Collaborative Social Group Influence for Event Recommendation

Authors:

Abstract
In event-based social networks, social groups refer to self-organized communities consisting of users sharing the same interests. Through data analysis on \emph{Meetup} (\textit{i.e.,} a typical event-based social network), we found that whether a user joins an event depends on the user's preference toward the social group that posts the event. However, existing event recommendation models generally consider users' individual preferences and neglect the influence of the social group. To this end, we propose a new Bayesian latent factor model \emph{SogBmf} that combines \emph{social group influence} and \emph{individual preference} for event recommendation. Experiments on five real-world data sets demonstrate the effectiveness of the proposed method.