Work, expression, manifestation, item … blog.
A weblog following developments around the world in FRBR: Functional Requirements for Bibliographic Records.
Maintained by William Denton, Web Librarian at York University. Suggestions and comments welcome at wtd@pobox.com.
Confused? Try What Is FRBR? (2.8 MB PDF) by Barbara Tillett, or Jenn Riley's introduction. For more, see the basic reading list.
Books: FRBR: A Guide for the Perplexed by Robert Maxwell (ISBN 9780838909508) and Understanding FRBR: What It Is and How It Will Affect Our Retrieval Tools edited by Arlene Taylor (ISBN 9781591585091) (read my chapter FRBR and the History of Cataloging).
Jonathan Rochkind noticed this schedule for the ALA Midwinter conference that starts this week lists the “FRBR Implementers Group” as meeting on Friday 19 January at 10:30 AM. It’s not listed on the ALA Midwinter wiki, though. If you know anything about it, please post a comment.
I don’t have my own blog, so I’ll put my report of this meeting here. (Judging by the lack of comments to William’s query, I may be the only person that attended that meeting who also participates in this corner of the ‘blogosphere’).
The FRBR Implementer’s Group.
This was an informal meeting–there is no membership in any organization of ‘FRBR Implementer’s Group’, it’s just a meeting of interested people, many working in FRBR implementations, as well as people working on FRBR and other standards themselves.
I sometimes had trouble keeping track of who was speaking or exactly what project they were speaking about, but here is my attempt at summarizing what was said. Forgive me for any mistakes, and sorry for all the names I missed! Much of what was talked about readers of frbr.org are likely already aware of, but there’s some new stuff.
Ed O’Neil talked about the IFLA FRBR working group and what they have been doing. There have been updates proposed to FRBR to add more flexibility to when an Expression difference occurs. The authority working group, working on a document now to be called Functional Requirements for Authority Data (FRAD; no longer FRANSAR or FRNSR or whatever it was) continues to work on that.
Kristin from NCSU whose last name I didn’t catch reports that they continue to try to put some notion of work-set into their famous Endeca-powered catalog. They’ve found this is very complicated, and it’s turned into more of a research project than the more straightforward implementation they originally optimistically thought they would do. They are interested in work-set grouping in part to de-duplicate records when they apply their new catalog model to the Triangle Research Library union catalog. They have been working with work-identifiers as supplied by OCLC, but have also found that the UC Melvyl algorithm is much more useful to them than the simpler OCLC work-identifier algorithm alone. Some of the problems they have run into with their bib data include missing OCLC #s and inconsistently entered headings.
NCSU is also using a notion of work-set in their homegrown ERM, where they assign homegrown work identifiers to serials in order to group different serial manifestation records together. This has already proved useful to technical services in keeping track of things, and they plan to use it in a future public display tool. They are using their own local ‘work identifiers’ now, but are interested in the idea of shared work identifiers.
Jennifer Bowen reported on RDA progress involving FRBR. In response to user commnity comment, they are looking at making references to FRBR in RDA more explicit. [Hooray says me!] The RDA draft will be slightly re-organized to make Part A correspond to FRBR, and Part B correspond to FRDA. They are re-organizing the chapter on Group 1 relationships, they aren’t sure whether to use the FRBR layout of Group-1-entitiy relationships, or Barbara Tillet’s (I think I’m remembering correctly that it’s Tillet who has done this work) taxonomy.
Bowen mentioned that a new RDA document had just been published on the JSC website on ‘scenarios’—different possibilities for record structures in FRBR implementations.
Bowen also reported on the XC project to create a collaborative open source catalog foundation. Next month they are having a by-invitation meeting for project partners, and have had lots of interest, they even had to turn some people away.
Two people whose names I missed reported on a project of a Belgian university and consortium whose name I missed (Sorry!), which has a catalog/ILS implementation based on FRBR which has a seperate record for each group 1 entity. For more information, email one of them at espleyj@vtls.com (I think!).
Someone announced that Arlene Taylor will soon have a book on FRBR published by Libraries Unlmiited.
Allison Carlyle, a professor at the University of Washington, announced that she will soon undertake a research project which is a user study of ‘substitutability’. Ask users who are coming into the library looking for a particular item, “If it’s not available, would this other item serve? How about this one?” To get an idea of user conception of ‘work set’. [Does this mean that 'work set' is defined by substitutability? Hmm. -Jonathan]. They will also look at if the user’s purpose (for a school assignment, etc.) has an impact on substitutabilty.
