社交网络分析关键技术研究

发布时间：2018-02-01 19:51

本文关键词： 社交网络节点中心性叠瓦算法社区挖掘环境节点　出处：《西安电子科技大学》2014年硕士论文　论文类型：学位论文

【摘要】：在Web 1.0时代,信息是由专业的网站发布,用户只能浏览、下载,不能参与信息的产生和传播。进入21世纪后,Web2.0逐渐兴起,并且发展势头迅猛,这种信息产生和传播模式也逐步发生了转变。用户由原来的只能浏览、下载转变成既可以产生信息,又可以传播信息,当然更可以阅读。用户也从原来的信息消费者一种角色转变成了消费者和生产者的两种角色。于是,人们将学习和生活以及工作的很大一部分时间转移到了网上,尤其是社交网络上。于是社交网络成了人们必不可少的一部分。在社交网络上,每天都有海量的数据产生。通过分析这些海量数据,可以得到人们的兴趣偏好、地域分布、生活状态以及一些事件的发展动态等等的潜在信息。这些信息有着特别巨大的社会价值和商业价值。政府可以参考相关信息进行政策调整,公司可以根据分析出的结果针对性的进行产品开发以及调整公司战略等等。因此,分析研究社交网络有着极大的理论意义和实际意义。本文以豆瓣网为社交网络实验数据,研究社交网络分析的关键技术,包括验证社交年网络的基本结构特性、分析社交网络节点性质以及社区发现,因此本文的研究工作可以分为三个部分。首先,本文分析和验证了社交网络基本结构特性的正确性,包括无标度性、小世界性和社区结构。其次,本文从“度中心性”和“紧密度中心性”两个方面对社交网络中的节点进行了研究。最后,本文深入分析研究了“递归叠瓦算法”这种社区发现算法;通过实验发现其缺陷,并分析了缺陷成因且进行了实验验证;在此其基础上,本文提出了“添加环境节点”的改进方法并用实验验证其正确性。在最后一部分中,除了社区发现的研究之外,通过对比分析节点时得到的“中心节点集”与社区挖掘得到的“社区节点集”,本文发现了“中心节点不一定在社区结构内”的现象,并对其形成原因进行了分析和实验验证;以此为基础,本文提出了社交网络节点分类方案,并为社交网络上的产品推广提出了建议。
[Abstract]:In the era of Web 1. 0, information is published by professional websites, users can only browse, download, can not participate in the generation and dissemination of information. And the rapid development momentum, this information generation and dissemination mode has gradually changed. Users can only browse from the original, download to not only produce information, but also can spread information. Of course, it is more readable. Users have also changed from one role of information consumer to two roles of consumer and producer. People move much of their study and life and work to the Internet, especially on social networks. So social networks become an essential part of people. On social networks. Through analyzing these massive data, we can get people's interest preference and regional distribution. Potential information about the state of life and the development of some events. This information has particularly great social and commercial value. The government can refer to the relevant information for policy adjustment. Companies can be based on the results of the analysis of targeted product development and adjust the company strategy and so on. It is of great theoretical and practical significance to analyze and study social networks. In this paper, we take Douban net as the experimental data of social networks, and study the key technologies of social network analysis, including verifying the basic structural characteristics of social networks. Analysis of the nature of social network nodes and community discovery, so this study can be divided into three parts. First, this paper analyzes and verifies the correctness of the basic structural characteristics of social networks, including scale-free. Secondly, this paper studies the nodes in social networks from the aspects of "degree centrality" and "compactness centrality". In this paper, the community discovery algorithm named "recursive imprint algorithm" is deeply analyzed and studied. The defects are found through experiments, and the causes of the defects are analyzed and verified by experiments. On this basis, this paper proposes an improved method of "adding environment nodes" and verifies its correctness by experiments. In the last part, in addition to the community discovery research. By comparing and analyzing the "central node set" and the "community node set" obtained by community mining, this paper finds that "the central node is not necessarily within the community structure" phenomenon. The causes of its formation are analyzed and verified by experiments. Based on this, this paper proposes a classification scheme of social network nodes, and gives some suggestions for product promotion on social networks.
【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】：2014
【分类号】：TP393.09

【相似文献】