Modeling and propagation analysis on social influence using social big data

Abstract

Although most existing models focus on the evaluation of social influence in online social networks, failing to characterize indirect influence. So we present a novel framework for modeling and propagation analysis on social influence using social big data. We design a method to transform the social big data into a social graph to characterize the connections between the social interaction and the spreading of short message service or multimedia messaging service (SMS/MMS) by using bidirectional weighted graph, and measure direct influence of individual by computing each node’s strength, which includes the degree of node and the total number of SMS/MMS sent by each user to his/her friends. Then, we present an algorithm to construct an influence spreading tree for each node using the breadth first search algorithm, and measure indirect influence of individual by traversing the influence spreading tree. We extend the susceptible-infectious-recovery (SIR) model to characterize propagation dynamics process of social influence. Simulation results show that influence can spread easily in contact social network due to the good connectivity. The greater the degree of initial spread node is, the faster the influence spreads in social network.