摘要
针对现有的网络社区挖掘算法在社区划分的质量不高及执行效率低的问题,提出了一种基于日志聚类的邮件网络社区挖掘算法LENCM(the log clustering based e-mail network community mining algorithm),算法根据日志聚类节点的密度变化确定核心节点,构成日志连通子图并确定邮件网络社区划分的初始社区中心点和个数,采用错误注入的方式构造算子,并把执行后的日志与关联规则进行比较,借助社区中心动态调整方法将非核心节点划分至所属社区。实验证明基于日志聚类的邮件网络社区划分挖掘算法有较高的划分质量和较快的执行效率,具有一定的有效性和可行性。
Research the quality and efficiency of network community partition. The paper puts forward A mail network community partition mining algorithm based on log clustering. The algorithm determines the core node by the change of the log cluster node density, constitutes the logs connected subgraph, determines the initial community center point and the number of e-mail network community, adopts the way of error injection to construct operator, and makes the imple-mentation of the log compared with association rules with the community center dynamically adjust the division of the non-core nodes to their respective communities. The experimental results show that the improved algorithm has higher divided quality and faster execution efficiency.
出处
《科技通报》
北大核心
2014年第2期96-98,101,共4页
Bulletin of Science and Technology
基金
基于云技术的远程教育系统的设计(LG2012-23)
关键词
日志聚类
社区挖掘
网络社区
动态中心
log clustering
community mining
online communities
dynamic center