Overview

Figure 3.1 is an overview of the planner. The state of the system is made up of the beliefs, and a record of the dialogue history, which is simply a list of acts. The choice of these state components was motivated by the structure seen in BDI architectures, where from the beliefs, desires, and dialogue history, each agent's intention can be inferred by plan recognition (see section 2.8). The processes of the system are the planner, which constructs the game tree from the act alternatives, and the evaluator, which evaluates the tree in the context of the agent's belief state. These processes together are used to choose the most efficient dialogue strategy given the agent's belief model. The belief revision process is used to revise the agent's beliefs in the light of executed acts as they are added to the dialogue history.

Figure 3.1: Overview of the Planner