Methods of social network texts analysis for a psychometric model of personal behavior

Abstract

The development of technologies for predicting personality behavior is one of the priority directions for improving the diagnostic apparatus of psychology. The integration of information technologies, mathematical methods and big data processing capabilities into the methodology of psychological research makes it possible to build and test formal psychometric models for their further use in creating software systems that can predict personal behavior. This paper presents a description of methods and technologies for qualitative analysis of social network texts used in the development of algorithms for predicting personality behavior types as part of the creation of a psychological model of the subject's behavior in the digital environment. Anonymized dataset was collected based on psychological survey on “Dark Triad” for students and their profiles on the VK social network as initial data for the analysis. Then were identified several cognitive behavioral predictors in form of most commonly used lexicon and themes, that are typical for persons with different levels of “Dark Triad” characteristics. The obtained results can later be used in training neural network models to predict personal behavior.