An Algorithm Of Authoritative Page Mining Based On K-L Transformation

Up to now ,the algorithm of web search engine can be divided into two classes: the first class is based on vector space model (VSM) by which algorithms matche degree between the key words and the documents by their content's relation; the second class is based on Web links by which algorithms estimate the rank of a web page.Based on the idea of the web link structure, Sergey Brin and Lawrence Page pioneered research with the Page Rank algorithm in 1998. In the same year, J. Kleinberg brought forward the HITS algorithm. Yet some other researchers bring forward a lot of algorithms based on web link structure, for example, SALSA, PHITS, Bayesian and so on. These algorithms have been used in practice, and been proved having good effect.It's well known that the discrete Karhunen-Loeve transformation is based on defining an optimal basis for orthogonal expansion of a set of signals so as to minimize the mean squared error in reconstruction from a truncated basis.Expressing the authority in the Web as a matrix, denoting the convexity between pages as the element of the matrix, authoritative pages can be viewed as the centralized part of its energy, in this way, the Karhunen-Loeve transform can be used to mine authoritative pages in the Web.Based on above consideration, this paper combine technology from relative domain, introduce a new algorithm which is called authoritative page mining algorithm (APMA) based on K-L transformation. A lot of tests show that the APMA can find authoritative pages which are interested by Web browers effectively.