efficient computation of pagerank（高效的网页排名的计算）.pdf

Efficient Computation of PageRank Taher H. Haveliwala∗ Stanford University taherh@ October 18, 1999 Abstract This paper discusses efficient techniques for computing PageRank, a ranking met- ric for hypertext documents. We show that PageRank can be computed for very large subgraphs of the web (up to hundreds of millions of nodes) on machines with limited main memory. Running-time measurements on various memory configurations are presented for PageRank computation over the 24-million Stanford WebBase archive. We discuss several methods for analyzing the con- vergence of PageRank based on the induced ordering of the pages. We present convergence results helpful for determining the number of iterations necessary to achieve a useful PageRank assignment, both in the absence and presence of search queries. 1 Introduction The dramatic growth of the world-wide web, now exceeding 800 million pages [11], is forcing modern web search engines to look beyond simply the content of pages in providing relevant answers to queries. Recent work in utilizing the link structure of the web for improving the quality of search results is promising. In particular, the Google search engine uses PageRank, an iterative algorithm that determines the “importance” of a web page based on the “importance” of its parent pages [5]. A related algorithm used by the IBM HITS system maintains a hub and an authority score for each page, in which the authority score for a page is determined by the hub scores of its parents, and the hub score of a