Аннотация
The detection of user emotions play an important role in human-machine-interaction. By incorporating emotion recognition, applications such as monitoring agents or digital companions can be improved. The recognition result is used to model the state of the user. This leads to an adaptive reaction of the system towards users needs and claims and thus to an enhanced system reaction.
Besides emotions, based on short events, personality and moods also play an important role. Standard emotion recogniser do not consider them adequately and therefore neglect a crucial part of user modelling.
The challenge is to gather a reliable prediction about the actual mood of the user and moreover represent changes in users' mood during interaction. This includes both the different user-dispositions and the observation of the inner user-state. Therefore conclusions from external conditions has to be drawn.
For this we suggest a model with few parameters that, based on a prevailing mood represents the actual mood of an user. Thereby recognised emotions are modelled as forces, appealing on the mood and shifting it within the VAD-space. Additionally by implementing a repulsing force a decaying mood can be modelled. Thereby it is possible to simulate the short-time effects and the volatility of emotions and also the slow and continuous development of moods.
Furthermore the model is able to incorporate different traits of personality. The prevailing mood, representing a certain user-disposition, will be shifted within the VAD-space. In Addition, recognised emotions are assigned to different forces, by what external observations according to personality-traits can represent different inner user-states.
Пользователи данного ресурса
Пожалуйста,
войдите в систему, чтобы принять участие в дискуссии (добавить собственные рецензию, или комментарий)