Microsoft BrowseRank Algorithm

BrowseRank is compared to PageRank, which rates a page's popularity by the number of links on that page. BrowseRank evaluates the importance of a page based on users' navigation and behavior.

The authors are Chinese and primarily teach or hold a research position at universities in Beijing: Yuting Liu, Bing Gao, Te-Yang Liu, Yin Zhang, Zhiming Ma, Shuyuan He, Han Li.

BrowseRank Calculation

Navigation graph: Dots are pages and relationships are user transitions between pages. It also takes into account the time spent by the user on the pages. It wants to be more efficient than the link graph on pages to determine their importance.  

Information about the behavior of the visitor is obtained from the browser. That is:

They are used to plot. It represents the process of randomly passing Internet users. It is assumed that when a visitor visits a page and remains on it, they implicitly vote for that page.

The algorithm is then based on a continuous Markov process, which is applied by taking the graph as a model to determine the stationary probability of the process distribution corresponding to the importance of the pages.
To estimate time, a distinction is made between a moment or page and loading another page. The last page of the session uses the average time that is observed if it is not the last or similar process.

Simplified algorithm

Enter Internet user behavior data.
Exit: assessing the importance of the page.
Algorithm:

  1. Create a navigation graph.
  2. Qii score for all pages.
  3. EMC Transition Probability Matrix Assessment
    Then, stationary probability distribution is obtained by degrees.
  4. Calculation of the stationary probability distribution.

Details of the algorithm are given in the document with reference.

Comparison with PageRank

PageRank relies on a graph of links between pages and believes that the more links to a page and the more important it is and is visible to many Internet users.
He uses the Markov process at an inconspicuous time on connections to assess their importance.
It has the following disadvantages:

Google uses not only the PageRank algorithm to determine the position in the results, but also the page rating algorithm, for which a patent was filed in 2007. PageRank is one of the criteria for determining a page score.

Disadvantages of BrowseRank:

Time criterion

In fact, it is impossible to know whether the user is reading the page or left the browser open and left to do something else.
A short, simple, clear page will be quickly read, but will not be less important than another, longer, confusing or difficult to decipher. For example, the page with the store address is read in seconds and, nevertheless, is crucial!

PageRank assesses the quality of links

The page score compiled by Google takes into account the number of criteria, much more than BrowseRank, as seen in the patent summary.
Spam on mutual or hard links is getting better and better controlled, which also makes this argument unfounded.

Concept of trust

PageRank depends not only on the number of links, but also on the relevance and weight of the links. Links from an important site matter more than links from an infrequent site.
These criteria are absent in BrowseRank, for which, in principle, all clicks stand.

Uncontrolled spam

Just as webmasters try to use PageRank to their advantage by creating links, they will try to hack BrowseRank, forcing robots (scripts) to work that will mimic human behavior and park on their pages. Such scenarios can be executed in millions of copies.