From Digital to Distant Diplomatics at DH2023
On July 10-14, the 2023 edition of the global DH conference by ADHO will take place in Graz. As our project is hosted with the local organizers, we obviously feel compelled to engage in this.
During the conference, we will present a paper titled “Probabilistic Modeling of Chronological Dates to Serve Machines and Scholars” (together with Andreas Habring and Martin Holler from the Mathematics Institute) and two posters titled “Digital contributions to a 300 years old methodology: Diplomatics & DH” and “Few Shot Classification for Labeling of Medieval and Early Modern Charter Texts“. The paper will cover the modeling of document dates such that it can allow scholars to express uncertainty when annotating data, it acts as a differentiable loss function for training models, and it allows for unbiased interpretable performance evaluation under uncertainty. The first poster will cover work in progress on showcasing which charter features are identifiable automatically and testing which other features are recognized by the machine that can be useful for diplomatics research; while the second poster will cover our strategy to support filtering the descriptive texts and the transcriptions in Monasterium.net which seeks to assign semantic categories for the legal acts they record, such as confiscation, donation, or property sale.
Additionally, our institute is leading the BeCoRe project, a sister project of DiDip. The BeCoRe team will present a poster titled “How to detect institutional and regional feature clusters in late medieval charters? Collaboration between more and less digital humanists in the project BeCoRe” (Sébastien Barret, Marlène Helias-Baron, Dominique Stutzmann, Niklas Tscherne, Georg Vogeler, Jacqueline Schindler), which will present the project and discuss what role machine learning methods can have for the human interpretation of late medieval charters as means to convey authority, especially regarding their graphical aspects.
If you’re a diplomatist interested in digital methods, you will probably share our interest in the “Programming Historian: Developing a Digital Humanities Tutorial” workshop, as well as the “Digital Editions” and “Network Analysis” sessions, in addition to the “Supervised vs. Unsupervised deep learning for medieval Hebrew manuscripts” paper.
Stay tuned for more updates, as we will be reporting on our further engagement in the conference and the presented material!