Tübingen RDM Days 2025

Talk

Advancing Science: Research Data Management and AI in the Digital Age

_{Research Data Management (RDM) is a key foundation of responsible science in the digital age. This talk explores how RDM contributes to the quality, transparency, and reusability of research – and how it intersects with the use of Artificial Intelligence (AI). It opens the “Research Data meets AI” event by providing context, institutional guidelines, and perspectives on good scientific practice.}

Dr. Holger Gauza (ZDV),

Jennifer Esslinger (Dez. II)

Talk

The EU AI Act and Research Data Management: Challenges and Implications for University Research
_{This talk highlights the main implications of the EU AI Act for university research and research data management. Key definitions, roles and exemptions of the regulation in the university context will be analysed. The aim is to provide a basic understanding of the new legal requirements and their implications for scientific practice.}

1h Lunch Break

Talk

Artificial Intelligence meets Research Data Management: LLMs in the NFDI

_{The talk shows how AI methods such as Large Language Models (LLMs) within the German National Research Data Infrastructure (NFDI) contribute to improving data quality, metadata availability as well as accessibility and how data from the NFDI is made available to improve these processes. Using the Text+ consortium as an example, data protection-compliant AI applications and specific application scenarios will be presented, which open up new possibilities for the reuse of research data.}

Short Talk
Pilot Project of the Specialized Information Services Theology: AI-assisted Research Data Analysis with Jupyter Notebook

_{The Specialized Information Services Theology pilot project addresses the growing use of AI in theological research. We are planning to provide executable Jupyter Notebook or Google Colab to overcome accessibility hurdles in computational methods. This fosters replicability and reproducibility, offering a low-threshold environment for theologians to easily apply AI tools to their research data.}

Short Talk

In-Class Test: AI-Based Content Analysis of the IxTheo Bibliography

_{For decades the IxTheo Classification has proven to be an established categorization system for the contents of the IxTheo bibliography. However the IxTheo has grown to a size where intellectual assignment is no longer feasible. The talk will present the current state of automatically classifying records using LLMs and an outlook on other planned steps to enhance the accessibility of the data of the Tübingen Specialised Information Services (FID) bibliographies by using AI-driven solutions.}

10min Break

Talk

FAIR Research Data – Challenges and Perspectives

_{Ten years after the introduction of the FAIR principles, the question remains: how FAIR is science today? This talk explores current challenges, existing standards, and future perspectives for researchers and institutions – with a focus on incentives, practical implementation, and the added value for the University of Tübingen.}

Talk

Central AI Services for Research Data Management at the University of Tuebingen

Talk

AI in the DFG's Funding Activities

_{Artificial intelligence (AI) is used for various purposes in research practice and is therefore also playing an increasingly important role in the field of RDM. With the "Guidelines for dealing with Generative Models for Text and Image Creation" and the development of initial funding opportunities, the DFG aims to address the great potential of AI in research and provide support for its use.}

Talk

PeriMyo – Prediction of Perioperative Myocardial Injury using Machine Learning

_{Mortality and morbidity after non-cardiac surgery is a major health problem, with perioperative myocardial damage playing a crucial role. Machine learning (ML) methods offer the potential to systematically identify relevant risk factors and enable individualized risk predictions for patients. In a retrospective analysis of 8347 patients, ML models with a particular focus on interpretability were developed and used to determine potential risk factors for perioperative complications.}

Talk

“And it felt really good!”: Generative AI and the Transformation of Everyday University Life of Students and Staff

_{The talk provides insights into the everyday use of generative AI by students and staff at the University of Tübingen. Based on a survey and qualitative interviews, it highlights practices, interpretations, and tensions that arise in the use of generative AI in the context of academic daily life.}

Talk

Machine Learning for Earth System Science
_{Many relevant processes in the dynamics of the Earth system are either not understood sufficiently well, or a computationally to costly to be simulated in terms of explicit physical models. Machine Learning approaches have recently been proposed to address this issue. In my talk, I will give an overview over some recent applications of Machine Learning, including state-of-the-art generative models, in different fields of Earth system science, including spatial field reconstruction, weather prediction, and climate projections.}

Talk (in English)

FAIR and verifiable: The role of FAIR Digital Objects in trusted AI
_{The many digital objects that are created or used by researchers throughout the Research Lifecycle are largely not FAIR. The resulting lack of provenance leads to fragmentation and loss of the scientific record, inefficient research programming and opaque academic publishing. In the Leiden academic community, we are experimenting with the use FAIR Digital Objects as FAIR and verified “metadata tags” to track the digital assets composing the Research Lifecycle (data, software and methods among them). Erik will report findings from these pilots and prospects for broader application in Leiden and beyond. Research data and metadata that adhere to the FAIR Principles enhance the ability of machines to Find, Access, Interoperate with and Reuse data making AI more trustworthy, equitable, and fit to purpose. This opens the door to virtuous feedback loops where FAIR data engenders trusty AI that is increasingly entrusted with the creation of FAIR data.}

10min Break

Panel Discussion

Using AI in Research Data Management – Opportunities and Challenges

Moderation

Dr. Rebecca Hahn (Public Relations Department)