Conference workshops 2026

7 October 2026: 9:00 – 13:00 (Workshop 1)

AI Tutor for Language Learning: From Dictionary Entry to Voice Assistant

• Workshop leader: Benedikt Perak

This hands-on workshop, comprising four 45-minute sessions, demonstrates how AI can be used to create, rather than merely consume, language resources. Working in Google Colab, participants will follow a complete pipeline from a raw dictionary entry to a voice-enabled AI tutor.

The workshop consists of five stages. First, participants will transform a raw dictionary entry into a structured JSON record containing definitions, examples, grammatical data, and translations. They will then identify semantic relations and visualise them as a lexical network, add synthesised pronunciation to headwords and example sentences, enrich lexical data through a retrieval-augmented generation (RAG) pipeline grounded in trusted sources, and generate complete language lessons incorporating grammar exercises, CEFR mapping, and spaced-repetition scheduling.

The workshop introduces tools and methods including large language models, structured output, prompt engineering, network graphs, text-to-speech synthesis, embeddings, vector storage, RAG, and agent-based lesson generation. Participants will receive five runnable Google Colab notebooks, JSON templates for structured lexical resources, and a working example of an enriched dictionary with a lexical network, audio, and automatically generated lessons.

No local installation is required. No prior technical knowledge is necessary, although familiarity with Python may be helpful. The exercises use Google Gemini and Google Cloud Text-to-Speech, while OpenAI is supported as an alternative.

7 October 2026: 9:00 – 13:00 (Workshop 2)

Exploring the Tezaurus.hr Platform Through the Multilingual Thesaurus of Performing Arts

• Workshop leaders: Kristijan Crnković, Leticija Crnković, Cécile Chantraine Braillon, Anamarija Žugić Borić

This workshop introduces digital thesauri as tools for structuring, connecting, and sharing knowledge, with a focus on terminology in the performing arts. Organised by the DARIAH-EU Working Group Theatralia and facilitated by ArhivPRO, the developer of the DARIAH-HR platform Tezaurus.hr, it combines introductory presentations, practical work, and discussion.

The workshop draws on the ongoing development of the second version of Theatralia’s Multilingual Thesaurus of Performing Arts within the DARIAH-supported project Performing Arts: Structuring Semantic Knowledge for the Digital Age. It explores the potential of Tezaurus.hr for publishing, consulting, and collaboratively maintaining this multilingual resource, while considering its alignment with open science principles.

Following an introduction to digital thesauri and a presentation of Tezaurus.hr, participants will work with a prepared set of performing arts concepts, manually entering them into the platform to explore its functionalities. The practical activities will address multilingual labels, definitions, synonyms, and semantic relations, including the challenges posed by terms without direct equivalents across languages.

The discussion will consider how the platform meets the needs of a multilingual scholarly community, including collaborative editing, public access, data reuse, and interoperability. Participants will also learn about transferring thesaurus data between platforms, using OpenTheso and Tezaurus.hr as examples. The workshop aims to identify opportunities and limitations, inform the further development of the thesaurus, and outline possible next steps for collaboration.

Participants are encouraged to bring their own laptops to take part in the practical activities. Online participation via Zoom will also be possible.

7 October 2026: 14:00 – 15:30 (Workshop 3)

From a Federated Research Infrastructure to a Critical Research Infrastructure: A “Resilience by Design” Approach for DARIAH-IT

• Workshop leaders: Alessia Spadi, Emiliano Degl’Innocenti, Federica Spinelli

The evolution of DARIAH-IT from a distributed research infrastructure into a federated ecosystem of data centres, computational resources, digital services, scientific workflows, and research communities raises a fundamental technical and scientific question: how can an SSH research infrastructure preserve its essential functions when individual nodes, services, data resources, or dependencies suffer degradation, become unavailable, or are compromised?

Addressing this question requires moving beyond component-level cybersecurity towards a systemic approach in which criticality, dependencies, functional continuity, recovery and adaptation become architectural properties of the infrastructure. Operational resilience must therefore be accompanied by epistemic resilience: the capacity of the infrastructure to preserve the conditions that make digital resources and computational processes trustworthy, interpretable, and scientifically reusable. 

The relationship between cultural memory and machine memory provides a particularly relevant perspective for this discussion. As cultural knowledge is increasingly encoded into structured datasets and processed through automated and AI-supported environments, research infrastructures mediate between human interpretations of cultural memory and their computational representations. Their role therefore extends beyond the preservation and accessibility of digital resources to ensuring that cultural knowledge remains contextualised, interoperable, traceable, and reusable throughout its digital lifecycle.

