[For some reason, my previous attempt at sending this doesn't seem to have
gone to the list... my apologies if you get it twice]

Harry said:

> The long-term solution is to not use the schemas as input to the
> registry.
>
> A better solution is to maintain the DC vocabulary in the vocabulary
> management system and provide an application interface for the registry
> to access this info.

I'm afraid I'm really misunderstanding something, because I don't
understand this comment, and it seems to me to contradict what I thought
the DCMI registry was aiming to provide.

Firstly, I _do_ strongly agree with your statement that we need appropriate
tools for managing structured information about the semantics of the Dublin
Core terms i.e. a Vocabulary Management Tool/System - whether those tools
are based on Tom's text files and scripts, XML files, or a relational
database application is an implementation-level decision.

The RDF Model and Syntax and the RDF Schema vocabulary provide means of
exposing/publishing _representations_ of the information managed/stored in
that Vocabulary Management Tool/System in a commonly understood, machine-
understandable form. Those W3C specifications allow DCMI to expose
information about its terms and vocabularies and they allow the managers of
other vocabularies to expose information about their terms and vocabularies
in a decentralised, distributed form.

There will be other registries (not to mention other applications) which
access the machine-readable representations of the DCMI vocabularies
(the "DCMI RDF schemas"). But perhaps more importantly for my argument
here, I believe that sooner rather than later it will be desirable for the
DCMI registry to be able to access/read/index/display the machine-readable
representations of vocabularies owned by other organisations.

It seems to me (apologies if I'm wrong) that you are arguing that the
registry should adopt a proprietary interface "hard-wired" to whatever
system DCMI adopts as a Vocabulary Management System. But that would make
the DCMI registry a "closed" system in terms of its inputs. I don't see how
this interface will work with other non-DCMI "Vocabulary Management
Systems/Tools".

Surely this is precisely what RDF and RDF Schema are for? This is why
Roland and others have worked _so_ hard to ensure that we _do_ have good
representations of the DCMI terms and vocabularies using the open, commonly-
understood forms provided by RDF, RDF Schema and XML.

Publishing representations of DCMI terms/vocabularies in this form makes
them available to other non-DCMI applications (including other registries);
likewise if the DCMI registry adopts this form as its input mechanism it
remains "open" and can read/index/display the machine-readable
represenations published by others.

I guess basically I feel that the view of the DCMI registry as a "closed
world" which takes its inputs _only_ from the DCMI "Vocabulary Management
System" is too narrow. The registry needs to be much more "open" in terms
of its potential inputs, and (it seems to me) support for reading/indexing
RDF Schema-based representations of terms and vocabularies is essential for
that.

Pete, not really at work but unable to sit this out... ;-)