A Scanning Method for Detecting Clustering Pattern of Both Attribute and Structure in Social Networksv

Abstract

Community/cluster is one of most important features in social networks. Many cluster detection methods are proposed to identify such an important pattern, but few of them are able to identify the statistical significance of the clusters, and take important attributes into account. In this study, we will review some definitions of networks and methods that are proposed to deal with the clusters in networks. Then, we will propose a scanning method originated for analyzing spatial data for identifying clusters in social networks. Since the properties of network are more complicated than those of spatial data, we verify the feasibility via simulation studies. The results show that the detection powers are affected by cluster sizes and connection probabilities, and the detection accuracy of our proposed method is better than that of the modularity method. In addition, we apply our proposed method to some empirical data to identify statistically significant clusters. Keywords: Community/cluster detection; Scanning window; Scan statistic; Structure and attribute cluster.?