Stephen Paling wrote:
<snip>
This is a pretty persuasive response. I would agree that a search that needs to be both specific and comprehensive often requires structured metadata. Full-text searching may not get you there because of differences in vocabulary, etc. But we nonetheless need to make some hard choices. How typical is a search such as, "Bulgneville, Battle of, France, 1431"? Is that sufficient for us to maintain a professional identity? If we drew a pie chart of the information ecosystem, how big would that slice be? For that matter, how big would the entire print slice be? Can we afford to focus only on that slice?
</snip>
Following the policy of "literary warrant" makes it fully justified to do this, i.e. an author devoted enough energy to writing 20% or more of a resource on this topic. On the other hand, if we wanted to follow the rule of "user warrant" (i.e. people want the term), that could be very interesting and even highly possible to implement today with current tools. It would have been much more difficult in the card environment. Literary warrant is based on "post-indexing" while user warrant is much more dynamic. Perhaps the two together could create something rather new, e.g. using dbpedia. An interesting idea...
<snip>
I've advocated elsewhere for letting catalogs be catalogs, i.e., sophisticated databases of a library's holdings. I stand by that, with the caveat that static MARC records with no provision for patron input is probably not ideal. But part of the Copernican turn I advocate involves acknowledging that the information solar system simply doesn't revolve around books and book-like objects, or BBLOs as I have taken to calling them in class. A huge majority of information is now available online, and I would rather emphasize what we can contribute to information architecture, harvestable metadata, exploitable document structures, search engine optimization, and algorithms for automation. One of the distinctions that I have the most difficulty explaining to colleagues and students is that contemporary information technologies aren't about ~organizing~ digital objects, they're about ~making~ digital objects. I want us to participate in that.
</snip>
Yes, librarians and catalogers must participate in this, otherwise they will be completely left behind. But it takes a different mindset, especially for a cataloger. As an example of trying to use simple algorithms, you may be interested in an experiment of own using Research Guides. This method uses Google by taking user-generated keywords and a rather complex query to limit results to relevant research guides created by librarians. See: http://www.galileo.aur.it/opac-tmpl/npl/en/pages/diyfinder.html This way I try to leverage as much cooperation from my colleagues out there as I possibly can. I have used this method on some other attempts as well. I find I often have to play with the query, however. Still, I think catalogers are some of the best people to improve these kinds of methods.
James L. Weinheimer j.weinheimer_at_aur.edu
Director of Library and Information Services
The American University of Rome
Rome, Italy
Received on Thu Jul 01 2010 - 02:51:08 EDT