This workshop presents a resilience-by-design framework for the gradual evolution of DARIAH-IT, the Italian node of DARIAH (Digital Research Infrastructure for the Arts and Humanities), towards a Critical Research Infrastructure. The approach begins with a criticality assessment aimed at identifying which combinations of data, services, workflows, computational resources, identities and organisational dependencies are necessary to maintain essential scientific and cultural functions. Inventories of critical services, dependency maps, risk-impact matrices and domain-specific threat models form the basis for analysing the propagation of failures and identifying individual points of failure within the federation.

Starting at an architectural level, the outlined framework requires investigating how geographically distributed resources can evolve from a set of interconnected nodes into a resilient federated infrastructure. Relevant mechanisms include containerised microservices, multi-cluster orchestration, service meshes, zero-trust API gateways, OAuth2/OIDC and mTLS, secure CI/CD pipelines, immutable backups, distributed replication, and business continuity and disaster recovery procedures. The central problem is therefore one of controlled interdependence: designing a federation in which services and resources can cooperate while preventing individual failures from propagating across the wider ecosystem.

A complementary dimension concerns the resilience of scientific workflows. Workflows create dependencies that span services, datasets, computational environments, authentication mechanisms and external resources. Their resilience therefore requires reproducible execution environments, explicit dependency management, provenance tracking, continuous monitoring and mechanisms to detect and respond to anomalous behaviour. The evolution of the DARIAH-IT orchestration environment towards native containerised and distributed execution provides an experimental context for the study of adaptive workflows, anomaly detection, agent-based orchestration, federated learning and semi-automatic recovery mechanisms.

For SSH infrastructures, however, operational continuity alone is not sufficient. A technically successful failover or recovery may still constitute a scientific failure if provenance, authenticity, semantic consistency, reproducibility or reliability are compromised. Resilience must therefore be assessed simultaneously at the infrastructural and epistemic levels. This requires measurable indicators that combine service availability and recovery times with workflow reproducibility, provenance traceability, metadata quality, interoperability and the integrity of digital cultural heritage. The resulting model links four complementary dimensions: accessibility and interoperability; quality, trust and credibility; security and resilience; transparency and explainability. 

By presenting the approach adopted by DARIAH-IT to address these challenges, we aim to provide a test bed for investigating a broader technical and scientific question: how distributed SSH research infrastructures can evolve from federated service architectures into resilient critical ecosystems, capable not only of providing access to computational resources and cultural data, but also of ensuring the integrity, reliability, and continuity of the scientific and cultural functions that increasingly depend on them.

7 October 2026: 16:00 – 18:00 (Workshop 4)

Documenting (DARIAH) events on Semantic Web with Commons and Wikidata.org

Workshop leader: Željko Blaće

Session type: Demonstration + Hands-on Workshop / Tutorial

Academic conferences produce rich scholarly output – speaker profiles, presentation slides, session recordings, and event media – yet this data is frequently monolingual, hard to find or even lost when conference websites go offline and/or pdfs go corrupt. Wikimedia infrastructure offers a free, sustainable, and linked open data ecosystem to preserve, document, and amplify scholarly events, their contributors and topics they address in multiple free media formats, multilingual descriptions and most importantly search engine/machine friendly, language agnostic linked open data of Wikidata.org.

This short, practical demonstration and workshop introduces organizers, research managers, and student/volunteers to Wikidata and Wikimedia Commons as open publishing and archiving services. Participants will explore practical workflows for structuring conference metadata (organizers, speakers, talks, venues) on Wikidata and depositing open-access event media and/or slides onto Wikimedia Commons (including practices of when not to do it). We will demo Data Reuse with basic querying (SPARQL) to generate automated conference lists and integrate event data back into web platforms.

What You Will Learn

Structured Event Metadata: How to create and link Wikidata items for conferences, sessions, speakers, and presentations. Open Media Archiving: Best practices for uploading images, presentation slides, and media to Wikimedia Commons using proper open licensing (CC BY / CC BY-SA). Volunteer Workflows: Practical guidelines for setting up lightweight “event documentation teams” or edit-a-thons alongside conference proceedings.

Target Audience

Academic conference organizers, Digital Humanities project coordinators, research communications staff, open science advocates, and event volunteers. No prior experience with Wikidata or MediaWiki is required.

Prerequisites & Requirements

Participants should ideally bring their own laptop or at least a smartphone with installed Wikipedia app. Prior creation of a free Wikimedia account in days before the session using your own smartphone is strongly recommended.