Influence Strength Aware Diffusion Models for Dynamic Influence Maximization in Social Networks

Social network plays a fundamental role as a medium for the spread of influence among its individuals. During the influence spreading process, one favorable goal is achieving influence maximization in social marketing. Thus diffusion model which identifies a set of individuals to initiate this spread so that more individuals can be triggered at last is very critical. However, to the best of our knowledge, all current diffusion models only consider the dynamics during diffusion and ignore the dynamically changing influence strength during information propagation. In this paper, taking account of both dynamics and influence strength during information diffusion, we propose two diffusion models in social networks for dynamic influence maximization. The first one is the Time-dependent Comprehensive Cascade (TCC) model, which considers that the activation probability between two individuals is dependent on previous activation trials. The second one is the Dynamic Variable Threshold (DVT) model, which considers that the activation threshold of an individual could be changed based on the individual´s attitude towards the propagated information. Theoretical analysis show that our proposed two diffusion models are more practical compared with previous diffusion models.