社交网络特征计算与关键节点识别的实验研究

发布时间：2018-01-04 12:10

本文关键词：社交网络特征计算与关键节点识别的实验研究　出处：《吉林大学》2015年硕士论文　论文类型：学位论文

【摘要】：随着web2.0的发展,社交网络也得到飞速的壮大,各类社交网站和服务的出现,不仅极大地丰富了人们的生活,同时也对人类的社交行为和生活方式产生了深刻改变。对社交网络相关领域的研究是当前网络研究的热点。因为社交网络的独特地位和作用,已经深刻影响了人们社会生活的方方面面，同时也关系到舆论导向、社会安全,甚至国家安全.因此,对社交网络的研究具有重大意义。本文针对社交网络,基于复杂网络理论和改进的ADWP (Activity degree weighted Pagerank)算法,使用图论与复杂网络建模工具Networkx,对社交网络的特征和关键节点识别进行了实验分析和研究。本文的主要工作包括： 1.基于复杂网络理论,针对社交网络的拓扑结构,以腾讯微博为代表,计算了相关的无标度和小世界特征,在Networkx平台,验证了这两个特征,并进行了实验分析。 2.基于变化的拓扑结构思想,采用删除法和收缩法对节点的重要性进行评估,提出了改进的ADWP关键节点识别算法。本文首先详细介绍和分析了基于PageRank的关键节点识别算法,然后,在此基础上,引入社交网络中用户“活跃度”这个重要因素,在权重分配上进行了改进。即在基于PageRank的依据链接进行分配的基础上增加了用户的“活跃度”作为权重分配指标,更好的刻画出社交网络的特征,改进和完善了原有的PageRank关键节点识别算法。最后,在Networkx环境下以腾讯微博的转发网络为例,验证了算法的基本思想和识别关键节点的效果。综上所述,本文基于Networkx环境,从社交网络的复杂网络特征计算和关键节点识别发现两个方面进行了实验分析和研究。本文的研究工作具有一定前沿性,对同类工作,也具有一定的理论参考价值。
[Abstract]:With the development of web2.0, social network has been growing rapidly. The emergence of various social networking sites and services has not only greatly enriched people's lives. At the same time, the social behavior and lifestyle of human beings have been profoundly changed. The research on social network is the focus of the current network research, because of the unique status and role of social networks. It has deeply affected all aspects of people's social life, but also related to the guidance of public opinion, social security, and even national security. Therefore, the study of social networks is of great significance. This paper aims at social network, based on complex network theory and improved ADWP activity degree weighted algorithm. With the help of graph theory and complex network modeling tool Networkx, the characteristics and key node identification of social networks are analyzed and studied experimentally. The main work of this paper is as follows: 1. Based on the theory of complex network, aiming at the topology of social network, taking Tencent Weibo as the representative, the scale-free and small-world features are calculated, and the two features are verified on the Networkx platform. Experimental analysis was also carried out. 2. Based on the idea of changing topology, the importance of nodes is evaluated by deleting and shrinking methods. An improved ADWP key node recognition algorithm is proposed. Firstly, this paper introduces and analyzes the key node recognition algorithm based on PageRank in detail, and then, on this basis. Introduction of the social network user "activity" is an important factor. The weight distribution is improved, that is, the user's "activity" is added as the index of weight allocation on the basis of link allocation based on PageRank. Better portray the characteristics of social networks, improve and improve the original PageRank key node identification algorithm. Finally, in the Networkx environment, Tencent Weibo forwarding network as an example. The basic idea of the algorithm and the effect of identifying key nodes are verified. To sum up, this article is based on the Networkx environment. The experiment analysis and research are carried out from the two aspects of complex network feature calculation and key node recognition of social network. The research work in this paper has some vanguard, and the same kind of work has been done. Also has certain theoretical reference value.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP393.09

【参考文献】