Identification of potential systems by the construction of system behavior
Problem solving in Distributed Artificial Intelligence involves the decomposition of tasks into subtasks that are then distributed to single autonomous agents. If the collection of these agents is viewed as a system, then several important problems exist involving the relationships between the individual agents and the system. Some of these problems are, deriving system behavior and understanding what the system is achieving as the individual agents achieve their goals. This thesis discusses how the concept of system behavior can be defined by considering the formation of system behaviors from the unification of rules, how a goal tree structure could be used to prune unnecessary behaviors, and how the derived system behavior can be used to identify potential systems.