Main menu:
Company Google became the first who has patented system of link analysis for ranging web pages. The algorithm has received name PageRank (PR). In this chapter we will tell about this algorithm and how it can influence ranging of search results.
PageRank calculate for each web page separately, and it is defined PageRank of pages referring to it. The main task consists in finding the criterion expressing importance of page. In a case with PageRank such criterion theoretical attendance of page has been chosen.
Let's consider model of travel of the user on a network by transition under links. It is supposed that the user begins viewing of sites with some casually chosen page. Then under links it passes to other resources. Thus there is a probability of that the visitor will leave a site and again will begin viewing of documents with casual page (in algorithm PageRank the probability of such action is accepted 0.15 on each step). Accordingly, with probability 0.85 it will continue travel, having passed on one of accessible links on current page (all links are thus equal in rights). Continuing travel indefinitely, it will visit on popular pages many times, and on little-known - it is less.
Thus, PageRank web pages it is defined as probability of a finding of the user on the given web page; thus the sum of probabilities under all web pages of a network is equal to unit as the user necessarily is on any page.
As to operate with probabilities not always it is convenient, after a number of transformations with PageRank it is possible to work in the form of concrete numbers (as, for example, we have got used to see it in Google ToolBar where each page has PageRank from 0 to 10).
According to described above model it is received that:
- Each page in a network (even if on it there are no deep links) initially has nonzero PageRank (though also very small);
- Each page having proceeding references, transfers a part of the PageRank to pages to which refers. Thus transferred PageRank inversely proportional to a reference number on page - the more links, the smaller PageRank it is transferred on everyone;
- PageRank it is transferred not completely, on each step there is an attenuation (that probability of 15 % when the user begins viewing with new, casually chosen, pages).
Let's consider now how PageRank can influence ranging of search results (we speak "can" as in the pure state PageRank for a long time already does not participate in algorithm Google as it was earlier, but about it more low). With influence PageRank all is very simply - after the search engine has found a number of relevant documents (using text criteria), to sort them it is possible agrees PageRank - as it will be logical to assume that the document having áîëüøåå number of qualitative deep links, contains the most valuable information.
Thus, algorithm PageRank forces out upward in search resul those documents which without search enginer are most popular.
Now PageRank it is not used directly in algorithm Google. It and is clear - after all PageRank characterizes only quantity and quality of backlinks to a website, but at all does not consider the reference text and information contents of referring pages - namely the maximum value is necessary to these factors at ranging. It is supposed that for ranging Google uses so-called thematic PageRank (that is considering only links from thematically connected pages), however details of this algorithm are known only to developers Google.
To learn value PageRank for any web page it is possible by means of Google ToolBar which shows value PageRank in a range from 0 to 10. It is necessary to consider that Google ToolBar shows not exact value PageRank, and only range PageRank to which the website gets, and range number (from 0 to 10) it is defined on a logarithmic scale.
Let's explain on an example: each page has the exact value PageRank known only Google. For definition of the necessary range and information output on ToolBar the logarithmic scale (the example is shown in the table) is used
Real value |
PR Value ToolBar |
1- |
1 |
10- |
2 |
100- |
3 |
1000- |
4 |
Etc. |
All figures are conditional, however visually show that ranges PageRank shown in Google ToolBar, are not equivalent each other. For example, to lift PageRank c 1 to 2 it is easy, and with 6 to 7 it is much more difficult.
In practice PageRank it is used basically for two purposes:
1. To determine the importance of a web page. PageRank does not give the exact information on referring pages, but allows quickly determine web page's importance. For websites it is possible to adhere to the following gradation: PR 4-5 - the most typical PR for the majority of websites, PR 6 - the very importance site, PR 7 - the size almost unattainable for the usual web designer, but sometimes meets, PR 8, 9, 10 - meet only at websites of the large companies (Microsoft, Google, etc.). Knowledge PageRank can be used at an exchange of links to estimate quality of the page offered to an exchange and in other similar situations.
2. An estimation of level of a competition by search inquiry. Though PageRank also it is not used directly in algorithms of ranging, nevertheless allows to estimate competition of the set inquiry indirectly. For example, if in search result there are sites with PageRank 6-7 the site with PageRank 4 has very few chances to rise in a top.
One more important remark - values PageRank shown in Google ToolBar are recalculated seldom enough (time in some months), therefore ToolBar shows to a certain extent out-of-date information. That is the Google search engine considers changes in deep links much faster, than these changes are displayed in Google ToolBar.
Friends: