A generic task strategy for solving routine decision making problems

Abstract

An algorithm to solve routine decision making problems is presented here. This algorithm combines plan-step generation to find actions that can be applied to solve the particular problem, plan-step assessment to evaluate how useful each action is within the context of the given problem, and plan-step assembly to select and sequence the best actions. These processes can themselves be implemented using the generic tasks of hierarchical classification, routine recognition and a variation of abductive assembly. This paper describes routine decision making problems, the routine decision making algorithm and demonstrates it through three knowledge-based systems.