正则化度修正随机块模型的演化网络社团发现

目前大多数用于社团发现问题的模型只适用于静态网络而忽视了时序信息,因此,无法较好地建模真实世界数据。针对该问题,提出一种基于度修正随机块模型的演化社团发现模型。根据演化聚类框架的原理,基于社团隶属矩阵将一个正则项引入到度修正随机块模型的目标函数中。利用网络交叉验证方法进行模型选择,处理社团个数随时间变化的演化网络,从而克服由于假定社团个数为常量而导致的与真实世界数据不相符合的问题。实验结果表明,与经典的动态随机块模型和Facet Net相比,该模型具有较高的准确性和较低的误差率。...

Full description

Saved in:
Bibliographic Details
Published in计算机工程 Vol. 42; no. 8; pp. 134 - 138
Main Author 王亭亭 戴维迪 焦鹏飞 李晓明
Format Journal Article
LanguageChinese
Published 天津大学计算机科学与技术学院,天津,300072%天津大学计算机科学与技术学院,天津300072 2016
天津市认知计算与应用重点实验室,天津300072
Subjects
模型选择
演化分析
演化网络
社团发现
节点特性
随机块模型
随机块模型
stochastic block model
node peculiarity
模型选择
evolutionary analysis
evolving network
model selection
演化网络
演化分析
节点特性
community detection
社团发现
Online AccessGet full text
ISSN1000-3428
DOI10.3969/j.issn.1000-3428.2016.08.025

Cover

More Information
Summary:目前大多数用于社团发现问题的模型只适用于静态网络而忽视了时序信息,因此,无法较好地建模真实世界数据。针对该问题,提出一种基于度修正随机块模型的演化社团发现模型。根据演化聚类框架的原理,基于社团隶属矩阵将一个正则项引入到度修正随机块模型的目标函数中。利用网络交叉验证方法进行模型选择,处理社团个数随时间变化的演化网络,从而克服由于假定社团个数为常量而导致的与真实世界数据不相符合的问题。实验结果表明,与经典的动态随机块模型和Facet Net相比,该模型具有较高的准确性和较低的误差率。
Bibliography:evolving network; evolutionary analysis;community detection; model selection;stochastic block model; node peculiarity
31-1289/TP
WANG Tingting ,DAI Weidi ,JIAO Pengfei ,LI Xiaoming （ 1. School of Computer Science and Technology, Tianjin University, Tianjin 300072, China; 2. Tianjin Key Laboratory of Cognitive Computing and Application,Tianjin 300072,China）
Nowadays,many models for community detection are designed only for static networks, which ignore the temporal information and are always not ideal to model the real world data. In order to solve this problem, an evolving community detection model based on the degree-corrected block model is proposed. According to the theory of the framework of evolutionary clustering, the model introduces a regularization term based on the community membership matrix into the objective function of the degree-corrected stochastic block model. The network cross-validation approach is utilized for model selection, so the proposed method is able to deal with evolving networks with
ISSN:1000-3428
DOI:10.3969/j.issn.1000-3428.2016.08.025