|Institution:||University of Windsor|
|Keywords:||Affect Control Theory; Cognitive modeling; Computational models of emotion; Fuzzy modeling|
|Full text PDF:||http://scholar.uwindsor.ca/etd/5255
Emotion modeling is a multi-disciplinary problem that has managed to attract a great deal of research work spanned to a wide spectrum of scholarly areas starting at humanistic science fields passing through applied sciences and engineering and arriving at health care and wellbeing. Emotion research under the umbrella of IT and Computer Science was extensively successful with a handful of achievements especially in the last two decades. Affective Computing is an IT originated systematic research area that strives to best model emotions in a way that fits the needs for computer applications enriched with affective component. A comprehensive Affective Computing based system is made of three major components: a component for emotion detection, a component for emotion modeling, and finally a component to generating affective responses in artificial agents. The major focus of this dissertation is on developing efficient computational models for emotions. In fact most of the research works presented in this dissertation were focused on a sub problem of emotion modeling known as emotion regulation at which we strive to model the dynamics of changes in the emotional response levels of individuals as a result of the overt or covert situational changes. In this dissertation, several emotion related problems were addressed. Modeling the dynamics for emotion elicitation from a pure appraisal approach, investigating individualistic differences in emotional processes, and modeling emotion contagion as a type of social contagion phenomena are a few to name from those conducted research works. The main contribution of this dissertation was to propose a new computational model for the problem of emotion regulation that is based on Affect Control Theory. The new approach utilized a hybrid appraisal-dimensional architecture. By using a fuzzy modeling approach, the natural fuzziness in perceiving, representing and expressing emotions was effectively and efficiently addressed. Furthermore, the combination of automata framework with the concept of bipolar emotional channels used at the heart of the modeling processes of the proposed model has further contributed to promote the behavior of the model in order to exhibit an accepted degree of human-like affective behavior.