Evaluating node importance in complex networks based on factor analysis

Objective and accurate assessment of each node importance is a basic and vital issue to research complex networks. Many algorithms and software tools have been developed, but most of them make use of single metric measurement, which is incomplete and limited to evaluate node importance in the real networks with large-scale nodes and complex relationships. In this paper, we propose a node importance evaluation method with multiple metric measurements. Factor analysis is used to explore the relationships of multiple measures, and avoid random subjective values in node importance calculation. Then the method proposed is applied to a complex network—“Les Misèrables” figure relations network, and compared with other typical algorithms such as PageRank and HITS. Experiment results show that the method has a good and reasonable value.