Enhanced preprocessed multi-step splitting iterations for computing PageRank

TL;DR

This paper introduces a new multi-step splitting iteration method and its modifications using Arnoldi techniques to accelerate PageRank computation, especially when the damping factor is near 1, demonstrating significant performance improvements.

Contribution

It proposes a novel multi-step splitting iteration approach for faster PageRank calculation and extends it with Arnoldi-based modifications, addressing convergence issues of traditional methods.

Findings

Significant speedup in PageRank computation with the new methods

Effective handling of high damping factor scenarios

Demonstrated improvements on large test matrices

Abstract

In recent years, the PageRank algorithm has garnered significant attention due to its crucial role in search engine technologies and its applications across various scientific fields. It is well-known that the power method is a classical method for computing PageRank. However, there is a pressing demand for alternative approaches that can address its limitations and enhance its efficiency. Specifically, the power method converges very slowly when the damping factor is close to 1. To address this challenge, this paper introduces a new multi-step splitting iteration approach for accelerating PageRank computations. Furthermore, we present two new approaches for computating PageRank, which are modifications of the new multi-step splitting iteration approach, specifically utilizing the thick restarted Arnoldi and generalized Arnoldi methods. We provide detailed discussions on the construction and theoretical convergence results of these two approaches. Extensive experiments using large test matrices demonstrate the significant performance improvements achieved by our proposed algorithms.