top of page

DaSCHCon 2026 – AI in Practice: Experiences, Workflows, and Platforms

Sep 25
3 min read

We are happy to invite you to the DaSCHCon 2026 – AI in Practice: Experiences, Workflows, and Platforms that will take place on Wednesday, 21 October 2026 at the University of Bern in Bern and online.


When: October 21, 2026 (09:45-16:45)

Where: University of Bern, Hochschulstrasse 4, Kuppelraum and online 

 

Artificial Intelligence technologies are increasingly shaping how research is conducted across disciplines – from text and image analysis to metadata enrichment, pattern detection, and knowledge extraction. But how are these tools actually used in research and cultural heritage environments? And to what extent do those algorithms benefit, at what cost (e.g. data sovereignty, environmental impact), and what are the potential traps?

 

Researchers, data specialists, and professionals from academia and cultural heritage institutions are invited to share how they integrate AI into their scholarly workflows and institutional practices.


Programme

9:30–10:00

Arrival and Coffee

 

10:00–10:15

Welcome

Rita Gautschy (DaSCH)

Session A – Seeing the collection: images, vision models, and what they cannot see

10:15–10:35

Face Recognition in Painted Portraits: What Works, What Fails, and Why It Needs Art

Historians

Sébastien Marcel (Idiap) & Marie-Eve Celio-Scheurer (MAH)

10:35–10:50

AI Poster Indexing

Ann-Kathrin Seyffer & Oliver Sievi (Swiss National Library)

10:50–11:05

An Image-based and AI-assisted Metadata Enrichment Pipeline for Print Collections

Cristóbal F. Barria Bignotti & Bénédicte De Donker (MAH)

11:05–11:20

AI Anomaly Detection in Non-Invasive Archaeological Datasets. Between Use Cases and Dead Ends

Piotr Wroniecki (Montefortino Prospection & Digitalisation)

Session B – Opening up: access, responsibility, and the human layer

11:20–11:40

Audits eat edits for breakfast? Experiences from scaled AI-supported metadata enhancements for a historical glass-slide collection

Johannes Alisch (Digital Collections Bern)

11:40–11:55

When the language model queries the catalogue: What GLAM institutions provide when they host an MCP server

Arno Bosse (RISE)

11:55–12:10

Human Infrastructure: Supporting Wikipedia in the Age of AI

Iolanda Pensa (SUPSI)

12:10–13:10

Lunch

 

Session C – From text to structure: editions, entities and semantic models

13:10–13:25

AI-assisted data extraction for FAIR-compliance: From a source-evidence approach to semantic modelling

Martina Diaz & Maarten Delbeke (ETH)

13:25–13:40

GraphChat: Grounded Conversational Access to Cultural Heritage Knowledge Graphs

Matteo Romanello, Harshdeep Singh, Yurui Zhu & Giovanni Colavizza (Odoma Ltd.

13:40–13:55

Towards an LLM-Driven Semantic Pipeline for Cultural Heritage Knowledge Graph Construction

Fabrice Tereszkiewicz (A/Z&T GmbH)

13:55–14:10

Linked Elites – ORD for the SSH (LESSH)

Francesco Beretta & Stephen Hart (University of Bern)

Session D – Choosing and running models: evidence, sovereignty, sustainability

14:10–14:30

Benchmark First: Making Model Choice a Routine Step in Research Data Workflows

Maximilian Hindermann (RISE & University Library Basel)

14:30–14:45

Domain Adaptation for Small Language Models

Tobias Steiner & Merlin Streilein (University of Bern & Dodis)

14:45–15:00

Transparent environment for a sustainable semantic exploration of rare books (“TESSERA”)

Roman Walt (ETH Library)

15:00–15:30

Coffee break

 

Workshop Part A – Messy sources, clean records: small workflows and their error budgets

15:30–15:48

3 Inputs: Chronoquill by Gentiana Rashiti & Nicole Graf (ETH Library); Artificial Intelligence: A Millstone Around Our Neck or a Valuable Tool for Scholarly Work? by Raphael Germann (ETH Zurich, USI and University of Basel); and Bots in the $SHELL: Modular, interactive workflows with a local LLM and Unix tools by Pierre-Louis Blanchard (Swiss National Museum)

15:48–16:05

Discussion

Workshop Part B – Beyond the specialists: access, mediation and AI literacy

16:05–16:23

2 Inputs: Artistic Intelligence by Sandra Becker (Wikimedia CH); and From Poor Scans to Research Hypotheses: The Digital Kerry Edition as an AI-Assisted Workflow by Romain Büchi (University of Geneva)

16:23–16:40

Discussion

16:40–16:45

Closing

Rita Gautschy (DaSCH)

Please register for the event by filling in this form until October 7th, 2026


 

Comments


bottom of page