Safranbolu House Typologies

Computational Methods for the Analysis of Traditional Safranbolu House Plan Typologies Using Generative Adversarial Networks

Safranbolu, a UNESCO World Heritage site, contains well-preserved Ottoman houses dating from the eighteenth century. The project addresses the limited computational research on their architectural typologies by developing a structured plan database and artificial-intelligence methods for studying their organization and variation.

The research analyzes architectural concepts including form, symmetry, and proportion through machine-learning workflows. Generative adversarial networks, categorical representations, and clustering methods support both qualitative and quantitative comparisons of house plans, while open datasets and applications extend the work to future interdisciplinary research in artificial intelligence and architectural heritage.