The document proposes methods to accelerate PageRank computations using extrapolation techniques. It discusses how PageRank works and is typically computed using an iterative power method. The authors' approach is to use successive PageRank vectors to estimate the components in the directions of the first few eigenvectors, subtracting them to remove their influence and speed convergence. Empirical results show quadratic extrapolation can significantly speed up PageRank convergence, though not enough for truly personalized computations. The extrapolation techniques may help accelerate other similar problems.