复杂网络的鲁棒性与中心性指标的研究

RESEARCH ON ROBUSTNESS AND CENTRALITY METRICS OF COMPLEX NETWORKS

网络鲁棒性是指网络遭到随机故障或蓄意攻击时仍能维持其功能的能力,理解复杂网络部分结构的失效对网络结构和功能的影响有着非常重要的意义。针对不同的开放数据集和爬取的新浪微博数据集,通过计算移除部分节点后的巨片和连通子图数目等指标,着重分析蓄意攻击对网络的影响,发现度攻击策略对不同网络结构影响均较大,紧密度和介数攻击策略对网络结构的影响有明显区别。实验表明,非微博网络的蓄意攻击中,采用度和介数攻击策略效果较好,而微博网络应采用度和紧密度攻击策略。

Network＇s robustness refers to the capability of network to remain its functionality unchanged when suffering random failures or malicious attacks,it is of important significance to understand the impact of partial structural failure in complex network on the structure and function of networks. Aiming at different open datasets and the Sina microblogging datasets which is derived by crawling,we concentrated on analysing the impact of malicious attacks on the network structure by calculating the indices of giant component and the number of connected subgraph after removing a portion of nodes,and found that the degree attack strategy had a great impact on different network structures,while closeness and betweenness attack strategies had distinct impact on network structure. Experiment showed that in malicious attacking against non-microblogging network,to adopt the degree and betweenness attacking strategies simultaneously has better effect,while for microblogging network the degree and closeness attacking strategies should be used.