LibraryThing and FRBR

[deborah]

I got to thinking today (during Julie Allinson's presentation on using FRBR to model e-scholarship) about LibraryThing and FRBR. Unusually for me, before I braindumped here, I looked around to see what others have said about this. And I found that of course of course, Tim is not unaware of how the LT works system overlaps with some of the goals of FRBR.

In November, he mentioned over at thingology that LT was following the OpenFRBR project -- but he said there "I also have philosophical problems with FRBR, for understanding relationships in mechanical, "binary" ways. (I've never managed to communicate this right.)" I wish he could find a way to communicate it, because I'd love to understand his problems with it. As a user who's been madly cataloguing my own library on LT -- and working on improving the works combinations, especially across languages -- I've been incredibly frustrated and see FRBR as having the potential to solve many of those problems. Right now LT really has the notion of

• a FRBR-style work (called, appropriately, a "work"),


• a FRBR-style manifestation (identified by ISBN, although about a quarter of my books are pre-ISBN, so I guess those are grouped by title/author in the free text fields?),


• and a FRBR-style copy (not called anything in particular, but identified to me by the string you have the following copy of this work , and distinguished only in that any changes I make to the metadata do not propogate to anyone else's copies. (This has actually been fairly frustrating to me, as I see so many erroneous metadata which break the LT works system. The anthology Firebirds Rising, for example, somehow got imported with four different authors (you only see three at that link; that's because I've fixed mine). The most common, Francesca Lia Block, is one of the authors who's submitted to the anthology. That second author on the list, Sharyn November, is the actual editor. My copy somehow imported with another of the submitting authors, Diana Wynne Jones, listed as the author. And of course one edition imported without any author at all. Would the Wikipedia model of a certain number of powerusers work here? Or maybe a notification system, where I could correct a work's metadata and send notes to any owners of that work who hadn't blocked such notifications, where the note was an explanation of the changes made and a simple "accept these changes to my copy" button? At a certain level, the original cataloguing sources (eg. Amazon, LOC) are being treated as more authoritative than any of the users, since most of the users won't bother modifying the source information of their own imports, and no other user has the power to fix it for them. Not a Wikipedia model, there.)

Anyway, what's missing here is the FRBR notion of an expression - LT works skip straight from work to manifestation. What's missing, then, is the grouping of Hamlet by quarto and folio, of To Kill a Mockingbird by Swedish and English editions, of The Ogre Downstairs by versions with tape/classical music instead of records/commercial pop. I don't know how easy this would be to implement in a hivemind cataloguing tool such as LT -- how would most readers know which expressions they have! -- but man, as I play with LT I wish the distinction were there.

Anyway, the other big problem with LT works is one that I don't know if FRBR solves. In LT, The Jungle Books, The First Jungle Book, and The Complete Works of Kipling have no relationship aside from sharing an author. Does FRBR allow a way to indicate that "excerpt of" or "compiled in" relationship?

Comments

Where do you want it?

[librarythingtim.livejournal.com]

Hey, I get Google Alerts on my name ;) Do you want a response here, or shall we do it in LibraryThing Talk?

Re: Where do you want it?

[deborah]

Hey, excellent! I'd love a response here, though if you feel it would be better to make it more widely available a conversation I don't mind if we move it over there.