On Thu, 7 Jun 2001, Thomas Baker wrote:
>
> I regard our current Dublin Core as a strawman pidgin -- a very helpful
> start. But over time a living pidgin should emerge as the set of
> elements most commonly used for interoperability. Over time I would
> expect to see some of the properties in the "lower-level" namespace to
> become more popular and salient than one or two of the properties in
> the "top-level" namespace.
This may be true regarding domain specific elements (tho it would imply
there has to be coherency between all domain specific extensions and
between all of them and 'top-level' which I have not seen stated). How
well will it work with other hierarchical relationships between
concepts? In DC we have one hierarchical structure i.e. qualifier.
A metadata creator may create instance metadata using a DC 'metadata
package' which includes this relationship. The owner/user of that metadata
may not want that relationship encapsulated in their metadata, as
expressed in the original metadata package, to be changed over time by
changes to a DCMI controlled vocabulary.
On the other hand I accept it is a powerful way of doing away with
conversions of metadata to new versions over time....! I guess to use your
analogy of dictionary, it would be interesting to consider whether the
dictionary ( ie DCMI vocabularies) will stay coherent, up to date, easy to
understand. On the other hand the language in a book written 50 years ago
is dated, can be difficult to understand (DC instance metadata may be
difficult to interpret after N internet years).
Recognizing and ratifying such natural
> evolutions in status should not require us to move terms unnecessarily
> between namespaces. Nor should the status of a term at the moment it
> is added to the DCMI vocabulary be fixed in the namespace URI it was
> assigned at the time -- while that status should change, the unique
> identifier should not.
I think you are assuming that one 'concept' will only ever have one unique
id (namespace URI). Already this is not the case between differently owned
metadata packages (e.g. the concept of creator appears in IMS and MARC and
DC etc). These will all have different namespace URIs. I think what we are
trying to solve within DC can be compared to this situation where the
relation between concepts to namespaces to schemas is possibly one to many
to many.
This follows brief reading of messages on xml-dev list from Michael
Sperberg-McQueen
http://lists.xml.org/archives/xml-dev/200105/msg00136.html
http://lists.xml.org/archives/xml-dev/200105/msg00157.html
I am sure one can 'make things work' whatever the decision re namespaces,
it is just useful to think where the 'cost' will fall on such things as
- effort in evolving domain specific vocabs
- effort in updating instance metadata
- ensuring integrity of instance metadata
- etc etc
Rachel
>
> Tom
>
> _______________________________________________________________________________
> Dr. Thomas Baker [log in to unmask]
> GMD Library
> Schloss Birlinghoven +49-2241-14-2352
> 53754 Sankt Augustin, Germany fax +49-2241-14-2619
>
---------------------------------------------------------------------------
Rachel Heery
UKOLN (UK Office for Library and Information Networking)
University of Bath tel: +44 (0)1225 826724
Bath, BA2 7AY, UK fax: +44 (0)1225 826838
http://www.ukoln.ac.uk/
|