Predicting information diffusion probabilities in social networks: A Bayesian networks based approach

Abstract

In past few years, social networking has significantly contributed to online presence of users. These social networks are hosts to a number of viral phenomena. This has fetched a lot of attention from various researchers and marketers all over the world. Major portion of the studies done in the field of information diffusion through social networks has focused on the problem of influence maximization. These methods demand the diffusion probabilities associated with the links in the social networks to be provided as inputs. However, the problem of computing these diffusion probabilities has not been as widely explored as the problem of influence maximization. In this paper, we tackle the problem of predicting the probabilities of diffusion of a message through the links of a social network. This paper presents a Bayesian network based approach for solving the aforesaid problem. In addition to the features related to the social network, this machine learning based Bayesian framework utilizes user interests and content similarity modeled using the latent topic information. We evaluate the proposed method using the data obtained from the well-known social network platform - Twitter.