The pagerank algorithm computes probability iteratively until convergence, which is measured by checking if the difference between the values computed in the current iteration and those from the previous is below a threshold.
I remember (could be wrong) seeing somewhere that this threshold is usually set 0.01, or maybe it was set 0.01 in the original publication, but I could not find any literature on this.
Can anyone give me some suggestions, is there such a `typical' value to use as convergence threshold, or is there any way to decide what the threshold should be, without empirically testing by experiments?
Many thanks!
