[OAI-implementers] multiple supporting filestreams
jose.borbinha at bn.pt
Thu Mar 2 12:19:47 EST 2006
We had to face already that problem at the National Digital Library in
Portugal. Our infrastructure is "METS based", so ultimately soon or latter
all those objects will be properly described in that way, but until now we
could not develop yet solutions for all the genres (we have thesis, but also
lots of other genres...). And sometimes the objects can not wait...
The pragmatic solution we are using now is simply to create an HTML
"home-page" for each of those objects, which only link to each file. When
possible, we add also some descriptive information to the top of the page
(title, author, ..). Than we create a metadata record pointing to that HTML
file. If the object has PDFs, sound, etc., we identify it as a HTML MIME
type object. If it is just a group of PDF files, or Word, etc., we identify
according to its specific MIME object. By default, it is a HTML object!!!
We use the same solution for objects that arrive to us in multiple CDs or
DVDs, for example, and that we want to put on-line.
Here is an example that I found fast (with a fancy webpage... usually they
are not so nice):
Prof. José Borbinha (jlb at ist.utl.pt)
DEI/IST/UTL - Departamento de Engenharia Informática (Department of
Information Systems and Computer Engineering)
INESC-ID - Grupo de Sistemas de Informação (Information Systems Group)
De: oai-implementers-bounces at openarchives.org
[mailto:oai-implementers-bounces at openarchives.org] Em nome de Jody DeRidder
Enviada: quinta-feira, 2 de Março de 2006 15:10
Para: OAI-implementers at openarchives.org
Assunto: [OAI-implementers] multiple supporting filestreams
I have no doubt that someone has already addressed this, so I would
being pointed to where it was hashed out...
I just discovered that we have a number of ETDs that have a ton of
filestreams (three of them have hundreds. Seriously!) There is nothing in
the incoming metadata to indicate that these filestreams exist, much less
their mime types are, or relationship to the primary filestream (a pdf). I
was informed that the students have been instructed to link these streams as
relative URLs within their pdfs, but there is nothing in place to ensure
all files are linked, or linked correctly. And if the pdf is "harvested"
then accessed out of context, of course none of those links will work.
In the OAI records I've been creating, I've simply referred to the pdf file
for type, format, and identifier. What *should* I be doing here?
I have no administrative, structural, or preservation metadata for the
supporting filestreams. It would be simple enough to create redirect links
(my local version of sustainable URLs) for each bitstream, and add those
the appropriate OAI records as dc:identifiers, but good gracious! Hundreds?
Best practice is one! Plus there would still be nothing to ID the
or relationship to the thesis itself.
OAI-implementers mailing list
List information, archives, preferences and to unsubscribe:
More information about the OAI-implementers