At 13:30 22-05-2003 -0400, Larry E. Dixson wrote:
>It wasn't clear to me if, after discussions, this break-out group was as
>enthusiastic about SRW/SRU as this quote, in a paper prepared
>before the meeting, would indicate. However, perhaps Sebastian has some
>thoughts about that.
I think "enthusiastic" would be an awfully strong word. Although we really
didn't get as far as recommending specific technologies, I would say that I
sensed three distinct schools of thought among the members of the search
a) SRW has most of what's needed, and since it harkens from Z39.50, it has
a good NISO pedigree -- it can be used if the SRW "community" is open to a
discussion about the specific requirements posed by the content vendors.
b) Let's roll a new IR protocol.
c) Let the content providers provide access to their content in any old way
they please, but let's recommend that they use web services technology and
clearly document their interface.
Because the chairman of the group believed that SRW would be a good option,
I think we were leaning slightly in that direction, but it's not a sure thing.
It's important to remember that this is a very different crowd of people
from your usual library interop folks... the content vendors are just now
grudgingly accepting the need for some form of standardisation in the area
because they find that the vast increase in popularity of metasearch
solutions over the past 18 months is negatively affecting the performance
of their systems.. but they don't necessarily come from a culture steeped
in open IR ideas.
The metasearch people, on the other hand, already have to search in all
kinds of different systems, some of which support Z39.50, but many of which
don't.. in some cases, brutal "screen-scraping" of HTML-based user
interfaces is the only way forward, and pretty much *any* protocol is going
to seem like a vast improvement over that... Z39.50, SRW, SRU, XQuery
subsets --- they all just become different back-ends in cross-searching
systems that, by design, are not picky about how they get hold of their
data... the reality is that the business case for pushing hard for a single
IR standard is reduced in that context... the users who're paying for the
party and hoping for maximum interoperability may feel different, but I
think they were a little under-represented at the NISO workshop.
Sebastian Hammer, Index Data <http://www.indexdata.dk/>
Ph: +45 3341 0100, Fax: +45 3341 0101