基于F模式识别的复杂网络社区发现及可视化

发布时间：2019-03-07 09:10

【摘要】：随着近年来计算机技术的迅猛发展,网络的复杂程度以及数据量逐年增加。因此,可视化在复杂网络的分析研究领域变得越来越重要。在许多的可视化方法中,依据社区发现进行可视化已经成为一种趋势,并且在复杂网络研究领域取得了良好的效果。本文对复杂网络中的社区发现及可视化进行了研究,主要完成的工作有三部分。第一部分是尝试在复杂网络社区发现的过程中通过引入模糊数学的思想,实现动态的网络社区划分,更好的展现网络特性。本文选取了节点的度中心性和接近度中心性作为节点重要性测度。在具体的实现过程中,应用第一次F模式识别确定社区的核心影响力节点,第二次F模式识别确定非核心节点应属的社区,两次F模式识别就可以完成社区的划分,识别方法都是F模式识别的直接方法,使用最大隶属度原则。第二部分是尝试应用模糊数学思想,将两种测度综合考虑来实现社区发现。与单测度主要区别在于第二次F模式识别需要同时考虑两个测度,因此需要采用F模式识别的间接方法,采用F集的贴近度进行识别。最后,本文提出了基于划分结果的可视化方案,通过Processing语言搭建了可视化平台。实现了以环形布局为基础的环形嵌套布局和以力引导模型为基础的牵引聚和布局算法。尝试实现社区划分结果的动态展示,方便对网络特性的研究,并最终取得了良好的效果。总而言之,本文不仅通过引入模糊数学理论成功实现了基于F模式识别的复杂网络社区划分,还设计并实现了可视化算法将划分结果展示了出来。本文算法具备实用性的同时,也为复杂网络可视化领域理论研究提供了一个崭新的视角。
[Abstract]:With the rapid development of computer technology in recent years, the complexity of network and the amount of data increase year by year. Therefore, visualization is becoming more and more important in the field of complex network analysis. In many visualization methods, visualization based on community discovery has become a trend, and has achieved good results in the field of complex network research. In this paper, community discovery and visualization in complex networks are studied. The first part is to try to realize the dynamic division of network community by introducing the idea of fuzzy mathematics in the process of discovering complex network community, so as to show the characteristics of the network better. In this paper, the degree centrality and proximity centrality of nodes are selected as the measure of node importance. In the concrete realization process, the first F-pattern recognition is applied to determine the core influence node of the community, the second F-pattern recognition determines the community to which the non-core node belongs, and two F-pattern recognition can complete the division of the community. The identification method is the direct method of F pattern recognition, and the maximum membership degree principle is used. The second part is an attempt to use fuzzy mathematics to consider the two measures to realize community discovery. The main difference from single measure is that the second F pattern recognition needs to consider two measures at the same time, so it is necessary to adopt the indirect method of F pattern recognition and the close degree of F set for recognition. Finally, this paper proposes a visualization scheme based on partition results, and builds a visualization platform through Processing language. The circular nested layout based on ring layout and the traction aggregation and placement algorithm based on force-guided model are implemented. Try to realize the dynamic display of community partition results, facilitate the study of network characteristics, and finally achieve good results. In a word, this paper not only successfully realizes the complex network community partition based on F-pattern recognition by introducing fuzzy mathematics theory, but also designs and implements the visualization algorithm to show the partition results. This algorithm not only has practicability, but also provides a new perspective for the theoretical research of complex network visualization field.
【学位授予单位】：东北大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：O157.5

【相似文献】