一种基于网络拓扑重合度的关键节点识别方法

关键节点识别是分析和掌握复杂网络结构和功能的重要手段,对于研究网络鲁棒性、维持网络稳定性具有重大现实意义.为了探索节点与邻居之间的关联性,提出了一种有关度中心性和公共邻居数量的关键节点识别方法,仅用局部信息就表征出了节点重要性,展现了网络拓扑重合度对关键节点识别的影响,网络拓扑重合度是指节点在通信过程中与其他节点可共用的部分.通过静态和动态攻击的方式对六个真实网络和三个人工网络进行节点移除攻击,以最大连通子图比例和网络效率作为节点识别准确性评价标准.实验表明蓄意攻击比随机攻击更有针对性,此外证明了所提方法与度中心性DC、K-shell分解法、映射熵ME方法、集体影响CI方法以及潜在增益EPG方法相比更能准确评估出节点的重要性.

A method for identifying key nodes based on the overlap of network topology

Critical node identification is an important means to analyze and master complex network structure and functions, and it has great practical significance for studying network robustness and maintaining network stability. In order to explore the correlation between nodes and neighbors, this paper proposed a key node identification method related to degree centrality and the number of common neighbors. This method only used local information to characterize the importance of nodes, and demonstrated the impact of network topology overlap on the identification of key nodes. Network topology overlap referred to the part that a node could share with other nodes in the communication process. Experiments performed node removal attacks on six real networks and three artificial networks through static and dynamic attacks. The experiments used the maximum connected subgraph ratio and network efficiency to evaluate node identification accuracy. Simulation experiments show that deliberate attacks are more targeted than random attacks. In addition, the proposed method can more accurately evaluate the importance of nodes than DC method, K-shell method, ME method, CI method and EPG method.