复杂网络传播学中重要节点的发现

发布时间：2018-04-15 06:32

本文选题：复杂网络 + 数据挖掘　；参考：《西安电子科技大学》2013年硕士论文

【摘要】：众所周知，复杂网络广泛存在于自然、生物、工程和人类社会领域。深入研究复杂网络可以揭示隐藏的大量复杂系统的共同规律。现阶段总的来讲，复杂网络进入了数学、物理学、生物学、计算科学、网络科学等高强度的跨界混搭状态。复杂网络传播模型的研究最初始于20世纪60年代，由于舆论的散布和病毒传播、扩散类似，因此现有的舆论传播模型大都借鉴了最早提出的传染病模型。而发现核心节点是传播控制中的重要手段，目前国内外对于复杂网络静态数据分析工作较多，而对于传播控制中动态分析的则较少。在网络演化模型生成机制的研究基础上，许多学者提出了很多改进模型。网络演化模型不仅可以捕捉网络生成的动态性，，而且对实际网络的设计合理性和结构特征研究具有十分重要的意义。根据信息传播中扩大和缩小效应，找出信息传播过程中主要关键点，可以很好地应用在舆情控制、广告效应、病毒传播控制等领域。本文的主要工作是研究复杂网络传播学中重要节点的发现，内容如下: 1.介绍了复杂网络演化的研究历史以及相关的基础理论，总结出典型的传播模型研究实现算法概况，并给出比较、分析，得出各个算法的适用对象及其范围。 2．本文针对已有模型中社区发现算法的不足点，提出了新的改进算法。即，基于边凝聚系数的简单图社区结构发现的研究提出了改进的边的凝聚算法。通过已有的两个数据集的实验数据证实由于边的凝聚算法在去边之后只需要重新计算局部的其他边的凝聚系数，所以时间复杂度大大降低。且通过实验对比分析得到了重要节点衡量的量化。 3.基于微博信息平台下，ROST软件和SPSS软件进行分析，如用户的话题相似度，相互转发、评论、关注与被关注等的频繁程度，做出聚类划分，并找出“兴趣圈子”，实现对复杂网络的动态数据分析。利用重要节点衡量标准分析得出实验结果与实际情况相比较具有很强的吻合性。
[Abstract]:As we all know, complex networks exist widely in the fields of nature, biology, engineering and human society.A deep study of complex networks can reveal the common laws of a large number of hidden complex systems.At the present stage, complex networks have entered high intensity cross boundary mixing states, such as mathematics, physics, biology, computational science, network science and so on.The study of complex network communication model started in 1960s. Because the spread of public opinion and virus spread is similar, most of the existing public opinion communication models draw lessons from the first proposed infectious disease model.The discovery of core nodes is an important means of propagation control. At present, there are more work on static data analysis of complex networks at home and abroad, but less on dynamic analysis in propagation control.On the basis of the research on the generation mechanism of network evolution model, many scholars have proposed many improved models.The network evolution model can not only capture the dynamics of network generation, but also be of great significance to the study of the design rationality and structural characteristics of the actual network.According to the expanding and shrinking effect of information dissemination, the main key points in the process of information dissemination can be found out, which can be well applied in the fields of public opinion control, advertising effect, virus transmission control and so on.The main work of this paper is to study the discovery of important nodes in complex network communication. The contents are as follows:1.This paper introduces the history of complex network evolution and related basic theories, summarizes the general situation of typical propagation model research and realization algorithms, and gives the comparison and analysis, and obtains the applicable object and scope of each algorithm.2.In this paper, a new improved algorithm is proposed to overcome the shortcomings of the community discovery algorithm in the existing models.That is, an improved edge aggregation algorithm is proposed for community structure discovery of simple graphs based on edge cohesion coefficient.The experimental data from two existing datasets show that the time complexity is greatly reduced because the edge aggregation algorithm only needs to recalculate the local coacervation coefficients of other edges after edge removal.The quantization of the important nodes is obtained through the comparative analysis of experiments.3.Based on Weibo information platform and SPSS software analysis, such as users' topic similarity, mutual forwarding, comments, attention and attention frequency, make clustering division, and find out "interest circle".The dynamic data analysis of complex network is realized.It is found that the experimental results are in good agreement with the actual situation by using the measurement standard analysis of important nodes.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：O157.5

【参考文献】