Joint Cluster Analysis of Attribute Data and Relationship Data: the Connected k-Center Problem.

摘要: Attribute data and relationship are two principle types of data, representing the intrinsic extrinsic properties entities. While attribute has been main source for cluster analysis, such as social networks or metabolic becoming increasingly available. It is also common to observe both carry orthogonal information in market segmentation community identification, which calls a joint analysis so achieve more accurate results. For this purpose, we introduce novel Connected kCenter problem, taking into account well data. We analyze complexity problem prove its NP-completeness. present constant factor approximation algorithm, based on further design NetScan, heuristic algorithm that efficient large, real databases. Our experimental evaluation demonstrates meaningfulness accuracy NetScan