:32PM +0200, Sigfrid Lundberg, NetLab wrote:
> I agree with Andy's two main points
>
> - Element refinement should apply to one element only
>
> - Of all things the agentRoles could dumb down to, dc:contributor
> is the most reasonable one
I would like to join Sigge, Thomas Habing, Carl, and Diane
in agreeing with these points. However, I lean towards the
more extreme variant suggested by Andy:
>4) We could go further and say that all agent 'roles' that have
>to do with contributing to the content of the resource become
>element refinements of dc:contributor. I.e. that dc:creator
>becomes one of a potentially long list of dc:contributor
>element refinements and simply sits alongside things like
>dcterms:author and dcterms:illustrator.
This would spare us from having to declare and explain parallel
terms such as "illustratorCreator" and "illustratorContributor"
-- I don't think we want to go there.
It is worth recalling that back in November 1998 the Data
Model group thought that all three CCP elements should be
qualified by a vocabulary of "roles" (see [1]):
>16 An element qualifier for the dc:Contributor, dc:Creator
> and dc:Publisher elements, which qualifies those elements by
> specifying the contributory role, is dcq:AgentRole.
Grammatically, these roles were seen as "adjective-like"
inasmuch they would adhere to any of the CCP elements.
They were to be encoded as extra information associated with
an element in the instance metadata -- information that could
simply be ignored or discarded for dumb-down. At the time,
one important issue was recognized to be [2]:
>The blessing of particular external controlled vocabularies
>for values of qualifiers. For example, do we 'bless'
>the USMARC Relator codes for use with the dcq:AgentRole
>qualifier? If so, how?
Since then -- perhaps influenced conceptually by RDF? --
we have moved away from seeing refinements as extra
adjectives towards seeing them as sub-properties -- i.e.,
elements in their own right -- and this leads us to posit
that each refinement should refine just one parent element.
This being the case, I should think we would want to avoid
declaring refinements more than one level deep, especially
if the basic distinctions between C, C, and P are at all
controversial. Making all AgentRoles sub-properties of just
dc:contributor makes sense in this regard. I should think it
would also make it simpler to solve the problem of blessing
external vocabularies.
Would doing this create any problems for existing
implementations?
Tom
[1] http://www.jiscmail.ac.uk/cgi-bin/wa.exe?A2=ind9811&L=dc-datamodel&T=0&P=1178
[2] http://www.mailbase.ac.uk/lists/dc-datamodel/files/issues.html
--
Dr. Thomas Baker [log in to unmask]
Institutszentrum Schloss Birlinghoven mobile +49-171-408-5784
Fraunhofer-Gesellschaft work +49-30-8109-9027
53754 Sankt Augustin, Germany fax +49-2241-144-1408
|