A good explanation about the convergence of various algorithms. This paper also describes an adaptive and on-line algorithm for computing the page importance. It can be used for focus crawling as well as for search engine's ranking.

HITs is a link-structure analysis algorithm which ranks pages by "authorities" (pages which have many incoming links and provide the best source of information on a given topic) and "hubs" (pages which have many outgoing links and provide useful lists of possibly relevant pages). Ranking is performed at query time. [PDF format]

This paper describes a prototype system, later known as the Teoma Search Engine. It performs a Link Analysis, loosely based on the Kleimberg method, and computed at query time.

A hierarchical approach for computing PageRank. The local PageRanks of page for each host are computed independently and then used to compute the global PageRank of Web Graph.

A paper about the computation of PageRank using the standard Power Method and the new Quadratic Extrapolation which computes the principal eigenvector of the Markov matrix representing the Web link graph with an increased speed up of about 50-300%.

A survey on PageRank, Hits and SALSA. It also describes two Bayesian statistical algorithms for ranking of hyperlinked documents and the concepts of monotonicity and locality, as well as various concepts of distance and similarity between ranking algorithms.

Given a typical user query to find quality documents related to the query topic. It uses an Hits variation.

It proposes a new weighted HITS-based method that assigns appropriate weights to in-links of root documents and combines content analysis with HITS-based algorithms.

A Kleimberg's algorithm improvement. [PDF format]

Postscript-format slides which introduces citation importance ranking by Larry Page, Google's founder.

Do Hits and PageRank (and some variations) give stable rankings under small perturbations to the linkage patterns? [PS format]

A general framework for measuring the quality of an index and providing the background on the PageRank and Random Walks. Imagine a Web surfer who wanders the Web. At each step, he/she either jumps to a page on the Web chosen uniformly at random, or follows a link chosen from those on the current page.

Describes efficient techniques for computing PageRank.

Lossy encoding for large scale PageRank calculation.

A poster paper by Stanford db group which describes iterative methods for calculating PageRank. [PDF format]

Lawrence Page's PageRank Patent.

PageRank's values on the Web follow a power law. An high in-degree of a node does not imply high PageRank, and vice versa. [PDF format]

It shows some theoretical results for understanding the distribution of the score in the Web according to PageRank. Seven golden rules for building good pages are presented. [PDF format]

It introduces a probabilistic model that integrates link topology (used to identify important pages), anchor text (used to augment the text of cited pages), and activation (spread to linked pages). Experiments are on MSN Directory. [PDF format]

A focused search algorithm (SALSA) based on Markov chains. It starts with a query on a broad topic, discards useless links, and then weights the remaining terms. A stochastic crawl is used to discover the authorities on this topic. [PS format]

Link Popularity algorithms biased according to a user-specified set of given interesting pages.

Information on the algorithm, how to increase PageRank, what diminishes it and how to distribute PageRank within a website.

The CLEVER search engine incorporates several algorithms that make use of hyperlink structure for discovering information on the Web. It is an extension of Hits method.

An eingenvalues algorithm for calculating reputation in P2P networks and isolating malicious peers. There is a relationship with PageRank algorithm.

This method uses query dependent importance scores and a probabilistic approach to improve upon PageRank. It pre-computes importance scores offline for every possible text query. [PDF format]

This paper describes a joint probabilistic model for modeling the contents and inter-connectivity of document collections such as sets of web pages or research paper archives. [PDF format]

First Stanford paper about PageRank. It is a static ranking, performed at indexing time, which interprets a link from page A to page B as a vote, by page A, for page B. Web is seen as a direct graph and votes recursively propagate from nodes to nodes. Ranking is performed at indexing time. Used by Google.

A mathematical paper about the convergence of methods used for solving the PageRank Matrix.

"Google's PageRank is an eigenvector of a matrix of order 2.7 billion"

Integrates ODP data in PageRank calculation for performing query time probabilistic ranking.

It describes how to compute incrementally PageRank when Web graph's link topology changes. [PS format]

"Random Surfer" model extension. At each step of traversal of the Web graph, the surfer can jump to a random node or follow a hyperlink or follow a back-link (a hyperlink in the inverse direction) or stay in the same node.

About the using of PageRank in Web Track 8 "large" and "small" datasets. [PDF format]

About the using of Link Popularity in Web Track 9 datasets. [PDF format]

PageRank and Hub and Authority generalization based on the topic of Web Pages. Definition of a model where a surfer can move forward (following an out-going link) and backward (following an in-going link in the inverse direction). [PS format]