[I've altered the subject line here as I think we've moved away from the
LDAP document and I'm splitting my response into separate threads
because I think the core issue of the "model" for simple DC is worth
focusing on separately. Hope that's reasonable.]
Roland,
> When i then look at dc.xsd there is an attributeGroup
> xlink:metadataLink specified for the CCP,
> dc:source/dc:relation elements, which is not reflected in the
> abstract model for dcsimple - nor in the one of DC with qualifiers.
>
> So i'm don't quite get, what the xml-guide-dcsimple actually
> is. Seems there is something aside of literal string values,
> what is important. Why the xlink's have no arcrole ?? Is
> there a relation with "Harvesting RDF statements from xlinks" ??
Andy has (just after your message, I think!) posted a new version of the
guidelines document in which we've removed the references to using
Xlink.
Yes, I did look at "Harvesting RDF statements from Xlinks" and it was
partly trying to reconcile what we had suggested with the sort of
mappings in that document which made me feel we were not on solid
ground.
> [By the way: Why the "qualified" version doesn't
> use xml-schema typing and instead introduces
> new dcxml xml-attributes?
I think this is probably the one aspect of the latest version of the
document which I feel slightly uncertain about (sorry, Andy!)
Based on some of your (Roland's) earlier comments, I have a feeling that
data typing is probably the way to go to express "encoding schemes" so
instead of the proposed convention of
<dc:subject scheme="LCSH">blah blah</dc:subject>
we'd have something like
<dc:subject xsi:type="dcterms:LCSH">blah blah</dc:subject>
(Can I just confirm that is your suggestion, please, Roland?)
I guess the reason that it isn't in this version is that
(a) using a "scheme" attribute (we've now abandoned the dcxml namespace
as unnecessary) seemed intuitive for a human author/reader. I must admit
I fear awful confusion over explaining the differences between xsi:type,
dc:type and rdf:type, _but_ having said that, if we need to argue that
case, then I think we should do so.
(b) to my frustration, I just haven't had time to pursue the "encoding
scheme as data type" option in the XML Schemas as I wanted to do (we've
removed the reference to the XML Schemas from this recent version)
> I'm rather puzzled about the "xml id" attribute mentioned in
> the draft. I can't find an explanation for dc-elements with
> the "xml id" attribute in the draft. some more miracles in
> xml-schema? ]
That's been removed from this version too, on the basis that it was
related to the use of Xlink.
Pete
|