An online paper recommendation system driven by user's interest model and user group

Abstract

How to recommend appropriate papers to researchers based on their research interest has already attracted lots of attentions. A research interest model based on several historical behaviors is proposed. A reduction function is proposed to adjust the different influences of the behaviors, and then the user group with similar interests is created based on the interest model. Two paper recommendation methods are finally explored, which based on the user's research interests and on the user group, respectively. Experiments show that the proposed research interest model performs well.