Info
Leibniz Research Alliance ‘Value of the Past’
Artificial Intelligence in Archives and Collections. Practices, Potentials, and Evidence Production in Dealing with Images and Multimodal Cultural Heritage
Conference
The Research Lab 1.3. “Digital Heuristics and Digital History” of the Leibniz Research Alliance “Value of the Past” is organising a conference in Marburg on artificial intelligence in heritage institutions, such as archives and collections, and how these new technologies are transforming archival institutional practices. The conference will bring together curatorial and archiving knowledge and new AI-based methods, and will also provide a forum for ethical reflections on the use of AI in academic and archival practices.
Topics will primarily focus on – but will not be limited to – visual sources, such as photography and graphic collections or those with mixed image-text sources and multimodal information processing.
The conference will provide a forum for researchers and practitioners from the humanities, archives and collections to connect with researchers and engineers in the fields of artificial intelligence, computer science and the digital humanities, to discuss new findings, and to exchange experiences. The event seeks to promote an interdisciplinary, cross-sectoral dialogue between research, development and practice.
We cordially invite stakeholders from all areas in the field. Attendance without contribution to the programme is welcome.
------------------------------------------------------
Hybrid format: in-person and virtual
Conference languages: German and English
Dates and Conference Venue:
Thursday, December 12th, 12:30 - 19:30
Friday December 13th, 9.00 - 15:45
Herder-Institut für historische Ostmitteleuropaforschung – Institut der Leibniz Gemeinschaft, Gisonenweg 5-7, 35037 Marburg, Germany.
Organizer:
- Leibniz-Forschungsverbund „Wert der Vergangenheit“, Lab 1.3. Digitale Heuristik und Historik
- Leibniz-Zentrum für Archäologie (LEIZA), Mainz
- Herder-Institut für historische Ostmitteleuropaforschung – Institut der Leibniz-Gemeinschaft, Marburg
Registration »
Key Topic: Dealing with Images and Multimodal Cultural Heritage
In the years to come, curatorial and archival processes in memory and heritage institutions (including key aspects of cataloguing such as description, classification and categorization) will be increasingly supported by automated systems and artificial intelligence. These practices attribute value to sources and link archival materials and collection objects to societal narratives. Collections and archives thus form an essential basis for memory-related discourses and shape our view of the past.
New technologies have now reached the stage where they are potentially suitable for the requirements of cultural heritage institutions. There is potential promise in the partially automated indexing and cataloguing of historical sources, and particularly of digital images, the semantics and meaning of which have until recently only been accessible to the human eye and not the machine. Images can now be automatically described in semantic terms and therefore made findable. AI methods – and most recently multimodal AI processing – are opening up new possibilities for automatic text and layout recognition, automated image annotation, and the analysis of visual sources and their contextualisation (e.g. text-image combinations or audiovisual sources).
At the same time, however, there is a lack of knowledge about how evidence – and facts – are generated and how AI processes affect the attribution of authenticity to archival documents and photos. At this moment, the humanities lack semantically high-quality and subject-appropriate training data sets. Similarly, there is hardly any agreement about what constitutes an acceptable outcome of computational classification processes, or what benchmarks should be used to evaluate the results. We therefore need to develop best practices (that may tap into explainable Artificial Intelligence (XAI) methods), benchmarks and goals. To increase the value that the information and knowledge held in our archives and collections have for future research, we need to deepen our understanding of processes and algorithms.
The conference will bring together curatorial and archiving knowledge and new AI-based methods, and will also provide a forum for ethical reflections on the use of AI in academic and archival practices. We will consider how automated processing and AI methods require detailed epistemic reflection and methodological-technical control to ensure that no false or tainted evidence is generated. Thus, the conference will discuss the effects of these new technologies on the production of evidence, thereby contributing to the crucial question of how the digital transformation is changing knowledge creation in the humanities and what this means for scholarship in historical disciplines.
Committee: Elke Bauer, Simon Donig, Annette Frey, Dominik Kimmel
Speakers & Abstracts (pdf download):
We welcome guests to the evening-impulse by Martin Andree on Thursday 17:45 (online). If you plan to attend this presentation only, please also register here.
---------------------------------------------------------------------------------------------------------------------
Programme
(it will be possible to attend all keynotes and sessions online).
