Qing Cheng and Costas Tsatsoulis
We present a unique methodology that combines case-based design, explanation-based reasoning, and decision theory for the evaluation of design alternatives. Our technique uses: (1) case-based reasoning to generate design alternatives, (2) explanation- based reasoning to assign subjective probabilities to the possible values of missing design specifications, and to assign subjective utilities based on multiple criteria to the design alternatives, and (3) decision theory to find the best alternative. Our technique integrates case-based design and decision theory, and offers a way to deal with missing and incomplete design specifications and to document and explain design decision making and intent. The methodology has been implemented in an intelligent, case-based system in the domain of pharmaceutical design.