Temporal Behavior of Social Network Users in Information Diffusion

The rapid development of online social networks (OSN) renders them a powerful tool for information diffusion. Understanding the temporal behavior of OSN users is critical in studying the diffusion process. While there is much work on building various diffusion models to characterize the information propagation process, the diversity of OSN users' behavior patterns is seldom addressed in these models. However, as revealed by previous measurement studies as well as our empirical observations, OSN users' temporal behaviors are quite different. In this paper, we investigate users' temporal behaviors based on collected traces from a popular OSN in China, covering the diffusion actions of around 2.8 million users for more than 3 years. In particular, distinct from existing studies which mine users' temporal behaviors, we focus on diffusion times and propagation latencies of social links. After clustering the OSN users into two groups with our proposed temporal features, we observe some interesting phenomena, like "Birds of a feather flock together," and specifically find that users from different groups behave heterogeneously. Motivated by this observation, we propose a continuous time diffusion model by incorporating users' heterogeneous temporal diffusion patterns. Comparison of simulation results between our model and a homogeneous model demonstrate the necessity of applying our proposed model for more realistic diffusion descriptions.