网络虚拟身份关系的提取和分析

发布时间：2019-03-08 08:41

【摘要】：随着互联网的普及和飞速发展,人们之间的交流方式从以前的纸质信件,到现在的电子邮件、微信、QQ、微博,发生了巨大的变化。而这些互联网应用的普及使用产生了一个新的概念--网络虚拟身份。从广义上讲,网络虚拟身份是指能标示出某个独立个体的一个身份ID,而从狭义上讲,网络虚拟身份是指人们在一些网站或者网络应用注册的账号,让网站区分出不同的用户。移动互联网让人们交流越来越紧密和频繁,两个虚拟身份进行交流通信就形成了一个关系对,这个关系可以是有方向或者是无方向的,取决于具体的关系定义。海量的关系对相互关联从而会产生一个社交网络关系图。为此,本论文将首先对这些海量的网络虚拟身份关系数据进行提取,这里的提取是建立在原始网络数据流的基础上。然后对提取出的网络虚拟身份之间的关系进行分析,主要目标是由已知的网络虚拟身份关系挖掘出未知的、可能的身份关系强度。本文首先介绍了网络虚拟身份及关系的基本概念,和社交网络关系分析的相关概念。其次,介绍了一些相关理论基础,复杂网络相关理论,社群发现算法理论,标签传播算法,并对Spark图计算框架GraphX进行介绍。本文设计和实现了一个通用的网络虚拟身份账号关系的提取系统,可以从实时网络流中还原出数据,并对网络虚拟身份进行提取,目标是提取出两个有交互关系的身份关系对。然后对这些提取的身份关系对数据进行关系挖掘分析。主要分为两部分,一是在粗粒度关系分析上,本文创新性地将经典的社群发现算法COPRA,利用GraphX计算框架进行并行化,在性能和准确性上达到预期效果,从而将虚拟身份关系图划分为一个一个单独的社群。二是在细粒度关系分析上,利用GraphX实现了 2跳邻居算法,并以此为基础实现了两个节点间的共同好友计算的算法。根据这两种粒度分析,可以将图关系网络中任意两点的关系的强度划分出来,从而挖掘出任意两节点之间的关系。最后,总结全文并对未来的工作进行展望。
[Abstract]:With the popularity and rapid development of the Internet, the way of communication between people from the previous paper letters, to the present e-mail, WeChat, QQ, Weibo, has undergone tremendous changes. And the widespread use of these Internet applications has produced a new concept-network virtual identity. In a broad sense, network virtual identity refers to an identity ID, that can mark an independent individual. In a narrow sense, network virtual identity refers to the accounts registered by people on some websites or network applications, so that websites can distinguish different users from each other, and in the narrow sense, the network virtual identity refers to the accounts registered by people on some websites or network applications. Mobile Internet makes people communicate more and more closely and frequently. The communication between two virtual identities forms a relationship pair, which can be oriented or not, depending on the specific definition of the relationship. Massive pairs of relationships relate to each other, resulting in a social network diagram. Therefore, this paper will first extract these massive network virtual identity relationship data, which is based on the original network data stream. Then the relationship between the extracted network virtual identity is analyzed. The main goal is to mine the unknown and possible strength of the network virtual identity relationship from the known network virtual identity relationship. In this paper, we first introduce the basic concepts of network virtual identity and relationship, and the related concepts of social network relationship analysis. Secondly, some related theories, such as complex network theory, community discovery algorithm theory, label propagation algorithm, are introduced, and the Spark graph computing framework GraphX is also introduced. This paper designs and implements a general network virtual identity account relationship extraction system, which can reduce the original data from the real-time network flow and extract the network virtual identity. The goal is to extract two identity pairs with interactive relationship. Then the extracted identity relationship of the data for relationship mining analysis. It is divided into two parts: first, in the coarse-grained relationship analysis, this paper innovatively parallelizes the classical community discovery algorithm, COPRA, using the GraphX computing framework to achieve the expected results in performance and accuracy. Thus, the virtual identity diagram is divided into a separate community. On the other hand, the two-hop neighbor algorithm is implemented by using GraphX in the analysis of fine-grained relationship, and the algorithm of mutual friend computation between two nodes is implemented on the basis of this algorithm. According to these two kinds of granularity analysis, the strength of the relation between any two points in the graph relation network can be divided, and the relationship between any two nodes can be excavated. Finally, the thesis is summarized and the future work is prospected.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP393.0;O157.5

【相似文献】