Locating the Source of Asynchronous Diffusion Process in Online Social Networks

Abstract

Locating the source of information in online social networks facilitates knowing the origins of events, verifying the authenticity of information and finding the initial spreaders of rumors. However, locating the source of information in online social networks is a challenge task. This is because the information diffusion process is dynamic and complex, and the observation of the process is restricted to a limited number of nodes. Many previous works have studied the source locating problem under synchronous diffusion models, but the information diffusion process in social networks is naturally asynchronous and stochastic. The source locating method designed for the synchronous diffusion process cannot be directly applied to the asynchronous diffusion process. Aiming at solving the source locating problem of the asynchronous diffusion process in online social networks, we propose a source locating method that consists of an estimator based on the correlation coefficient and a matrix to represent the diffusion time delay between nodes approximately. Five sampling strategies for choosing observable nodes are presented to cooperate with the source locating method. Numerical simulations show the proposed method is superior to the state-of-the-art method on the asynchronous diffusion process in three types of networks.