基于标签传播概率的重叠社区发现算法

发布时间：2018-06-14 09:02

本文选题：重叠社区 + 标签传播概率　；参考：《计算机学报》2016年04期

【摘要】：发现高质量的社区有助于理解真实的复杂网络,尤其是动态地分析社区重叠结构,对社区管理和演化具有重要意义.文中提出一种基于标签传播概率的LPPB(Label-Propagation-Probability-Based)重叠社区发现算法,该算法首先为每个结点赋予一个独立的标签,然后根据结点的影响力大小将结点进行排序;在标签传播的过程中,综合网络的结构传播特性和结点的属性特征计算标签传播的概率,同时利用结点的历史标签记录修正标签更新结果;最后将传播后具有相同标签的结点划分为同一社区,社区间的重叠结点构成了社区重叠结构.作者在基准数据集和带时间维度的C-DBLP网络上进行实验,结果验证了该算法具有较高的准确性和稳定性,并且通过对重叠结构的动态分析,揭示了社区重叠结点的行为特性和C-DBLP网络处于高"耦合度"的发展趋势.
[Abstract]:It is important for community management and evolution to find out that high quality community is helpful to understand the real complex network, especially to dynamically analyze the overlapping structure of community. In this paper, a new LPPB-Label-Propagation-Probability-Based-based community discovery algorithm based on the probability of label propagation is proposed. Firstly, an independent label is assigned to each node, and then the nodes are sorted according to the influence of the nodes. The probability of tag propagation is calculated by synthesizing the structural propagation characteristics of the network and the attribute characteristics of the nodes. At the same time, the update results of the labels are corrected by using the historical label records of the nodes. Finally, the nodes with the same label after propagation are divided into the same community. Overlapping nodes between communities constitute overlapping structures of communities. The results of experiments on datum data set and C-DBLP network with time dimension show that the algorithm has high accuracy and stability. The behavior characteristics of community overlapped nodes and the development trend of high coupling degree of C-DBLP network are revealed.
【作者单位】：武汉大学计算机学院;
【基金】：国家自然科学基金(61272277) 中央高校基本科研业务费专项基金(274742)资助
【分类号】：TP311.13

【相似文献】