| Abstract [eng] |
Computational Thinking is considered one of the key competencies in modern education, yet in practice it is usually assessed only through the correctness of the task’s final solution. This dissertation presents a process-oriented automated assessment model of computational thinking based on behavioural interaction data, enabling the analysis of learners’ problem-solving strategies and the identification of recurring behavioural patterns. The study is based on interaction logs collected from interactive tasks of the Bebras Challenge. Solution logs were transformed into structured features, including action sequences, click counts, solution duration, and other behavioural indicators. Clustering methods were applied to identify groups of similar problem-solving strategies, while a neural network classifier was used for automatic classification of solution types. The results demonstrate that process-oriented behavioural data can support more detailed and adaptive assessment of computational thinking. |
