Abstract
The tasks of Natural Language Processing (hereinafter - NLP) are very topical today, in particular in the context of the application of machine learning and artificial intelligence models. In this work the approach allowing to carry out classification of texts with the minimal required manual markup is considered. The problem of prediction of resonance of news in the Kazakhstani media space where resonance is understood as abnormal interest of the public to the publication within the limits of a concrete information source is considered. The proposed model is implemented and tested within an informational system.