抄録
This paper proposes methods of knowledge representation and acquisition for conceptual design of physical systems by introducing the notion of determination rules. From Value Engineering perspectives, physical systems are interpreted by their functions, structural attributes, and the means-ends relations relating with them, i.e., functional compositions, physical causal laws, and structural compositions. These knowledge are used as the initial domain theories of EBG (Explanation-Based Generalization), then they are refined through compilation. Through examining the fundamental structures of determination rules they are then introduced to the domain theory of EBG. They reflect relations among functions structures and their causes/effects by which various types of empirical design knowledge can be reprsented. Using determination rules as the domain theory of EBG, they are once instantiated for an exampler object, and then generalized by identity elimination. Through these processes, they are reformed into efficient and rationalized design knowledge utilizable for systematic design processes. This method is implemented to a system in a logic programming environment.
