Emotional dectection and reasoning for student support

The consortium are working upon two aspects of the iTalk2Learn support mechanism: The emotion detector and the emotion reasoner.

iTalk2Learn partners, Birkbeck College (BBK) and the Institute of Education (IOE), are continuing to analyse data that was collected from the Wizard-of-Oz studies that took place in March. These studies relate to the work on task-dependent support for Fractions Lab, together with task-independent support.

Initial results suggest that emotions expressed during learning should be considered while providing adaptive feedback, as they have an impact upon the effect that feedback has on students. These results also provide hints on the kinds of feedback that are more effective, depending on the emotional state of the student. Based on the results of these Wizard-of-Oz studies, BBK and IOE are currently implementing the iTalk2Learn support mechanisms. The system analyses what it knows about the students (detection of their interaction with the system, student’s speech) in order to support them. Two different components are included within the support:

  1. An emotion detector
  2. An emotion reasoner

Emotion detector:

The emotion detector is based on the student emotions that were detected in the Wizard of Oz studies. The input for the emotion detector is the result of analysis by the Sail Labs (Sail) speech recognition software. This software was specifically trained with children’s voices (in both German and English).

Emotion reasoner:

The emotion reasoner attempts to reduce negative emotions by changing the environment via adaptive support. It uses rules on transforming negative emotions into positive emotions by aligning the student’s reasoning process with the learning task. The rules are based on the Wizard-of-Oz studies, where the platform was used as a tool to investigate the types of support that are effective for particular emotions. Further rounds of Wizard-of-Oz studies are planned in the UK and Germany in order to gather further data for the support. Based on the results, the consortium will refine and develop the support offered by the iTalk2Learn platform.

To find out more about the support mechanisms in iTalk2Learn, get in touch today!