(1115) MODELLING OF AN EDUCATIONAL PROFILE OF A STUDENT BY ANALYZING PUBLIC USER DATA FROM SOCIAL NETWORKS

摘要

The present paper discusses the prospects of using the VK (VKontakte) social network for identifying psychological traits, interests and professional hobbies, that are important for spotting gifted senior high school students. Methods: psychological testing and questionnaire survey (?#Careerguidance? method), academic record analysis, structural and content analysis of a social network profile, content analysis, percentile normalization, machine learning. Using a sample of 1692 senior high school students giftedness was defined as a combination of intelligence (analogies, convergent and divergent thinking styles), creativity (fluency, flexibility, originality, elaboration, independence), conative and personal features (conation for knowledge, selfactuating, proactivity, determination, resoluteness, social intelligence). The paper further presents ?psychological profiles? of gifted senior high school students based on the analysis of their VK profile data. Female users mostly join communities with educational (informative), social and commercial content, male users mostly with entertaining and educational (informative) content. Female users mostly post and share in order to inform and entertain friends and subscribers, to encourage action. Male users do the same in order to inform and entertain friends and subscribers. Machine learning proved to be useful in researches with a large number of participants and attributes. To solve the problem a binary classification was used. So, gifted and non-gifted students were compared. Support vector machine proved to be the most efficient model to solve this problem. It allows to identify participants with highly developed psychological features, calculate correlations between participants with different degrees of giftedness features and VK communities. Predictive modeling is performed based on regional and federal communities as markers. A part of marker communities is targeted on the participants’ gender. The novelty of the research lies in the fact that social network user data is examined specifically in order to identify giftedness. Practical application: recruitment of prospective students with significant cognitive and personal potential; express diagnostics of cognitive, personal features and soft skills of students in order to develop individual educational paths.