摘要
为挖掘当前社交网络中具有相同内在因素、特定组织结构的群体,提出一种基于特定用户的群体关系挖掘与分析方法。首先,以特定用户为切入点,改进基于最短路径的图聚类算法,以此关联它们之间的关系,构建初级群体;然后,构造初级群体属性特征比对集合,利用动态权重相似性对其进行扩展,挖掘它们所处的群体;最后,对挖掘出的群体进行聚类效果评估。实验结果表明,该方法聚类效果良好,能够有效挖掘相关群体,为社交网络中的群体发现提供了新思路。
In order to mine the groups with the same internal factors and specific organizational structure in the current social network, a method of mining and analyzing group relationships based on specific users is proposed. Firstly, with the specific user as the entry point, the graph clustering algorithm based on the shortest path is improved, and the relationship between them is established to construct the primary group. Then, the primary group attribute feature comparison set is constructed, and the dynamic weight similarity is utilized. It is extended to mine the groups in which they are located;finally, the clustering effect is evaluated on the excavated population. The experimental results show that the method has good clustering effect and can effectively mine relevant groups, which provides a new idea for group discovery in social networks.
作者
陈志扬
曹金璇
聂世民
CHEN Zhi-yang;CAO Jin-xuan;NIE Shi-min(Information Technology & Network Security Institute,People’s Public Security University of China;CIC of Security & Law for Cyberspace,People’s Public Security University of China,Beijing 100038,China)
出处
《软件导刊》
2019年第9期183-187,共5页
Software Guide
基金
国家重点研发计划项目(2016YFB0801100)
国家自然科学基金项目(61602489)
“十三五”国家密码发展基金密码理论研究重点项目(MMJJ20180108)
中国人民公安大学基本科研业务费重大项目(2019JKF108)
关键词
群体发现
图聚类
最短路径算法
特定用户
group discovery
graph clustering
shortest path distance
specific users
