In recent decades, a large amount of materials science data has been generated. Most of these data have never been made publicly available, either in repositories or otherwise, so the quality of the data or information is uncertain. Hence, the challenge for the digital transformation of material science is nowadays not only to answer urgent questions using digitalization methods, but also to investigate all existing and methodically relevant data from decentralized data sources and evaluate them with regard to their usefulness for these new methods. In order to make the knowledge of data and their quality fully accessible for digitalization, curation concepts and solutions are being developed and implemented in a demonstrator as part of the STREAM project (Semantic Representation, Networking and Curation of Quality Assured Material Data, funded by the Federal Ministry of Education and Research). These curation solutions enable the validation of existing data with respect to conformity to FAIR principles and support a subsequent upgrading or extending to fulfill the FAIR requirements (FAIR-ification).
A curation concept is presented that makes existing cross-domain data from different repositories accessible to users. The chosen approach to realize this is ontology-based. A major focus of this work is on the cross-domain uses of ontologies such as the EMMO and MatVoc for materials science.
As shown by a prototype implementation, the functionality and benefits of the concepts for cross-domain exchange of material data based on semantic technologies will be exemplarily presented. Hereby it is achieved that data from different data silos are centrally accessible for all users. The material repositories NOMAD and a repository with data of the macroscopic mechanical characterization of materials of the Fraunhofer-IWM serve here as an appropriate data source.
Finally, an example is used to show how the newly developed ontology-based curation approaches can be used in the future to improve the digitization of material development on the basis of existing and subsequently FAIR-ified data in a computer-aided manner.