Karen Coyle at ALA Midwinter: Knowledge Organization and its Future in Libraries

By Greg Borman


On January 23, 2012, Karen Coyle took on the role of guest speaker at the ALCTS CCS Subject Analysis Committee (Cataloging and Classification Section) meeting as part of ALA Midwinter in Dallas, Texas. With a presentation accompanied by slides titled “KO: Knowledge Organization”, Coyle, a Berkeley, California-based digital library consultant, writer, and speaker, gave a passionate and often humorous talk on topics such as keyword access and searching, the Library of Congress and its linked data work, and websites that have forged ahead with using linked data in exemplary ways.

Coyle began by pointing out that, generally speaking, keyword access is not truly knowledge organization. While acknowledging that this kind of access is useful for “well-named nouns”, more ambiguous terms may present problems. For example, if a user enters the term “python” in a Google search box, it’s more likely that this will generate a list of links to items about the Python programming language, as opposed to the predatory reptile. After providing this example, Coyle exclaimed that while humans are capable of utilizing their intelligence when keyword searching, computers are not really equipped for dealing with many of the subtleties of language.

After speaking about keyword access, Coyle moved on to how libraries use classification. In her view, libraries tend to place things together that mean the same thing in the stacks environment. This is all well and good, but Coyle sees major problems for users: she believes that the Library of Congress doesn’t do a stellar job of explaining its classification system to the layperson. As a result, patrons often end up confused, wandering the stacks while failing to grasp the finer details of how and why materials are organized the way they are.

Following this, Coyle took on the topic of linked data. Indeed, Coyle has been speaking and writing about linked data and the Semantic Web for several years; examples of her talks and writings can be found at http://kcoyle.net/index.html. In her view, there have not yet been truly effective online tools developed for the public to navigate the Library of Congress Classification and Subject Headings, two key ways for people to more effectively use online catalogs. However, she did mention that the Library of Congress currently has queries to some of their linked data available online at http://id.loc.gov/.

As for non-library websites that offer exemplary uses of linked data, Coyle singled out Freebase (http://www.freebase.com/) and Open Library (http://openlibrary.org/). According to the Freebase website, they currently have significant data for about 20 million topics or entities, with each topic/entity given a unique ID. This is an effort to grapple with the problem of searching ambiguous terms. As for Open Library, which is a project of the San Francisco-based Internet Archive, their stated goal is to have data for all books ever published. Many book records link to electronic versions of the publication or to WorldCat, from which users can borrow a print copy from their local library. Relying on contributions from users to add to and correct the information on books, Coyle noted that Open Library’s discoverability features (related links on book pages, etc.) are particularly admirable.

In her keynote address to the Library and Information Technology Association (LITA) in October of last year, Coyle stated that 2011 was the “year when [linked data] went from being a wild, harebrained idea to becoming accepted as mainstream.” Yet after her ALA talk, I asked her if she could give an example of an academic library that has used linked data effectively. She couldn’t offer any such example. However, it appears that the tools that will allow libraries around the globe to embrace the Semantic Web ideal of unbounded data sharing are gradually coming together.