Benjamin A Abrahamse wrote: <snip> I raised this question at a FRBR pre-conference last summer in Chicago: Do we really expect catalogers to spend their time establishing "works"? Or is the question of "workhood" -- if indeed it needs to be answered -- something that is better left to literary and historical scholarship?
The answer I was given was, "Well that's what they've always been doing with uniform titles." But is it? To my mind, a uniform title is basically an instruction to collocate items under a fixed, but essentially arbitrary label. ... </snip> This is correct but I think we can illustrate it more clearly using subjects (where the function is exactly the same) because we now have http://id.loc.gov/. Within a local relational database that uses an authorities module, the text for "Aircraft accidents" is replaced in each bibliographic record by a link to a separate table of subjects. Because of this relationship, the text for the subject "Aircraft accidents" is entered only one time and although the text appears to be in each bibliographic record, it is only a link for searching and display. The actual link in the bibliographic record may be "SUB.34568" (totally made up) or whatever the internal mechanics of the database uses. Therefore, when someone clicks on "Aircraft accidents" in such a database, they are actually searching "SUB.34568" while the textual display would also come from that record in the authorities module. This is why, if it changes to "Airplane accidents" it needs to change only once and through the links, the display would change everywhere. (I won't talk about subdivisions here) The idea is to replace the links within each separate relational database with a standardized URI, e.g. instead of "SUB.34568" there would be http://id.loc.gov/authorities/sh85001323. From here it would work exactly the same as above, except "web-enabled," i.e. so long as everywhere there is the concept of "Aircraft accidents" they would include http://id.loc.gov/authorities/sh85001323 in some way. This is the backbone of the Semantic Web, as I perceive it, and shows how important we could become. This method could probably work fairly well right now for subjects since there is supposed to be an authority record for every subject. (Ha!) The same goes for names, etc. once they are put online. The problem is that FRBR posits the existence of many things that do not exist, e.g. the "work" record, the "expression" record, which as Benjamin points out, with the exception of titles such as Hamlet, have mostly been only a collocation device without an authority record although the text is strictly determined by authorized forms. To enable the collocation/textual display for these parts to work in a similar fashion as the subjects mentioned above, means making lots and lots and lots(!) of URIs for currently non-existent works and expressions. This will drain a lot of our resources, even using automated means, and I fear that there will be far too much room for those tiresome, obscure theoretical disputes that will demand our time, but will be of practically no benefit to our users. So, I completely agree with Benjamin. Do we embark on such an epic journey to create all of these URIs because we need to shoehorn it into theoretical models that were figured out almost 20 years ago? (As some have pointed out, the WEMI model was actually figured out in the 19th century, if not before) Or do we do we say that there is something wrong with the WEMI model itself? I think we still have a choice. I'm a practical kind of guy. Open the data and link what can be linked. Just doing that would improve our status, and information retrieval, substantially. James Weinheimer [email protected] Director of Library and Information Services The American University of Rome via Pietro Roselli, 4 00153 Rome, Italy voice- 011 39 06 58330919 ext. 258 fax-011 39 06 58330992
