Weighted clustering of attributed multi-graphs
Date
2017ISSN
0010-485XSource
ComputingVolume
99Issue
9Pages
813-840Google Scholar check
Keyword(s):
Metadata
Show full item recordAbstract
An information network modeled as an attributed multi-graph contains objects described by heterogeneous attributes and connected by multiple types of edges. In this paper we study the problem of identifying groups of related objects, namely clusters, in an attributed multi-graph. It is a challenging task since a good balance between the structural and attribute properties of the objects must be achieved, while each edge-type and each attribute contains different information and is of different importance to the clustering task. We propose a unified distance measure for attributed multi-graphs which is the first to consider simultaneously the individual importance of each object property, i.e. attribute and edge-type, as well as the balance between the sets of attributes and edges. Based on this, we design an iterative parallelizable algorithm for CLustering Attributed Multi-graPhs called CLAMP, which automatically balances the structural and attribute properties of the vertices, and clusters the network such that objects in the same cluster are characterized by similar attributes and connections. Extensive experimentation on synthetic and real-world datasets demonstrates the superiority of the proposed approach over several state-of-the-art clustering methods. © 2016, Springer-Verlag Wien.