Full Article:

This research seeks and proposes the integration of architects' common stock of knowledge to leverage artificial neural networks, given the lack of active tools that allow for human-computational interaction and that can assist in the preliminary stage of design in an informed manner. The research was conducted as experimental and qualitative research, and its methodology design was divided in two stages, working first in a qualitative approach, capturing extractable implicit knowledge with semantic analysis, and secondly, by delivering the knowledge to an ANN through the processing of semantic information by a NLP model to create matrix and node graphs. Results demonstrate the potential of ANNs to generate representations from declarative database adjacency relations and show that the leveraging of AI with architects' declarative knowledge could provide a way to improve the initial phases of architectural design and decision-making through an iterative and knowledge-guided approach.