融入H指数的局部拓展类重叠社区发现算法

Local Extension Class Overlapping Community Discovery Algorithm with H-index

本文针对传统的局部拓展算法具有较强的随机性和社区发现质量不高的缺陷,在种子拓展和H指数的基础上,提出了一种基于节点重要性和改进适应度的重叠社区发现算法(HDOC).算法采用H指数和局部影响力计算节点重要性,按此对节点排序和选种,使算法具有稳定性;且本文采用了一种新的适应度计算方法,兼顾考虑了节点自身和邻接点重要性,来对社区进行扩展,提高了社区的发现质量.通过在真实网络和LFR生成的人工网络中进行测试,并与其他社区发现算法对比实验表明,HDOC的社区识别能力较强,并且具有不错的时间效率.

Aiming at solving the defect that the traditional local extension algorithm has strong randomness and lowquality of community discovery,on the basis of seed expansion and H-index,an overlapping community discovery algorithm( HDOC) based on node importance and improved fitness is proposed. The algorithm uses the H-index and the local influence to calculate the node importance,and then sorts and selects the nodes according to them,which makes the algorithm stable;A newfitness calculation method is proposed to expand the community,taking into account its own and neighbor importance. It improves the quality of community discovery.Through testing in the real network and LFR generated artificial network and comparing with other community discovery algorithms,HDOC has strong community recognition capabilities and good time efficiency.