Weinheimer Jim schrieb:
>
> The way we achieve these goals today (our basic tools)
> hasn't really changed from what they were in the 19th
> century, but those tools are completely obsolete today.
> For example, the "heading" no longer needs to be a text
> string, but a relational number.
Well, "relational" is a misleading term here, it is the
IdNumber of an authority record. This has nothing to do
with relational databases, and esp. you don't need that kind
of tool to make use of such numbers - in fact, RDBs are
not any better at that than other database models.
> to our users today, which will allow them to search materials
> *reliably* by authors, titles, and subjects?
>
Which indeed is far from possible in current G.Booksearch.
I'd like to add: The catalog should reliably bring the parts
of a series together. Or should we say, all parts of a
multipart publication? LibraryThing has learnt the importance
of this the hard way but they found their own solution, with
their own concept of "series", and not a bad one, at least one that
shows what users want and apparently enjoy to have. The solution,
however, was not to be achieved without input labor, and that is
something Google is not willing to invest. The (MARC?) data they get
from their library partners is probably insufficient.
Another thing is the collocation of editions. Also something LT
peferred to find out for themselves, and did, inventing the "Canonical
title" rather than the (more formally defined) "Uniform title".
GB is trying to collocate editions by automated means, with doubtful
success. There can be no speaking of *reliable* collocation. Viewed
with FRBR aspirations in mind, horrible.
B.Eversberg
Received on Tue Sep 30 2008 - 03:26:11 EDT