The difficulties in CAD data interoperability arise from the need for using heterogeneous CAD systems and the lack of a proper notion for describing CAD designs [Raghothama and Shapiro, 2002]. Existing CAD systems each have their data formats, and the information recorded in these formats vary significantly. Some systems have the ability to record design histories for representing the design rationales, but others do not.
We present a semi-automatic system that gives the user 1) the flexibility to define and use a rich semantic model that can go beyond the current standards and 2) the capability to efficiently “teach” the system to learn design intent and objectives. Given a user-defined semantic model (e.g., a set of relations between CAD sketches) and a few examples of the desired variations of a CAD design, our system learns the design rationales automatically and selects the semantic descriptions that best represent the design rationales. We recreated the ambiguous CAD designs described in Raghothama and Shapiro’s previous work (Raghothama and Shapiro, 2002) and tested our system with these CAD designs using a set of user-defined relations between CAD sketches (e.g., dimensions). In the experiment, our system successfully learned the design rationales from a few examples of design variations and generated constraints between sketches in the CAD models to prevent possible ambiguities.