在线社交网络的动态消息传播模型研究与应用

发布时间：2019-01-20 18:44

【摘要】：在线社交网络的盛行改变了人们的交流方式,丰富了人们的社交关系网络。互联网开放便捷的特性给人们的交流带来便利的同时,使得一些流言蜚语、病毒、谣言等在社交网络平台中流传开来难以控制。当前国内外对复杂网络传播学的研究方兴未艾,在线社交网络这种新兴媒体的传播研究正面临着巨大的挑战。研究在线社交网络中消息的传播规律、分析网络中舆论传播的机理、找出舆论扩散的关键因素以及关键节点可以有效地监控舆情并及时阻断不良言论在社交网络中扩散。目前学术界针对消息传播问题的研究主要是基于复杂网络传播动力学所建立的模型仿真进行的,其中的一些理论、模型和方法都有助于更好的理解不同网络的传播行为。然而,传统的传播模型由于其简单的理论特性往往难以描述真实在线社交网络的消息传播过程。因此,本文首先探讨了传统传播模型的均场假设和接触式退化机制在仿真在线社交网络消息传播的不足,分析了在线社交网络的在线活跃行为特征、消息传播的有向性特征以及节点的非均匀特性,在深入研究和探讨复杂网络传播动力学的基础上,提出一个基于在线社交网络的动态消息传播模型。模型引入一个退化函数使传播者自发地退化成免疫者,避免了网络核心节点的提前退化；动态指定节点的权威度和免疫力,使模型可以描述在线社交网络中节点间的拓扑差异；采纳有向图作为传播网络,并可扩展外部社会加强的影响因素,提高了模型的适用性和可扩展性。为了验证模型的有效性,本文分析了采集的三个新浪微博消息传播网络的基本拓扑性质以及度分布特征,发现采集的实验网络同样具有小世界和无标度特性。利用模型在该网络中仿真消息的传播,结果表明,不同参数变化下的仿真结果皆符合现实在线社交网络中的消息传播过程。最后,我们将模型应用于识别网络中有较高影响力的节点,利用模型的单传播源仿真实验评估了每个节点的传播影响力,分析节点传播影响力与中心性特征的相关性,结果显示,有向社交网络中节点的影响力并不能由κ-核的大小表征,而出度和紧密中心性为更好的描述标量。该结果有助于识别消息传播网络中的关键节点,并为进一步研究在线社交网络中消息、谣言等的传播机理打下基础。
[Abstract]:The prevalence of online social networks has changed the way people communicate and enriched their social networks. The open and convenient nature of the Internet brings convenience to people's communication, at the same time, makes some gossip, virus, rumor and so on in the social network platform to be difficult to control. At present, the research on complex network communication is in the ascendant at home and abroad, and the communication research of online social network is facing a great challenge. This paper studies the law of message dissemination in online social network, analyzes the mechanism of public opinion spread in the network, finds out the key factors of public opinion diffusion and the key nodes can effectively monitor public opinion and block the spread of bad speech in social network in time. At present, the research of message transmission in academic circles is mainly based on the simulation of the model established by the complex network propagation dynamics. Some of the theories, models and methods are helpful to better understand the propagation behavior of different networks. However, because of its simple theoretical characteristics, the traditional communication model is difficult to describe the message propagation process of real online social networks. Therefore, this paper first discusses the shortcomings of the traditional propagation model, such as the average field hypothesis and the contact degradation mechanism, in simulating the message transmission of the online social network, and analyzes the characteristics of the online active behavior of the online social network. On the basis of deeply studying and discussing the propagation dynamics of complex networks, a dynamic message propagation model based on online social networks is proposed. The model introduces a degradation function to make the communicator spontaneously degenerate into immune person, which avoids the degradation of network core nodes, dynamically designates the authority and immunity of nodes, so that the model can describe the topological differences between nodes in online social networks. By adopting directed graph as communication network and extending the influence factors of external social strengthening, the applicability and expansibility of the model are improved. In order to verify the validity of the model, this paper analyzes the basic topological properties and the degree distribution characteristics of the three collected Sina Weibo message dissemination networks. It is found that the collected experimental networks also have the small-world and scale-free characteristics. The model is used to simulate the propagation of messages in the network. The results show that the simulation results with different parameters are consistent with the process of message propagation in the real online social network. Finally, we apply the model to identify the nodes with high influence in the network, and use the single source simulation experiment to evaluate the propagation influence of each node, and analyze the correlation between the node propagation influence and the central characteristics. The results show that the influence of nodes in directed social networks can not be represented by the size of 魏-core, but the degree of outlier and compactness is a better representation of scalars. The results are helpful to identify the key nodes in the message dissemination network and lay a foundation for further research on the spreading mechanism of messages and rumors in online social networks.
【学位授予单位】：广东工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：G206;TP393.09

【参考文献】