[OAI-implementers] ListMetadataFormats problem
Wed, 31 Jan 2001 11:56:49 +0000 (GMT)
On Tue, 30 Jan 2001, Young,Jeff wrote:
> I'm working on an OAI repository for a group that wants to define it's own
> metadata format. The trick is, this format will be a combination of Dublin
> Core elements and some new elements yet to be defined. The examples in the
> OAI specs, however, all assume that the metadata will consist of elements
> from a single namespace. I believe, for the most part, that the OAI spec
> doesn't preclude the use of multiple namespaces. For example, I imagine that
> the following XML fragment is likely to be acceptable:
> <combined xmlns:dc="http://purl.org/dc/elements/1.1/"
> <dc:title>This is the title</dc:title>
> (As far as I can tell, specifying multiple namespaces in the
> xsi:schemaLocation attribute is perfectly valid.)
I wonder whether the following is reasonable (perhaps not appropriate to
the issue at hand):
<identifier>(journal) Nucl. Phys. xyz</identifier>
Be more sensible?
My reasoning is adding new types of metadata sets means that a harvester
must be specifically written to understand that data set to extract any
useful information. If a backwards compatible system can be used with DC
then harvesters need not be specialised, but archives are given the
opportunity to qualify their responses. For example a harvester could
amalgamate all <description> fields, or, if the harvester supports
differential description fields it can take advantage of archives that
provide that information.
<relation type="shoesize"> anyone?
Computer Science, University of Southampton