Presented at the ACRL 10th National Conference: Crossing the Divide, Denver, Colorado, March 15-18, 2001.
Presentation Slides (Microsoft PowerPoint)
Draft of Full Paper (Microsoft Word 97)
ABSTRACT:
The past decade has seen an explosion in the numbers of electronic information resources available to identify and retrieve articles and other information relevant to a scholar's research needs. Library users are confronted with a myriad of online and CD-ROM bibliographic and full-text databases. Past practices used by academic libraries -- i.e., those which rely on in-person consultation with librarians or which expect end-users to select from simple menus of available resources – are not enough. Improvements are needed in the way libraries facilitate the online database selection process, especially for the increasing number of end-users who access library resources remotely.
Academic libraries are trying a variety of strategies of varying sophistication to address this need. Our effort has focused on analyzing and then emulating in a web-based expert system the techniques most commonly used by reference librarians when helping library users identify databases to search. In depth characterization data describing each of more than 140 databases currently available to University of Illinois at Urbana-Champaign (UIUC) Library patrons were gathered and/or generated and put into an SQL database. Partial or complete database controlled vocabularies were added to this SQL database to further characterize 80 of these 140 resources. (The extensive use of each database’s own controlled vocabulary to better distinguish databases from one another is one of the more unique features of this approach to the problem.) Using a database selection algorithm that mimics the database selection strategy used by librarians, our prototype expert system searches against this SQL database to generate recommendations. Users are able to select resources by any of three methods: 1) searching on free-text keywords and phrases taken directly from their topics; 2) browsing available databases by subject categories; or 3) selecting desired databases just by characteristics and chronological coverage. When using either method one or two, database characteristic filters may be employed to help narrow the selection.
The prototype system has been undergoing testing by UIUC library faculty and staff and by UIUC teaching faculty and students since January 2000. Detailed transaction logs tracking queries submitted, the ways in which the prototype system is being used, and information resources finally selected using the prototype system are being taken. Usability testing of the prototype system also is being done.
Our presentation will describe lessons learned in during test deployment of the prototype system. What worked and didn’t work, based on transaction log and usability-testing results, will be discussed. Upgrades to the UIUC Library’s website will be implemented during the summer and fall of 2000. Ways in which this research informs these updates also will be described. The audience will learn about ways in which web-based expert technologies can be applied to help improve library reference services and deal with new service needs created by the expanding availability of online information resources. They also will learn about how both users and librarians react to and exploit such technologies, and about issues that arise when implementing such technologies in a library setting.
Back to Timothy W. Cole, Publications & Presentations
Last updated 13 August 2000 (twc).