I gave a report on the progress of the DC CD WG to a recent meeting of
the NISO Metasearch Initiative. As part of that report I highlighted
that

(a) several of the terms (properties, classes/encoding schemes, terms
within classes/encoding schemes) proposed for use in the DC CD AP have
not yet been assigned persistent URIs and they can not be deployed until
this is resolved;

(b) for some of the terms that have been proposed to the Usage Board,
the DCMI Usage Board has indicated that it is probably not prepared to
assign DCMI-owned URIs to those terms - not because they are not
well-modelled or there is no evidence of their usefulness, but because
they were considered to be beyond the scope of the Usage Board's
responsibility. (This applies to the three vocabulary encoding schemes,
and their constituent instances, that were proposed in October 2004; and
also to the CLD Type Vocabulary class and its constituent
classes/types.)

I presented a few slides on the requirements for identifiers for
metadata terms, which I've broken out from the original presentation for
ease of reference:

HTML: <http://www.ukoln.ac.uk/metadata/nisomi/tg2/20050323/termid/>
Powerpoint 2000:
<http://www.ukoln.ac.uk/metadata/nisomi/tg2/20050323/termid/termid.ppt>

In those slides I was trying to separate out the two questions of

(a) the assignment of identifiers for the terms
(b) the long term maintenance of the relationship between those
identifiers and the terms that they denote i.e. the management of the
term "semantics" through time (clarifying/refining definitions, adding
additional information etc), ensuring they continue to resolve to some
appropriate representation etc etc etc

I think it may be possible to make some progress on (a), while
recognising that (b) is not yet resolved.

We could, for example, request a PURL top-level domain from OCLC's PURL
service and use that as the basis for assigning PURLs to terms, on the
basis that, in the short term, this WG - or some individual or
organisation on behalf of the WG - will manage those identifiers, with
the intention that we will seek another organisation to take on this
role in the longer-term.

A term to which we assign a URI (or a variants of such a term) now may
turn out to be the subject of a proposal to the Usage Board at a later
date, in which case the first term would continue to exist, but its use
might be deprecated in favour of the DCMI-owned term at that point.

I'm conscious that missing from step (a) above is the sort of rigorous
examination of the "modelling" of terms to which the Usage Board submits
proposals for terms. I must emphasise that the approach I propose here
is not intended to be a short circuit for that sort of rigour. We may
choose to assign URIs for terms that the UB has indicated that it
considers out of its scope. But it would be relatively easy for us to
assign a PURL to a term that was ill-defined or did not fit within the
framework of the DCMI Abstract Model, and we must seek to ensure that
does not happen. We should certainly not assign PURLs to terms which are
still "under development".

And we need to be careful not to give impression that these terms are
part of the DCMI metadata vocabularies/DCMI Namespaces.

I have no strong preferences for the string that should be used for the
PURL top-level domain. Given that we do not know who might be the
long-term owner we are not in a position to use that organisation's name
for this purpose, but PURL top-level domains can be any string (subject
to some character restrictions).

So (subject to availability - I haven't checked) possibilities might be:

dccdwg, as the basis of "namespaces" something like
http://purl.org/dccdwg/terms/ , http://purl.org/dccdwg/type/ , and term
URIs like http://purl.org/dccdwg/terms/dateContentsCreated and
http://purl.org/dccdwg/type/catalogue
collection (similarly)
colldesc (similarly)
cld (similarly)

The disadvantage of the first is that this WG is not going to exist in
perpetuity. That doesn't reduce the usefulness of the string as the
basis for constructing unique identifiers, or the continued usefulness
of those identifiers once this group no longer exists, but it may become
slightly misleading to a human reader.

As for the short-term maintenance, I am willing to ask for a PURL
top-level domain on behalf of this WG and to set up suitable redirects
to RDF/XML representations from a UKOLN server as terms are assigned
URIs. In principle, representations could be served from a DCMI server,
even if they aren't purl.org/dc PURLs, but I'm not sure whether this
might cause confusion (and/or contradict any of the small print in Usage
Board documentation!)

For the long-term maintenance... as I said in the slides, it requires a
group with awareness of the DC context, of the DCMI Abstract Model, and
probably of RDF and RDF Schema. We had some discussion of this in the
context of the NISO MI TG2 meeting, but I don't think we reached any
definite conclusions.

So....

(i) does this approach ("assign URIs now, while seeking long-term
maintenance agency") seem appropriate/feasible?
(ii) does it seem reasonable for us to request a PURL top-level domain
for this purpose?
(iii) do you have any preferences for the name of that top-level domain?
(iv) do you have any suggestions for the long-term maintenance agency
for the terms?

Pete
-------
Pete Johnston
Research Officer (Interoperability)
UKOLN, University of Bath, Bath BA2 7AY, UK
tel: +44 (0)1225 383619    fax: +44 (0)1225 386838
mailto:[log in to unmask]
http://www.ukoln.ac.uk/ukoln/staff/p.johnston/