A multi-agent extension of Negoplan and its application to a business strategy game.

Abstract

Negoplan is a domain-independent decision support system which can be customised for different application domains. Negoplan enables the representation, simulation and support of sequential, context-dependent decision-making problems that involve the interaction between two participants and their decision environment. This thesis presents extensions that increase the expressiveness of Negoplan's constructs. They enable Negoplan to represent decision-making problems that involve more than two participants. This enlarges substantially the class of Negoplan's applications. The usefulness of the extensions is demonstrated on a managerial decision-making application that involves more than two participants. This Negoplan application concerns self-training of a Business Administration student to test his managerial decision-making skills in an environment in which Negoplan represents two other autonomously behaving participants. The application also demonstrates the suitability of customised Negoplan as an automated training tool.