Fragile link structure in PageRank computation
Hideaki Ishii, Roberto Tempo · 2009
In this paper, we study the so-called PageRank algorithm used at Google that determines the importance of each webpage. We first provide a brief overview of this algorithm. Then, we consider the problem of computing the changes in PageRank values when the web data contains erroneous link information. Based on interval matrix techniques, we develop an efficient centralized algorithm using linear programming. The theoretical results are verified by means of extensive simulations for a small web with 150 pages and a larger web of 1,200 pages.