Humanist Discussion Group, Vol. 38, No. 353.
        Department of Digital Humanities, University of Cologne
                      Hosted by DH-Cologne
                       www.dhhumanist.org
                Submit to: humanist@dhhumanist.org




        Date: 2025-02-04 11:53:22+00:00
        From: James Cummings <james@blushingbunny.net>
        Subject: ATNU Virtual Speaker Series - Hannah Busch - "Matching Medieval Manuscripts with Machine Learning" - 2025-02-19

Animating Text: Hannah Busch (Newcastle)
https://research.ncl.ac.uk/atnu/news/atnuvirtualspeakerseries-hannahbusch-2025-02-19.html

Our next speaker in the ATNU Virtual Speaker Series is Hannah Busch from
the CCeH at the University of Cologne who will talk to us about "Matching
Medieval Manuscripts with Machine Learning".

Join us on Wednesday 19 February 2025 at 4pm UK time. (We will send the
zoom link to all registered attendees shortly before the event.)

"Matching Medieval Manuscripts with Machine Learning"
Hannah Busch (University of Cologne)
Wednesday 19 February 2025
4pm (GMT) (8am PT, 11am ET)

Abstract:  Large-scale digitization projects of the past twenty years and
the possibility of exploitation with the help of the International Image
Interoperability Framework (IIIF) have substantially contributed to
reaching a critical mass which allows the application of deep learning for
the study of medieval book scripts. In the past years, not only the number
of digitized medieval sources increased significantly, but also the quality
of the image data. Parallel to this development, the computation of images
is becoming more powerful, and—more importantly—affordable.

During my presentation I am going to talk about the possibilities of dating
and localizing the origin of medieval Latin manuscripts with the help of
Deep Machine Learning/Artificial Intelligence. I will be giving insights in
how to approach such an undertaking of building an image similarity search
based on palaeographical features of medieval Latin scripts. In particular,
I’ll be focusing on the reuse of existing scholarly manuscript descriptions
for the training of Artificial Neural Networks and the challenges that come
with relying on those new technologies. How is the palaeographic
information encoded in descriptive metadata? Can manuscript metadata be
read and processed by the machine? Can it be used to teach Artificial
Neural Networks which manuscript samples are similar by means of Latin
palaeography? To conclude my presentation, I’d like to discuss how we can
build a bridge between the output of the artificial palaeographic eye and
the human readable descriptive metadata.

Bio: Hannah Busch studied German-Italian Studies (BA) and Textual
Scholarship (MA) in Bonn, Florence, and Berlin. She worked as a research
associate at the Trier Center for Digital Humanities, and as a PhD
candidate within the project “Digital Forensics for Historical Documents”
at the Huygens Instituut (KNAW) in Amsterdam and at Leiden University. In
her doctoral thesis, she is working on the application of deep machine
learning methods for the dating and localization of medieval Latin
manuscripts. Her research interests also lie in various areas of digital
medieval studies, in particular the (mass) digitization of medieval written
documents and experimentation with computer-aided methods for manuscript
research. Since June 2023, Hannah is a research associate at the Cologne
Center for eHumanities (CCeH) within the academy project Formation of
Europe.
===

If you missed our previous talks you can see recordings of them at:
https://research.ncl.ac.uk/atnu/speakers/


_______________________________________________
Unsubscribe at: http://dhhumanist.org/Restricted
List posts to: humanist@dhhumanist.org
List info and archives at at: http://dhhumanist.org
Listmember interface at: http://dhhumanist.org/Restricted/
Subscribe at: http://dhhumanist.org/membership_form.php