From Notes to Nodes – Develop an AI-driven exploration tool for musicological data in the Culture Knowledge Graph
Data Challenge 2026
We are thrilled to announce the third data challenge hosted by the Data Competence Center HERMES. This time we are collaborating with NFDI4Culture – the consortium for research data on material and immaterial cultural heritage within the German National Research Data Infrastructure (NFDI).
The Challenge
Knowledge graphs are increasingly used to store, interlink, and contextualize complex research data. By integrating heterogeneous distributed data sources and describing them semantically, they enable structured, interoperable access to cultural heritage information. This makes them particularly valuable for research in the humanities – including musicology.
The Culture Knowledge Graph (CKG) is a large and continuously growing example of such an infrastructure. It aggregates data from multiple independent projects and institutions, connecting musical works, sources, persons, performances, places as well as historical contexts.
Our question:
How can musicologists easily interact with a knowledge graph and retrieve relevant information without being SPARQL experts?
Your Objective:
Develop an AI-driven tool, allowing researchers to explore musicological knowledge in the Culture Knowledge Graph and beyond conversationally.
The Dataset
This challenge is based on the CKG developed within NFDI4Culture. The full graph currently contains approximately 106m RDF triples. You can retrieve data via a public SPARQL Endpoint, but you will also get a full RDF dump (N-Triples). The NFDI4Culture Ontology (CTO) and the NFDIcore Ontology are the semantic backbone of the CKG.
In the domain of musicology, the CKG has so far integrated extensive research data from the following sources:
- RISM (~28m triples)
- musiconn.performance (~2m triples)
- APSearch (~9k triples)
- Detmolder Hoftheater (~ 1,7k triples)
Although the data sources are independent, the knowledge graph could reveal rich connections between research objects – such as shared persons, works, places, institutions, incipits – that could open up new horizons in musicological research.
Beyond NFDI4Culture, there exists an abundance of further RDF-based data that potentially could be connected in a joint musicological discovery and exploration system, for example:
Your Mission
Your task is to design and implement an AI-based discovery and exploration tool that uses the Culture Knowledge Graph as its core data source and supports musicologists in their research.
You are free to choose your technical approach. You may also integrate external data sources, such as the list mentioned above, and enrich the graph-based answers using federated queries, especially where shared authority data is available. Take inspiration from existing projects such as SPINACH, but above all, put your creative ideas into practice.
Your solution should:
- Answer musicological research questions (e.g. about composers and their works, verified sources of works in RISM, repertoire studies at various performance venues, references to opera libretti in CORAGO, or historical contexts)
- Provide a user-friendly interface
- Rely exclusively on non-paid / openly available models and tools
- Be scientifically transparent, e.g. by exposing the generated SPARQL queries or reasoning steps
- Use the musicological data from the Culture Knowledge Graph as base data
Who can register?
Participation in the challenge is open to anyone with an interest in the topic and the objectives of the challenge. You may take part individually or collaborate with others as a team, depending on your preferred way of working. Participants who would like to join a team but have not yet found collaborators are encouraged to get in touch with us ahead of time or during the kick-off event. We will support networking among participants and, where possible, facilitate the formation of teams during the preparation phase and at the kick-off.
The Challenge Timeline
- Kick-off (18.03.2026) - This online event is a chance to get to know each other, ask questions, and explore initial ideas and approaches. If you’re looking for teammates or collaborators, you’ll also have the opportunity to form teams.
- Register Your Team (until 31.03.2026) - Use the registration button provided below to sign up.
- Develop Your Solution - You’ll have 3 months to tackle the challenge. During this time, we’ll host optional check-in meetings for teams to share progress and ask questions.
- An in-person check-in meeting will take place on 4–5 May as part of the event “Innovation mit ‚KI‘? Hack Day für maschinelles Lernen in den Humanities” at the Academy of Sciences and Literature Mainz. It offers you the opportunity to come together in one place, work collaboratively on your projects, and exchange ideas related to the challenge with others. Participation is flexible and may also include joining other workshops or hackathon activities taking place during the event, depending on your interests. Food and drinks will be provided.
- A remote check-in meeting at the beginning of June.
- Submit Your Work (until 30.06.2026) - Upload your packaged code (developed in the programming language of your choice) to a Git repository and send the link, together with the names of your team and team members, to hermes.challenges@uni-marburg.de.
- Review - A panel of experts in information science & musicology will evaluate your submission based on user experience, originality, maturity and interoperability.
- Final Presentation & Recognition (9. - 11.11.2026) - Winning teams will be awarded at the 6th Culture Community Plenary in Bochum (travel expenses covered for winning teams) and can present their solution at the plenary to the 4Culture Community. Additionally, all successful submissions will get a blogpost and presentation on the Culture Information Portal.
Evaluation Criteria
Your submissions will be evaluated across five complementary categories in order to recognize different strengths of the proposed solutions. User Experience assesses how intuitively and effectively users can interact with the tool and explore the data. Originality focuses on the novelty of the idea and the creativity of the chosen approach. Maturity reflects how well developed, coherent, and complete the overall solution is. Interoperability considers how effectively the solution makes use of standards, linked data principles, and connections to external datasets or services. Musicological Adequacy evaluates the extent to which the solution addresses authentic research questions in musicology and appropriately reflects domain-specific concepts and distinctions.
Key dates
- Kick-off: 18 March 2026, 4:00 PM — Zoom link
- Registration deadline: 31 March 2026, 11:59 PM
- In-person Check-in meeting: 4/5 May 2026, Mainz
- Remote Check-in meeting: Beginning of June 2026
- Submission of solutions: 30 June 2026, 11:59 PM
- Results Announced: September 2026
- Final Presentation & Award Ceremony at the NFDI4Culture Plenary: 9-11 November 2026, Bochum
