| 13h30 | Welcome and check-in |
| 14h00 | Workshops (1st session) |
| 15h30 | Break |
| 16h00 | Workshops (2nd session) |
| 17h30 | Break and Flying Dinner start |
| 18h30 | Opening night Welcome |
| 18h35 | these burning bridges aren’t as bright as I had hoped Keynote Talk Fernando Obieta |
| 19h00 | Curating Access to a Globalized World Panel Discussion Sophia Prinz & Helmut Grabner |
| 19h30 | Presentation Wikimedia CH GLAM Program |
| 19h45 | Born-Digital Cultural Heritage – How the Demoscene (re)imagines traditional art Shana Marinitsch |
| 20h30 | Exchanges, Networking, Flying Dinner |
| 22h00 | Closing |
Workshops Descriptions
Apertus and PublicAI
Oleg Lavrovsky, Technical Community Manager, EPFL AI Center
Our workshop is an invitation to digital humanities researchers and members of the OpenGLAM community about leveraging Apertus and related open source LLM technologies for cultural data analysis, curation, and storytelling, while fostering a community around open, collaborative, and responsible AI use in cultural heritage projects.
Understand how to design LLM workflows that augment human research and curation processes (e.g., keyword extraction, metadata enrichment, summarization, sentiment analysis). We will explore a small demo that uses Apertus to assist in generating contextually relevant summaries for historical texts. This could be used to design digital exhibitions based on archival content.
Participants will be given free access to a Swiss inference provider, where they can proceed to set up their hackathon challenge. We will also explain how to run large language models locally (this is possible today even without a high-powered computer) for maximum privacy and flexibility.
In the last part of the workshop, we will discuss how to fine-tune models for specific domains (e.g., historical texts, local dialects) using open datasets. Understanding how to evaluate LLM outputs for performance, topical sensitivity, historical accuracy, and potential biases should be helpful for you regardless of which LLM you use.
Bio: the speaker, Oleg (“O”) Lavrovsky, is well known in the GLAM community as long-time champion of the Open Knowledge Network in Switzerland. O has worked on data engineering projects with non-profits, government, universities and startups, and now works at the EPFL / ETH AI Centers as a member of the Apertus team.
Datasheets for Cultural Heritage Datasets
Christiane Sibille, Digital Scholarship Services, ETH Library
A central element of GLAMHacks is the use of cultural heritage collections as data. In recent years, there has been a debate in the wider cultural heritage community about how such data sets should be described to ensure responsible computational re-use. Building on broader considerations regarding Collections as Data (Padilla et al. 2023) and Datasheets for Datasets (Gebru et al. 2021), cultural heritage experts recently developed the idea of “Datasheets for Cultural Heritage Collections” (Alkemade et al. 2023) and developed “A Checklist to Publish Collections as Data in GLAM Institutions» (Candela et al. 2023).
During the workshop, we will first contextualise these approaches and work on practical examples.
Coding with AI for Non-Developers
Birk Weiberg, Managing Director, Digitalization Initiative of the Zurich Higher Education Institutions (DIZH)
One no longer has to be a developer to start turning ideas into working code with AI – this workshop is about understanding what that actually involves. It is aimed at people who work with or are interested in cultural heritage data and are curious about what AI-assisted coding can do for them – no coding experience required.
The session covers different approaches and tools for AI-assisted coding – what they can do, what their possible shortcomings are, and how they can still be useful. Through a live demo and a hands-on exercise with open cultural data, you’ll experience how describing an idea in plain language can result in working code.
The goal is not to turn you into a developer. It’s to give you enough understanding to build rough prototypes, recognize their limits, and find a common language with developers you’ll meet at the hackathon. Think of it as a translation tool for interdisciplinary collaboration.
Uploading Image Collections to Wikimedia Commons with OpenRefine
Joris Burla, Head Collection Documentation, Museum Rietberg
Museums and Collections that want to share their images on Wikimedia Commons often face two challenges: First, how to upload a lot of images, and secondly how to upload them with the metadata, that they already have in their databases. In this workshop we will learn how to use OpenRefine to tackle the two challenges in one go. We will look at the whole process from start to finish, learn about the quirks of Wikidata and the potential considerations before uploading. It would be good for participants to have some previous knowledge of OpenRefine, as the course will not be a beginners course on OpenRefine, but everyone is welcome. Due to the lack of time, we can’t work with everyone’s data. We will look at the process with some exemplary data.
Keynote Talk
these burning bridges aren’t as bright as I had hoped
Fernando Obieta, Conceptual media artist and artistic researcher, https://fernando-obieta.com
As it is unlikely that the analogue world will be flooded with generative AI in the foreseeable future, we should use this time to explore the craftsmanship and artistic abilities of our species and beyond. Our institutions hold a wealth of situated knowledge(s) and perspectives that can help us to foster and preserve critical thinking. But how can we foster appreciation and accuracy in a rapidly changing (digital) landscape? From a feminist theoretical perspective, we will examine the responsibilities and moral challenges that lie ahead as we endeavour to build new bridges.
Fernando Obieta is a media theorist, artist and SNF Doc.CH PhD candidate, with a background in computer science, interaction design, and transdisciplinary studies in the arts. His artistic work and research focus on the critique and reflection on (digital) media, artefacts and systems, and how these are contingent and situated.
Panel Discussion
Curating Access to a Globalized World
Sophia Prinz, Professor for Design Theory and History, Zurich University of the Arts
Helmut Grabner, Professor of Data Analytics and Machine Learning, ZHAW Zurich University of Applied Sciences
Moderation : Sonja Gasser and Birk Weiberg
Digital archives (in the best case) present alternative ways to navigate the world and to structure what it contains. They create curiosity in their users and gain relevance not just through interesting interfaces but also through their bond to the analog world. The panel gives insights into two such projects that take different approaches to the data they work with and the role they foresee for their users.
Sophia Prinz (ZHdK) presents the project “Müliweg Eins”, which was developed with the PHZH and SKKG as an educational tool for high school students. The project starts with a small selection of digitized objects from the SKKG collection and challenges students to conduct research on and curate them. Helmut Grabner (ZHAW) has been working as a computer vision researcher with the artist Kurt Caviezel on a website that provides users with access to more than 11,000 webcams primarily through text-based interfaces. And while “Watching the World” presents itself like an archive, its database consists of an ever-changing set of current images without memory.
At the same time, both projects address the complexities of a globalized society and try to empower non-experts to navigate this terrain. Crucially, both approaches ask what becomes visible — and what remains hidden — when we design the conditions of seeing. And both are only possible because of the accessibility of their data and the digital tools they use.
Born-Digital Cultural Heritage – How the Demoscene (re)imagines traditional art
As the first of its kind, the Demoscene has been recognised as UNESCO Intangible Cultural Heritage by 9 European nations, Switzerland being one of them. The application for World Cultural Heritage is now underway.
More than 2.000 of these small, realtime-rendered programs are being created every year by the community – we are presenting a selection of those which fittingly Bridge The Gap between realtime animation and traditional cultural heritage.
Shana Marinitsch is a digital artist, cyber feminist and community organiser living in Switzerland. She is member of the board at Echtzeit – Digitale Kultur and the Chaos Computer Club Zurich, part of the Haecksen collective and main organiser of the digital culture festivals MountainBytes and Revision Demoparty.