一种基于完全子图与标签传播的重叠社区检测算法

One overlapping community detection algorithm based on complete subgraph and label propagation

提出了一种基于完全子图和标签传播的重叠社区检测CLPOA算法。该算法首先搜寻完全子图,并为每个子图分配唯一标签,实现快速标签预处理;然后根据每个节点的邻接节点标签来更新该节点的标签,同时提出接触频数优化标签选择策略降低标签随机传播概率;最后,通过网络标签分布情况进行社区划分。选取两个小规模标准数据集和两个大规模网络数据集进行实验,结果表明CLPOA算法能保持和COPRA算法相同社区划分质量,同时具有更好的算法稳定性和时间性能。

As detecting overlapping community can analyze and understand complex networks,it has become a hot topic in data mining.In this paper,an overlapping community detection algorithm based on complete subgraph and label propagation is proposed.This algorithm firstly searches complete subgraphs and assigns a unique label for each sub-figure to realize fast label preprocessing.This can update their labels on the basis of the adjacent nodes of each node.This paper also presents the concept of contact frequency to help selection label,to reduce the probability of random selection.Two small-scale standard datasets and two large-scale network datasets are applied to the experiment,the results show that CLPOA algorithm has better algorithm stability and time performance,while maintaining the same community dividing quality as COPRA algorithm.