Predicting new user's behavior in online dating systems

Abstract

Predicting new user's reaction behavior to its recommended candidate partner correctly is critical to improve recommendation accuracy in online dating systems. However, new user (cold start) problem and data sparseness problem in the online dating system make this task very challenging. In this paper, we propose a hybrid method called crowd wisdom based behavior prediction to solve the two problems and achieve good prediction accuracy. By this method, old users who have been recommended partners before are first separated into groups. Users in each group have similar preference for partners. Then, we propose a novel measure to combine a group user's collective behavior to predict one user's behavior, which can solve the data sparseness problem. By calculating the probability a new user belongs to each group and utilizing the group's behavior we can solve the new user problem. Based on these strategies, we develop a behavior prediction algorithm for new users. Experimental results conducted on a real online dating dataset show that our proposed method performs better than other traditional methods. © 2011 Springer-Verlag.