CAREER: Collaborative Knowledge Spaces for Filtering Information

Oregon State University, Corvallis OR

Investigators

Abstract

This research project will develop techniques and systems that answer the following question, independent of content type: "What items have the highest probability of being both interesting and relevant to my information need?" Whether or not an item (document, resource, etc) is "interesting" for a given information need will depend on the user's background and taste. Book and web page recommenders will be built and evaluated that predict relevance and interest based on content and interest ratings entered by each member of a large Internet community. A content rating indicates how much a user believes that a relationship exists between an item and a content attribute while an interest rating specifies the user's level of interest in an item. Affinities between users' interests and users' perception of content relationships will be leveraged to transfer recommendations implicitly between users. The career development plan involves a) the development of a graduate focus in "Intelligent Information Systems," which allies approaches from artificial intelligence, information retrieval, and human computer interaction against problems of information overload, and b) development of a summer REU program with the goal of increasing the number of broadly educated US Citizens (particularly women) in Ph.D. computing graduate programs.