User representation in social bookmarking websites

Abstract

As social networking websites become more and more popular, the necessity to serve users of these websites is increased. Since all such social websites are user-centric, representing users with a good model directly changes the results of the applications. There are many studies which consider users in their models. In this paper, we analyzed such papers and collected different user representation models. Although there is no study which considers time parameter in user representation, we included the time parameter in one of the models. We evaluated the models for two social bookmarking website datasets by visualizing them as a graph with social network analysis, and by calculating f-measure scores for a tag recommendation system. As a result, tag property is considered as the best property to represent a user in such a website.