No abstract

Influence maximization for opinion formation (IMOF) is an important problem in social networks, which aims to select most influential nodes and obtains the maximal propagation of the most ideal opinions. The existing studies on the IMOF primarily concentrate on the effective selection scheme of the most influential nodes (ie, seed nodes) and the improved opinion formation models. However, there is little work describing and defining the IMOF mathematically. In this paper, we formulate the IMOF problem mathematically and propose the weighted cooperation model to compute the opinions of network nodes. To determine the most influential nodes, an effective scheme (including the score of potential nodes (SPNs) and elimination of overlapping influence (EOI)) is proposed. By using the reverse propagation process of public opinion and dynamically adjusting the influence score of each node, SPN determines the sorting of network node within the finite iterations. Moreover, to better determine the initial seed nodes, the EOI is adopted to deal with the similarity and overlapping influence among these nodes with high influence score. Finally, to evaluate the performance of SPN, the number of positive nodes with the change of seed nodes and the times of iteration by SPN is compared with three mainstream algorithms in social network data sets. Simulation results show that the proposed scheme obtains superior number of positive nodes than the comparison algorithms.