Multi-factor information matrix: A directed weighted method to identify influential nodes in social networks

Abstract

Identifying super-propagators on social networks takes on critical significance in guiding and controlling public opinions on the internet, information retrieval, viral marketing and so on. The influence of nodes is enhanced under the dual role of network structure and dynamic interactions, while existing research tends to attach great importance on the network structure. This means the interaction behaviors between nodes are ignored, which indicates that the performance of algorithms still needs to be improved. Moreover, only a minority of work has been conducted on directed weighted networks. To address these problems, from the perspectives of the network topology and dynamic interaction, a multi-factor information matrix centrality algorithm is developed in this study in accordance with the characteristics of information propagation process and three degrees of influence rule, with the node influence, neighbor influence, and mutual influence exerted by information feedback considered simultaneously. Based on this, the subjective and objective weighted methods are combined to enhance the accuracy of evaluation results, then the multi-factor matrix can be obtained. Four comparative experiments are employed to verify the effectiveness of proposed method with ten networks and nine existing centralities applied. The results of the experiments confirm the validity and superiority of the proposed method in terms of differentiation, connectivity and propagation performance. Moreover, its low time complexity makes the algorithm can be used in large-scale social networks. In general, methodological and multidimensional support is provided for the accurate identification of super-propagators in online social networks.