To Appear In: Journal of Knowledge Based Systems U SING ADAPTATION K NOWLEDGE TO R ETRIEVE AND ADAPT D ESIGN C ASES Barry Smyth Hitachi Dublin Laboratory, Trinity College Dublin, Dublin, IRELAND. EMail: barry.smyth@hdl.ie Mark T. Keane Department of Computer Science, Trinity College Dublin, Dublin, IRELAND. EMail: mark.keane@cs.tcd.ie ABSTRACT Two critical stages in case-based design are the retrieval of a suitable design and the adaptation of that design. When a new design problem is presented to the system, the retrieval stage locates a similar design solution that is reusable in the current context. During adaptation, retrieved designs are modified to meet the specific demands of the target problem. In this paper, we address both of these stages in the context of a case- based software design system called Dj Vu. In particular, we argue that there is a close relationship between the reusability of a design and its adaptability. We present an approach that allows the adaptation requirements of design cases to be accurately determined during retrieval and subsequently exploited during adaptation. We show that this approach results in improved retrieval accuracy, flexibility, and greater overall problem solving efficacy. In short, our position is that this "adaptation-guided retrieval" will ensure optimal reusability in case-based design. Keywords: Case-Based Reasoning; Retrieval; Adaptation; Software Design 1 Introduction Case-Based Reasoning (CBR) is a reasoning method that exploits experiential knowledge, in the form of past cases, to solve new problems (see [1]). When faced with a new problem, a CBR system will retrieve a case that is similar, and if necessary, adapt it to provide the desired solution. To date, there has been some success in applying the CBR paradigm to complex design tasks [2,3,4,5]. A case-based design (CBD) system will be successful if it is cost-