基于广义接近中心性识别网络中多个有影响力的传播源

发布时间：2018-01-20 06:09

本文关键词： 复杂网络多传播源广义接近中心性 K-means　出处：《安徽大学》2017年硕士论文　论文类型：学位论文

【摘要】：近年来,对复杂网络的研究已经受到计算机、数学、经济学、传播学和生物学等不同学科领域的关注,网络的结构与动力学是复杂网络科学的两个最基本问题。对于网络结构的探测包括:网络的社团划分,关键节点的识别,链路预测等,其中许多学者都致力于网络中有影响力传播源识别的研究,即找到网络中的一个或者多个节点使得这些节点对网络的影响最大。这一课题在实际生活中对抑制疫情扩散,加速信息传播和推广新产品等都具有重要战略意义。本文主要研究的是寻找一组节点使得这一组节点对网络的影响最大。一个节点到网络中所有节点距离和越小则这个节点越重要,由此我们想到当一组节点到网络中所有节点距离和最小时这组节点对整个网络来说比较重要。基于此思想本文主要有以下两方面工作:1.由节点的接近中心性推广到一组节点到所有节点的距离越短越好,进而提出广义接近中心性。因此寻找一组最重要节点的问题转化为寻找目标函数的最优解,之后我们证明可以用K-means聚类模型近似求目标函数的最小值。2.用single-contact SIR模型和all-contact SIR及谣言传播模型在实际网络上进行模拟实验,并与度、介数、K-核、着色、最优渗流等方法进行比较分析。结果表明:广义接近中心性指标在signal-contact SIR模型、all-contact SIR模型和谣言传播模型上都表现出较好的结果。
[Abstract]:In recent years, the research on complex networks has been concerned by computer, mathematics, economics, communication and biology. The structure and dynamics of network are two basic problems in complex network science. The detection of network structure includes community division of network, identification of key nodes, link prediction and so on. Many of them are devoted to the research of influential source identification in the network. That is to find one or more nodes in the network to make these nodes have the greatest impact on the network. This problem in real life to curb the spread of the epidemic. It is of great strategic significance to accelerate the dissemination of information and promote new products. In this paper, the main research is to find a set of nodes to make this group of nodes have the greatest impact on the network. A node to all nodes in the network and the smaller the distance. The more important this node is. From this, we think that when a group of nodes to all nodes in the network and minimum distance, this group of nodes for the entire network is more important. Based on this idea, this paper mainly has the following two aspects of work:. 1. The shorter the distance from a group of nodes to all nodes, the better. The problem of finding a group of most important nodes is transformed into finding the optimal solution of the objective function. Then we prove that the K-means clustering model can be used to approximate the minimum value of the objective function. 2.Using single-contact SIR model and all-contact model. SIR and rumor propagation model are simulated on the actual network. The results show that the generalized approach to centrality is based on the signal-contact SIR model, and is compared with the methods of degree, medium K- kernels, coloring and optimal percolation. Both all-contact SIR model and rumor propagation model show good results.
【学位授予单位】：安徽大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：O157.5

【参考文献】