基于节点多阶邻居递阶关联贡献度的重要性辨识

Importance identification method based on multi-order neighborhood hierarchical association contribution of nodes

为更精细化地辨识节点重要性,扩展节点有效信息集聚范围和类别,对网络节点的空间位置属性信息与其直接及间接邻居节点间关联结构信息进行融合、集结,提出复杂网络多阶邻居递阶关联贡献度的节点重要性辨识方法。依据网络节点空间位置层级差异及层间交互信息给出节点层级位置属性贡献度定义;构建复杂网络目标节点多阶邻居递阶关联贡献度矩阵,表征直接邻居节点、间接邻居节点与目标节点间关联度对其影响力的递阶贡献;提出节点跨层跨级空间拓扑位置贡献度与多阶邻居递阶关联贡献度融合的节点重要性辨识方法。仿真试验表明:在6个真实网络上所提方法有效提升节点重要性辨识的精细性和准确性。本研究通过探究网络节点间多阶递阶交互行为,为深入探索网络背后的动态演化机理,进而做出预测和调控提供科学的理论基础。

In order to identify the node importance more finely and extend the scope and category of effective information gathering of nodes,the spatial location attribute information of network nodes and their direct and indirect neighbor nodes were fused and clustered,a node importance identification method of multi-order neighbor hierarchical association contribution of complex networks was proposed.The definition of the contribution of node level location attributes was given based on the network node spatial location hierarchical differences and inter-layer association information.A complex network target node multi-order neighbor hierarchical association contributions matrix was constructed to characterize the hierarchical contribution of the associations between direct neighbor nodes,indirect neighbor nodes and target nodes to their influence.A node importance identification method that fused node topological location contribution across layers and levels of space with multi-order neighborhood hierarchical association contribution was proposed.The simulation experiments showed that the proposed method could effectively improve the precision and accuracy of node importance identification on six real networks.This study provided a scientific theoretical basis for in-depth exploration of the dynamic evolution mechanism behind the network,and then made prediction and regulation by exploring the multi-order hierarchical interaction behaviors among the network nodes.