Abstract:

Rapid development of information technology, mathematical methods and the possibilities of big data processing makes it possible to build and verify formal psychometric models for use in creating of software systems that can predict personal activity success. This paper was prepared within the problem framework of a project to develop a psychometric model of success, which is based on a selected set of cognitive behavioral predictors of personal activity. One task of this project is to develop an automated system for predicting the academic success of students based on data from the information-analytical system of the University and from their profiles on social networks. Some of the important sources of personalized data that can be converted into psychometric characteristics of a person are posts and reposts texts extracted from personal webpages. The paper presents the description of a software module in which classical methods of information retrieval are used to process these texts, namely: text indexing and word frequency characteristics analysis. After processing, topic extraction is applied, that is, the extraction of main topics that the student raises in his texts on social network. According to results of the study, the most typical text topics for groups of relatively successful, average and unsuccessful students were presented and identified as one of the cognitive behavioral predictors of academic success.

Keywords:

Natural language processing, text indexing, data analysis, topic extraction, school psychometry.