Abstract: Many efforts have been made to evaluate node importance in complex networks. However, some traditional methods based on node position in networks do not take into consideration the influence derived from multiple domain topology features, which leads to the low evaluation precision about node importance. To solve this problem, based on a deep analysis of such traditional methods as mixed degree decomposition (MDD) algorithm, a new method, named cluster and neighbor mixed decomposition method(CNMD),is proposed, which combines the global and local features of the complex network topology structure, and adopts in kinds of three-degree influence principle to represent the local features of the node.Extensive experiments on ten kinds of network datasets in different field show that the average resolution, the lowest and the highest resolution of all experimental datasets are 98.73%, 92.44% and 99.99%, respectively,which is obviously better than traditional methods, like MDD, Eksd and MCDWE algorithms.Therefore, CNMD method is not only suitable for multi-scale undirected network topology, but also applicable for evaluating node importance under all circumstances.

Key words: complex networks, multi-domain, global features, local features, three degrees of influence