Re: [RDFa] ISSUE-8: RDF containers in RDFa

On Thu, Jul 12, 2007 at 10:03:23AM -0500, Dan Connolly wrote:
> On Thu, 2007-07-12 at 07:03 -0700, Ben Adida wrote:
> > <ul about="#paper" rel="dc:creator">
> > <li>Mark</li>
> > <li content="Steven">Steve</li>
> > <li>Ralph</li>
> > </ul>
> > 
> > yields:
> > 
> > <#paper> dc:creator ["Mark", "Steven", "Ralph"] .
> 
> This design space can be tricky, so having good concrete
> examples is really valuable.
> 
> I wonder if dc:creator is a good use case. Does
> its range include collections? A quick search
> yields "An entity primarily responsible for making the resource."
> not "an entity or list of entities...".
> 
> http://dublincore.org/documents/dces/

>From a DCMI point of view, I see no room for doubt: the
definition refers to "an entity", not "or list of entities".

It is worth recalling that, in the early years of Dublin Core,
there was discussion about whether multiple values should be
represented in a single "field" (separated by punctuation)
or in multiple, repeated fields.  Both methods were used,
though the repetition of "fields" soon became the recommended
method.

Ever since the "core" Dublin Core properties such as dc:creator
were declared as RDF properties in the late 1990s, their
ranges have remained unspecified.  DCMI now wants to do the
right thing and assign ranges.

DCMI is currently holding a Public Comment [1] on a proposal
for domains and ranges for DCMI properties [2].  Part of the
proposal is to define a class, dcterms:Agent -- "A resource
that acts or has the power to act" -- to be declared as
http://purl.org/dc/terms/Agent.

Since dc:creator has for years been used with
literal values, DCMI does not want to contradict
these triples with a range specification.  Therefore,
the proposal on the table is to create dcterms:creator
(http://purl.org/dc/terms/creator) as a subPropertyOf
dc:creator (http://purl.org/dc/elements/1.1/creator) and
to assign the range of dcterms:Agent to dcterms:creator,
leaving the range of dc:creator unspecified.  This action
also has the intended beneficial side effect of putting all
general-use DCMI properties into one namespace.

Following this proposal,

    <#paper> dc:creator "Mark" 

would remain valid, though the intention, over time, is
to promote the use of the more precisely defined dcterms:
properties.  It is worth considering whether the RDFa specs
should continue to use the well-known legacy dc: properties
or help promote the dcterms: properties by using them in
the examples.

> A common misconception is that
> 
>   <#paper> dc:creator ("Mark" "Steven" "Ralph").
> 
> somehow automatically corresponds to
> 
>   <#paper> dc:creator "Mark", "Steven", "Ralph".
> 
> so that "Mark" would get bound to ?who in
> queries like
>   SELECT ?who WHERE { <#paper> dc:creator ?who }.
> 
> This is not so.

This is an important point generally, and of course is not
specific to dc:creator.

I leave for vacation soon so may not be able to respond to
further discussion in this thread for awhile.  If anyone has
general feedback on the DCMI proposal, I would encourage them
to please suggestions or reactions during the current comment
period [1] for consideration at DC-2007 in Singapore [3].

Tom

[1] http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0707&L=dc-architecture&P=291
[2] http://dublincore.org/documents/2007/07/02/domain-range/
[3] http://www.dc2007.sg/

-- 
Tom Baker - tbaker@tbaker.de - baker@sub.uni-goettingen.de

Received on Wednesday, 18 July 2007 12:26:18 UTC