Martha Yee talked about a FRBR catalog implementation at the UC(?) Cinema (Studies?) library. She said they use a FRBR model when cataloging, although they do it in a traditional ILS/catalog system. They consider an authority record to be a work record, a bib record to be an expression record, and a holdings record to be a manifestation record. Their ILS supports item records hanging off the holdings records too. They have a new FRBR based user interface which isn’t yet ‘live’, but you can see a preview at http://cinema.library.ucla.edu.
Someone whose name I missed from OCLC mentioned that they have done some work and are going to do more with better algorithms of identifying work sets, involving work with xISBN, FictionFinder, as well as WorldCat (all of which apparently currently use somewhat different algorithmic techniques of identifying work sets?). It was news to me personally that OCLC does not strictly use their publicized simple work-identifier algorithm in the xIBSN service. Although they start from that algorithm, they have other algorithms which try to do better. I asked if the descriptions of these other algorithms would be made public, but he wasn’t sure. He also said that xISBN will soon become an officially supported production service, rather than the experimental research service it is now.
There was some back and forth discussion about the idea of Super Work. Interestingly, it seemed as if most people present who have done some kind of FRBR implementation had seen the need to create some kind of Super Work representation in their implementations, for an entity collecting a number of works together. Among other things, this allows one to create bibliographic relationships that point to that ‘super work’ rather than to a particular work. To me personally, this indicates the need for a Super Work entity in the FRBR model. (I personally continue to be concerned with the lack of attention on updating the FRBR model in reaction to things learned from practice like this—maybe once there are more people practicing, this will start up! I personally think this needs to be done with the attributes and relationships too, not just the entities). Martha Yee mentioned that in their work with creating a FRBR model from existing MARC data, there are big problems with ambiguity in existing recorded work relationship (‘added entries’) in 730 (and other 7xx). These added entires can be to related works, or to constituent works in an aggregate (I think I have the two main possibilities right!), and while there is a MARC indicator to represent which of these sorts of things a title added entry is, it is seldom used in the actual records one finds. This leads to obvious difficulties.
Someone asked about ideas for best practices in cataloging in order to create cataloging records that were ‘FRBR ready’. Paul Weiss said the biggest things people had recognized was: 1) Use Uniform Titles whenever possible, use them consistently. 2) Use Role Indicators. There was some discussion of HOW to construct uniform titles effectively, and mentions of current inconsistencies in how this was done (especially between music catalogers and ‘traditional materials’ catalogers), and trouble mapping existing uniform titles to FRBR information. Some expressed hope that FRAD and RDA Part B would clear some of this up. People talked about the need to fix our uniform title practices. There was also some mention of the need to clearly encode ‘role indicator’ type information on Group 1 entity relationships (as per Martha Yee’s comments above), not just on ‘people’.
Someone said they would schedule another “FRBR Implementer’s Group” meeting for the Friday prior to ALA annual this summer.
Comment by Jonathan Rochkind — 19 January 2007 @ 9:45 pmAn excellent report! Thanks very much. Sounds like a very interesting discussion with lots of good people.
Comment by wtd — 23 January 2007 @ 12:05 amI just wanted to clarify what I said about “FRBR-ization” algorithms. What is interesting about the Melvyl approach is its incorporation of fuzzy matching, represented by assigning weights to different full or partial matches on fields. We are not using it, so it hasn’t been ‘useful’ to us per se. I also mentioned that, as the UC project report acknowledged, they did not have time to incorporate authorities, which would improve the matching. I was encouraged that Tom Hickey said OCLC is also working on fuzzy matching for their work set algorithm, which already employs authorities data.
Comment by Kristin Antelman — 25 January 2007 @ 10:00 amMore from Thom Hickey (who is apparently the guy who’s name I missed at the meeting, sorry!), about some of the ways OCLC is dealing with work clusters.
http://outgoing.typepad.com/outgoing/2007/01/frbr_implemente.html
Comment by Jonathan Rochkind — 30 January 2007 @ 2:17 pmAnd sorry for missing Kristin Antelman’s last name. I am glad people are chiming in to correct my ‘report’ in any way, I am so pleased at any public cross-institution discussion of this sort of stuff, I think not enough occurs.
Comment by Jonathan Rochkind — 30 January 2007 @ 2:19 pm