Abstrakt

In 2004 a group of affective computing researchers proclaimed a manifesto of affective learning that outlined the prospects and white spots of research at that time. Ten years passed by and affective computing developed many methods and tools for tracking human emotional states as well as models for affective systems construction. There are multiple examples of affective methods applications in Intelligent Tutoring Systems (ITS). This paper revises whether the white spots from the 2004 manifesto have been covered as well as explores the progress in affective tutoring systems construction. The article provides also a brief comparison of selected affective tutoring systems. When reviewing affective computing literature in 2014 one might be impressed – there are algorithms that recognize affect from diverse input channels with accuracies reaching over 90%. There are emotional virtual characters and some of them are used in e‐learning environments. However, although there is much progress in affective computing research, it seems that more white spots are found than covered, which is typical for relatively young scientific domains. Highest emotion recognition accuracies are obtained for distinguishing two emotions, which is obviously not enough for learning systems. Some of the future challenges include: improvement of both accuracy and granularity of emotion recognition, methods for emotion representation models mapping and comparison, integration of emotion recognition results from multiple input channels, effective intervention models for learning, design patterns and frameworks for affective tutoring systems, models for quantifying and integration of uncertainty related to emotional states recognition, affect‐adaptive control flows and more. The thesis of this paper can be summarized as follows: Affective computing grew up from infancy, however it is still far from maturity especially when applied to learning support. During the decade of diverse investigations, affective‐cognitive imbalance in ITS has changed in research, however has not changed in learning support tools.

Autor (1)