Howard Noble wrote:
>As a slight aside, why might an OAI repository object to opening a public
>z39.50 or SRW interface?
>
>
Well in Z39.50 there's the whole stateful connection thing.
SRW addresses that issue (note that it wasn't available to us when
eduSource was being developed). It actually can look and feel to a user
very much like an OAI interface.
But now it becomes a question like this: if you have a web server, why
would you install a second web server to serve the same pages?
And I think at this point it now becomes a question of philosophy.
- The SRW approach feels like the 'live search' approach - somebody
conducts a search centrally, and the search is propogated through the
network (in other words, it feels like a federated search approach (I
say 'feels like' because I don't think it entails such an approach, just
that its design and structure seem to suggest it)
- The OAI approach, by contracst, is geared much more to a harvest
approach. You grab the records, then you search them at your place.
Insofar as search exists, it's almost as an afterthought.
Now, in addition to the hassle of running and maintaining extra code
(the second repository engine), creating a federated search imposes a
significant burden on a repository, espcially if the repository is small
and the federation is large. It's a lot less of a load to simply make
records available for occasional harvesting than to facilitate every
search that comes your way through the network.
Moreover, as the purpose of a federated search is typically not merely
to create a network but also to control access to parts of that network,
the host of a repository may not be able to provide any access to the
remainder of the network without providing user authentication. Thus,
the OAI repository is now in the position of either contributing to a
network it can't really benefit from, or assuming an overhead that is
not only more difficult but also adds an extra burden to its existing
users.
None of these bad things *follow* - you can run SRW like OAI. But it
seems to me that it would feel like the thin edge of the wedge. Your
mileage may vary.
>Stephen - was anything written up on the experiences in Canada?
>
>
Not really. Most of these debates occurred in closed meetings and much
of what was said can't be substantiated through links or references,
which makes writing them up a marginal proposition at best.
Some good links to the planning documents are here (I was involved in
writing most of these):
http://www.edusource.ca/english/documents_key_eng.html
-- Stephen
>Howard
>
>----- Original Message -----
>From: "Phil Barker" <[log in to unmask]>
>To: <[log in to unmask]>
>Sent: Friday, October 07, 2005 11:31 AM
>Subject: Re: registries of repositories
>
>
>
>
>>Hello,
>>
>>Stephen Downes wrote:
>>
>>
>>>OAI repositories already expose their metadata through the widely known
>>>and widely used OAI Harvesting Protocol, and are registered on sites
>>>like OAIster. They might reasonably ask, why don't you simply harvest
>>>metadata from them using their existing protocol?
>>>
>>>
>>>
>>The SPP project's cross search makes use of metadata in OAI-able
>>repositories that have no Z39.50 interface by harvesting the metadata
>>and storing it in a local database that has a Z39.50 interface. I don't
>>know that the Z target of the harvested metadata can be made available
>>to third parties, other people on this list might know whether the
>>technology and harvesting agreements would cover that.
>>
>>Phil
>>
>>
>>More on SPP at http://www.portal.ac.uk/spp/
>>An example of a cross search using the SPP portal software at
>>http://www.eevlxtra.ac.uk
>>
>>--
>>Phil Barker Learning Technology Adviser
>> ICBL, School of Mathematical and Computer Sciences
>> Mountbatten Building, Heriot-Watt University,
>> Edinburgh, EH14 4AS
>> Tel: 0131 451 3278 Fax: 0131 451 3327
>> Web: http://www.icbl.hw.ac.uk/~philb/
>>
>>
>>
>>
--
Stephen Downes ~ Research Officer ~ National Research Council Canada
http://www.downes.ca ~ [log in to unmask] __\|/__ Free Learning
|