[OAI-implementers] oai_dc: Multiple identifiers and formats
within a single record?
Wed, 18 Dec 2002 23:38:22 +0000 (GMT)
On Wed, 18 Dec 2002, Tim Brody wrote:
> Probably the best solution would be to create a "jump-off" page with links
> to the abstract, and full-text formats. And it is this page that you include
> in your oai_dc identifier.
I tend to disagree... on two counts. 1) this makes it
difficult/impossible for harvesters to gather full content. 2) it is
questionable whether it is good practice to describe resource A (the
full-text or whatever) but provide a URL for resource B (the jump-off
My view is that the jump-off page is 'related' to the full-text, therefore
the URL for the jump-off page belongs in dc:relation (rather than in
> Alternatives might be to just include the URLs in identifier tags (any
> harvester can get the file type by performing an HTTP query), but this
> wouldn't include the "Full Text"/"Abstract" descriptions.
This seems to be a better approach to me.
My preferred approach (as suggested in the original message) would simply
be to make available separate oai_dc records for each format of the full
text. The downside is that the multiple records would carry significant
duplication (title, description, subvject, etc.), but each could carry the
correct format/identifier pair.
> Or you could do some kind of string encoding (e.g. eprints.org 2.0):
> <format>pdf http://xxx/yyy</format>
This usage does *not* fall within the semantics of dc:format. I would
therefore discourage it.
The bottom line is that we need to stick within the limitations of simple
DC when using the oai_dc format - which means that there will be some
things we cannot convey because the format is too simple! If we want to
provide richer metadata, then we have to use a richer metadata syntax -
whether that's RDF or METS or whatever.
Distributed Systems, UKOLN, University of Bath, Bath, BA2 7AY, UK
http://www.ukoln.ac.uk/ukoln/staff/a.powell +44 1225 383933
Resource Discovery Network http://www.rdn.ac.uk/