一种基于图数据库的虚拟身份关系挖掘算法

A Virtual Identity Relation Mining Algorithm Based on Graph Database

基于虚拟身份的关系挖掘是关系挖掘的重要途径,但虚拟身份存在一定的不真实性。鉴于此,将虚拟身份一一映射到真实身份,再针对真实身份进行关系挖掘,并采用图数据库存储强关联的关系数据。该关系挖掘算法包括3部分:首先基于公共场所卡口设备和审计设备采集到的日志数据,抽取手机终端MAC、卡口设备MAC及微信构建虚拟身份库,将虚拟身份微信反向映射到真实身份手机终端MAC;然后找出单节点手机终端MAC在某段时间内的同行人作为一个关系团体,或者直接找出虚拟身份库中微信对应映射到的手机终端MAC最大度数节点作为核心节点;再利用同行人分析算法找出该节点在某段时间内的同行节点作为一个关系团体。研究结果表明,相比单纯基于虚拟身份的关系挖掘,基于图数据库的虚拟身份关系挖掘算法准确率可提高至100%。

Relationship mining based on virtual identity has become a very important part of relationship mining.However,virtual iden⁃tity has some inauthenticity.In view of this,firstly,virtual identity is mapped to real identity,and then relationship mining is carried out for real identity.This relational mining algorithm consists of three parts.Firstly,based on log data collected by bayonet devices and audit devices in public places,MAC,MAC and WeChat mobile terminals are extracted to build a virtual identity database,and then the virtual identity of WeChat is reverse-mapped to MAC,the real identity mobile terminal.Next we take the fellow pedestrians of the single-node mobile terminal MAC in a certain period of time is found as a relational group,or directly find the core node of the mobile terminal MAC node with the largest degree mapped to WeChat in the virtual identity library.Peer analysis algorithm is used to find the peer node in a certain period of time as a relational group.In this paper,graph database is used to store strongly correlated relational data.The experimental results show that the accuracy of this paper is improved to 100%compared with the pure relationship mining based on virtual identity.