Analogical reasoning plays an important role in design. In particular, cross-domain analogies appear to be important in innovative and creative design. However, making cross-domain analogiesis hard and often requires abstractions common to the source and target domains. Recent work in case-based design suggests that generic mechanisms are one type of abstractions useful in adapting past designs. However, one important yet unexplored issue is where these generic mechanisms come from. We hypothesize that they are acquired incrementally from design experiences in familiar domains by generalization over patterns of regularity. Three important issues in generalization from experiences are what to generalize from an experience, how far to generalize, and what methods to use. In this paper, we describe how structure-behaviorfunction models of designs in a familiar domain provide the content, and togetherwith the problem-solving context in which learning occurs, also provide the constraints for learning generic mechanismsfrom design experiences. In particular, we describe the model-based learning method with a scenario of learning of feedback mechanism.

Sambasiva Bhatta & Ashok Goel. Model-Based Indexing and Index Learning in Engineering Design. International Journal of Engineering Applications of Artificial Intelligence, Special issue on Machine Learning in Engineering, 9(6):601-610, December 1996.