I was honored to be a panelist at the LITA/ALCTS CCS Authority Control in the Online Environment Interest Group presentation of “Authority Control Meets Faceted Browse.”

What is faceting? Why is it (re)emerging in use? Where can I see it in action? This program is intended to introduce the audience to facet theory, showcase implementations that use faceted approaches for online catalogs, and facilitate discussion on the relationship between structured authority data and this type of navigation.

Kathryn La Barre of University of Illinois at Urbana-Champaign explained the theory, while NCSU’s Charley Pennel, Vanderbilt’s Mary Charles Lasater, and I each described its implementation in Endeca, Primo, and Scriblio respectively. Scriblio is an open source project that has less than one FTE working on it, so it’s an honor to see it compared against commercial offerings, especially NCSU’s groundbreaking work.

My slides are online in QuickTime and PDF form, and I was proud to be able to show off the new public beta of the Lamson Library website and catalog, based on Scriblio.

I should be careful to point out that faceting is a theory of cataloging and classification, while clustering is the technical process of aggregating and reporting relevant metadata in search and browse screens. The difference is that Scriblio doesn’t impose rules on our cataloging practice, it simply supports clustering the metadata to make it easier to find the resources we’re looking for.

If anything, the importance of authority control increases in faceted/clustered search and browse systems, but it is a matter of exchanging one set of technological constraints for another. Card catalogs, with their alphabetical access and physical affordances (or limitations) demanded cataloging practice that is in some ways at odds with the very different affordances and limitations of faceted/clustered search and browse.

Among current implementations, clustering does well with subjects, but poorly with authors. Looking at the cardinality of those facets, it’s easy to understand the problem:

(statistics from the Lamson Library (beta) catalog.)

The number of unique authors compared to the number of total authors is very high, while a large number of subjects are represented by a small number of unique headings. Still, some authors are well suited to faceted browse, and their emergence in a result set could be mined to help users further refine their searches. Example: J. K. Rowling is an obvious top author in searches for both “harry potter” and “j k rowling”. Her statistical “pop” in the results might be worth looking at and worth leveraging elsewhere.

As currently implemented, however, clustered results most help the user who doesn’t know the proper terms for her field of interest. A user searching “sociology of education” is likely to be interested in materials cataloged under “educational sociology,” and clustered search navigation works well in that and similar circumstances.

clustering, faceting, faceted search and browse, authority control, library, library 2.0 lib20, libraries, Scriblio, presentation, LITA, ALCTS

ALA Midwinter 2007, ALCTS Future of Cataloging presentation: Collaboration, Not Competition. (slides: QuickTime & PDF.)

Stir my writings on The Google Economy and Arrival of the Stupendous post with frame four of the ALCTS And The Future Of Bibliographic Control: Challenges, Actions, And Values document:

In the realm of advanced digital applications, we are interested in collaboration, not competition.

We take as axiomatic the idea that library catalogs and bibliographic databases on the one hand, and Web search engines on the other, have complementary strengths. No matter what their respective popularity may be among the general population, neither of these broad categories of tools can compete with the other, on the other’s own ground. Realizing this, we maintain that “future catalogs” discussions based on the idea of “competition between the catalog and search engines” have become passé, leading to redundant sets of questions and answers. Such discussions lead to foregone, dead-end conclusions which tend to ignore points 1-3 above. The interesting questions about “the future of the catalog” now have to do with collaboration, not competition. Collaborations with librarians and nonlibrarians who operate social networking sites, implement “Web 2.0” or “Library 2.0” services, and pursue creative mashups of the most heterogeneous types of metadata, will invigorate both our practice and theory, as well as strengthen our relationships with our user groups. These collaborations will also be fueled by our expertise in metadata creation, of the traditional library type as well as in newer forms. Because recent and future data mining products, such as Endeca, will continue to require sources of rich metadata, the value of bibliographic metadata itself is likely to increase.

alamw2007, alcts, collaboration, competition, future of bibliographic control, future of cataloging, google economy, midwinter, presentation

ALA Midwinter 2007, ALCTS Cataloging Norms Discussion Group presentation: Metadata and faceted searching: an implementation report based on WPopac. (slides: QuickTime & PDF.)

Faceted searching such as that made possible by WPopac (look for the new name soon) improves the usability of our systems and findability of our materials, but also puts new demands on how we catalog them.

My favorite search example is sociology of education, both because it’s a common search in our logs, but also because it demonstrates how our systems can help bridge the gap between what our users know and what our catalogs know. That is, the user doesn’t have to know that the proper term is educational sociology, the aggregated facets (or clusters) gently reveal that to the user.

Cookery may be an even better example. Not because it’s the subject term everybody loves to hate, but because a search for cookbooks has a high likelihood of returning cookery as a facet.

The problem, though, is that cookery is often cataloged as cookery, French (650 $a cookery, French) instead of cookery — French (650 $a cookery $a French).

And all those other regional divisions of cookery are there too, each one trying to claim its own place in the cluster of subject headings. The result is that, while educational sociology easily rises to the top of the results for a relevant search, cookery does not.

There are, of course, other headings. Art is divided almost the same as cookery is, giving us art, Indic, cookery, Indic, sculpture, Indic, and probably a dozen others.

If properly used, the metadata in our collections will set libraries apart from other information services in terms of usability and findability. The improvements to searching that WPopac offers rest directly on the body of good cataloging data in the collection. I look forward to even more improvements in the findability of library materials as we start to consider the new ways the metadata we’re creating can be used.

alamw2007, alcts, cataloging norms, midwinter, presentation, subject headings, wpopac