Toward a paradigm for validating decision-making in man-in-the-loop simulations

Simulations, as a tool to answer various questions in both commercial and military areas, have proliferated in the past decade. They are no longer just used by the "computer expert" or "operations research specialist" but rather by novice, non-simulation experts needing answers to critical questions within time constraints. Man in the loop simulations are a class of computer models which are typically used for training students in command type of decisions. Man in the loop simulations are distinguished by the decision making process being controlled dynamically by the trainee.; Quality control of computer simulations is controlled through verification and validation throughout the life cycle of the model. Verification is the process that determines that the model and simulation functions as it was originally conceived, specified, and designed. Validation is the process that addresses the credibility of the model and simulation in its depiction of the modeled world. In the simulation and modeling community, there are a lack of tools for the modeler to use to assist in the verification and validation process. The objective of this research is to develop a methodology which allows the user to validate the decisions of the man in the loop dynamically in real time.; This research explores the use of automated tools that can operationally validate the decisions of the student during real time. Case-based reasoning is explored as a modeling paradigm for incorporating a validation mechanism for a man-in-the-loop simulation. The approach taken by this research is to reduce the model users risk (type II error) by reducing the number of invalid solutions. This method holds the potential to greatly increase the model users confidence in the results of a training exercise. A proof of principle demonstration has been developed that operationally validates the decisions made by a military commander in a high resolution simulation. This demonstration establishes the ability of an artificial intelligence system to operate in real time and assist in the validation process.; The contribution to the body of knowledge is primarily in the development of a generic methodology for validating the operational decisions for a man in the loop simulation. Secondary contributions have been made in the areas of artificial intelligence and modeling of the command and control process.