| I was curious as to whether you had sufficient influence
| over/co-ordination with your targets that you had been able
| to get them
| all to implement a common interface (or a number of common interfaces),
| but it sounds like that isn't the case.
No, and in the global info world is not very likely - its hard enough to do
it within a single large (in terms of data held!) academic institution.
| Where you take the "screen-scraping" approach, do you do co-ordinate
| that with the target service providers? i.e. they are happy that your
| service gets data in this way (and they don't change their query
| interface or the HTML output every few months without warning you!)
Obviously, an agreement and notification periods are needed. We'd always ask
for a more efficient interface such as SOAP of course, but in many cases
that's not likely to happen on 'legacy systems' - i.e. anything not actually
being built!
| Also, I have in interest in collection-level description. I'm
| aware that
| the NHM did a lot of work on CLD and I can see your
| collection navigator
| service as well as the data navigator service. I was wondering if you
| made any use of CLD to allow users to select targets for searching (or
| to allow the portal to present them with a "landscape" of relevant
| collections)?
You can select databases in the Data Locator. In the Navigator the plan is
to embed direct references to Data Locator searches within the CLDs of the
relevant collection, so yes, indirectly.
The old Data Locator did have a high level browse, but it had many problems
so we dumped it. Too many collections crossed our artificial categories.
However, not many of the databases in the Locator have CLDs: that's a little
strange, but given resources we need to focus the CLD project on areas
/poorly covered/ by other information.
| > Thus of course the dynamic search is
| > limited to the speed of the remote web application
|
| Right. Is that a problem? How many remote targets do you search at a
| time? I think you said the service used OAI and cross-search so do you
| aggregate some of the remote data?
Untested beyond prototypes, but searches are concurrent. Currently we don't
use the '3 tier' search (see the presentation) on remote databases, we have
no need to. However, the next stage will include these. When we do include
our Library system in the dynamic search it slows down a LOT, due to the
speed of Unicorn, so we will use regular OAI harvesting on that one.
For the upcoming 'global potatoes' project we are in the 'get agreements
phase'...
| If the answer is, "read my paper/presentation" over there, that's fine
| ;-)
I did do an overview / intro to this for the Institutional Web Manager's
conf last year but I was too ill to deliver and so its not on the IWMW site.
Here it is for those who have read this far...
http://internt.nhm.ac.uk/nof/crossdb/
(link to 830k PowerPoint)
cheers
Mike
|