《计算机应用研究》|Application Research of Computers

基于节点从属度的加权网络重叠社区划分算法

Weighted network overlap community partition algorithm based on node dependency degree

免费全文下载 (已被下载 次)  
获取PDF全文
作者 付立东,郝伟,李凡
机构 1.西安科技大学 计算机科学与技术学院,西安 710054;2.西安电子科技大学 计算机科学与技术学院,西安 710071
统计 摘要被查看 次,已被下载
文章编号 1001-3695(2021)02-009-0377-05
DOI 10.19734/j.issn.1001-3695.2020.02.0014
摘要 针对传统社区划分算法忽略现实世界网络特征导致社区划分准确率低的问题,提出了一种基于节点从属度的加权网络重叠社区划分算法。该算法提出加权网络模型,通过模型得到了能刻画出真实网络结构的加权网络;通过网络拓扑结构定义了核心社区,核心社区对社区划分的准确性有着重要作用。该算法计算节点与核心社区间的从属度,并与从属度阈值进行比较进行核心社区扩展,根据扩展模块度优化思想,通过不断地调整从属度阈值直到获得最优的社区结构,完成重叠社区划分。在人工网络数据集和真实世界网络数据集上与已有算法进行实验对比,实验结果验证了所提算法能够准确、有效地检测出重叠社区。
关键词 复杂网络; 加权网络; 节点从属度; 重叠社区; 扩展模块度
基金项目 国家自然科学基金资助项目(61432010,61502363)
陕西省自然科学基础研究项目(2020JM-526,2020JM-533)
西安科技大学博士后科研启动项目(2018QDJ049)
本文URL http://www.arocmag.com/article/01-2021-02-009.html
英文标题 Weighted network overlap community partition algorithm based on node dependency degree
作者英文名 Fu Lidong, Hao Wei, Li Fan
机构英文名 1.College of Computer Science & Technology,Xi'an University of Science & Technology,Xi'an 710054,China;2.School of Computer Science & Technology,Xidian University,Xi'an 710071,China
英文摘要 Aiming at the problem that the traditional community dividing algorithms ignore the real-world network characteristics and the accuracy of community dividing is low, this paper proposed a weighted network overlapping community dividing algorithm based on node membership. The algorithm proposed a weighted network model and obtained a weighted network that could describe the real network structure through the model. It defined the core community by the network topology structure, and the core community played an important role in the accuracy of community division. The algorithm calculated the degree of membership between the node and the core community, and compared it with the threshold of membership to expand the core community. According to the optimization idea of the extended modularity, by continuously adjusting the thres-hold of membership until obtained the optimal community structure, it divided the overlapping community. Compared with the existing algorithms on the artificial network dataset and the real-world network dataset, the experimental result verifies that the proposed algorithm can accurately and effectively detect overlapping communities.
英文关键词 complex network; weighted network; degree of node dependency; overlapping communities; extended modules
参考文献 查看稿件参考文献
 
收稿日期 2020/2/8
修回日期 2020/3/26
页码 377-381
中图分类号 TP399
文献标志码 A