Analyzing the Digitized Roman de la Rose: The Evolution of a Digital Humanities Project

The Roman de la Rose was the most popular secular work in the European Middle Ages, with over 300 extant copies, 146 of which are in the Roman de la Rose Digital Library (RDL). The size of the corpus allows the Rose to serve as an intriguing window into late medieval manuscript production and economies of culture. I am particularly interested in exploring trends in illustration, text density, and parchment color to situate the Rose as an object of economic value over place and time. I have used the RDL over the past decade to experiment with data visualization and computational image analysis, focusing on the evolution of my approach as new technologies have become available and more accessible. The RDL publicly shared codicological and manuscript data, which I synthesized and expanded to create an interactive visualization of the extant corpus. The next evolution of the project used Imageplot software to visualize a sample of 9000 manuscript images and visualized them over time by saturation level as a way to see trends in parchment and color density over time. The project continued to evolve as I developed a recommender system for the images through the use of convolutional neural networks (CNN) - something we now simply call AI. The programming threshold for doing this work was high, so the work stalled until the February 2025 launch of Distant Viewing Explorer, a browser-based interface for object detection and other ways of exploring large sets of images with AI, which is reinvigorating the project. I am experimenting with computer vision on manuscript images. All research necessarily evolves as new opportunities become apparent and technologies arise. This project is intentionally emblematic of that process, as it builds on an earlier digital humanities project and engages with a particularly fast-developing computational methodology.