Vectorization and Optimization of User Behavior Data in E Learning Systems

Abstract

At the first stage, an applied scientific research developed a procedure for collecting data on the parameters of user interaction with the user interface. This input procedure receives many heterogeneous messages about the actions of a particular user in the interface, while the output represents a vector that describes the user in aggregated form. The set of vectors for different users, in turn, was then used as input for the k-means clustering algorithm, the result of which is the user's attitude to one of the k clusters that distinguish the user by the type of behavior. User interface interaction data is available to 67.8% of GlobalLab platform users. There is no such data for the Diary.ru electronic diary. Considering that not all users of the GlobalLab platform took measures to create a project, ideas, work with questionnaires and educational materials, the total number of students for whom the value of all 4 variables differs from the neutral one was 9.7 thousand.