Case-based reasoning (CBR) is a learning paradigm that classifies new instances by analyzing similar past instances, ignoring dissimilar ones. It uses symbolic representations rather than real-valued points. CBR has been applied to conceptual design, legal reasoning, and planning by reusing solutions to similar past problems. The CADET system employs CBR to design simple mechanical devices like faucets, using a library of 75 past designs. It represents functions with qualitative relationships between input and output water levels and temperatures. CADET searches the library for exact functional description matches or partial subgraph matches to suggest solutions for new design problems.