[OAI-implementers] Aggregators and querying by repository identifier?
Wed, 06 Mar 2002 18:55:35 -0500
i don't know how other people do this, but i prepend the repository id
to all set names. so, for example, if i harvest the set "All" from VTETD
and everything from CaltechETD, i end up with the set structure:
i interpret sets as being a method of structuring the collection, and
with a plain vanilla union collection (as opposed to one that does
merging of categorical sets) there is probably no better structure than
the individual sub-collections
of course this is not a general solution, but it works for us :) and our
"aggregator" has been running for about 9 months now. see the ODL
website (http://oai.dlib.vt.edu/odl) for more details - specifically,
read the tech report.
Alan Kent wrote:
> My understanding of OAI currently is that you can query by from/until
> and by set, but can you query by repository identifier?
> With the advent of aggregators, I am wondering if people are going
> to want to be able to harvest from an aggregator selectivly based
> on repository identifiers. Before aggregators it was not an issue
> (a repository contained records of that repository). With aggregators,
> my understanding is records will retain the OAI-identifier (and
> hence repository identifier?) of the original repository, so a
> aggregator will hold records from lots of different repositories.
> I was thinking that sets should not be used, as the aggregator should
> probably try to preserve the sets of the original data (have not
> looked into if this is possible yet).
> Anyone have any thoughts on this? Is there a need, or is it an overkill?
> Just thinking about how I might try to implement the DP part of OAI
> (on leave next week, so don't expect anything soon!), and if I did
> so, I would probably make it dish up the OAI data that I have harvested
> from other sites (ie: and aggregator).
> OAI-implementers mailing list
hussein suleman - email@example.com - vtcs - http://www.husseinsspace.com