ABSTRACT
Many common design tasks involve reasoning with incomplete domain theories and imprecise or dynamically changing problem specifications. We refer to such tasks as plausible design. A system for providing plausible design advice should be able to help elaborate a design problem, propose plausible values and solutions, create variations on a design, and guide the design process. The challenge for any computational model is that the problem spaces are not easily enumerable, problems may entail tradeoffs and compromises, and advisory systems must flexibly respond to users, rather than follow a single rigid strategy for control. We describe a model of plausible design and its implementation in JULIA, a program that designs the menu and presentation of a meal.
