Cultural overfitting: the architectural canon behind algorithmic survey
VLC arquitectura. Research Journal · 30.10.2026 · 13 (2) · 247-261
DOI: 10.4995/vlc.2026.27391 ↗Kurz gefasst
Fachartikel in der Zeitschrift VLC arquitectura. Research Journal, online erschienen am 30.10.2026. Verfasst von Michele Russo. Die Arbeit nennt 45 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).
Abstract (Originalsprache)
Architectural survey is a process of knowledge that combines interpretation, established methods of data collection and analysis, and the critical representation of results. Advances in instruments and 3D data management have already profoundly altered this process; the introduction of algorithms for automated or semi-automated classification now risks weakening the link between survey, interpretation and understanding of architecture. Every classification presupposes a theory of architecture: a set of categories, a canon of reference buildings, a definition of the boundaries between elements and materials. This article argues that the main epistemological risk of algorithmic survey lies less in the accuracy of models than in the narrowness of the architectural knowledge they embed, a condition defined here as cultural overfitting. It distinguishes four levels of the phenomenon (corpus, taxonomy, morphological canon, stratification over time), with attention to what taxonomies fail to represent, such as designed voids and transformations. A critical reading of the main public datasets reveals a prevalence of European religious and monumental architecture and formal taxonomies shared across datasets; fortified architecture, virtually absent, is discussed as a limit case. The article concludes with principles for culturally situated models that preserve the relationship between architecture and knowledge.
Zitieren
APA
Russo, M. (2026). Cultural overfitting: the architectural canon behind algorithmic survey. VLC arquitectura. Research Journal, 13(2), 247-261. https://doi.org/10.4995/vlc.2026.27391
Vancouver
Russo M. Cultural overfitting: the architectural canon behind algorithmic survey. VLC arquitectura. Research Journal. 2026;13(2):247-261. doi:10.4995/vlc.2026.27391
BibTeX
@article{Russo2026,
author = {Russo, Michele},
title = {Cultural overfitting: the architectural canon behind algorithmic survey},
journal = {VLC arquitectura. Research Journal},
year = {2026},
volume = {13},
number = {2},
pages = {247--261},
doi = {10.4995/vlc.2026.27391}
}
EUVIMED