Revealing shape semantics from morphological similarities of a collection of architectural elements: The case study of the columns of Saint-Michel de Cuxa

Over the last three decades, the introduction of digital technologies in the field of architectural documentation has profoundly changed the tools and the acquisition techniques. Most of the developments concerns metrical and colorimetric characteristics of the objects studied. These developments, surrounding the practice of architectural survey, tend to respond primarily to the requirements of completeness. In this context, it seems necessary to assess the impact of these instruments on the cognitive value of architectural representation. With a strong technological presence, the study of the built heritage is facing a problem of "overload" of information. Indeed, it fails to strengthen the representation in its role as vehicle of knowledge. Confronted with the intelligibility deficit, this article propose an original approach for reading morphological features of an artifact by using a bottom-up approach: the meaning of the elements comes from the analysis of low-level geometric properties of a collection of instances related to the same theorical model. The idea is to rely on data accumulation in order to make emerge high-level semantic features from the comparative analysis of common low-level geometric features. The introduced principles are illustrated by the comparative analysis of 31 columns of the cloister of the abbey of Saint-Michel Cuxa, aimed to identify a common semantic layout.