Clustering Networks by Structures

This paper presents some new ways of uncovering underlying structures, including the roles that vertices play in the network. We begin by defining a property we call structural similarity. We then define a structure-connected cluster (SCC) as a collection of vertices they have strong structural similarity. Finally, we propose an algorithm that only finds SCCs that represent groups of peers in networks, but also identify vertices that play special roles such hubs that bridge clusters and outsiders that are marginally connected to clusters. Identifying vertex roles is useful for applications such as viral marketing and epidemiology. For example, hubs are responsible for spreading ideas or disease. In contrast, outsiders have little or no influence, and may be isolated as noise in the data. An empirical evaluation of the algorithm using both synthetic and real datasets demonstrates superior performance over other methods such as modularity-based algorithms.