Knowledge Reuse in SEED Exploiting Conceptual Graphs

In terms of conceptual structures theory, the objective of our project is the development of a pattern directed inference systemm7]. It is to operate over conceptual graph terms extended to express constraints on the referent values, and the matching mechanism is to combine order sorted uniication with the action of a constraint solver. This combination leverages type representation technology from the conceptual graph communityy3] and constraint technologyy10]. In terms of the objectives of the SEED projectt6], this system will animate the elaboration and specialization of problem representations. Furthermore, a formal description of the representation and the mechanisms of its generative capabilities will clarify the theoretical basis for case retrieval and design reuse. 1 Overview of SEED SEED is an acronym for Software Environment to support the Early phases in building Design..6] The intent of the SEED project is to create software which will support preliminary design of buildings. This includes using the computer as an active tool which helps to generate designs. The speciic goal of SEED is to aaord the user greater exibility and expressiveness in the design of recurring building types. SEED realizes a computational model of design based on the trial and error exploration of a search space generated by constructive steps. Selective reuse of familiar design elements and approaches gives shape to this space as patterns of constructive operations in a design media. Earlier systems, such as Genesiss8, 9], have demonstrated these concepts: annotated solid models have allowed the formal representation of designs, and shape grammars operating over these annotated solid models have been interpreted as formalizations of the constructive operators. However, SEED generalizes the representation introducing the opportunity for greater reuse of the artifacts of design practice. SEED is built around the idea of a design space, a set of partial or complete solutions to an architectural design problem. This idea is roughly equivalent to the AI term \search space" | the design space is deened by starting states and operators, which allow the derivation of one state from another, and will include some acceptable goal states. The current state is the current focus of interest of the system. Since SEED is always restricted to the domain of building-design problems, it is suucient to call the search space a design space. So, SEED works by growing the design space during the elaboration of a design. To achieve the goal of design experience reuse, SEED allows …