Quoting Jonathan Rochkind <rochk...@jhu.edu>:
You _can_ do things this way, out of neccesity, but it's definitely
not preferable from a data mangement point of view, right?
Jonathan, I didn't mean it as data management structure -- I meant it
as a data resource, in the same way that xISBN is a resource. You
don't expect to query xISBN for every transaction against your
database, but you can use it to enhance your data store and help you
make decisions. If at that point you want to add the "foreign key" to
your data, that depends on all of the usual criteria one applies to
such decisions.
We have a lot of information, collectively, that shouldn't have to be
re-done by every cataloger. In part, this is what folks are hoping
FRBR will help with, creating a pool of information about works and
expressions so catalogers don't have to supply that information
manually -- it will be accessible to the cataloging system and the
information can be pulled in to the record the cataloger is working
on, or can be accessed by an application that is doing record merging.
Whether or not it then makes sense to incorporate that information
locally, or to access it when needed, is a systems decision.
kc
We're
talking about the difference between a a single 'foreign key' in
each record stating that it's part of a certain work (preferable
from data management point of view), compared to basically
heuristics for guessing from as-written-on-title-page (or as entered
by a user) title/author combinations (less preferable from data
management point of view, but possibly neccesary to avoid the
expense of human data control), compared to this idea of a
"switching file" that is sort of just a human-controlled enhancement
to the heuristics (but if you're going to spend human time doing
that, why not just spend human time doing it right, the "foreign
key" approach? The "switching file" approach is to my mind a less
efficient encoding, not a more efficient one.)
On 8/7/2011 11:32 AM, Karen Coyle wrote:
Quoting James Weinheimer <weinheimer.ji...@gmail.com>:
if the purpose is to get the FRBR-type results
to show what works, expressions, manifestations and items exist.
For those records that do not have the uniform title entered, they
fall outside, and there is nothing to do except to add the uniform
titles (or URIs or whatever),
In the Open Library, where they decided to gather manifestations
under works (as usual, expression was harder to do), all it took
was one record for the manifestation to have a uniform title. I'll
illustrate:
Mann, Thomas
[Der zauberberg]
Magic Mountain
Mann, Thomas
[Der zauberberg]
Montagna incantata
Mann, Thomas
Magic Mountain
Mann, Thomas
Montagna incantata
These give you the information you need to bring them together into
a single work even though some records don't have a direct link to
the work. I could imagine a kind of "switching file" with links
between original and translated titles that would remove the need
for uniform titles in the process of "work-ifying" a set of bib
records. (Not unlike OCLC's xISBN service, BTW, only based on
titles not identifiers.)
Not every bit of information has to be in every record. We can have
information outside of individual bib records that helps us make
decisions or do things with the records. One of the benefits given
for FRBR is that it makes it easier for us to share this common
knowledge, and to make use of it. I think that even without a
formal adoption of FRBR we could gain efficiencies in bib record
creation and system functionality by having a place (undoubtedly on
the web) where we share this knowledge. If you look at what DBPedia
is doing with general information from Wikipedia and other
resources, then you get the idea. DBPedia is messy and rather ad
hoc, but a LIBPedia could be made up of authoritative sources only.
kc
--
Karen Coyle
kco...@kcoyle.net http://kcoyle.net
ph: 1-510-540-7596
m: 1-510-435-8234
skype: kcoylenet