Internet Marketing Glossary

Definition: TrustRank

TrustRank is a system for ranking web pages based on the "trustworthiness" of the various sources of links to those pages. TrustRank is also the numeric measure of this trustworthiness, used in search engine ranking and spam-detection algorithms. TrustRank has come to be viewed as the successor to PageRank, or at least a major refinement of it. It allows search engines to automatically gauge the probability that a particular page is legitimate, and to adjust rankings based on that probability.

TrustRank recognizes the superiority of human judges over algorithms in the task of separating "good" from "bad" web pages. The problem with using humans is one of scale — it's much too expensive to employ humans for large-scale spam-detection operations. The TrustRank algorithm makes some use of human judgement to select and evaluate a set of "seed" pages (the humans make a determination of whether or not each page in this seed set is spam or not, and pages identified as spam are removed from the set). The algorithm then examines the extended link structure of the "good" seed pages to determine if pages to which they link are also likely to be "good."

Good pages pass on some TrustRank to the pages to which they link. Like PageRank, TrustRank is attenuated at a distance, so that the measure of trust passed to each linked page is divided by the number of links on the linking page. PageRank is an element of the calculation of TrustRank, and some formulations of the TrustRank algorithm order sites by PageRank because "it is more important for us to correctly detect spam in high PageRank sites, since they will more often appear high in query result sets." This implies that the TrustRank algorithm is more effective at filtering out spam from websites with high PageRank values, and that TrustRank would not have as much application for judging websites with low PageRank (few inbound links). The authors themselves state as much.

TrustRank, at least as it was evaluated in the original paper on the topic (from which the quotes on this page are taken), seems to be biased in favor of websites run by large corporations, governments, and universities, among others. This is because the authors were taking care to use only "highly trustworthy" websites in the seed set, so all the seed sites were subjected to the human spam review and also confirmed to be controlled by a "clearly identifiable authority (such as a governmental or educational institution or company)." In practice, this bias explains why links from certain "privileged" categories of websites are worth so much more than ordinary links.

Click here to return to the ClickScience Internet marketing glossary.