Handcrafted Metadata

Last week I went to WebWise in LA, now a conference sponsored jointly by IMLS, the Getty and OCLC, where about 350 folks had gathered to explore the theme “Inspiring Discovery – Unlocking Collections.” Part of unlocking collections, as we all know, is describing them so others can find them, and the conference exposed some interesting takes on that theme. Dan Greenstein (CDL), for example, questioned “the value of hand-crafted metadata” and called for more automation in generating descriptive metadata. His plenary was followed by a panel on which Elisa Lanzi (Smith College Imaging Center) called for more and better metadata, only to hand the microphone to Bill Moen (U of North Texas SLIS), who presented a study which showed that of the existing 2000 MARC fields and subfields (up from a measly 278 in 1972), only 36 are used in 80% of the records he analyzed. As you can see, a clear case of mixed messages about the whole metadata thing (ramp up or demolish?), which the audience eagerly jumped on.

In retrospective, I think the confusion stemmed from the fact that everybody was talking about the same thing (metadata), yet as applied to different materials in different communities. Dan could comfortably call for more automation, because the materials on the forefront of his mind (books) lend themselves to that approach – a lot of the metadata going into a MARC record comes straight from a book’s cover so to speak, and if you digitize the cover and the rest of the book to boot, you have enough data for a record and a full-text search – voilà. Elisa, on the other hand, spoke about visual resources collections, which (so far) have not suffered from the overblown treatment a MARC record (according to Bill Moen’s data) could afford them – and, no surprises here, of course Murtha Baca (Getty) seconded her during the discussion, since museum records tend to suffer from a lack of standardized description rather than from an excess of it. Furthermore, it’s hard to see how you could do anything but handcraft when describing a painting or a sculpture – where else should the description come from?

I cherished this discussion not only because description and descriptive practices in different communities are a long-standing hobby-horse of mine (see the RLG Descriptive Metadata Guidelines I had the distinct pleasure of working on), but also because we recently announced our 2006 RLG Member Forum with the title “More, Better, Faster, Cheaper: The Economics of Descriptive Practice.” I think the real question isn’t “to hand-craft or not to hand-craft,” but rather how to best invest our scarce resources of people and time to create descriptions which truly serve our audience. Bill Moen wants to take his study one step further to look at which MARC tags genuinely support FRBR User Tasks (Find, Identify, Select, Obtain), and I think that’ll start adding a critical dimension to this debate which is all too often only present in our speculation about them, and that’s our users.

2 Comments on “Handcrafted Metadata”

  1. A belated thanks for the comment, Eric. I’ve also heard from some of the folks who participated in this discussion during WebWise (and whom I’ve named above), and they found the summary accurate and the conclusion helpful.
    In the meantime, I’ve been having 2nd thoughts on the whole FRBR thing – while I’m by no means an expert on FRBR, it’s no secret that FRBR was written by librarians for librarians, and I’d conclude that there’s a risk in using it as representative of what users really want. It’s yet another stipulation of what librarians think about users, and it may be a very good one at that, but that’s my cautionary tale for the day!

  2. Günter – A useful presentation of what you heard interwoven with your well-informed opinion, opinion I’m inclined to second. Somehow we need to build the right metadata for the right purpose at the right price.

Comments are closed.