Multi-stage seed selection for viral marketing

In this paper, a multi-stage seed selection approach is proposed, that performs seed selection in several stages to optimize viral marketing in online social network. In each stage, seeds are selected among the central nodes that have not yet been activated. Thus, the proposed approach exploits the information that is acquired during the previous stages in order to improve its performance in the forthcoming stages. Experimental results using data from a real-world social network show that the proposed approach brings significant improvement in performance (up to 30% compared to the single-stage approach).