Abstract. Automated semantic analysis and topic modelling methods were first applied to a large corpus of public online communities and discussions to analyse strategies for responding to digital risks in modern society. A large array of public data on profiles of the social network “Vkontakte” and their subscriptions to online communities was used to train a predictive model of individual adaptation to risks in the field of education and professional self-determination. Based on the results of thematic modelling, three topics of subscriptions to online communities were identified as signs of psychological adaptation: preparation for exams, self-development and “popular” psychology, education and science. The developed version of the predictive model of psychological adaptation showed high accuracy and the importance of such factors as age, gender and place of residence for individual adaptation to digital risks in the field of education and professional self-determination.