Reflection-based artificial intelligence in art history – explainable hybrid models for image search and analysis


The project aims to provide both practical and theoretically oriented
reflections on the use of image similarity assessments in art history.
These considerations are to differ from existing approaches in art history
in two respects.
First, art historical expertise in the form of digitized or digitally
available textual collections is to be used. For this purpose, knowledge
graphs shall be created in a semi-automatic and interdisciplinary process
and used to train hybrid AI models.
Second, we want to make the AI-generated results explainable. In
particular, this will be done by incorporating expert texts and
representing them in machine-interpretable knowledge graphs to break the
black-box property of data-driven deep learning models. The practical part
is therefore accompanied by a reflexive investigation in four research
scenarios, where we explore the impact of different text resources and
knowledge graphs on AI-generated results and finally on art historical
research processes.
The perspective of our project is to increase the acceptance of AI, which
is generally highly criticized in the humanities, by addressing the
methodological problems of its use.


Prof. Dr. Ralph Ewerth, Universität Hannover

Prof. Dr. Hubertus Kohle, Ludwig-Maximilians-Universität München

Matthias Springstein, Universität Hannover

Julian Stalter, Ludwig-Maximilians-Universität München

Stefanie Schneider, Ludwig-Maximilians-Universität München

Dr. Eric Müller-Budack, Universität Hannover

Maximilian Kristen, Ludwig-Maximilians Universität München