Abstract:

The goal of the present research is to test the hypothesis that it is possible to predict educational interests of students by analyzing social profiles data from the VK (VKontakte) social network. In order to test the hypothesis a heterogeneous data set was used: 126 000 user profiles, 600 MB of text content of thematic communities, over 100 000 most popular pages and groups in the social network. Research methods used are as follows: content analysis, classical psychological diagnostics methods, mathematical statistics, machine learning. As a result, a correlation has been found between educational interests and cognitive abilities of a user on the one hand and this user’s social media profile on the other. Based on this data, a model to predict educational interests and attributes of giftedness (intelligence, creativity, motivation) was developed. Achieved prediction accuracy was between 0.62 and 0.8.

The authors present ways of applying the research results in National Research Tomsk State University: modelling a profile of a prospective student’s educational interests and individual recommendations on educational programs available in the university; analyzing a student’s educational needs, developing a system for recommending educational resources based on educational needs; identifying prospective students with high cognitive and creative abilities in social networks and recruiting talented students to participate in the university's research programs; running express diagnostics based on a student’s social media profile in order to recommend of one of the three ways of a personalized learning in TSU; identifying students with specific educational needs.

Keywords:

Social networks, data mining, data analysis, individualization of learning.