Collaborative Filtering: Possibilities for Digital Libraries Janet Webster Associate Professor, Head Librarian-Guin Library Oregon State University Jon Herlocker Assistant Professor Oregon State University Seikyung Jung Doctoral candidate Oregon State University At Oregon State University (OSU), the OSU Libraries and the School of Electrical Engineering and Computer Science are collaborating on a project to improve the effectiveness and accessibility of digital collections and Web information portals. The project goal is to make digital resources more accessible through an innovative search interface that incorporates collaborative filtering (CF). Large digital collections and federated portals such as the National Science Digital Library (NSDL) bring enormous quantities of diverse information to users via the Web. New approaches to search interfaces are needed to make the wealth of online content more accessible and useful. We utilize CF--a process whereby each user of the information benefits from the experience of previous users. Users queries are matched against previous questions asked by other users. Then, the system recommends documents, pages, or resources that these other users found useful. The portal 'learns' what resources are valuable for which questions by observing the users' behavior and recording the recommendations. This powerful approach, developed and implemented in entertainment (e.g., MovieLens.org) and commercial settings (e.g., Amazon.com), incorporates the results of human analysis of content on a massive scale. We are testing a search and recommendation interface for the OSU Libraries' extensive Web site in a way that enlists users to recommend pages and/or databases and e-journals to others asking similar questions of the site. This briefing will examine the challenges of developing and testing the merits of a recommendation system in this diverse setting, including resolving issues of integration with existing library systems and library tradition; dealing with noisy and untrustworthy data; computation, display and explanation of recommendations; and inferring recommendations from user behavior. Web Site: http://dl.nacse.org/osu (test site) Presentation: Collaborative Filtering: Possibilities for Digital Libraries (PPT)