论文部分内容阅读
为进一步提高有向无环图(DAG)的社区发现算法性能,降低算法计算复杂度,提出一种随机网络空模型负特征谱平分 DAG 模块化相似分区的社区发现算法.通过定义近似空模型,建立 DAG 网络的模块化相似分区,并考虑随机网络节点顺序基础上,通过负特征谱平分模块化求解方法,获得模块矩阵所具有的负特征最小值特征向量,并据此进行社区划分.通过分析发现,在给定DAG网络结构下,则所得相似性分区接近最优分区.最后,在合成及真实测试网络上进行了实验,并与典型算法进行比较,验证所提算法的可行性和有效性.“,”In order to further improve the performance of the directed acyclic graph (DAG) community discovery algorithm and reduce the computational complexity of the algorithm, a random network null model negative characteristic spectral bisection DAG modular similar partition of the community discovery algorithm is proposed here. By defining the approximate null model, here establish the DAG network modular similar partition, and consider the random network node order, then, through the negative characteristic spectral bisection method of the modular solution, the negative features of the minimum feature vector of the matrix module is obtained, and accordingly the community is divided. Finally, experiments are carried out on the synthetic and real test network, and compared with the typical algorithms, the feasibility and effectiveness of the proposed algorithm is verified.