SMAC: Subgraph Matching and Centrality in Huge Social Networks

Abstract

Classical centrality measures like betweenness, closeness, eigenvector, and degree centrality are application and user independent. They are also independent of graph semantics. However, in many applications, users have a clear idea of who they consider important in graphs where vertices and edges have properties, and the goal of this paper is to enable them to bring their knowledge to the table in defining centrality in graphs. We propose a novel combination of sub graph matching queries which have been studied extensively in the context of both RDF and social networks, and scoring functions. The resulting SMAC framework allows a user to define what he thinks are central vertices in a network via user-defined sub graph patterns and certain mathematical measures he specifies. We formally define SMAC queries and develop algorithms to compute answers to such queries. We test our algorithms on real-world data sets from CiteSeerX, Flickr, YouTube, and IMDb containing over 6M vertices and 15M edges and show that our algorithms work well in practice.