An Emotion Transition Model using Fuzzy Inference

This paper proposes an emotion transition model considering internal emotions and external inputs. The proposed model has internal emotion values representing each intensity of multiple emotions as internal states. Each intensity of emotions estimated from the input texts is given as the external inputs. These values of intensity are used in the antecedent part, and emotion transition is estimated by fuzzy inference. The proposed model can treat the same external inputs differently considering the internal states and it employs the idea of a psychological phenomenon, the mood congruency effect: The proposed model can express emotion transition more naturally. According to subjective evaluation experiments, high correlation between the proposed model and humans is obtained.