复杂网络大数据中重叠社区自动检测仿真
|
摘要
针对现有方法检测复杂网络重叠社区时受网络规模和社区规模,以及重叠节点比例影响较大,导致检测结果准确性不高、稳定性较差、效率较低的问题,提出了基于模糊谱聚类的重叠社区自动检测方法,将复杂网络中的每个节点用三元组形式描述,在社区结构划分的每次迭代过程中,根据贡献函数加权处理的方法计算复杂网络中各个节点的从属系数,根据给出的社区结构划分规则对各个节点的从属系数做规范化处理,采用相似系数法构建模糊相似矩阵用于评价节点从属系数规范化处理结果,同时将复杂网络描述成一个无权无向图,并利用模糊谱聚类算法和预先设置的判断阈值,检测无权无向图中是否存在重叠社区,如果图中社区结构之间的重叠程度大于该设置阈值,说明图中存在重叠社区,将发生重叠的社区合并为一个大社区。仿真结果表明,所提方法能够实现复杂网络大数据中重叠社区的自动检测,且具有准确性较高、稳定性较好、耗时较少的优点。
Currently, methods for detecting complex network overlapping communities are affected by the network scale, the community size and the proportion of overlapping nodes, resulting in low accuracy, poor stability and low efficiency. Therefore, a method to automatically detect the overlapping community based on fuzzy spectral cluster was presented. Firstly, each node in complex network was described with triple forms. In each iterative process of community structure division, the weighting contribution function was used to calculate the membership coefficient of each node in complex network. According to the given community structure division rule, the membership coefficient of each node was normalized. On this basis, the similarity coefficient method was used to construct the fuzzy similarity matrix which evaluated the normalization processing result of membership coefficient of node. Meanwhile, the complex network was described as an unweighted undirected graph. Moreover, the fuzzy spectral clustering algorithm and the preset judgment threshold were used to detect whether there was an overlapping community in the unweighted undirected graph. If the degree of overlap between the community structures in graph was greater than the set threshold, the overlapping communities existed in graph. Finally, all the overlapping communities were combined into a big community. Simulation results prove that the proposed method can realize the automatic detection of overlapping community in complex network big data. Meanwhile, this method has the advantages of high accuracy, good stability and less time consumption.
Currently, methods for detecting complex network overlapping communities are affected by the network scale, the community size and the proportion of overlapping nodes, resulting in low accuracy, poor stability and low efficiency. Therefore, a method to automatically detect the overlapping community based on fuzzy spectral cluster was presented. Firstly, each node in complex network was described with triple forms. In each iterative process of community structure division, the weighting contribution function was used to calculate the membership coefficient of each node in complex network. According to the given community structure division rule, the membership coefficient of each node was normalized. On this basis, the similarity coefficient method was used to construct the fuzzy similarity matrix which evaluated the normalization processing result of membership coefficient of node. Meanwhile, the complex network was described as an unweighted undirected graph. Moreover, the fuzzy spectral clustering algorithm and the preset judgment threshold were used to detect whether there was an overlapping community in the unweighted undirected graph. If the degree of overlap between the community structures in graph was greater than the set threshold, the overlapping communities existed in graph. Finally, all the overlapping communities were combined into a big community. Simulation results prove that the proposed method can realize the automatic detection of overlapping community in complex network big data. Meanwhile, this method has the advantages of high accuracy, good stability and less time consumption.
引文
[1] 吴永亮,等.一种基于信息扩散的复杂网络重叠社区发现算法[J].计算机工程与科学,2016,38(4):792-799.
[2] 龙浩,汪浩.最优节点复制的复杂网络重叠社区发现算法[J].信息与控制,2017,46(2):178-185.
[3] 贺超波,等.一种集成链接和属性信息的社区挖掘方法[J].计算机学报,2017,40(3):601-616.
[4] 辛宇,杨静,谢志强.一种面向语义重叠社区发现的Link-Block算法[J].软件学报,2016,27(2):363-380.
[5] 陈俊宇,等.一种半监督的局部扩展式重叠社区发现方法[J].计算机研究与发展,2016,53(6):1376-1388.
[6] 王冠男.大数据环境下网络异常数据准确检测仿真[J].计算机仿真,2017,34(8):437-440.
[7] 许加书,韩忠愿,顾惠健.基于节点拓扑结构和属性的重叠社区检测算法[J].计算机应用研究,2016,33(12):3615-3619.
[8] 张黎烁,高继勋.基于节点尺度特征的重叠社区检测算法[J].河南理工大学学报:自然科学版,2016,35(5):706-712.
[9] 彭焱,溪利亚.局部优先的动态网络重叠社团及其演变模式检测[J].计算机工程,2016,42(12):188-195.
[10] 刘世超,朱福喜,冯曦.复杂网络的重叠社区及社区间的结构洞识别[J].电子学报,2016,44(11):2600-2606.
[2] 龙浩,汪浩.最优节点复制的复杂网络重叠社区发现算法[J].信息与控制,2017,46(2):178-185.
[3] 贺超波,等.一种集成链接和属性信息的社区挖掘方法[J].计算机学报,2017,40(3):601-616.
[4] 辛宇,杨静,谢志强.一种面向语义重叠社区发现的Link-Block算法[J].软件学报,2016,27(2):363-380.
[5] 陈俊宇,等.一种半监督的局部扩展式重叠社区发现方法[J].计算机研究与发展,2016,53(6):1376-1388.
[6] 王冠男.大数据环境下网络异常数据准确检测仿真[J].计算机仿真,2017,34(8):437-440.
[7] 许加书,韩忠愿,顾惠健.基于节点拓扑结构和属性的重叠社区检测算法[J].计算机应用研究,2016,33(12):3615-3619.
[8] 张黎烁,高继勋.基于节点尺度特征的重叠社区检测算法[J].河南理工大学学报:自然科学版,2016,35(5):706-712.
[9] 彭焱,溪利亚.局部优先的动态网络重叠社团及其演变模式检测[J].计算机工程,2016,42(12):188-195.
[10] 刘世超,朱福喜,冯曦.复杂网络的重叠社区及社区间的结构洞识别[J].电子学报,2016,44(11):2600-2606.