Competitive Influence Maximization with Uncertain Competitor Sources and the Bandwagon Effect in Social Networks

Abstract

In today’s networked society, various sources of competitive information vie for attention and strive to enhance their social influence. Agents with valuable information aim to select a strategic set of influential individuals for information dissemination. This study delves into the Competitive Influence Maximization problem based on the independent cascade model, where an agent selects a seed set to maximize their social influence while contending with uncertain sources of competition. When adopting information, individuals exhibit a bandwagon effect. We begin by demonstrating that the objective function in our problem is neither supermodular nor submodular. Subsequently, we introduce upper and lower bound problems using the Sandwich approach, showcasing their objective functions as submodular and monotone non-decreasing. We propose a greedy algorithm to solve these bounds effectively. A theoretical analysis of the Sandwich approach’s performance is presented, followed by experimental evaluations to assess its effectiveness.