Pre-partitioning based distributed algorithms for graph pattern queries

Abstract

Graphs are everywhere from a biological network to the World Wide Web, and they are usually enormous. As graph patterns represent various characteristics of the graph, diverse interesting graph mining applications rely on graph pattern analyses. In this dissertation, we propose scalable distributed algorithms for three graph pattern queries that are the basic building blocks of many other graph pattern queries. The three pattern queries are the triangle pattern query, the subgraph pattern query, and the connected component pattern query. The proposed algorithms are based on vertex coloring and pre-partitioning for balancing the workload and minimizing the size of intermediate data and the number of rounds.