Revisiting Local Computation of PageRank: Simple and Optimal

TL;DR

This paper analyzes and improves the complexity bounds of local PageRank computation algorithms, showing that a simple existing method is already optimal and providing tighter bounds for estimating PageRank centrality.

Contribution

It provides new worst-case complexity bounds for local PageRank algorithms, proving the optimality of a classic method and improving existing upper and lower bounds with simple techniques.

Findings

ApproxContributions algorithm is optimal in worst-case complexity.

New upper bound for local PageRank estimation: $O(n^{1/2} imes ext{min}( ext{in-degree}^{1/2}, ext{out-degree}^{1/2}, m^{1/4}))$.

Matching lower bounds confirm the tightness of the bounds.

Abstract

We revisit the classic local graph exploration algorithm ApproxContributions proposed by Andersen, Borgs, Chayes, Hopcroft, Mirrokni, and Teng (WAW '07, Internet Math. '08) for computing an $\epsilon$-approximation of the PageRank contribution vector for a target node $t$ on a graph with $n$ nodes and $m$ edges. We give a worst-case complexity bound of ApproxContributions as $O(n\pi(t)/\epsilon\cdot\min(\Delta_{in},\Delta_{out},\sqrt{m}))$, where $\pi(t)$ is the PageRank score of $t$, and $\Delta_{in}$ and $\Delta_{out}$ are the maximum in-degree and out-degree of the graph, resp. We also give a lower bound of $\Omega(\min(\Delta_{in}/\delta,\Delta_{out}/\delta,\sqrt{m}/\delta,m))$ for detecting the $\delta$-contributing set of $t$, showing that the simple ApproxContributions algorithm is already optimal. We also investigate the computational complexity of locally estimating a node's PageRank centrality. We improve the best-known upper bound of $\widetilde{O}(n^{2/3}\cdot\min(\Delta_{out}^{1/3},m^{1/6}))$ given by Bressan, Peserico, and Pretto (SICOMP '23) to $O(n^{1/2}\cdot\min(\Delta_{in}^{1/2},\Delta_{out}^{1/2},m^{1/4}))$ by simply combining ApproxContributions with the Monte Carlo simulation method. We also improve their lower bound of $\Omega(\min(n^{1/2}\Delta_{out}^{1/2},n^{1/3}m^{1/3}))$ to $\Omega(n^{1/2}\cdot\min(\Delta_{in}^{1/2},\Delta_{out}^{1/2},m^{1/4}))$ if $\min(\Delta_{in},\Delta_{out})=\Omega(n^{1/3})$, and to $\Omega(n^{1/2-\gamma}(\min(\Delta_{in},\Delta_{out}))^{1/2+\gamma})$ if $\min(\Delta_{in},\Delta_{out})=o(n^{1/3})$, where $\gamma>0$ is an arbitrarily small constant. Our matching upper and lower bounds resolve the open problem of whether one can tighten the bounds given by Bressan, Peserico, and Pretto (FOCS '18, SICOMP '23). Remarkably, the techniques and analyses for proving all our results are surprisingly simple.