The PageTrust algorithm: How to rank web pages when negative links are allowed?

Abstract

The paper introduces a novel algorithm derived from the PageRank algorithm of Brin and Page. The PageRank algorithm interprets an hyperlink from page a to page b as being a positive vote from a to b. Starting from this interpretation, it attributes a rank to each page. However, it does not offer the possibility to take into account negative votes. The PageTrust algorithm includes negative links and converges to a trust value for each page. The PageTrust algorithm appears as a natural extension of PageRank and it preserves good properties of robustness against possible spammers. Moreover several parameters allow us to strengthen or weaken the role played by negative links.