Knowledge about other agents in heterogeneous dynamic domains

Abstract

The authors are concerned with heterogeneous agents in dynamic domains, where an agent may have to interact with varying number and various types of agents. To assist the inevitable computational limitations involved in reasoning and deciding within interactions, a focusing mechanism is essential in the dynamic update of an agent's knowledge about other agents. The authors claim that this focusing should be led by motivations and intentions of an agent. At the preliminary stage they identify the possible relationships that exist among agents. From these relationships they draw a framework within which an agent reasons about those agents who are relevant to it and influence its persistence in holding on to its goals, and further, identify the decision making points involved in this process.<>