关于增量式PageRank的Monte Carlo方法的复杂性
On the Complexity of the Monte Carlo Method for Incremental PageRank
摘要:
本文扩展了[B]中对增量PageRank的分析。Bahmani,A. Chowdhury,和A.戈埃尔快速增量和个性化PageRank。VLDB 2011]。在这项工作中,作者证明了一个运行时间的$O(\frac{nR}{ε^2} \ln(m))$保持PageRank更新超过$m$边缘到达图中的$n$节点时,该算法存储每个节点的$R$随机游走和PageRank传送概率为$ε$。为了证明这个运行时间,他们假设边缘以随机顺序到达,并将其留给未来的工作来扩展他们的运行时间保证对抗边缘到达。在这篇文章中,我们通过展示一个图和一个运行时间为$Ω\left(R n m^{\lg{\frac{3}{2}(1-ε)}}\right)$的对抗边缘到达顺序来证明随机边缘顺序假设是必要的。更一般地,对于任何整数$d \geq 2$,我们构造一个图和对抗边顺序,其中运行时间是$Ω\left(R n m^{\log_d(H_d (1-ε))}\right)$,其中$H_d$是$d$的第三谐波数。
英文摘要:
This note extends the analysis of incremental PageRank in [B. Bahmani, A. Chowdhury, and A. Goel. Fast Incremental and Personalized PageRank. VLDB 2011]. In that work, the authors prove a running time of $O(\frac{nR}{ε^2} \ln(m))$ to keep PageRank updated over $m$ edge arrivals in a graph with $n$ nodes when the algorithm stores $R$ random walks per node and the PageRank teleport probability is $ε$. To prove this running time, they assume that edges arrive in a random order, and leave it to future work to extend their running time guarantees to adversarial edge arrival. In this note, we show that the random edge order assumption is necessary by exhibiting a graph and adversarial edge arrival order in which the running time is $Ω\left(R n m^{\lg{\frac{3}{2}(1-ε)}}\right)$. More generally, for any integer $d \geq 2$, we construct a graph and adversarial edge order in which the running time is $Ω\left(R n m^{\log_d(H_d (1-ε))}\right)$, where $H_d$ is the $d$th harmonic number.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。