Fine-Grained Detection of Programming Students’ Frustration Using Keystrokes, Mouse Clicks and Interaction Logs

Prolonged frustration leads to loss of confidence and eventual disinterest in the learning itself. The modelling of frustration in learning is thus important as it informs on the appropriate time to intervene to sustain the interest and motivation of students. To automatically detect learner’s frustration in a naturalistic learning environment, the novel use of keystrokes, mouse clicks and interaction patterns of students captured within the context of a tutoring system was proposed. The modelling approach was described and a comparison was made between the proposed model using Bayesian Network and the baseline Naïve Bayes model. With the formulation of an overlapped sliding window mechanism, the granularity of detection was also investigated. The results confirm the hypothesis that a combination of keystrokes, mouse clicks and interaction logs can be used to accurately distinguish affective states of frustration and non-frustration amongst novice learners of computer programming in a granular fashion.