Architectural Design Knowledge Modeling

We have presented an approach for automated knowledge modelling and re-use in design through the development of an algorithm that operates in conjunction with the machine learning technique. The approach presented has the advantage that such automated knowledge modelling system works in conjunction with the problem-solving process as the knowledge representation is in an embedded/distributed structural form with fast response. The model has the features of an artificial intelligence (AI) system. The AI approach adapted belongs to the methodology of soft computing and it is most convenient to retain the knowledge from architectural data, for instance, that are rather soft as well. Through the use of domain specific knowledge a designer determines attribute values based on dependency among the design variables to support the design development process. The present research is carried out in relation to the determination of the architectural design quality of underground stations in the Netherlands. The methodology employed is a multivariable functional approximation structured in a neuro-fuzzy knowledge representation form. The paper deals with the details of the novel design knowledge modelling and highlights its important merits for architectural design.