Computer Methods for Thematic Modeling of Textbooks Collection in Natural Language

Abstract

The paper presents the development results of computer methods for analyzing text data and assessing classification inaccuracies at the stages of thematic modeling. This study uses as an example the task to process textual data of a collection of graduate qualification works prepared and defended by students of Altai State University, Faculty of Mathematics and IT in recent years.
 The main results obtained in the paper are listed as follows. Relevant application areas and directions for computer methods and thematic modeling in the educational process are identified. Justification of the general algorithm for solving the problem of the thematic analysis of collections of educational materials is carried out. Information technologies for thematic modeling are developed, and estimation of analysis errors on a set of test documents is obtained. It is shown that computer-based methods of thematic modeling and information technology to support them can be used both in the educational process and in the development of educational and methodological documents.