Search engines have a vested interest, only users relevant to their search results tailored to deliver. They are constantly confronted with the problem of aggressive online marketing methods exposed to, for example, through automated webspam comment.

Before the Trust Ranks discusses the implementation of the first term must be clearly identified. In general, is in Combating Web Spam with Trust Rank algorithm presented meant when Trust Rank is mentioned. One of the co-authors of this paper was Jan Pedersen, a Yahoo! employee a year later the patent application link-based spam detection submitted. therefore It is not in the Trust Rank to a patent filed by Google. It is likely, however, that Google uses a very similar principle to the Trust Rank. Caused some confusion in this regard, the fact that Google is almost at the same time held the trademark on the term "Trust Rank", but that an anti-phishing filter called. See also the following video from Matt Cutts :

The basic idea of the Trust Ranks is the division into good and bad websites. In good websites mean those that are regularly maintained and monitored their content. It is not differentiated from these pages within a domain. Good websites are also notable in that it with very low probability on bad sites link, but with a high probability of high-quality pages. Bad websites are spam pages that follow as illegal or fraudulent or exist solely for the purpose of search engine manipulation. The Trust Rank now represents a criterion reference to which good and bad websites can be identifed:

The problem at this point is the inability to make the distinction completely automated in good and bad sides. Therefore, the Trust Rank algorithm is based on a so-called oracle function , in which a human author makes this distinction. Since it is impossible for an ever-growing number of websites, each website individually equip with a value for the Trust Rank, which is a PageRank -like algorithm used inheritance principle. This first one is automated pre-selection of sites hit that should have as many good websites. For this pre-selection, for example, the PageRank serve as a selection criterion. The selected sites are categorized as seed and get called by a man known as a Trust score assigned. This TrustScore is essentially the value of the Trust ranks and then as the PageRank passed on linked websites. Through the above-mentioned property of good websites, with only a very low probability to bad, but a high probability of good websites to link the Trust Rank but provides greater security against tampering as the PageRank. However, since it is not assumed that all linked pages are also the same content monitoring and care subject as the original seed, is a damping factor in the inheritance of Trust Ranks used. The following figure verdeulicht this statement:

The Trust Rank is an effective approach to minimize proliferation of spam in search engine results. Furthermore, it can also be used as a ranking factor, as it will be just like the PageRank iteratively and can assign a value to all pages of the Internet, which in turn creates a metric that allows a comparison of different websites. A high Trust Rank may therefore be a better ranking result. This can be achieved by being linked to from a page with high trust.

The problem with the use of the Trust Rank algorithm is the one choosing the right seed to other websites and the calibration of the different parameters (such as the damping factor).

Finally, once the most important facts about Trust Rank in brief: