Computational situated learning in design

This paper presents the development of a computational system of Situated Learning in Design (SLiDe). Situated learning is based on the notion that knowledge is more useful when it is learned in relation to its immediate and active context, ie its situation, and less useful when it is learned out of context. The usefulness of design knowledge is in its operational significance based upon where it was used and applied. SLiDe elucidates how design knowledge is learned in relation to its situation, how design situations are constructed and altered over time in response to changes taking place in the design environment. SLiDe is implemented within the domain of architectural shapes in the form of floor plans to capture the situatedness of shape semantics. SLiDe utilises an incremental learning clustering mechanism not affected by concept drift that makes it capable of constructing various situational categories and modifying them over time. The paper concludes with a discussion of the potential benefits of using SLiDe during the conceptual stages of designing.