Educational behavior: tracking the activation of knowledge, integrative skills and abilities in the higher education system

the monitoring of knowledge has advanced significantly thanks to the methods of deep learning, there is a gap in determining the actual level of knowledge of students - it does not consider the significant impact of the educational behavior and abilities of students, which may better reflect their knowledge acquisition and provide a high degree of credibility for research. This paper presents an extended model of tracking knowledge about the characteristics of behavior and abilities of students in learning, which addresses this gap by including in its structure models of behavior in the process of learning considering individual capabilities. The model consists of four structural levels: construction, learning, forgetting and forecasting. Embedding level expands the information about the exercises by means of the Exercise Embedding module (considering skills, knowledge level, tasks and difficulty) and the Learning Ability Embedding module (considering answer time, answer options, hint time), forming a fundamental learning construction. The learning level evaluates the degree of acquisition of knowledge by analysing differences between current and previous acquisition, time intervals and comparative states of knowledge. The level of forgetfulness model’s behavior through functions of erasure and retention of forgotten, based on previous knowledge states, current achievements in learning and time intervals. The results of experiments show that the analyzed knowledge tracing model can serve as a predictor of educational results, significantly exceeds existing models on some indicators, for example, in the accuracy when predicting the state of students' knowledge.

learning dynamics’ modelling, knowledge tracing, directed learning, learning behavior, learning ability, deep learning