Co-evolution of Viral Processes and Structural Stability in Signed Social Networks

TL;DR

This paper explores how the polarity of social links influences viral spread and user alertness in signed social networks, introducing a novel energy model based on Heider's balance theory to analyze structural stability and contagion dynamics.

Contribution

It presents a new energy model linking social balance with epidemic dynamics in signed networks and analyzes the role of hostile links in forming user clusters.

Findings

Hostile links influence cluster formation of alerted and infected users.

The proposed algorithm efficiently computes network energy states.

User awareness and initial network setup affect structural balance outcomes.

Abstract

Prediction and control of spreading processes in social networks (SNs) are closely tied to the underlying connectivity patterns. Contrary to most existing efforts that exclusively focus on positive social user interactions, the impact of contagion processes on the temporal evolution of signed SNs (SSNs) with distinctive friendly (positive) and hostile (negative) relationships yet, remains largely unexplored. In this paper, we study the interplay between social link polarity and propagation of viral phenomena coupled with user alertness. In particular, we propose a novel energy model built on Heider's balance theory that relates the stochastic susceptible-alert-infected-susceptible epidemic dynamical model with the structural balance of SSNs to substantiate the trade-off between social tension and epidemic spread. Moreover, the role of hostile social links in the formation of disjoint friendly clusters of alerted and infected users is analyzed. Using three real-world SSN datasets, we further present a time-efficient algorithm to expedite the energy computation in our Monte-Carlo simulation method and show compelling insights on the effectiveness and rationality of user awareness and initial network settings in reaching structurally balanced local and global network energy states.