CBR for Design

Abstract

Design research has a number of goals, including a better understanding of design, the development of tools to aid human designers, and the potential automation of some design tasks. Computer-aided design is concerned with using computers to assist in the design process to produce better designs in shorter time. The reuse of well-tested and optimized designs is an important aspect for decreasing design times, increasing design quality, and improving the predictability of designs.

This chapter summarizes the main results of applying CBR to support human problem solving in design. Its special focus is on research carried out in Europe and in particular in Germany. We start by providing a general characterization of design tasks. An overview of applicable reasoning methods, as well as a survey of case-based systems for design assistance, will be provided in Section 8.3. Following this, Section 8.4 discusses characteristics of case-based design (CBD) using the domain of architectural design for illustration. Reoccurring problems in CBD are listed. Sections 8.5 and 8.6 describe a number of methods that have been developed to handle the complexity of design tasks. Section 8.5 introduces an algorithm for flexible case retrieval that allows for the combination of several similarity measures at query time dynamically. In Section 8.6 three approaches to structural similarity assessment and adaptation are presented and discussed. Section 8.7 introduces Eadocs, a multi-level and hybrid expert system for conceptual design tasks that achieves structural adaptation by case combination. To conclude, approaches for automating configuration tasks, as introduced in Chapter 6, are contrasted with approaches aimed at the assistant-like support of design tasks and directions for future research are pointed out.