METHOD AND SYSTEM OF RANKING OF A VARIETY OF DOCUMENTS ON THE PAGE OF SEARCH RESULTS Russian patent published in 2019 - IPC G06F17/30 

FIELD: information technology.

SUBSTANCE: invention relates to a method and server for ranking a plurality of documents on a search result page (SERP). Method performs a calculation using a machine learning algorithm (MLA), ranking estimates for each document from a variety of documents indicating the calculated relevance of the relevant document to the request and the original rank of the corresponding document in the SERP, calculating the class communication score for each document from a variety of documents, which is the probability of the corresponding document being associated with a predetermined document class and used when increasing the level of change in the original rank of the relevant document, which is the difference between the original and the changed document rank, calculating a modified ranking estimate for each document from a plurality of documents based on a corresponding ranking estimate and a corresponding class relationship estimate indicating the modified relevance of the corresponding document to the request, and initiating a SERP mapping with documents that were ranked based on the corresponding modified ranking score.

EFFECT: technical result is to increase the efficiency of the ranking of the documents found.

20 cl, 13 dwg