Study of Automated Command Support Systems.

Abstract : This study addresses techniques for the design of automated support systems for naval management. It seeks to apply advanced computer techniques developed by the artificial intelligence community to real-world command environments. The managerial environment exhibits two major differences from most other application environments. First, the problems encountered by the manager are not routine or predictable. It is important that it be possible to tune the system to the actual needs as the problem develops. Second, the manager is the expert; he must be able to understand the system's behavior and the reasons for the actions it takes. In combination, these requirements indicate that the commander, or his delegate, should be able to modify the extend the system even though they may have little knowledge of programming or system design. The crucial design issue is to provide means by which the human user can exercise deep control without requiring a correspondingly deep knowledge of the system's implementation. The primary device we have exploited to provide the user with effective control is to base the design on the explicit encoding of the system's knowledge as a set of models each of which contains the constraints and goals that describe necessary relations in the information that describes the situation in a specified component of the application environment. These models remain available to the user for his study and modification. The implementation principles used to develop this design have been demonstrated in an experimental system called ACS.1 for automated command support.