Thursday, 12th December 2024
12.00 pm Warm up
12.30 pm Welcome & Introduction
Peter Haslinger, Herder-Institut für historische Ostmitteleuropaforschung – Institut der Leibniz-Gemeinschaft
Achim Saupe, Leibniz-Forschungsverbund Wert der Vergangenheit
Annette Frey, Leibniz-Zentrum für Archäologie
Simon Donig, NFDI4Memory
Dominik Kimmel, Leibniz-Zentrum für Archäologie
Keynote 1
12.45 pm Tayler Arnold, University of Richmond (online)
Explainable and Auditable Search and Discovery of Visual Cultural Heritage Collections
Panel 1 Exploring and Analysing Collections from Textual and Multimodal Contexts
1.30 pm Günther Mühlberger, Universität Innsbruck
AI as an employee. How the digital transformation is (not only) changing archives
2.00 pm Frank Puppe, Universität Würzburg
Pipeline for Digital Indexing of Archaeological Record Cards
Moderation: Dominik Kimmel
2.30 pm Coffee Break
2.45 pm Mahsa Vafaie, FIZ Karlsruhe – Leibniz Institute for Information Infrastructure
Separation of machine-printed and handwritten text in archival documents
3.15 pm Markus Huff, Leibniz-Institut für Wissensmedien, Tübingen
ArchiveGPT: Psychological and technological perspectives on the AI-supported archiving of image material
3.45 pm Elisabeth Mödden, Deutsche Nationalbibliothek, Frankfurt a. M.
Automatic Indexing with Large Scale Vocabulary - Automatic Subject Indexing as Extreme Multi Label Learning Problem (Presentation in German)
Moderation: Kimberly Coulter, Leibniz-Institut für Länderkunde
4.15 pm Coffee break
Panel 2 Computer Vision: Semantic Segmentation, Analysing and Understanding Images
4.45 pm Ralph Ewert, TIB – Leibniz Information Centre for Science and Technology / Leibniz University Hannover
Unlocking Cultural Heritage: Computer Vision for Art and History Archives
5.15 pm Karsten Tolle, Goethe University Frankfurt am Main
Potpourri of Computer Vision in Cultural Heritage
Moderation: Simon Donig (Herder- Institut /NFDI4Memory)
Evening Impulse
5.45 pm Martin Andree, Universität Köln (online)
The Digital Monopolisation of Information and Knowledge
6.15 pm Evening Reception
Friday, 13th December
9.00 am Warm up
Continuation of Panel 2: Computer Vision: Semantic Segmentation, Analysing and Understanding Images
9.30 am Angelica I. Aviles-Rivero, Tsinghua University, Beijing (online)
Dusting Off the Unlabelled Data: Graph Semi Supervised Learning for Large-Scale Datasets
10.00 am Erik Radisch, Leibniz Institut für Länderkunde, Leipzig
A new Approach to Semi-Automated Annotations with Segment-Anything (Meta AI)
10.30 am Yury Korolev, University of Bath
Image filtering based on total variation spectral decompositions: an overview of the method and an application in medieval paper analysis
Moderation: Simon Donig (Herder- Institut /NFDI4Memory)
11.00 am Coffee Break
Panel 3: The Effects of Computational Methods on Image Analytical Research and Visual Studies in Cultural Heritage
11.15 am Carola-Bibiane Schönlieb, University of Cambridge (online)
Unveiling the invisible - mathematical imaging for cultural heritage
11.45 am Christopher Kermorvant, TEKLIA, Paris
How to use controlled vocabularies to describe early Japanese photographs using deep learning models
12.15 pm Peter Bell, Philipps Universität Marburg
Ways of Seeing in AI and art history
Moderation: Mila Oiva, University of Turku
12.45 pm Lunch Break
Keynote 2
1.30 pm James Evans, The University of Chicago (online)
The Geometry of Culture: Analyzing meaning through embeddings of text and images
Panel 4: Opportunities and Challenges for the Automated Indexing and Cataloguing of Visual Sources in Archives and Collections
2.15 pm Nicole Graf, ETH Zürich (online)
Between ritual and relief – when the computer squints: analysis of Al-based indexing in the Image Archive of the ETH Library
2.45 pm Larissa von Bychelberg, Uppsala University
Artificial Intelligence Technology’s Influence on the Authenticity of Digital Intangible Cultural Heritage Archives
Moderation: Peter Haslinger
3.15 pm Plenary Discussion & Summary
3.45 pm End of the conference
