Link-based ranking for Web pages is a technique that orders Web pages based on their linkage information. A popular application for link-based ranking is search engine technology. Google, which is currently considered the best search engine, owes much of its success to its PageRank method, which is a link-based algorithm to approximate the global popularity of Web pages. Considering the current size of the Web, calculation of the popularity of each web page (which is also called PageRank) can take many hours or days on a powerful computer.

        In this thesis, we designed and implemented the PageRank algorithm for distributed memory parallel computers. Our aim is to make PageRank calculations for the growing number of Web pages in a fast and scalable way. Thus, it will be possible to provide central personalized search service that is a feature of PageRank method. PageRank is currently not used to provide personalized search service to a group of subscribers. As can be seen from our results, our implementation can be used to give such a service with an appropriate infrastructure.