Fair Influence Maximization in Social Networks: A Group-Fairness-aware Multi-Objective Grey Wolf Optimizer

Abstract

The task of selecting a highly influential set of seed nodes within a social network, known as the Influence Maximization problem, has been extensively studied in recent years. However, the integration of fairness into this problem has yet to be thoroughly investigated. Traditional group-based fairness metrics have some significant limitations. These metrics aim to equalise the final activation probability between small and large groups but neglect the need to provide equal initial activation opportunities to each group. In this paper, we focus on group fairness by ensuring that information is fairly distributed among different groups during the initial propagation stage. We introduce a novel fairness evaluation metric, the Initial Activation Proportion Difference (IAPD), and model the Fairness-Aware Influence Maximization Problem (FIMP) as a multi-objective optimization problem. To address this, a Group-Fairness-aware Multi-Objective Grey Wolf Optimizer (GFMOGWO) is proposed to achieve the dual objectives of maximizing influence and ensuring fairness, thereby promoting fair and effective influence propagation. Extensive experiments on real-world datasets validate the competitive effectiveness and efficiency of the proposed GFMOGWO algorithm. We also report implicit relationships among different network attributes, experiment parameters, and fairness concepts.