The RSI temporal-cognitive model for believable and introspective interpretation in intelligent tutoring systems

摘要

We approach the characterization of believable, higher education-oriented agents for assisting humans immersed in the problem of distance (i.e. virtual) education (Espinosa, 1994) on a highly interactive course (Espinosa, 1995). Intelligent behavior is modeled by temporally describing emotions that affect the outcome of decisions. We assume that introspection leads to better decision making in complex human activities. A set of Enclosed but Open Worlds (Espinosa, 1996) allows for the specification of a computable structure representing such activities. The introspection mechanism includes temporal, educational, and cognitive modeling of user interface issues, and was first described as a monitor in (Espinosa, 1996). We characterize tutoring techniques in terms of multi-modal (i.e. temporal and emotional) operators, and causal relationships, and present a computationally oriented merger of cognitive and artificial intelligence theories. It should be viewed as a multidisciplinary effort.