|
MatrixNet is a machine learning algorithm with which Yandex builds its query ranking formula. Other search engines use different algorithms.
Since there are millions of website pages on the Internet, search robots do not access these pages, but rather a specially created database, or index. The pages found in this database are shown to the user when he enters a query.
New pages are constantly added to the index and unnecessary pages are deleted, so search engines update the index or make updates that are invisible to users. But the search engine must know by what rules the index is updated. Previously, only the search engine assessor was responsible for placing a relevant page in the index. But due to active website building, assessors simply do not have enough time to check all the pages found by the robot. Then, with the advent of the Matrixnet learning algorithm, some of the functions of assessors were assigned to his shoulders.
Peculiarities
Matrixnet appeared in 2009. with the introduction of the new Snezhinsk algorithm in order to increase the number of ranking criteria and improve search, as a result of which insignificant search criteria lost their importance. Also, thanks to this algorithm, it was possible to adjust the ranking of the site for certain queries without spoiling the search for other queries. The Matrixnet formula can contain thousands of values, and its precision allows you to filter out all non-matching pages. The number of pages viewed is constantly growing, but thanks to the Matrixnet algorithm, Yandex results are now more relevant.
|
