> Tim Spalding wrote:
> >
> > Interesting, though as it's titles not works, hard to compare. What is
> > the real definition of "titles" anyway? ISBN and if not ISBN,
> then
> > title string plus author string?
You need to be careful asking questions like this, because it can literally blow up in your face! When we talk about titles/editions, etc. there is a basic, almost philosophical problem with determining their differences. To put it simply: work and expression are defined in conceptual terms, while manifestation and item are defined in physical terms. There may, or not be, much correlation between the two.
I think defining separate works and separate expressions is fairly easy for people to understand, e.g. Homer's Iliad, the different textual versions represented in the various manuscripts, different textual versions represented in the early printed editions, the "definitive version with notes," then in Russian or Chinese, then Troy with Brad Pitt. I think people can conceptualize these differences fairly readily.
The considerations that define a separate manifestation are of a different order: dates of publication or printing, different publishers, the wording of the titles, number of pages, series statement, explicit edition statement. As you see, these considerations have nothing to do with comparing the text itself. Therefore, two different physical items with the same title/statement of responsibility, number of pages, series, edition statement and most importantly, as determined by Library of Congress Rule Interpretation 1.0 (at http://www.loc.gov/catdir/cpso/1-0rev3.pdf), will be considered the same manifestation, although the text inside may be quite different. (Other rules define editions differently, e.g. Differences between, changes within http://www.ftrf.org/ala/mgrps/divs/alcts/resources/org/cat/differences07.pdf) Therefore, if you follow LCRI 1.0 you will determine titles/editions/manifestations in one way, while if you follow the second rules, you will determine them s!
omewhat d
ifferently.
Conversely, if the areas mentioned above are different, e.g. a different date of publication, it will be considered a different edition even though the text itself may be exactly the same. Catalogers "trust" that the publishers and printers will point out differences in these areas.. When you do place these different "copies" side by side, you can often be amazed at how different they are. For early printed books made by hand, there may be *a lot* more work done on determining various "editions" in this sense, and of course manuscripts are of an entirely different nature.
I don't question the necessity of the catalogers' methods--you can't have catalogers running into the stacks to perform textual analysis on every item that may or may not be a duplicate. But this has always been based on the limitations of technology. With computerization, other options arise. There can be all kinds of comparisons made automatically using checksums, file compares and who knows what else. One site that does something like this is the Wayback Machine. Here is what you get when you look at the different version (editions) of the White House site: http://web.archive.org/web/*/http://www.whitehouse.gov
The asterisks denote changes (editions), so there are was change in April 17, 1999, but not June 8, 2004. Wikis have other means of showing and denoting changes.
This is done automatically and is completely beyond the ability of humans to achieve this level of accuracy. Obviously, it could be improved, e.g. there could be a list of changes for the different versions of the White House site, or all kinds of other methods could be considered.
The main thing is: when the full-text, full-version is online, new possibilities appear and we need to consider these. Unfortunately, FRBR and (I believe) RDA focus strictly on traditional methods and are silent concerning these new possibilities, which could increase accuracy almost exponentially and at the same time, save human labor.
Jim Weinheimer
Received on Tue Feb 17 2009 - 03:30:15 EST