基于三元组模体的网络核心-外围结构检测
Detection of core-periphery structure in networks by 3-tuple motifs
摘要:
近年来,网络的核心-外围结构作为一种中尺度结构受到了人们的关注。CP结构由密集的核心和稀疏连接的外围组成。本文受模体思想的启发,提出了一种基于三元组模体的CP结构检测算法。该算法首先根据节点的性质定义一个3元组模体,然后根据定义的模体构造一个模体邻接矩阵,最后将CP结构的检测转化为寻找一个最小模体电导的聚类。我们的算法可以检测不同的CP结构:包括单个或多个CP结构;和局部或全局CP结构。综合网络和经验网络的计算结果表明,该方法是有效的,可以应用于大规模网络。更重要的是,我们的算法是无参数的,其中核心和外围设备的检测,而不需要任何预定义的参数。
英文摘要:
Recently, the core-periphery (CP) structure of networks as one type of meso-scale structure has received attention. The CP structure is composed of a dense core and a sparse connected periphery. In this paper, we propose an algorithm to detect the CP structure based on the 3-tuple motif, which is inspired by the idea of motif. In this algorithm, we first define a 3-tuple motif by considering the property of nodes, and then a motif adjacency matrix is formed based on the defined motif, finally, the detection of the CP structure is converted to find a cluster that minimizes the smallest motif conductance. Our algorithm can detect different CP structures: including single or multiple CP structure; and local or global CP structures. Results in the synthetic and the empirical networks indicate that the method is efficient and can apply to large-scale networks. More importantly, our algorithm is parameter free, where the core and periphery are detected without the need for any predefined parameters.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。