一种基于局部回路的标签传播改进算法

AN IMPROVED LABEL PROPAGATION ALGORITHM BASED ON LOCAL CIRCUITS

社区发现是复杂网络分析领域的重要研究内容。针对标签传播算法中标签传播的随机性问题,提出一种基于局部回路的标签传播改进算法,即当节点邻居中标签出现频率存在多个最高时,不是随机选择一个邻居的标签作为该节点的标签,而是选择与该节点构成最短局部回路的邻居标签作为该节点标签,避免了标签在社区之间的任意传播。最后,用真实网络进行测试,结果表明,与传统的标签算法相比,提出的改进算法使社区划分结果得到了提高。

Community detection is the important research content in complex networks analysis field. In response to the randomness of label spreading in label propagation algorithm, we give in this paper an improved local circuits-based label propagation algorithm, i. e. , when a node has more than one neighbours with highest label occurrence frequency, it does not randomly select the label of one neighbour as its own but chooses the label of a neighbour which forms the shortest local circuit with itself as the label of its own, and this avoids arbitrary propagation of labels among the communities. Finally, we test it with actual networks and the result shows that the improved algorithm proposed achieves the meliorated community division outcome compared with traditional label algorithms.