Discovering User Similarities In Social Behavioral Interactions Based On Bayesian Network

Whether users in a social network are similar or not depends on the local topological structure of the social network as well as users' behaviors reflected by social interactions or user-generated contents. As the basis of latent social links, user similarities are uncertain by a quantitative degree. In this chapter, we adopt Bayesian network (BN) as the underlying framework and propose a data-intensive probabilistic approach for discovering user similarities. For the massive social behavioral interactions, we give the MapReduce-based algorithm for measuring direct similarities between users. We then construct a BN to describe these similarities by a graphical model with probabilistic properties called user Bayesian network and abbreviated as UBN. To measure indirect similarities between users, we give the method for measuring the closeness of user connections in terms of the UBN's graphical structure, and the MapReduce-based algorithm for measuring the dependence degrees by probabilistic inferences of UBN. Finally, we give experimental results and show the efficiency and effectiveness of our method.