Levels of FRBR

By levels I don't mean the FRBR group 1 WEMI (work-expression-manifestation-item) entity hierarchy, but more the amount of effort devoted to using FRBR.  At the high end you get specialized databases such as AustLit, the Australian literature catalog (which unfortunately you need a log-in to use), where they've gone to a lot of work to make sure they know who everyone is and understand the different editions of an author's work.  Just below that there are systems like FictionFinder here at OCLC that work with existing bibliographic information, but do extensive processing of the data to come as close as possible to a real FRBR understanding of the records.

It is possible, though, to step down from there and still make use of the FRBR hierarchy.  Probably about the minimal level is what Open WorldCat has been doing the last year: for each work a representative manifestation is chosen and that's what is exposed in Yahoo and Google.  This leaves much to be desired, but since the search engines aren't very intelligent about metadata, it's a lot better than giving them multiple manifestations for popular works and confusing everyone.  And, as Open WorldCat has shown, if someone does get into the system via that manifestation, you can then show them the other possibilities. (This week, Open WorldCat has gone a step further.  When you land on a Find-in-a-Library page (e.g. The Broker in hardcover, or The Broker in paperback) the libraries that display are rolled-up based on FRBR matching and icons for other formats we're aware of are displayed.)

The interface I've been working on with Phoenix Public is an attempt at something in between Open WorldCat and FictionFinder.  It starts with the same FRBR matching that those use, but, like FictionFinder, all the manifestations are searchable.  The search results, though, only show one manifestation from a work.  Of course, since the data is about Phoenix's collection, the item level is important and can easily have two dozen copies in a dozen different branches.  This brings up the same problem Open WorldCat faces--how to show the patron the possibilities without confusion.  The solution is probably to add some sort of annotation to the displayed manifestation to lead users to other manifestations if they so desire, similar to how Open WorldCat is giving access to various formats.

--Th

Comments

[Adam Chandler]

The part of the "Find it in a Library" that bugs me everytime I come back to it is this:

I am associated with an institution, Cornell University. Cornell subscribes to OCLC products and so therefore OCLC could (and should recognize) my affiliation (by IP address) easily, at least when I am on campus. So why does the service point me to copies of books at nearby universities? Does the service expect me to drive down the road for a copy? Instead the service should highlight the Cornell holdings. Maybe there could be a simple behind the scenes circ status ping protocol so you could tell me if it is on the shelf. Or easier yet, just make a big link to the Cornell link resolver and take me there. Alternatively, the service could check the Cornell holdings in WorldCat. If we hold the title, offer me a link to our local catalog; if we do not have the title, offer a big link to our ILL office, one that would send an OpenURL to our Illiad interface.

Thoughts?