基于网络结构的社交网络稳定性研究

发布时间：2018-03-18 19:25

本文选题：带容忍度k核心　切入点：坍塌　出处：《浙江工业大学》2014年硕士论文　论文类型：学位论文

【摘要】：近年来,随着在线社交网络的迅猛发展,网络稳定性已经成为一个备受关注的研究课题。在社交网络中普遍存在一种“网络坍塌”现象：用户会因为其好友的离开而离开这个网络,并进而引起其他好友的相继离开,从而导致整个网络用户数量级联式的减少,甚至解体。因此,如何有效地控制和减少这种网络坍塌现象已经成为一个关键的理论和实践问题。本文主要工作如下： 1.本文针对社交网络中存在的这种“网络坍塌”现象,在已有的锚点模型(The Anchored K-core Problem, AKP)基础上提出了容忍度k-core模型(The Tolerance K-core Problem,TKP)：具有容忍度的节点能够容忍好友数量低于设定的阈值k,但又不像锚点一样永远被保留在网络中。相比锚点模型,容忍度k-core模型更细致地描述了用户对于网络坍塌的响应机制,从而更准确地模拟实际社交网络中的坍塌过程,并更好地分析抗毁机制的效果。 2.分析了容忍度k-core模型中可容忍性对网络坍塌的影响和节点可容忍性在不同网络结构上产生的效应。相对锚点模型,容忍度模型花费较小的代价能有效地阻止网络坍塌发生,容忍度模型存在相变阈值,能够有效的判断选择何种容忍度可以使得网络具备较高的稳定性。在容忍度节点选取比例从0到1的递增过程中,ER, WS、BA三种网络中均存在一个使k-core节点数量突然增大的阈值,容忍度值越大阈值越小。此外相比ER与WS网络,k-core节点数量的变化在基于优先连接的BA网络中更加稳定。 3.探索不同容忍度节点选择策略对防止网络坍塌的效果。在优先选取高度数、优先选取高介数、优先选取高核心、随机选取节点四种策略中,研究显示优先选取高核心节点策略优于其它策略,优先高度数与优先高介数节点策略反而劣于随机选取节点策略。本文实验验证TKP模型在不同结构网络中均能通过调整容忍度大小、比例与选取策略来更好地维持社交网络的稳定性。
[Abstract]:In recent years, with the rapid development of online social networks, Network stability has become a subject of great concern. In social networks, there is a "network collapse" phenomenon in which users leave the network because of their friends' departure, which in turn leads to the departure of other friends. Therefore, how to effectively control and reduce the network collapse phenomenon has become a key theoretical and practical problem. The main work of this paper is as follows:. 1. This paper aims at the phenomenon of "network collapse" in social networks. Based on the existing anchor model, the Anchored K-core problem (AKP), a tolerance k-core model, the Tolerance K-core problem model, is proposed: the node with tolerance can tolerate the number of friends below the set threshold k, but not always remain in the network like anchor points. Compared with the anchor model, The tolerance k-core model describes in more detail the response mechanism of users to network collapse, thus more accurately simulating the collapse process in real social networks and better analyzing the effect of anti-destruction mechanism. 2. The effect of tolerance on network collapse in tolerance k-core model and the effect of node tolerance on different network structures are analyzed. Compared with anchor point model, tolerance model can effectively prevent network collapse at a lower cost. The tolerance model has phase transition threshold. In the process of increasing the tolerance ratio from 0 to 1, there exists a threshold to increase the number of k-core nodes in each of the three networks. In addition, the change of the number of nodes in ER and WS networks is more stable in BA networks based on preferential connections. 3. To explore the effect of node selection strategy with different tolerance on preventing network collapse. The research shows that the high core node strategy is superior to other strategies, and the priority high degree and high medium node strategy is inferior to the random selection node strategy. The experimental results show that the TKP model can better maintain the stability of social networks by adjusting tolerance ratio and selection strategies in different networks.
【学位授予单位】：浙江工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.01

【参考文献】