METHOD AND SYSTEM FOR CLASSIFYING AND FILTERING PROHIBITED CONTENT IN A NETWORK Russian patent published in 2020 - IPC G06F16/35 G06F21/00 G06N20/20 

FIELD: computer equipment.

SUBSTANCE: method of classifying and filtering content in a network, performed on a computing device comprising at least a processor and memory, which comprises instructions for executing a preparatory step, on which a collection of HTML documents is formed, wherein collection is formed so that each of documents included in it can be related to different classes of content; converting obtained from previous step data from HTML document into text; generating a token matrix for training the ensemble of classifiers; based on the produced token matrix, creating an ensemble of classifiers, comprising at least four classifiers, wherein for each classifier a decision priority is predetermined; a working step of obtaining a URL and downloading an associated HTML document; converting HTML document into text; generating a vector of pure tokens for the classifier ensemble; starting the ensemble of classifiers trained at the preparatory stage; outputting an analysis result comprising a result of classifying content contained in the received document; content is filtered based on the obtained class.

EFFECT: technical result consists in improvement of accuracy of classification and filtration of prohibited content in a network.

12 cl, 6 dwg