AFAIK, when a document gets added to a collection, by default a HTML
version of it is generated, which it's included as a tag in the xml file
along with the metadata. This is what we call "the Greenstone version"
(GV). The collection maintainer decides if this version is to be viewable
on the browsers, or should be kept hidden from the masses.
The OAI-PMH protocol harvests this metadata and provides total access to the
full document, when applicable, through the dc.identifier tag. I think
that, when metadata describe a digital library resource, a link to the
full document should always be included.
But then a question arises: Should this link refer to the original source
document (doc, pdf, etc), to the GV (this implies the referral happens in
the context of the collection), or to both of them?
According to DCMI specification, the dc.identifier tag is, I quote, "An
unambiguous reference to the resource within a given context". If this is
so, the context would be the collection (where the given resource is
included) and then, the URL should be the GV one.
I think the better option is to include in the Greenstone xml two
additional tags that will each map to it's dc.identifier when configuring
the OAI server, both to refer to the GV and to the source document. That
way, we let on the maintainer the decision, for each one of them, as to
include it or not. For example, if you use a external URL for link the
digital resource, then you have no need to map anyone of them...
I hope this be as painlessly doable as possible... It's better said than
done, I've heard (and since my english is on their way, I'm not even sure
of my saying 8-0).
Lic. Mariana Pichinini
BIBHUMA - Biblioteca Profesor Guillermo Obiols
Facultad de Humanidades y Ciencias de la Educaci?n
Universidad Nacional de La Plata
Calle 48 entre 6 y 7 - 1er subsuelo
B1900AMW LA PLATA, Argentina
Telefax: +54-221-4230125 interno 162 (l?neas rotativas)
> Hi there
> Several people have wanted URLs to the document in a dc.identifier
> I have been working on this, and have added code to generate a URL for
documents that have a source doc, eg word, pdf etc - i.e. ones where in
greenstone you'd use srclink to link to the original document.
> What do people want for documents where the source doc is not stored, eg
text and html? Three options I can think of:
> * no link - but then does this mean it won't validate in some
> * Link to the greenstone version? eg
> * option when building to store the original (in index/assoc), then link
to it like for word/pdf docs.
> Any other ideas?
> Any feedback appreciated.
> greenstone-users mailing list