Utilisation du Clustering Conceptuel pour l'Apprentissage Automatique des Notionsde Fonction, Structure et Comportement en Conception (Using Conceptual Clustering to Learn About Function, Structure, and Behavior in Design)

摘要

Modeling design knowledge is an active research topic among engineers,architects, and artificial intelligence researchers interested in knowledge based design. The primary concerns are knowledge representation and knowledge acquisition. Knowledge acquisition for knowledge based design is difficult because humans acquire design knowledge primarily through experience resulting in a diverse and unstructured body of knowledge. Conceptual clustering, an area of the broad field of machine learning, provides techniques for structuring observations into generalized concepts. Specifically, concept formation shows promise for learning design concepts. A conceptual clustering program is used to learn about design by clustering function, structure, and behavior. These clusters are then used as the basis for learning associations between function, structure and behavior. The resulting representation of design knowledge can be used to provide assistance during the synthesis stage of the design process.