Social psychology meets online information diffusion

• Main Author: Li, Hui
• Other Authors: Sourav S Bhowmick
• Format: Thesis-Master by Research
• Language: English
• Published: Nanyang Technological University 2022
• Subjects: Engineering::Computer science and engineering
• Online Access: https://hdl.handle.net/10356/154776
• Institution: Nanyang Technological University

Social networks provide online platforms for individuals to express their opinions and interact with others, and thereby generate huge volumes of social activities (e.g. tweets, retweets, comments, like). Such social activities give rise to online information diffusion. Hawkes processes, in which the social event arrival rate explicitly depends on previous events, is a well-known statistical technique utilized to model information diffusion. However, existing information diffusion framework are oblivious to the impact of social psychology (e.g. conformity) on the diffusion process. In this thesis, we explore the interplay between conformity of individuals and online information diffusion under the framework of Hawkes processes. Intuitively, conformity refers to the inclination to align our attitudes and behaviors with those around us. We propose two novel probabilistic models, BRUNCH and PICTURE, to explore the triggering relations among social events to describe ``which activity triggers which activities''. BRUNCH augments multivariate Hawkes processes (MHPs) by incorporating heterogeneous link functions, referred to as hybrid multivariate Hawkes processes, to cope with diverse effect of previous social activities on future activities. PICTURE models the phenomenon that the greater the influence of the preceding activity to the following activity, the more likely there is a triggering link between them. Based on these models of interactions between social activities, we propose a novel conformity-aware Hawkes process-based framework called chassis to characterize online information diffusion. The work bridges the classical online information diffusion problem in data analytics with conformity from the domain of social psychology. The key challenge is to quantitatively capture and exploit two flavors of conformity, informational conformity and normative conformity, hidden in activity sequences by utilizing the above mentioned trigger relations (i.e. diffusion trees or branching structure) constructed from the activities. In summary, this thesis revolves around the vision of the role of social psychology (e.g. conformity) in modeling online information diffusion.