Nonlinear Algorithms for Static-Rank Computation

摘要: Mainstream link-based static-rank algorithms (e.g. PageRank and its variants) express the importance of a page as linear combination in-links compute scores by solving system in an iterative way. Such algorithms, however, may give apparently unreasonable staticrank results for some link structures. In this paper, we examine computation problem from viewpoint evidence build probabilistic model it. Based on model, argue that nonlinear formula should be adopted, due to correlation or dependence between links. We focus examining simple formulas which only consider links same domain. Experiments conducted 100 million web pages (with multiple quality evaluation metrics) show higher could yielded new algorithms. The convergence is also proved paper functional analysis.