I think what Karen is saying about the importance of relationships and the arbitrariness of four levels makes a lot of sense. On the other hand, I think the kind of inheritance of information that you get from the WEMI hierarchy is potentially very useful and to the extent that the library community can come to an agreement on definitions of FRBR entities (or at least compatible definitions), it would be beneficial.
Performances always seem to me to be complicated in the FRBR model and don't fit neatly into four levels. Someone asked about videos of operas and that has been a long-standing point of contention between music and moving image catalogers (is it a version of the work by a composer or is it a new work of mixed responsibility based on the work of the composer, but created by a great many people?). I think there is a spectrum here and however we define our terms, hard-and-fast lines will be elusive.
I do think it's important to figure out what relationships and attributes are important to make this work and also to take a pragmatic approach towards encoding them. We can't know everything about a given item or work even if we wanted to and we're creating a simplified model of the bibliographic universe not trying to be Borges' "Funes the Memorious" and recreate everything.
I sometimes wonder if, for most materials, we couldn't get a lot of bang for the buck with a combination of fuller work records (especially for literature, film, music, religious works and the like), better links between works, and explicit, consistent, and machine-comprehensible coding of the pivotal characteristics of manifestations (i.e. the differences that matter to most users when selecting a particular item). This maybe wouldn't be enough for Shakespeare or the Bible, but for most materials in most contexts, I'm not sure that proactively creating separate expression-level records is practical or useful. Expression-level records seem to me to be most useful where particular expressions are likely to repeat themselves a lot (i.e. be instantiated in a lot of different manifestations).
To take the example of moving images:
Work records could provide information that we know users are interested in and that would apply to all version of the work. This includes things like original title, language, release date, standardized names of directors, cast members, etc, summaries, reviews, subjects, genres, settings, and so on.
Manifestation level records could provide the usual information, but they would provide (which we don't always do now) in a consistent, standardized, computer-manipulable form, the following types of information:
* Version: theatrical release vs. unrated, director's cut, etc.; also things like special or collector's ed., although these tend to have more to do with the extras than with the content of main film
* Physical characteristics: aspect ratio (widescreen vs. fullscreen, etc.), color, sound
* Language access: soundtracks, subtitles, captions
* Playback method/system requirements: DVD, VHS, 16 mm. film, QuickTime, Internet access, etc.
Some of these are usually thought of as expression and some as manifestation level info in FRBR, but depending on the user's needs, any of these could be the most important bit of information for organizing a group of manifestations.
Real expression-level records could be reserved for complicated situations or for situations where more that level of detail is required (Martha Yee's patrons at the UCLA Film & Television Archive may need more expression-level detail than the majority of users at the medium-sized academic library where I work or the typical public library).
Kelley McGrath
Received on Fri Dec 07 2007 - 15:55:35 EST