Analysis Of SNS User Features Based On IGGN Algorithm

Attribute and relation data are two types of data in many social network data. At present, the group detection analysis for social networks usually only focus on the relationship among the entities while ignore the attribute of entity. For the reason, if a group detection algorithm taking into account attribute data as well as relation data, it must be achieve more accurate results than those methods that only using relation data. In this paper, we introduce a Weighted Informative Graph model to solve the problem of merge the attribute and relationship and then propose a Cluster Mixed Index to measure the purity of clusters. Finally, we experiment on a coauthor network. The experimental evaluation demonstrates that IGGN which considers both attribute and relation has higher accuracy than the GN algorithm based on the simple link structure, and the Cluster Mixed Index is suitable to be used as the termination condition to control clustering process.