Digital resources in the Social Sciences and Humanities OpenEdition Our platforms OpenEdition Books OpenEdition Journals Hypotheses Calenda Libraries OpenEdition Freemium Follow us

DiDip Workshop at ICARus Convention (Venice, 13. Nov. 2025)

The ICARus community is meeting every semester in archives all over Europe. This autumn’s edition—the ICARUS Convention #35 (12–14 November 2025)—happened at the Archivio di Stato in Venice and had charters as a main topic. Under the theme “Bridging Centuries”, the gathering brought together archivists, digital humanists, and heritage specialists to discuss digitization, AI applications, and new approaches to archival research, in which our associate Zarko Vujosevic was strongly involved.

Thea Schaaf, Franziska Decker and Giuseppe Consolo at the entrance to the Archivio di Stato Venezia

Obviousy, DiDip had to be present—and we organised a workshop showing the recent developments of the project and contributed to a workshop organised by ICARus Italia targeted at the Italian community on the digital services provided by ICARus: matricula, topotheque and—certainly—Monasterium.net. On Thursday morning, we held a workshop dedicated to DiDip. In this, we showcased the shift from “Digital Diplomatics” to “Distant Diplomatics”, moving from classical browsing and full-text search toward large-scale quantitative analysis and machine learning. The team also introduced the transition from the traditional MOM-CA system to MOM-NG (Monasterium Next Generation), and the modular Virtual Research Environment around it that is designed to integrate archival data, user collections, digital editions, and new analytical tools in a more flexible and scalable way. Two aspects of our work received particularly positive feedback from the audience: HTR and layout prediction. Together, they form robust additions to our Virtual Research Environment and MOM-NG. Florian could even run a live show of the current state of MOM-NG, browsing, searching, and filtering 730,000 charters in milliseconds. Thea explained how she is implementing (and beautifying) all of it through her ReactJS components, and how the data basket feature will support the users.

two images of damaged charters with color coded lines
Bridging the gaps: State of the art line recognition (left) and MOM-NG line segmentation (right)

We presented our—i.e. mainly Nicolas’s—Handwritten Text Recognition (HTR) pipeline and compared it to competitive solutions like Transkribus. Our segmentation architecture can handle specific challenges of charter material, such varying image quality, while remaining robust and scalable for the hundreds of thousands of documents in Monasterium.net. This adds significant changes in downstream tasks using the NLP tools for identifying formulaic patterns, semantic structures, and text reuse across large corpora. Giuseppe introduced the audience in examples of these. He also showed our experiments in (semi-automatically) georeferencing places of issue and producing large-scale historical maps.

Notarial subscription from Landeshauptarchiv Koblenz, Best. 55 A 4, Deutscher Orden, Ballei Lothringen 660, in: Monasterium.net, URL <https://www.monasterium.net/mom/DE-LHAKO/55A4/660/charter>
faces looking through a barred window in a illustration of a charter (see https://monasterium.net/mom/IllUrk/collection)

We showed how their layout segmentation enables new forms of exploration. Franziska could present her first results in detecting notarial signs with the help of Anguelos’ FRAT app. The system detects various layout classes that, among others, cover writing areas, plicae, seals, calibration cards, and crucially, ornamental decoration. Martin Roland (Austrian Academy of Sciences) demonstrated how researchers can leverage these predictions to discover illuminated charters across collections, even in poorly indexed holdings. His examples showed both the potential and current limitations: the system successfully surfaced beautiful finds, while sometimes misclassifying elements (a decorated initial as a seal, archival stamps as decoration). These “errors” can actually lead to serendipitous discoveries; misclassification reveals previously unknown works!

You can have a look at our slides.


OpenEdition suggests that you cite this post as follows:
Georg Vogeler (December 19, 2025). DiDip Workshop at ICARus Convention (Venice, 13. Nov. 2025). DiDip. Retrieved January 20, 2026 from https://didip.hypotheses.org/4255


You may also like...