Improved Biogeography-based Optimization Algorithm for Social Influence Maximization

Abstract

The influence maximization issue in social network analysis determines a small selection of seed nodes that effectively improves aggregated influence in a hierarchical propagation model. The process of seed identification and find the best-known seeds are an NP-hard task. In this paper we have proposed novel idea that adding a few links to a small structural change can significantly boost cumulative influence. Therefore, to accomplish coverage as an aim, we used multi-objective meta-heuristic optimization to find initial seeds while taking aggregated influence and time step into account. The suggested approach then determined the lowest number of required missing links against every uninfluenced node to sustain propagation for acceptable non-dominated seeds. Finally, end vertices for the links recommendation are identified using meta-heuristic optimization approach known as locally and globally tuned Biogeography-based optimization. The suggested links connect non-influenced components to the influenced component and allow further influence propagation.