Analysis of local influence in social networks / Noura Hassan Badr Eldin Eissa ; Supervised Mohamed E. Elsharkawi , Ehab Hassanain , Neamat Eltazi

Thesis (M.Sc.) - Cairo University - Faculty of Computer and Information - Department of Information Systems

Among different phenomena in social networks, social influence has received great attention in research during the last few years, for its importance and implication on various applications like information diffusion, recommendation and marketing. Though lots of researches focused on studying the aspects of social influence, distinguishing influencers within the crowds and quantifying their influence, yet, the dynamics underlying this important phenomenon has mostly been ignored. Research assumed that influencers leave uniform effect across network users, regardless of the properties of these users neither their resistance to influence. While in fact, the role of the target user of influence is very important to the success or failure of the influence process. Moreover, most of the existing work lacks a comprehensive overview on the different aspects that represent pairwise influence; they use narrow definitions like retweeting, interaction frequency or topic contagions only. Also, most of these models assume that topic-contagions between users are enough to describe influence; overlooking that causality is not confirmed in these contagions. In this work, we suggest that target user{u2019}s features and personal characteristics affect her readiness to become influenced. We extract a set of user metrics from social interactions and use them to propose a new metric: {u2018}susceptibility to influence{u2019} that assesses the user{u2019}s chance to get affected by influence received from friends. Furthermore, we propose a new model for measuring pairwise influence; attempting to discover the top-k influentials from the point of view of the target user of influence

Issued also as CD