Achieving Fair Influence Maximization: Integrating Privacy, Diversity, and Temporal Fairness

Influence maximization (IM) is a well-known problem that changed the way to spread influence in the network. IM aims to find the most influential nodes, called the seed set, that maximize influence in the network. The existing algorithms solve IM with different approaches and diffusion models. The proposed work defines and addresses three fairness principles in IM. First, to ensure privacy concerns in the diffusion process, this work proposes a new diffusion model that preserves privacy on the sensitive attribute while the existing work ignores it. Second, the proposed work suggests an objective function to achieve diversified influence with an equitable distribution among the clusters. Third, this work analyses the impact of the temporal change of the network on the seed set and the number of influenced nodes, and suggests a novel adaptive algorithm that finds the seed set effectively and efficiently with respect to the temporal change in the network. Further, experiment results on five datasets show that the proposed algorithm is comparable and has even better results than the benchmark algorithms.