070370 UE Workshop on Methodology (2026S)

This course aims to give students experience in thinking about the history of digital collections and the effect of that history on AI models. The course will include lectures by the instructor, discussion of readings, and students' writing project proposals in the digital humanities.
Students will read papers on the design and auditing of language model training corpora. They will learn how to analyze training data sets and how to probe models. Students will also read papers and get hands-on experience analyzing large datasets to understand their structure and development, the "generative process" that gave rise to the data we have.
At the end of the week of the Blockseminar, students will propose a topic for a project involving analyzing the structure and measurement of large datasets and/or the effect of these modeling choices on model behaviour. Guided by feedback from the instructor, they will then write a project proposal to perform these analyses and experiments. This written proposal will be the basis for the main evaluation of student learning.

Students in this course will benefit from having taken other courses in the DH program, although these are not strictly required. In particular, students who have completed or are taking "GenAI for Humanists", "Doing Data Science", "Data Ethics and Legal Issues", "Visualization of Humanities Data", "Digitale Edition und Analysemethoden", or "Netzwerkanalyse in der Literaturwissenschaft" will find that the content of the current course complements concepts and methods covered in those courses.

The current course will allow students who have specialised in any of these areas of Digital Humanities to put their knowledge to work in proposing an open-ended project and getting initial feedback.