The Good, Bad and the Indifferent: Explorations in Recommender System HealthProceedings of the International ACM Intelligent User Interfaces Beyond Personalization Recommender Systems Research Workshop
AbstractOur work is based on the premise that analysis of the connections exploited by a recommender algorithm can provide insight into the algorithm that could be useful to predict its performance in a fielded system. We use the jumping connections model defined by Mirza et al. , which describes the recommendation process in terms of graphs. Here we discuss our work that has come out of trying to understand algorithm behavior in terms of these graphs. We start by describing a natural extension of the jumping connections model of Mirza et al., and then discuss observations that have come from our studies, and the directions in which we are going.
CopyrightCopyright © Saverio Perugini et al. | ACM, 2005.
PublisherAssociation for Computing Machinery
Citation InformationBenjamin J. Keller, Sun-mi Kim, N. Srinivas Vemuri, Naren Ramakrishnan, et al.. "The Good, Bad and the Indifferent: Explorations in Recommender System Health" Proceedings of the International ACM Intelligent User Interfaces Beyond Personalization Recommender Systems Research Workshop (2005)
Available at: http://works.bepress.com/saverio_perugini/4/