17.08.2011 19:56, James Weinheimer:
> On 17/08/2011 11:33, Bernhard Eversberg wrote:
> <snip>
>> We need to think in broader and more general terms, obviously, but
>> not too general and too grandiloquent either. Down to earth but
>> off the beaten track into all the accessible territory. But I
>> can't think of an easy catchphrase to sum it all up.
>> Therefore, to get real now, how about these headlines, for a beginning:
>>
>> What should catalogs do?
>> -- Produce reliable results ...
>> ... etc.
> </snip>
>
> While I agree with this, I keep telling myself that I am a librarian,
> and I ask: is this what the *patrons* really want?
They want ever more.
What I was addressing was more the technicalities and the infrastructure
rather than further-reaching wishful thinking. We will have to clear all
the technical stuff away first, and then build on the results. We will
first need more machine-actionable data (and less amorphous textual stuff)
and we need more linkable, web-serviceable foundations, and then build
something new on top of all that. Your ideas, IOW, populate a higher
level than my more basic requirements list.
Or has someone proposed a "Catalog without rules" yet? Or a format-free
database? That's what GBS already is? Not really! Some of the reliability
of searching that it features is grounded in the library data they have
integrated.
And this brings us to your remark:
> For instance, everyone will agree that they want "reliable" results
> (who would prefer "unreliable results"?)
I should have made that clearer. "Reliable search" means that one can
always know exactly *how* to search in order to get a definite
"yes" or "no" for an answer. The card catalog had that quality, but
the *how* of searching was just much too restricted and arcane.
You may say that's a librarian's demand, but known item searches
are quite common and need to be supported in better ways than just
ISBN access, now that we have megatons of old material and other
stuff without such numbers. Web services that start from something
else, found somewhere on the web, to find some library connection,
depend exclusively on ISBN/ISBN now, and that must change.
Maybe I should have said "bull's-eye" searching or something.
The human searcher, the naturally intelligent one, can do highly
successful known-items on GBS, making use of phrases and "intitle:"
and "inauthor:" and so on and so further, depending on the situation.
But the artificially intelligent web service cannot do this because
it cannot analyse the situation in the same way. In the absence of
more and better machine-actionable and machine-analysable data.
B.Eversberg
Received on Thu Aug 18 2011 - 02:51:41 EDT