PageRank is an interesting alogorithm developed by Google's founders LarryPage and SergeyBrin at Stanford University. Using the hyperlink structure information of the Web, it computes an authority value for each Web page, which can be later used to improve the ranking process. The authority value of a Web page is based on the authority values of the Web pages linking to it. Google claims that PageRank continues to be the basis of its ranking mechanism; [WWW]

Unlike JonKleinberg's HITS algorithm which, for a given query, finds authority and hub documents in the Web, the PageRank algorithm is run during indexing time, and is query-independent.

The AbsorbingModel; [WWW] developed within the Terrier project, takes a different approach and proposes a new model for hyperlink analysis, which can be used in either a query-independent, or a query-dependent way.

Experiments conducted within the TREC initiative showed that there were no compelling evidence to suggest that HyperlinkStructureAnalysis improves the retrieval performance.

last edited 2005-02-12 00:42:42 by IadhOunis