I'm convinced that Jane is right in stating that the schema are not independent. A base schema, expressing the set of 15 dc-elements and their base types, should be developed with both unqualified "appropriate literal" Dublin Core and qualified Dublin Core - whatever that means in its xml expression. I'm certainly not qualified to come up with the exact syntax of such as schema, having only dabbled in xml schema land. Jane, you have much more experience in this area than most of us and probobly can help us come up with something.
As for an interim schema, allow me first to put on my OAI hat - Our main interest in the OAI protocol is to have a schema that restricts a "dc record" to the 15 elements with values restricted to strings. This is our mandatory pidgin metadata format that all data providers must export. (We make no policy statement regarding qualified DC or applcation profiles, which may be provided as parallel metadata forms in OAI). Version 1.x protocol did quite well creating an OAI-specific version of such a schema. When we started work on version 2.x of the protocol, now in alpha, we thought it would make sense politically to derive the definitions of the DC elements from a DCMI supported schema, realizing that we'd have to import such definitions into our specific record format schema - thus, the beginning of these discussions. If defining an interim base schema for DC means that we'll need an interim deriving schema in OAI, resulting in interim conformant DC records in OAI - I don't think that makes sense. We need to release version 2.x of OAI-PMH on May 1 and really want it to be stable. I'd much rather have OAI define its own schema for DC and give up on this joint politically motivated effort. On the other hand, if it is possible to build a stable OAI schema on top of an interim (unstable) dc schema, then my OAI hat says "who cares". Jane, I leave it up to you to tell us whether this is possible.
Now, putting my DC hat on - interim things have a funny habit of becoming legacy. As I stated in an earlier message, if we define an interim schema that says in effect "dc elements may have any uncontrolled substructure as their values" we have effectively let the cat out of the bag. People will be able to create well-formed and valid dc instances that are not logical from the dc point of view. Thus, unless Jane you have some other way of scoping this value space (you may since you know much more about this world than I), I don't think that the interim schema approach just for OAI makes sense. After all, as said above, the import by OAI from a dc controlled schema is largely a political rather than a technical issues.
Hope this all makes sense.
Carl
> -----Original Message-----
> From: Jane Hunter [mailto:[log in to unmask]]
> Sent: Thursday, March 07, 2002 7:22 PM
> To: [log in to unmask]
> Subject: Re: Public Comment on DC-simple XML Schema declaration within
> OAI
>
>
> I think we all agree that we need to define XML Schemas for both the
> DCMES and also Qualified DC - but its becoming obvious that the two
> schema designs are not independent of each other and so should be done
> in parallel. The availability of the Qualified DC XML Schema will also
> effect the approaches taken by implementors of application profiles -
> and hopefully act as a guide to appropriate dumbing-down.
>
> But if we do the design of both schemas in parallel, then its unlikely
> we can have them both right for OAI by March 15.
>
> But I'd be happy if we could come up with an interim DCMES XML schema
> for OAI by March 15, providing its fairly broad and innocuous.
>
> jane
>
>
> > To me it seems that a lot of people are screaming for advice on "How
> > should we do qualified DC in XML", and since people don't seem to be
> > content (don't understand why, though) with doing that in
> RDF/XML in spite
> > of the fact that this encoding is of very good quality even
> if you ignore
> > the RDF side of it, I suppose we owe them that.
>
>
> jane
>
|