At 23:19 26-05-2003 +0100, Matthew Dovey wrote:
>What worries me about the multiple database requirement (rather than the
>technical rights, wrongs and technicalities) is whether this will actually
>be implemented. When I was working with the JISC "Clump" project back in the
>late-90's, the toolkits being used tended not to make use of the capability
>of searhcing multiple databases in a single query, and few, if any, of the
>servers being search actually supported it.
The notion in the context of the NISO workshop was that the commercial
vendors, along with their target info, would post usage guidelines which
might include instructions that multiple-database searches should happen
using the appropriate protocol mechanism rather than through separate
accesses. The assumption would be that the metasearch engines, playing by
something similar to the "robot ethics" rules of web harvesters, would
respect any such requests from the server operator...
I agree the multiple-database function was never widely supported by
servers in Z39.50, and for clients it was frequently not used because there
was a user requirement to show individual hit-counts for individual databases.
The difference may be that today, large server operators are probably more
likely to implement such a feature because they are very interested in
reducing the load on their servers from the metasearch engines.
The question is whether we can specify a sufficiently powerful
multi-database mechanism without throwing the simplicity of the protocol
Sebastian Hammer, Index Data <http://www.indexdata.dk/>
Ph: +45 3341 0100, Fax: +45 3341 0101