社交网络中信息传播影响最大化研究

发布时间：2018-06-22 13:50

本文选题：社交网络 + 信息传播　；参考：《国防科学技术大学》2014年硕士论文

【摘要】：目前,社交网络中信息传播影响最大化研究大多是通过选取网络中最有影响的K个节点作为初始激活节点,通过它们进行信息扩散。然而对于一个具体的社交网络用户来说,整个网络拓扑结构对其是非公开的,用户对于网络的认识仅局限于与其有直接联系的邻居用户,因此当用户希望扩散自己的某一条信息时,一个更合理的模型是用户通过挑选有限个邻居节点进行信息的级联扩散,而不是通过挑选网络中的任意K个节点。传统的基于网络全局的信息传播影响力最大化算法关于如何挑选网络中最有影响力的节点已经有了丰硕的成果,因此我们考虑设计一种算法,使得挑选的邻居节点能够与传统算法中挑选的具有影响力的节点联系起来。通过间接激活这些影响力较大的节点,使网络中接收到用户所发布信息的节点数最大。本文的主要创新如下:1.提出了接近依赖度指标用于评估用户邻居节点与网络中具有大影响力的节点集合在网络中的接近程度,并将该指标同传统的贪心算法结合起来形成新的算法PIMCD并在数据集上进行了实验,实验结果显示PIMCD算法在影响范围和时间复杂度上都比贪心算法和现有的启发式策略更有优势。2.设计了CAC评价指标用于评估节点的影响扩散能力,形成新的算法PIMCAC,算法同样在数据集上进行了实验,实验结果证明了该算法可以有效解决结构相对稀疏的网络中信息传播影响最大化问题。本文在用户个性化影响最大化领域做了初步的探讨,为后续同领域研究的开展奠定了基础。
[Abstract]:At present, the study of maximizing the impact of information dissemination in social networks is mostly by selecting the most influential K nodes in the network as the initial activation nodes, through which the information diffusion is carried out. However, for a specific social network user, the whole network topology is closed to them, and the user's understanding of the network is limited to the neighbor user who has direct connection with the network. So when the user wants to spread a certain piece of information, a more reasonable model is that the user selects a limited number of neighbor nodes for cascading diffusion of information, rather than selecting any K node in the network. The traditional algorithm of maximizing the influence of information dissemination based on the global network has been fruitful on how to select the most influential nodes in the network, so we consider designing an algorithm. It enables the selected neighbor nodes to be associated with the influential nodes selected in the traditional algorithms. By indirectly activating these influential nodes, the number of nodes receiving information published by users is the largest in the network. The main innovations of this paper are as follows: 1. An approach dependency index is proposed to evaluate the proximity between the user neighbor nodes and the influential nodes in the network. The index is combined with the traditional greedy algorithm to form a new algorithm, PIMCD, and the experimental results show that the PIMCD algorithm has more advantages than the greedy algorithm and the existing heuristic strategy in terms of influence range and time complexity. The CAC evaluation index is designed to evaluate the influence diffusion ability of nodes, and a new algorithm, PIMCAC, is formed. The algorithm is also experimented on the data set. Experimental results show that the algorithm can effectively solve the problem of maximizing the influence of information propagation in relatively sparse networks. This paper makes a preliminary discussion in the field of maximizing the impact of user personalization, which lays a foundation for the subsequent research in the same field.
【学位授予单位】：国防科学技术大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.09

【相似文献】