The objective of this interdisciplinary study was to identify informative signs of behavior of Russian-speaking users of the social network VKontakte in connection with the severity of their signs of depression. The study used data from 1268 Vkontakte users who filled out the Beck Depression Inventory (BDI), and also provided access to their profiles information. There were three groups of respondents with different levels of severity of signs of depression. Using machine learning methods, the support vector method (SVM) and the random forest algorithm (Random Forest), informative linguistic and behavioral signs of depression were revealed among users of the VKontakte social network, comparable to data obtained by researchers of English-speaking respondents from other social networks.