Graph Partitioning Algorithm with LSH: Poster Extended Abstract

Abstract

In this paper, we propose a graph partitioning algorithm which is based on the locality-sensitive hashing (LSH). Its time/space complexity is O(n), n is the number of vertices in graph. For all kinds of hyperscale graphs, it works at the speed of random partitioning method approximately. Compared with the mainstream graph partitioning algorithms, the new algorithm owns a simpler processing pipeline and avoids irregular memory access generated by graph traversals. The experimental results show that the new algorithm achieves 10× faster than Metis and 2× faster than label propagation algorithm at the cost of a reasonable precision loss.