METHOD AND SYSTEM OF SELECTION FOR RANKING SEARCH RESULTS USING MACHINE LEARNING ALGORITHM Russian patent published in 2020 - IPC G06F16/951 G06N20/00 

FIELD: information technology.

SUBSTANCE: invention relates to search results ranging means. Method of ranging search results using a machine learning algorithm (MLA), which has been trained to determine similarity of requests based on their text content, includes: obtaining a new request, extracting a plurality of search requests, calculating a corresponding similarity parameter between the new request and each query from the plurality of search requests, selecting the past request, associated with the highest corresponding similarity parameter, retrieving a set of search results associated with the given past request, wherein each of the search result set is associated with the annotation, including corresponding search requests, and calculation for each corresponding request from a plurality of corresponding requests, which were used to obtain access to the corresponding search result, a second similarity parameter, and using a corresponding second similarity parameter as a ranking factor for ranking corresponding search results as search results for a new query.

EFFECT: technical result is higher efficiency of ranking search results.

26 cl, 7 dwg