Modelling situations in intelligent agents
John Thangarajah, Lin Padgham, Sebastian Sardiña · 2006
BDI agent systems and languages such as PRS, JAM, JACK, 3APL, and AgentSpeak have been widely used in developing robust and exible applications in dynamic domains. However, one criticism of these systems is that the modelling of how agent reasoning progresses is too reliant on the rather low level notion of individual events. In our own work in a number of application areas, we have consistently noticed a need for a more abstract concept, which we call a situation. Recognition by the agent that it is in a particular situation may affect the goals that it has, may place overarching constraints on how it operates, or may influence the way that it chooses to achieve its goals.