基于离散模型的符号网络及动态网络社区检测
发布时间：2019-03-25 14:10
【摘要】：复杂网络社区检测是复杂性科学研究中受到广泛关注的方向,在信息科学、生物学、数学以及社会学等邻域都有着重大贡献和持续影响.近年来,针对不同类型的复杂网络,人们提出了很多寻找社团结构的算法,也称为社区检测算法.基于复杂网络社区检测在当今社会运用的广泛性,本文以复杂网络中基于相似度的符号网络社区检测以及动态网络社区检测作为主要研究内容,研究如何根据符号网络的特点来定义一个合理的相似度模型,以及如何将动态网络的实时信息进行数量化,建立合理的数学模型.具体如下:(1)基于离散模型的符号网络社区检测.本文首先考虑符号网络中存在正连接和负连接的特点,定义了新的节点相似度计算公式.将其加入到动力学演化模型中,使得符号网络中节点状态按照网络模型演化,理论证明该模型可以达到Lyapunov稳定.通过对真实网络以及人工合成网络进行仿真,并与已有算法对比,在时间和精度上优于已有算法.(2)基于离散模型的动态网络社区检测.本文针对动态网络随时间变化的特性,对不同时间步的网络邻接矩阵进行加权处理,既考虑上一时间步的网络结构,又考虑当前时间步的网络结构,得到新的邻接矩阵.通过时变的邻接矩阵并应用动力学网络模型来实现动态符号网络的社区检测.经实验仿真得出该算法不仅适用于小规模动态网络,还适用于节点数目较多且社区结构不均衡的大规模动态网络.
[Abstract]:Community detection based on complex network is the research direction of complexity science, which has great contribution and continuous influence in information science, biology, mathematics and sociology. In recent years, for different types of complex networks, many algorithms for finding community structures, also called community detection algorithms, have been proposed. Based on the extensive application of complex network community detection in today's society, this paper focuses on symbolic network community detection based on similarity and dynamic network community detection in complex network. This paper studies how to define a reasonable similarity model according to the characteristics of symbolic network and how to quantify the real-time information of dynamic network and establish a reasonable mathematical model. The details are as follows: (1) symbolic network community detection based on discrete model. In this paper, we first consider the characteristics of positive and negative connections in symbolic networks, and define a new formula for computing node similarity. It is added to the dynamic evolution model to make the node state in the symbolic network evolve according to the network model. The theory proves that the model can achieve Lyapunov stability. The real network and synthetic network are simulated and compared with the existing algorithms in time and precision. (2) dynamic network community detection based on discrete model. In this paper, according to the characteristic of dynamic network changing with time, the network adjacency matrix with different time steps is weighted to obtain a new adjacency matrix by considering both the network structure of the previous time step and the current time step network structure. Community detection of dynamic symbolic networks is realized by time-varying adjacency matrix and dynamic network model. The simulation results show that the algorithm is suitable not only for small-scale dynamic networks, but also for large-scale dynamic networks with large number of nodes and unbalanced community structure.
【学位授予单位】：内蒙古工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：O157.5
本文编号：2447045
[Abstract]:Community detection based on complex network is the research direction of complexity science, which has great contribution and continuous influence in information science, biology, mathematics and sociology. In recent years, for different types of complex networks, many algorithms for finding community structures, also called community detection algorithms, have been proposed. Based on the extensive application of complex network community detection in today's society, this paper focuses on symbolic network community detection based on similarity and dynamic network community detection in complex network. This paper studies how to define a reasonable similarity model according to the characteristics of symbolic network and how to quantify the real-time information of dynamic network and establish a reasonable mathematical model. The details are as follows: (1) symbolic network community detection based on discrete model. In this paper, we first consider the characteristics of positive and negative connections in symbolic networks, and define a new formula for computing node similarity. It is added to the dynamic evolution model to make the node state in the symbolic network evolve according to the network model. The theory proves that the model can achieve Lyapunov stability. The real network and synthetic network are simulated and compared with the existing algorithms in time and precision. (2) dynamic network community detection based on discrete model. In this paper, according to the characteristic of dynamic network changing with time, the network adjacency matrix with different time steps is weighted to obtain a new adjacency matrix by considering both the network structure of the previous time step and the current time step network structure. Community detection of dynamic symbolic networks is realized by time-varying adjacency matrix and dynamic network model. The simulation results show that the algorithm is suitable not only for small-scale dynamic networks, but also for large-scale dynamic networks with large number of nodes and unbalanced community structure.
【学位授予单位】：内蒙古工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：O157.5
【参考文献】
相关期刊论文 前4条
1 孙鹤立;黄健斌;田勇强;宋擒豹;刘怀亮;;Detecting overlapping communities in networks via dominant label propagation[J];Chinese Physics B;2015年01期
2 陈建芮;洪志敏;汪丽娜;乌兰;;Dynamic evolutionary community detection algorithms based on the modularity matrix[J];Chinese Physics B;2014年11期
3 张立升;谷伟凤;胡岗;弭元元;;Network dynamics and its relationships to topology and coupling structure in excitable complex networks[J];Chinese Physics B;2014年10期
4 吴建设;李力光;王晓华;于昕;焦李成;;Network evolution driven by dynamics applied to graph coloring[J];Chinese Physics B;2013年06期
，本文编号：2447045
本文链接：https://www.wllwen.com/kejilunwen/yysx/2447045.html
