Efficient sequential and parallel algorithms for popularity computation on the World Wide Web with applications against spamming

Abstract

When searching for information on the World Wide Web, it is often necessary to use one of the available search engines. Because the number of results are quite large for most queries, it is very important to have some measure of relevance of the result Web pages. One of the most important relevance factors is the popularity score which indicates how popular the page is among the users. We propose a modified version of status index by Katz and present efficient sequential and parallel algorithms that solve the problem. The high flexibility in our algorithm results in resilience to spamming as we show by experiments. © Springer-Verlag Berlin Heidelberg 2004.