With the tremendous increase in the amount of data in materials science, new ways to store and annotate data are necessary to ensure fulfilling the FAIR principles – and to do efficient, good, and new science. Consequently, ontologies have been of increased interest as they do not only allow storing and annotating but also semantically linking data even across domains. This way data are represented in a machine-readable fashion which opens up new application possibilities, for example in interdisciplinary research, to increase reusability of data, or asking complex questions requiring knowledge from different domains. The European Materials and Modeling Ontology, EMMO (http://emmo.info/), is an attempt to provide a representational framework for the physical sciences. However, appropriate ready-to-use domain ontologies are so far completely lacking in the field of materials science. There are several large databases for computational material data each adopting their own meta data schemes for data annotation. The largest is the NOMAD Repository that has most other relevant databases in the field included. Furthermore, the NOMAD Archive provides a normalized form of these data independent of their source using the NOMAD Metainfo [1] as metadata schema. The NOMAD Metainfo includes a number of relations between concepts and therefore already goes beyond the simple metadata concept. We have converted it to the ontology format OWL and demonstrate on this poster how this enables connecting multiple sources of knowledge. Within the NOMAD ecosystem, we have created a NOMAD Structure Ontology (NSO) in order to represent materials, in particular crystaline solids, as well as a NOMAD Properties Ontology (NPO) that semantically describes concepts used by materials scientists. One example is the electronic band structure and its relations to various physical, chemical, and elastic properties. We demonstrate a first application of this NOMAD ontology triad (Metainfo Ontology, NSO and NPO) and showcase interoperability with external ontologies. As an outlook we discuss ideas how to connect with experimental data through ontologies.

[1] L. M. Ghiringhelli et al., npj Comput. Mater. 3, 46 (2017).