Automating Iconclass: Ground Truth and Model Predictions for Early Modern Religious Woodcuts
Abstract
This dataset accompanies the article: Thomas, Drew B. (2026). “Automating Iconclass: LLMs and RAG for Large-Scale Classification of Religious Woodcuts.” Digital Culture & Education, 16(3). It contains bibliographic metadata, ground-truth Iconclass annotations, and model predictions for a curated corpus of early modern religious woodcut illustrations printed in the Holy Roman Empire before 1601. The corpus includes: Illustrations from the 1534 edition of Martin Luther’s German Bible Illustrations from the 1551 edition of Luther’s Bible A selected thematic subset of biblical scenes (Adam and Eve, Noah’s Ark, Annunciation, Nativity, Last Supper, Crucifixion) Each image record includes a manually assigned ground-truth Iconclass code and predicted Iconclass codes generated using two Retrieval-Augmented Generation (RAG) model configurations: RAG with hybrid (keyword + vector) retrieval over a hierarchical Iconclass database RAG with vector retrieval over a basic Iconclass database Predictions are evaluated using a hierarchical match framework that accounts for full, partial, and over-specified matches. The dataset preserves: Book-level bibliographic metadata Image-level Iconclass annotations Model predictions for two configurations Match-type evaluation categories Digitized page images are held by the respective libraries and are not redistributed in this dataset. This dataset supports research in digital humanities, computational art history, and automated iconographic…
Read the paper
Related studies
- Knowledge Management Perspective of Generative Artificial Intelligence
- Navigating Generative Artificial Intelligence Promises and Perils for Knowledge and Creative Work
- Characterizing the evolving research on enterprise content management
- Large language models and multimodal foundation models for precision oncology
- From Generative AI to Extended Reality: Multidisciplinary Perspectives on the Challenges, Opportunities, and Future of Educational Computing
- Generative Artificial Intelligence: A New Engine for Advancing Environmental Science and Engineering
Go deeper in Evidence Hub
Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.