Authors:
Ernani Gottardo
1
and
Andrey Ricardo Pimentel
2
Affiliations:
1
Instituto Federal de Educação and Ciência e Tecnologia do RS - IFRS, Brazil
;
2
Universidade Federal do Paraná, Brazil
Keyword(s):
Affective Computing, Emotion Inference, Emotion and Learning, Adaptive Systems.
Related
Ontology
Subjects/Areas/Topics:
Emotional and Affective Computing
;
Enterprise Information Systems
;
Human-Computer Interaction
;
Multiple Sensory Devices
Abstract:
Adapting to users’ affective state is a key feature for building a new generation of more user-friendly, engaging and interactive software. In the educational context this feature is especially important considering the intrinsic relationship between emotions and learning. So, this paper presents as its main contribution the proposal of a hybrid model of learning related emotion inference. The model combines physical and cognitive elements involved in the process of generation and control of emotions. In this model, the facial expressions are used to identify students’ physical emotional reactions, while events occurring in the software interface provide information for the cognitive component. Initial results obtained with the model execution demonstrate the feasibility of this proposal and also indicate some promising results. In a first experiment with eight students an overall emotion inference accuracy rate of 60% was achieved while students used a game based educational softwa
re. Furthermore, using the model’s inferences it was possible to build a pattern of students’ learning related affective states. This pattern should be used to guide automatic tutorial intervention or application of specific pedagogical techniques to soften negative learning states like frustration or boredom, trying to keep the student engaged on the activity
(More)