A Number Of Studies Basesd On PageRank Sort Algorithm Improvenment

Google's global information quantity is biggest, it's inquiry number of times are most and it is the most famous search engine. Google uses the most long ranking algorithm is the PageRank ranking algorithm. The PageRank algorithm is widely used measure of the importance of web-pages, which according to the link structure between web-pages to web-pages to give each grade, based on scores of high or low rankings given successively. Explained from mathematics angle, may regard as a Markov random walk model, based on the web-page next step link information computation web-page transition probability, with a smooth distribution of Markov chain as the ultimate sort of value to the pages.The traditional PageRank algorithm uses the structure information of the web to judge the importance of web-pages, moreover in the computation process,consider the page's all chains as the equally important, namely all chains are gained equally PageRank value. This article thought that was a fatal defect of traditional PageRank algorithm. Inthis paper, by analyzing the means of the traditional PageRank algorithm——each chainenters the page regards as "the vote" page, the important page's vote has high importance, the number of votes of the page decided the importance of it. Therefore the page should according the different importance of each chain to give different weights, which are given to improve the algorithm targeted. Theoretically, the improved algorithm in line with the mind of the traditional PageRank algorithm much more; Through the eration result of matlab experiment, compared with the computed result of the traditional PageRank algorithm and improved algorithm's, proved that the improved algorithm promotes the PageRank value of the important pages, reduces the PageRank value of the unimportant pages, moreover it uses more less value of d can achieves the compute result of the traditional PageRank algorithm.In the numerical analysis,the condition number of a numerical problem is the measurement of the difficult of calculate, that is, this well-posedness of this problem. A problem has low condition number is known as the "well- conditioned",but has high condition number is known as the "ill- conditioned".Since there have hundreds of millions of pages, calculates the PageRank value is not to set the equation, but with computer to iterate , educe the approximate solution. If a equations has higher condition number, it is more difficult to use common method calculate the result. In the fourth chapter of this article, discussed the condition number of matrix, analysised of the error of it's approximate solution, proved that reduces the damping factor's value can reduce the condition number of the limit distribution of Markov chain, namely the approximate solution of the limit distribution of Markov chain is more precise.