摘要
针对传统算法社团划分精度较低以及模块度函数分辨率低的问题,提出一种基于相关拓扑势的社团发现算法,简称BITP算法。该算法考虑节点的相关性因素,引入相关拓扑势来衡量节点的影响力,寻找出其中的极大势值点,采用标签传播的思想对社团的规模进行控制。在人工合成网络和真实网络上,与多种算法进行实验对比,结果表明该算法多次运行结果相对稳定且社团划分精度较高。算法时间复杂度为O(n),且不需要先验知识,更适合大规模复杂网络上的社团结构挖掘。
Since the traditional methods obtain low precision in division and low resolution in module function,an algorithm of community detection BITP is proposed based on the interrelated topological potential. The algorithm introduces the interrelated topological potential to evaluate the influence of nodes by considering the correlation factor between nodes. The nodes with extreme potential are searched at first. The sizes of the local communities are controlled by adopting the method of label propagation. The experimental results on synthetic and real-world networks show that the proposed algorithm is relatively stable and achieves higher precision. It is more suitable for detecting community structure in large-scaled and complex networks with a time complexity of O（ n） and no prior knowledge.
出处
《计算机应用与软件》
2017年第1期258-262,269,共6页
Computer Applications and Software
基金
河南省科技攻关计划项目(142102210435)
河南省高等学校矿山信息化重点学科开放实验室开放基金项目(ky2012-02)
关键词
社团结构
复杂网络
相关拓扑势
标签传播
Community structure
Complex network
Interrelated topological potential
Label propagation
