Saturday 14 July 2007

TrustRank

TrustRank


TrustRank is a major factor that now replaces PageRank as the flagship of parameter groups in the Google algorithm. ( Note that a similar system is being used by Yahoo! Search as well ). It is of key importance for calculating ranking positions and the crawling frequency of web sites. It may also extend the grace period before penalties for accessibility and usability problems take place. There is no public information on a TrustRank score of a web site or URL, the only indications of trust are what you can perceive from a user point of view. TrustRank technology was based on the identification of web sites that have been an important resource on the Internet for a long time, web sites that are seen as "authorities" of a certain area of interest, web sites that are historically spam and error free, are provided by a long trusted source, promoted or maintained by a service that is seen less or nearly completely unlikely to show offensive material. The technology uses the most reliable network of references, and is - as PageRank - relying on links for many of its calculations, but also includes a lot of other factors that are next to impossible to be manipulated. Major portals, reference pages, national and international communities, long time conventional media publishers, governmental and educational pages, official pages of institutes, organizations, nonprofit organizations and the longest standing commercial web sites are among those that have been set or "voted" into the position of being a TrustRank hub. The algorithm for calculating TrustRank takes many parameters into consideration, from which the ones that are relatively new to Google ( not covered previously by PageRank and relevance calculations ) are examining linking patterns, the link profile, referral link age and history, and web page history. Trust can be accumulated by means common sense would indicate, including references from people at other trusted sources, refraining from using questionable and borderline business or optimization or methods, having no offensive content or major usability/accessibility problems, overall, "natural" growth and a clean web site history. Note that the TrustRank network evaluates all web pages, thus any and all inbound links carry this parameter ( with the sole exception of sources that are banned or penalized in the Google Index ). The system is currently the closest automatic quality factor, thus one can more or less deduct the impact of a new referrer from the source web site's popularity, status and quality. The algorithms were based on examining both user and web site behavior, translating common indications of trust and distrust into parameters, and are being fine-tuned constantly for an even higher accuracy in identifying them. Note that TrustRank is a technology that has been patented to utilize the implementation of human reviews, and that both positive and negative trust, while can be accumulated automatically over time, can be and is sometimes overruled by manual evaluation. Trust can be lost if abused, and distrust can be restored with correcting problems and optionally a goodwill request for re-evaluation.

Note: The term "TrustRank" is not used officially. You may address the parameter group simply as "trust".

As for battling the few manipulative methods aiming to artificially boost the TrustRank of a URL, Google has made many changes to its calculation rendering it virtually impossible. Trust is now closely tied to relevance calculations, and thus an off-topic link from a trusted source will not pass / negate the effects of this parameter ( and relevancy may be tracked back to as far as the home page of the source web site, thus eliminating references that are not within its original areas of interest ). All hubs and popular queries are being constantly monitored for any irregularities. From a diagnostic perspective, two of the changes are important. One is that in order to keep the rankings on search results pages a fair competition, TrustRank is being passed from page to page within any domain that is not a "set" hub. The other ( that also applies to "set" hubs ) is the effect of trust being tied to thematic relevancy. Read more on Anchor Text and Website Navigation.

Known issues


Case 1,
The effect of a certain threshold of TrustRank required to appear on the search results for generic searches, led some to believe that their web sites have been held back, penalized or banned from Google. While these pages were shown for less generic and obscure searches, single, or two word phrases would not list them in the results. For queries that are broad enough to assume the user is looking for the most general, most official information, only web sites with a certain minimum amount of TrustRank are shown for the results. See the unofficial term the "sandbox".

+ Resolution: As your web site accumulates more references to it from other web sites, its TrustRank will build gradually. Once it is perceived as a trusted resource for the given query, it will most likely appear on the search results for generic queries as well. Most phrases require only a minimum amount of this parameter to pass the threshold, while some closely monitored, highly competitive queries show only the most trusted results.


Case 2,
While TrustRank is a parameter applied to pages in general, an entire domain, IP range or even network may be affected by certain penalties that are connected to its calculation. A web site may lose trust to all pages for hosting or linking to ( or in special cases being excessively linked from ) other domains, which have been breaching Google policies, hosting spam, malware, spyware, or offensive material. Web sites with low or nonexistent TrustRank that virtually are unable to accumulate this parameter are usually under an automatic or manual penalty / ban for practices not only unethical, but sometimes even illegal. Becoming a part of link networks associated with such web sites may also lower or negate trust entirely. After checking many parameters, a following evaluation may conclude that the source and the target domains are affiliated. ( The detection of such affiliations is done with very good accuracy ). In such cases, losing trust and thus rankings would only be the first sign of a coming complete ban of all pages found on the domain.

+ Resolution: Make sure your web site does not host or link to pages that contain malware, spyware, offensive material, spam, or that are already penalized or banned for being marked as a phishing site, are or were at some point involved in illegal activities. If your site is being excessively linked to a 3rd party, you should contact the Google Web Spam team, or file a spam report through Google Webmaster Tools. Such 3rd party inbound links will not get the target pages banned, but may lower their trust ( and get a warning penalty issued ) if the linking pattern, ownership or IP range would indicate that the 3rd party may be affiliated with the targeted web site. Read more on Link Schemes and Historic Domain Penalty.

No comments:

Enter your email address:

Delivered by FeedBurner