High-Order Structure Exploration on Massive Graphs

Modeling and exploring high-order connectivity patterns, also called network motifs, are essential for understanding the fundamental structures that control mediate behavior of many complex systems. For example, in social networks, triangles have been proven to play role communities; online transaction detecting directed looped transactions helps identify money laundering activities; personally identifiable information star-shaped may correspond a set synthetic identities. Despite ubiquity such structures, existing graph clustering methods either not designed or suffer from prohibitive computational cost when modeling large-scale networks. This article generalizes challenges multiple dimensions. First ( Model ), we introduce notion conductance, define diffusion core, which is based on random walk induced by user-specified structure. Second Algorithm propose novel structure-preserving framework named HOSGRAP , partitions into structure-rich clusters polylogarithmic time with respect number edges graph. Third Generalization generalize our proposed algorithm solve real-world problems various types graphs, as signed bipartite multi-partite graphs. Experimental results both real graphs demonstrate effectiveness efficiency algorithms.