基于Kullback-Leibler距离的二分网络社区发现方法
发布时间：2018-10-18 14:13
【摘要】：由于二分网络特殊的二分结构,使得基于单模网络的现有社区发现算法无法适用。提出一种基于Kullback-Leibler距离的二分网络社区发现算法,该算法将异质节点间的连接关系转换为其在用户节点集上的连接概率分布,并建立基于概率分布的KL相似度衡量节点连接模式的差异性,从而克服二分结构对节点相似性评估的不利影响,实现对二分网络异质节点的社区发现。在人工网络和真实网络上的实验和分析表明,该算法能够有效挖掘二分网络社区结构,改善二分网络社区发现的准确性和效率。
[Abstract]:Due to the special dichotomy structure of binary networks, the existing community discovery algorithms based on single mode networks can not be applied. A community discovery algorithm for binary networks based on Kullback-Leibler distance is proposed. The algorithm converts the connection relationship between heterogeneous nodes into its connection probability distribution on the user node set. The KL similarity based on probabilistic distribution is established to measure the difference of node connection patterns, so as to overcome the adverse effect of binary structure on node similarity evaluation and realize community discovery of heterogeneous nodes in binary networks. Experiments and analysis on artificial network and real network show that the algorithm can effectively mine the binary network community structure and improve the accuracy and efficiency of binary network community discovery.
【作者单位】： 河南工学院计算机科学与技术系;河南师范大学网络中心;
【基金】：河南省高等学校重点科研资助项目(15A520063,16A520083)
【分类号】：O157.5
，
本文编号：2279377
[Abstract]:Due to the special dichotomy structure of binary networks, the existing community discovery algorithms based on single mode networks can not be applied. A community discovery algorithm for binary networks based on Kullback-Leibler distance is proposed. The algorithm converts the connection relationship between heterogeneous nodes into its connection probability distribution on the user node set. The KL similarity based on probabilistic distribution is established to measure the difference of node connection patterns, so as to overcome the adverse effect of binary structure on node similarity evaluation and realize community discovery of heterogeneous nodes in binary networks. Experiments and analysis on artificial network and real network show that the algorithm can effectively mine the binary network community structure and improve the accuracy and efficiency of binary network community discovery.
【作者单位】： 河南工学院计算机科学与技术系;河南师范大学网络中心;
【基金】：河南省高等学校重点科研资助项目(15A520063,16A520083)
【分类号】：O157.5
，
本文编号：2279377
本文链接：https://www.wllwen.com/kejilunwen/yysx/2279377.html
