改进蚁群算法优化重叠社区发现方法

Local extension approach through ant colony algorithm for overlapping community detection

为更好地挖掘社区结构,提出一种改进蚁群算法优化重叠社区发现算法(LEAC-OCD)。采用局部扩展的社区识别方法,将少数核心节点构成三角形模体,判断和量化社区结构的稳定性,实现粗糙划分社区结构;根据蚁群算法在转移机制的启发下自由移动,改变重叠节点的位置归属;通过社区量化稳定性的阈值,得到高质量的重叠社区结构。实验结果表明,LEAC-OCD算法在检测重叠社区结构方面优于其它经典算法。

To better explore the community structure,a local extended ant colony algorithm to optimize overlapping community detection(LEAC-OCD)was proposed.A non-overlapping community discovery method with local extension was adopted,so as to form a small number of core nodes into triangular motif to determine and quantify the stability of community structure and achieve the rough division of community structure.Subsequently,according to the ant colony algorithm,free movement was inspired by the transfer mechanism to change the position of overlapping nodes.Through the threshold value of community quantization stability,a high-quality overlapped community structure was obtained.Experimental results show that the LEAC-OCD algorithm is superior to other classical algorithms in detecting the overlapping community structure.