Towards Affective States Detection in Educational Contexts
Advisors: Dr. Jesús González Boticario y Dra. Olga C. Santos Martín
The relation between emotions and learning has been deeply studied by psychologists, who have pointed out possible improvements in learning when affective aspects are taken into account. Nowadays, affective state detection is a challenging and increasingly investigated problem in computer science, even more so in educational contexts, where emotions are spontaneous and usually have a low intensity (what makes its detection even harder). Bearing these issues in mind the progress of the work carried out in the first year of this Doctorate has focused on introducing a multimodal detection approach, which was previously proposed in the TFM in a real world context experimentation. Here kids were offered a series of mathematical problems while being monitored with different kinds of sensors. The purpose is to generate different emotion prediction models from different data mining algorithms applied to data collected from multiple data sources (from physiological signals to webcam recordings, and including mouse and keyboard interactions) which have been labelled by experts. Results obtained are being analysed and will be taken into account when designing the affective feedback to be provided to participants in a forthcoming experiment.
Sergio Salmerón Majadas