Exploring Architectural Design Cases1

information. Instead one has to cope with the problem of different theories and typological approaches competing with each other of time. Analogies in architecture depend more on the perception devoted to some theoretical approach than on physical properties of an architectural case. Architects use old solutions as source for inspiration in the design process. We present a model for the representation of architectural design cases. We face the problem of formalizing ill-structured domain knowledge without a consistent theoretical base. We further investigate the relationship between architectural design information and the languages used to represent that information. The theoretical model of language-game abstractions (LGAs) is presented as a means to study these relationships. Based on that model we propose an information structure that supports intelligent reasoning. This is used to retrieve appropriate case material about some design goal. The definition of similarity between design cases is a crucial point in the design of the retrieval algorithm. We use AI-techniques, which are also interesting for other Hypertext applications. In the following we want to present our approach to modelling the representation of architectural design cases. In [Kühn93] we introduced language game abstractions (LGAs) as means for representing architectural design knowledge. Section 2 gives an overview of this model. Section 3 emphasizes briefly the information structure of our current approach. In section 4, retrieval issues are approached and examples of exploration strategies are given. Section 5 summarizes the results and gives further research perspectives. 2 Language Game Abstractions Newell and Simon [Newe63] describe the design process as a search problem. Well defined goal states can be resolved starting from an initial state and using transitions between states. In this model, design is the path from the initial state to the a priori defined goal state.