FAU prepares research data for AI and future innovations

Photon and neutron research provides important insights for the development of new technologies from materials to pharmaceuticals to energy technologies, but equally critical as the results is the handling of the research data generated in the process. Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU) is participating in the DAPHNE4NFDI consortium project, which is now entering its second funding phase. What the project entails is explained by Prof. Dr. Tobias Unruh from the Chair of Crystallography and Structural Physics at FAU in our interview.

Photon and neutron research is a foreign concept for many people. Where do we encounter the results of this research in everyday life and why is this research important?

The results of our research are reflected in diverse new materials for 3D printing, solar cells, batteries, sensors and detectors, fusion reactors, pharmaceuticals, aerospace, and many other applications. Only when the atomic and microscopic structures of these materials and how they change over time are precisely understood can these materials be developed and optimized for their specific applications. However, this goes far beyond materials development and enables the investigation of fundamental relationships in countless subfields of modern natural sciences, forming an important pillar of basic research.

The DAPHNE4NFDI project does not deal with new experiments but with research data. Why are well-organized data a crucial building block for scientific breakthroughs today?

Research data are the foundation of all scientific work. The volume of data we generate with our methods is growing rapidly, but data is also becoming more complex. This increases the importance of automated data evaluation on the one hand, and on the other, we can significantly improve the efficiency of our investigationsif the measured data are so well documented and stored that they can later be reused in other contexts and do not need to be regenerated . Moreover, our data only become meaningfully usable for a much larger number of researchers and in particular for artificial intelligence (AI) applications through standardized data formats, documentation, processing, and provision.

The project aims to make research data “FAIR – that is, findable, accessible, interoperable, and reusable. Can you explain with a concrete example how this changes the work of researchers?

If all the data from the members of my team are FAIR in the best sense, then I can not only trace the results of their work at my workplace all the way back to the raw data, but also examine, compare, and evaluate these results in all their details with the results of international research teams that also provide FAIR data. For researchers, all this means at first is that the already highly advanced digitalization of the collection and processing of research data will become even more comprehensive and standardized. It is important to create an efficient work environment for this, and here too, AI applications can provide helpful support.

Artificial intelligence plays an important role in the second funding phase. What can AI achieve with well-prepared research data that was previously difficult or even impossible?

AI systems have the potential to generate complex evaluations of measurement data and visual representations of the results in real time so that measurement conditions and protocols can be optimized even during the measurement process. This not only makes measurements much more efficient but also includes rapid automated feedback in the measurement processes. Even the recognition of patterns and relationships, or simply an efficient, targeted search and selection of relevant data from among irrelevant datasets in vast amounts of data is barely conceivable without AI applications.

If DAPHNE4NFDI has achieved its goals in five years: How would researchers – and perhaps even all of us – notice that the project was a success?

The goals of DAPHNE4NFDI are essentially milestones, as the handling of research data is subject to continuous development, just like the research for which these data are collected. An important milestone for us is achieved, for example, when a doctoral researcher or even an AI application can efficiently and comprehensively use the full scope of research data from a previous doctoral researcher. The next milestones are further use across institute, national, and disciplinary boundaries, as well as the integration of research data that complement photon and neutron data.

Contact