Doing the LibraryThing in an Academic Library Catalog

TITLE: Doing the LibraryThingTM in an Academic Library Catalog BACKGROUND: Many libraries and other cultural institutions are incorporating Web 2.0 features and enhanced metadata into their catalogs. These value-added elements include book jacket images, reviews, and user-generated tags. One such site that libraries are looking at as a model is LibraryThing. LibraryThing is a social networking site that allows users to “catalog” their own book collections. Members can add tags and reviews to records for books, as well as engage in online discussions about them. In addition to its service for individuals, LibraryThing offers a fee-based service to libraries, where institutions can add LibraryThing tags, recommendations, and other features to their online catalog records. OBJECTIVE: To determine the potential value of importing user-contributed data from LibraryThing into a large academic library catalog, by analyzing the overlap between the two databases, and the quality and quantity of metadata available from each source. METHODS: A random sample of 383 titles from the University of Minnesota Libraries catalog (MNCAT) was searched in LibraryThing. The characteristics of matched and non-matched records were examined. For this project, metadata analysis focused on access via subject terms. Comparisons were made between the user-supplied tags in LibraryThing (social tags) and the Library of Congress subject headings (LCSH) and Medical Subject Headings (MeSH) in the library catalog records (controlled vocabulary system). RESULTS: Of a random sample of 383 titles from MNCAT, 80 works, or 21 percent of the sample, had corresponding records available in LibraryThing. Recent publications in English in the book format were strongly favored. There were a total of 227 LCSH/MeSH terms applied to the 80 catalog records, an average of 2.84 per record. In LibraryThing, 698 tags were applied to the same 80 titles, an average of 8.73 per title. Only one record had no subject metadata in either source. Popular user tags closely correlated with controlled terms roughly 31 percent of the time.