基于图嵌入和多标签传播的重叠社区检测算法

Overlapping community detection based on graph embedding and multi-label propagation algorithm

为进一步优化重叠社区检测算法,提出了一种新的基于度和节点聚类系数的节点重要性定义,按照节点重要性降序更新节点,固定节点更新策略,提高社区检测的稳定性。在此基础上,提出了一种基于图嵌入和多标签传播的重叠社区检测算法(overlapping community detection based on graph embedding and multi-label propagation algorithm,OCD-GEMPA)。该算法结合node2vec模型对节点进行低维向量表示,构建节点之间的权重值矩阵,根据权重值计算标签归属系数,据此选择标签,避免了随机选择问题。在真实数据集和人工合成数据集上对该算法进行实验验证。实验结果表明,与其他重叠社区检测算法相比,OCD-GEMPA在EQ和NMI这两个指标都有明显提升,具有更好的准确性和稳定性。

In order to further optimize the overlapping community detection algorithm,this paper proposed a new definition of node importance based on degree and node clustering coefficient,and the nodes were updated in descending order of node importance,and the node update strategy was fixed to improve the stability of community detection.On this basis,this paper proposed an OCD-GEMPA.The algorithm combined the node2vec model to represent the nodes in a low-dimensional vector,constructed a matrix of weight values between nodes,calculated the label attribution coefficient according to the weight values,and selected labels accordingly,avoiding the problem of random selection.Experimental verification of the algorithm on real data sets and synthetic data sets shows that compared to other overlapping community detection algorithms,the OCD-GEMPA algorithm has significant improvements in both EQ and NMI indicators,with better accuracy and stability.