A. A. Korepanova’s research while affiliated with St. Petersburg Research Center of the Russian Academy of Sciences and other places

Numerical data retrieved from the accounts of users of a popular Russian-language online social network have been used to automate the prediction of the PEN test (temperament test) results. This study aims to automate the assessment of personality traits of online social network users by comparing the test results and the content posted by the user on his or her account, using machine learning methods. Classifiers are constructed with CatBoost and random forest models for predicting the scores of extraversion–introversion and neuroticism. The theoretical significance of this result is the development of an approach to automating the assessment of human personality traits. The practical significance is the development of a program module to create an automated system for assessing the human personality traits through online social networks.

The subject of the research is the predictors of the intensity of interaction between users in online social networks, which can be formalized in probability estimate of the spread of multistep social engineering attacks. The research methodology is based on comparing, using the methods of statistical analysis, the observed online interaction of users and the fixed personal choice of friends in the proposed interviewing situation, simulating an example of the real social engineering attack. The main result is the identification of the predictors of the users’ interaction intensity. Based on the results of the analysis, the following predictors were identified: the number of mutual friends, the number of common communities, the number of likes and the number of gifts. The significance of the research is the development of an integrated approach to increase the level of organization’s security against the multistep social engineering attacks. This forms the basis for the timely identification of the most vulnerable parts of the information system, and indirectly serves as the basis for creating an intelligent decision support system.

The goal of this article is to propose an approach to linguistic values quantification and to consider an example of its application to the relationship types between users in the popular social network in Russia “VK”. To achieve this aim, we used the results of a sociological survey, by which were found the frequency of the order, then the probability theory apparatus was used. This research can be useful in studying of the influence of the types of users’ relationships on the execution of requests, also finds its use in building social graph of the organization’s employees and indirectly in obtaining estimates of the success of multi-pass Social engineering attacks propagation.