DF* PageRank: Improved Incrementally Expanding Approaches for Updating PageRank on Dynamic Graphs

TL;DR

This paper presents improved incremental algorithms, DF* and DF-P, for efficiently updating PageRank scores on dynamic graphs, significantly outperforming existing methods in real-world and large static graph scenarios.

Contribution

Introduction of the DF* and DF-P approaches that efficiently update PageRank on dynamic graphs with minimal overhead and high scalability.

Findings

DF* and DF-P outperform static and dynamic traversal methods by up to 15.2x.

Approaches achieve 1.8x speedup with each doubling of threads.

Significant improvements on real-world and large static graphs with batch updates.

Abstract

PageRank is a widely used centrality measure that assesses the significance of vertices in a graph by considering their connections and the importance of those connections. Efficiently updating PageRank on dynamic graphs is essential for various applications due to the increasing scale of datasets. This technical report introduces our improved Dynamic Frontier (DF) and Dynamic Frontier with Pruning (DF-P) approaches. Given a batch update comprising edge insertions and deletions, these approaches iteratively identify vertices likely to change their ranks with minimal overhead. On a server featuring a 64-core AMD EPYC-7742 processor, our approaches outperform Static and Dynamic Traversal PageRank by 5.2x/15.2x and 1.3x/3.5x respectively - on real-world dynamic graphs, and by 7.2x/9.6x and 4.0x/5.6x on large static graphs with random batch updates. Furthermore, our approaches improve performance at a rate of 1.8x/1.7x for every doubling of threads.