Attribute based diversification of seeds for targeted influence maximization (Q2055595)

This paper studies attribute based diversification of seeds for targeted influence maximization. The emphasis is on the categorical attribute-based diversity of the seeds. The problem of attribute-based diversity-sensitive targeted influence maximization is proposed. The node set is associated with a categorical dataset describing the profiles of nodes. A class of non-decreasing and sub-modular functions are designed to represent the categorical diversity. A solution to the attribute-based diversity-sensitive targeted influence maximization problem is provided under the reverse influence sampling approach. Both the diversity awareness and the targeted nature of the attribute-based diversity-sensitive targeted influence maximization problem have been tackled. Based on the triggering model, a algorithm regarding attribute-based diversity-sensitive targeted influence maximization has been proposed which gives a \(k\)-seed with certain approximation ratio with high probability. Numerical experiments have been conducted to illustrate the proposed method.