An emotion information processing model based on a mental state transition network

A machine that lacks of emotion computing ability cannot realize artificial intelligent sufficiently and cannot meet the increasing demanding of human-computer interaction as well. Though most current research is focusing on physical components of emotions, rarely are they carried out from the view of psychology. In this paper an emotion information processing model based on the mental state transition network and a corpus of common sense are proposed to detect human emotion. By a series of psychological experiments, we present a new way to predict future human´s emotions depending on the various current emotional states under various conditions. Besides, people in different sexes, characters and ages are taken into consideration in our experiments. From the psychological experiments data that is abstracted from 250 questionnaires, a Bayesian network for describing the transitions in distribution among the emotions and relationships between internal mental situations and external reinforcements are concluded. Further more, comparing seven relative evaluating experiments we found that the model provided a higher precision average rate of 0.843 respectively for the 50 random data examples.