Does the given number of walks on the given graph and returns the mean time per walk, in milliseconds.
Does the given number of walks on the given graph and returns the mean time per walk, in milliseconds. Used to autmatically balance forward and reverse work.
Estimates the personalized PageRank score from the given source to the given target.
Estimates the personalized PageRank score from the given source to the given target. The the true score is greater than the given minimumPPR, then the result will have mean relative error less than the given bound. If guaranteeRlativeError is set, then the relative error bound will be guaranteed (at the cost of slower execution); otherwise the relative error is only smaller than the given relativeError bound on average (based on experiments in the above publication).
Computes personalized PageRank from a source node to a target node.
Computes personalized PageRank from a source node to a target node. (Convenience method that calls the above method)
Given a target nodeId, returns a map estimates such that estimates(v) approximates ppr(v, target) with additive error pprErrorTolerance.
Samples from the PageRank distribution personalized to the given source distribution by doing a single random walk.
Samples from the PageRank distribution personalized to the given source by doing a single random walk.
Contains methods related to personalized PageRank estimation. All methods operate in the context of the given graph, teleportProbability, and random. See the paper "Personalized PageRank Estimation and Search: A Bidirectional Approach" by Lofgren, Banerjee, and Goel for more information on the algorithm implemented.