Reference: Gruber, T. R. Model-based Explanation of Design Rationale. Boston, MA, 1990.
Abstract: Engineering is a knowledge-intensive process. People need access to knowledge about the devices and products throughout the life cycle. In particular, knowledge about how the product works and why it is designed as such is very valuable, for collaborative work within groups of designers, and to support communication with people in other parts of an organization who make decisions about manufacturing, operations procedures, diagnostic procedures, configuration, testing, and marketing. Today that knowledge is fragmented and not in a form that can be interpreted by computer programs. Toward a vision of comprehensive, shared knowledge bases of engineering knowledge, we have begun a program of research aimed at developing representations for building machine-intelligible models of physical devices and the ways the devices interact with the environments of use. The project is called How Things Work, and it is a multi-year, collaborative effort involoving AI researchers, engineers, and physicists from within the KSL and from other centers at Stanford. The goal is to build knowledge-based models about how engineered devices work, and to support human and machine use of the knowledge problem solving and cooperative work. A key objective is to create models that are both machine-understandable and human-usable. Currently we are developing a software environment called DME (Device Modeling Environment) for acquiring and using models of the structure, behavior, and intended function of electromechanical devices. In this abstract we will briefly describe how DME can be used to address the problems of design knowledge capture and interactive documentation, using computer-generated explanations of how things work.
Notes: Updated May.
Full paper available as hqx, ps.