Projects

This project is dedicated to two central questions. The first is: How can information about the production of thin-film materials from a wide variety of sources be brought together and organized in computers in such a way that this knowledge can be reused and expanded by future knowledge? The solution is to use ontologies. The great advantage of this technology is that it enables the reuse of existing data sets in new projects, thereby saving costs. Another advantage is that it enables information from different data sets to be linked, thereby creating synergies.

The second question is: Can the data sets linked by the ontology be used to predict how changes in the manufacturing processes (e.g. a lower substrate temperature) affect the properties of the resulting thin film? To answer this question, we use artificial intelligence methods that combine artificial neural networks with a logical representation. If successful, this technology will accelerate the development of new materials.
This text was translated with DeepL on 28/11/2025

View project in the research portal

In the Urban-Rural-Energy project, we are developing open and transferable methods and tools that make it possible to calculate and suitably prepare robust, regionally interlinked and sector-coupled energy transition scenarios for the urban-rural nexus. Our aim is to promote intercom- munal cooperation and accelerate the local energy transition. Researchers benefit from the innovative methodology for robustness analysis in energy system models, the improvement of model solution times and the further development of efficient and open data management. The sub-project 'Data model, ontology and workflows for transferability' focuses on qualitative methods that enable and improve the organization and transferability of the data and processes used in the Urban-Rural-Energy project. We will link terms from different areas important for urban-rural energy to the Open Energy Ontology (OEO), namely from the data model, the areas of robustness, uncertainty and urban-rural nexus, as well as from the energy system models. In this way, we can make the terms used more comprehensible (especially for stakeholders), make the data and models easier to find, better structure the analysis of uncertainties and improve the transferability between models. Another focus of the OVGU concerns the preparation of input data for the new model calculations planned in Stadt-Land-Energie. The effort involved in processing heterogeneous input data is often underestimated. We are therefore using a graph-based workflow tool to create an automatic processing pipeline that converts different scenario data into the developed format and makes it available on the Open Energy Platform (OEP) for easy use.
This text was translated with DeepL

View project in the research portal

With the introduction of the CHEBI ontology and the associated web lexicon, a structure has been created that can be used to illustrate the logical relationships between different chemical substances and their functional properties. The classification of chemicals can be based on a wide variety of characteristics and is a highly manual and time-consuming process. In the course of this research work, possibilities to automate the classification of chemicals are being investigated. For this purpose, not only the latest findings and models from deep learning and especially neuro-symbolic integration are used, but also the rich logical annotations of the CHEBI ontology.
This text was translated with DeepL

View project in the research portal

Automatic reasoning systems have undergone rapid development in recent years. The integration of machine learning techniques has made it possible to develop effective heuristics for reasoning. Nevertheless, large logical theories, as found in many ontologies, often lead to problems. Therefore, in this research we explore possible machine learning approaches that allow to automatically select those axioms from a large theory that are needed to fulfill a given proof goal.
This text was translated with DeepL on 28/11/2025

View project in the research portal