Building a Physical Feature Database for Qualitative Modeling and Reasoning

Building a large-scale knowledge base of engineering common sense is indispensable for the development of intelligent CAD systems. At The University of Tokyo, we have started a project to build a knowledge base of physical features in the domain of mechanical design. A physical feature is a qualitative representation of a physical phenomenon and related attributes. The physical feature database is intended to be used for model building, automatic model generation, consistency management, and qualitative behavioral reasoning. This paper presents the knowledge representation scheme for the physical feature database, current state of the development, and future research directions.