Architectures of Meaning: Linking Images and Texts for Integrated Cultural Heritage Analysis

This project brings together philologists, architects, art historians and computer scientists with a shared goal: to build a common framework for analysing how images and texts interact within Cultural Heritage documents. Architectural drawings, artistic details, and technical descriptions often refer to one another, yet they are preserved across different media, languages, and historical contexts. By creating a unified digital environment where these materials can be encoded, compared, and linked - and where the gradual evolution of images over time can also be traced - we aim to make these relationships visible and analytically accessible for the first time. Guarino Guarini’s treatises and architectural drawings provide a representative and locally significant use case, while serving as a scalable model for international datasets. Two complementary approaches guide the analysis: annotated images are compared with other images both within Guarini’s works and in external sources such as European stereotomic treatises, e.g. those written by Derand, Jousse, Desargues; and text–image relations are modelled on two levels, linking texts to images described in the same document and to images in other works, including materials in different languages. This enables the creation of glossaries, synoptic views, and multilingual networks of relations. The project will produce multimodal XML/TEI-encoded datasets - including image variants, text-image alignments, and named entity structures - suitable for computational analysis and future AI training. Instead of creating new tools, we reuse and extend existing technologies such as EVT (Edition Visualization Technology), integrating them into a research infrastructure that supports multidirectional linking, exploration of heterogeneous materials, synchronic and diachronic analysis at scale. It will establish reusable standards for image philology, provide training-ready corpora for AI, and contribute to TEI-based knowledge networks.