基于赋权有向图的邮件网络亲缘关系研究与实现

发布时间：2018-01-10 08:31

本文关键词：基于赋权有向图的邮件网络亲缘关系研究与实现　出处：《国际关系学院》2015年硕士论文　论文类型：学位论文

【摘要】：社会网络分析、复杂网络研究已是当今社会各学科领域争相研究的热点问题,相关的研究成果层出不穷,并得到了广泛的应用。电子邮件是当今社会人们沟通和交流的重要媒介,邮件发送者构成的网络也符合复杂网络的特征,这使得电子邮件网络成为了社会网络研究的一个重要分支,也是文本在网络结构研究工作中的重点方向。通过对电子邮件之间的通联关系进行研究,发掘邮件网络的社团结构等性质,能够帮助我们更深层次挖掘邮件背后实体人之间的关联关系。本文以邮件网络节点间关系为主要研究方向,主要工作和创新点如下:1、介绍了现有的社会网络分析的相关概念、研究情况,总结了社区发现的一些经典算法。2、对邮件网络的通讯关系模式进行分析,根据邮件通讯有方向性,有权重(邮件数量)的特点,构建赋权有向的邮件通讯网络,用于亲缘关系的研究。3、本文以有向赋权图为基础,提出了一个基于社团边界度和邮件信息交互数的局部社团发现算法,并综合改进的Page Rank算法计算出的网络中节点权重,进行邮件网络目标节点的亲缘关系挖掘。4、最后在项目中实现了该邮件网络亲缘关系挖掘模块,根据需求最终推送三类值得关注的邮件用户。将算法应用于随机加权有向网络和在随机加权有向网络和安然邮件数据集上进行了实验验证,实验表明,本文算法在邮件网络中能够较为准确发现社团结构及目标节点的亲密节点。证明了此算法具有一定的实用性。通过对邮件网络的研究,能够分析出邮件用户之间的关联性以及在邮件网络中所处的地位。本文的工作为深入研究邮件实体关系挖掘领域的相关课题提供借鉴。
[Abstract]:Social network analysis, research on complex networks has been a hot issue in today's competing research disciplines of the social field, the related research results emerge in an endless stream, and has been widely used. E-mail is an important medium for people in today's society of communication and exchange, which is also in line with the sender network characteristics of complex networks, which makes the network become a e-mail an important branch in the research of social network, is also the text in the research work of network structure in the focus direction. Through the study of the relationship between email communications, explore e-mail network community structure and other properties, can help us more deeply into the relationship between entities. Behind the mail to the main mail between network nodes. The direction of research, the main work and innovation are as follows: 1, introduces the related concepts of the existing social network, research situation Situation, summarizes some classical algorithm of.2 community discovery, analyze the communication patterns in relation to email network, according to the email communication with direction, a weight (number of messages) features of e-mail communication network to construct a weighted and used to study the phylogenetic relationship of.3, this paper is based on a directed graph. This paper proposed a new algorithm of local society and boundary information interaction based on the number of messages, Page Rank algorithm and node weight calculated by the improved network, mail network relationship of target node mining.4, finally realized in the project the email network related mining module according to the demand of the final push three types of concern email users. The algorithm based on random weighted network and weighted network and at random on the Enron email dataset were verified. The experimental results show that the algorithm in this paper Can be found in close email network node in the community structure and the target node is more accurate. It is proved that the algorithm is practical. Through the study of the mail network, we can analyze the relationship between users and mail status in the mail network. This work as the research topic mining entity relationship mail in the field of reference.

【学位授予单位】：国际关系学院
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP393.098;O157.5

【参考文献】