基于多核心标签传播的复杂网络重叠社区识别方法

发布时间：2018-04-24 20:43

本文选题：复杂网络 + 社区识别　；参考：《通信学报》2017年02期

【摘要】：针对传统基于标签传播的重叠社区识别方法存在较强的随机性,以及需要预设相关阈值来辅助完成社区识别等缺陷,提出基于多核心标签传播的重叠社区识别方法(OMKLP)。在分析节点度以及节点与邻居节点的局部覆盖密度后提出核心节点评价模型,并在此基础上给出局部核心节点识别方法;基于局部核心节点,提出新的面向重叠社区的异步标签传播策略,该策略能够快速地识别出社区内部节点与边界节点,以获得重叠社区结构;提出重叠节点分析方法,进一步提高识别重叠节点准确度。OMKLP算法无需掌握任何先验知识,仅在掌握网络基本信息(点、边)基础上,便能够准确识别出重叠社区结构,从而有效解决了传统标签传播算法所存在的缺陷。在基准网络和真实网络上进行测试,并与多个经典算法进行对比分析,实验结果验证了所提算法的有效性和可行性。
[Abstract]:In view of the shortcomings of traditional overlapping community identification method based on label propagation and the need of preset correlation threshold to assist in community identification, an overlapping community identification method based on multi-core label propagation is proposed. After analyzing the degree of nodes and the local coverage density of nodes and neighbors, the evaluation model of core nodes is put forward, and the identification method of local core nodes is given, which is based on the local core nodes, the local core nodes, the local core nodes, the local core nodes and the local core nodes. A new asynchronous label propagation strategy for overlapping communities is proposed, which can quickly identify the nodes within the community and the boundary nodes to obtain the overlapping community structure, and propose an analysis method for overlapping nodes. Further improve the accuracy of identifying overlapped nodes. OMKLP algorithm does not need to master any prior knowledge, only on the basis of the basic network information (points, edges), can accurately identify the overlapping community structure, Thus, the defects of the traditional label propagation algorithm are effectively solved. The proposed algorithm is tested on the benchmark network and real network, and compared with several classical algorithms. The experimental results show that the proposed algorithm is effective and feasible.
【作者单位】：嘉兴学院数理与信息工程学院;哈尔滨工程大学计算机科学与技术学院;
【基金】：国家自然科学基金资助项目(No.61672179,No.61370083,No.61402126) 教育部人文社会科学研究青年基金资助项目(No.15YJCZH088) 浙江省自然科学基金资助项目(No.LY16F020027) 浙江省教育厅科研基金资助项目(No.Y201636127,No.Y201533771)~~
【分类号】：O157.5

【相似文献】