A dynamic interactive kansei user model

The ability to feel emotion has been recently recognized as a sine-qua-non condition for intelligence, but how can we give such a capability to a computer? How can we get users to successfully communicate their feelings to the machine? The research presented in this paper focuses on the emergence of a multimedia language for human and system to communicate about highly subjective concepts (“kansei” in Japanese), such as visual impressions. We suggest that interaction and dynamics are key factors for addressing this issue. Indeed, it is through social interaction that two people are able to reach a mutual understanding over emotional responses. Subjective responses are not static but rather follow from continuous dynamical processes. We propose a multimedia language for the user and the system to communicate highly subjective impressions. Multimedia definitions corresponding to a subjective impression result from intensive user-machine interaction. The user model is implicitly encoded in the emerging language