Daejeon, July 27–31
Knowledge graphs have become one of the central paradigms in data provision and analysis because they can represent research artifacts – such as literature, objects, persons, places, events – and their relationships in a formally structured, interlinked and machine-understandable way [1, 2]. In cultural heritage, they support data integration across heterogeneous collections, enable cross-domain discovery, and make it possible to pose complex questions that would be difficult to answer within isolated databases. This complexity creates a recurring problem: the mechanisms that make knowledge graphs machine-actionable often reduce accessibility for humans. Access commonly requires familiarity with formal query languages (notably SPARQL), ontologies, identifiers, and semantics. For many humanities scholars and cultural heritage practitioners, this produces a barrier to engagement: they may be very competent in domain knowledge but lack the time or incentive to acquire specialised technical expertise. As a consequence, knowledge graphs risk becoming “infrastructures without audiences”, where advanced capabilities exist but remain underused in everyday research and teaching [3].
NFDI4Culture is the German National Research Data Infrastructure consortium for tangible and intangible cultural heritage. It brings together a large network of partners from disciplines including architecture, art history, musicology, performing arts, and media studies, with the goal of making cultural heritage research data systematically manageable, sustainably reusable, and interoperable. NFDI4Culture follows a federated approach: data remains with providers, while shared services enable harmonisation, enrichment, discovery, and reuse [4]. A central service is the Culture Knowledge Graph [5, 6, 7], a semantic index over decentralised resources that interlinks metadata across partners using the NFDIcore ontology [8] and the NFDI4Culture ontology [9]. The graph currently comprises more than 100 million RDF triples [10] and is designed to enable cross-domain discovery, federated queries, and transparent semantic integration.
Within NFDI4Culture, we developed the Culture Data Story Laboratory [11] to mediate between infrastructure complexity and human engagement. The format draws inspiration from data journalism platforms such as The Upshot by The New York Times [12], where visualised data supports narrative exploration. Comparable tools exist in the Digital Humanities, including the CLARIAH Data Stories Editor [13] and the Carnegie Hall Data Lab Experiments [14].
A data story weaves a written narrative with interactive elements and live data queries. Instead of confronting users with a technical approach, data stories begin with a thematic or research-driven entry point and progressively reveal how relevant entities are represented and connected in the graph. They combine explanation (why this matters), inspection (what the data shows), and method (how the claim can be reproduced). Data stories therefore operate at the intersection of scholarship, teaching, and science communication [15].
In the context of the DH2026 conference theme “Engagement”, we present data stories as an engagement method: they connect scientific communities to research infrastructures by translating semantic complexity into guided pathways through data [16].
We suggest this workshop as an interactive, experimental laboratory: a structured environment for trial, revision, comparison, and collaborative critique. Participants will treat queries as testable hypotheses, visualisations as interpretative instruments, and narrative framing as a scholarly choice that can be revised in response to what the data reveals. In the workshop, we encourage participation, shared problem-solving, and reflection on how emerging technologies (including generative AI) can be used responsibly and transparently.
Our distinctive methodological contribution is the implementation of the Data Stories Laboratory as an easy-to-use, interactive environment based on the Shmarql Linked Data Publishing Platform [17]: Shmarql enables knowledge-graph-based data stories to be written in plain Markdown with embedded, directly executable SPARQL queries whose results are rendered as tables and visualisations. This makes data stories “executable narratives”: readers can follow not only the argument, but also the computational steps that produce the evidence. The workflow is comparable to a computational notebook (such as Jupyter) in that narrative, code, and output are co-located; however, it is natively grounded in Linked Open Data.
1. Introduction: Engagement, Infrastructure, and Data Stories
Introduction to NFDI4Culture, the Culture Knowledge Graph, and the rationale behind data stories as a mode of scholarly communication and engagement.Introduction to NFDI4Culture, the Culture Knowledge Graph, and the rationale behind data stories as a mode of scholarly communication and engagement.
2. From Knowledge Graphs to Executable Narratives
Conceptual foundations of shmarql-based data stories; Markdown, SPARQL, and live visualisations; comparison with computational notebooks and basic data journalism practices.Conceptual foundations of shmarql-based data stories; Markdown, SPARQL, and live visualisations; comparison with computational notebooks and basic data journalism practices.
3. Exemplary Data Stories in Practice
Guided walkthrough through example data stories such as the Italian Data Journey [18] and Experiments in Data Alchemy [19]; discussion of narrative structure, query design, and interpretative framing.Guided walkthrough through example data stories such as the Italian Data Journey [18] and Experiments in Data Alchemy [19]; discussion of narrative structure, query design, and interpretative framing.
4. Hands-On Session: Data Story Laboratory
Participants work in small groups to modify existing data stories or draft short narrative sections with embedded SPARQL queries using prepared templates.Participants work in small groups to modify existing data stories or draft short narrative sections with embedded SPARQL queries using prepared templates.
5. Reflection, Discussion and Wrap Up
Joint reflection on data stories as scholarly artefacts; discussion of sustainability, reuse, and the role of AI-assisted narration.Joint reflection on data stories as scholarly artefacts; discussion of sustainability, reuse, and the role of AI-assisted narration.
After completing the workshop, participants will be able to:
Our workshop follows an interactive didactic model that allows for participation at the conference venue as well as online (hybrid setup). Past experience has shown that the workshop works best with 20 to 40 participants. Short conceptual inputs alternate with guided exploration and hands-on exercises. By working directly with executable narratives rather than abstract examples, participants experience first-hand how engagement emerges at the intersection of data, narrative, and interpretation. The workshop does not assume prior expertise in SPARQL or ontology engineering. Instead, it positions data stories as an engaging scholarly format that enables meaningful engagement with complex infrastructures while preserving analytical depth and transparency.