Hits:
Affiliation of Author(s):计算机科学与工程学院
Journal:Modern Physics Letters B
Funded by:省、市、自治区科技项目
Key Words:Overlapping community detection;ensemble learning;line graph
Abstract:Overlapping community detection is a hot topic in the research of data mining and graph theory. In this paper, we propose a link community detection method based on ensemble learning (LCDEL). First, we transform graph into line graph and construct node adjacency matrix of line graph. Second, we calculate node distance of line graph through a new distance metric and get node distance matrix of line graph. Third, we use PCA method to reduce dimensions of node distance matrix of line graph. Then, we cluster on the reduced node distance matrix by k-means clustering algorithm. Finally, we convert line graph back into original graph and get overlapping communities of original graph with ensemble learning. Experimental results on several real-world networks demonstrate effectiveness of LCDEL method in terms of Normalized Mutual Information (NMI), Extended Modularity (EQ) and F-score evaluation metrics.
Co-author:刘致华,wangguishen,周玉,王贵参
First Author:wanghongmei
Indexed by:Journal paper
Volume:34
Issue:27
Page Number:1
Translation or Not:no
Date of Publication:2020-09-30