Vulnerability assessment in social networks under cascade-based node departures

Computer Science Laboratory, École Polytechnique - 91120 Palaiseau, France

^(a) fragkiskos.malliaros@polytechnique.edu

Received: 11 March 2015
Accepted: 24 June 2015

Abstract

In social networks, new users decide to become members, but also current users depart from the network or stop being active in the activities of their community. The departure of a user may affect the engagement of its neighbors in the graph, that successively may also decide to leave, leading to a disengagement epidemic. We propose a model to capture this cascading effect, based on recent studies about the engagement dynamics of social networks. We introduce a new concept of vulnerability assessment under cascades triggered by the departure of nodes based on their engagement level. Our results indicate that social networks are robust under cascades triggered by randomly selected nodes but highly vulnerable in cascades caused by targeted departures of nodes with high engagement level.