Local Higher-Order Graph Clustering

Local Higher-Order Graph Clustering

Hao Yin (Stanford University)
Austin R. Benson (Stanford University)
Jure Leskovec (Stanford University)
David F. Gleich (Purdue University)

Local graph clustering methods aim to find a cluster of nodes by exploring a small region of the graph. These methods are attractive because they enable targeted clustering around a given seed node and are faster than traditional global graph clustering methods because their runtime does not depend on the size of the input graph. However, current local graph partitioning methods are not designed to account for the higher-order structures crucial to the network, nor can they effectively handle directed networks. Here we introduce a new class of local graph clustering methods that address these issues by incorporating higher-order network information captured by small subgraphs, also called network motifs. First, we show how to adapt the approximate personalized PageRank algorithm to find clusters containing a seed node with minimal motif conductance, a generalization of the conductance metric for network motifs. We also generalize existing theory to maintain the properties of fast running time (independent of the size of the graph) and cluster quality (in terms of motif conductance). For community detection tasks on both synthetic and real-world networks, our new framework outperforms the current edge-based personalized PageRank methodology. Second, we develop a theory of node neighborhoods for finding sets that have small motif conductance, where the motif is a clique. We apply these results to the case of finding good seed nodes to use as input to the personalized PageRank algorithm.

More on http://www.kdd.org/kdd2017/.