To accelerate collection, accumulation, distribution, and utilization of materials data, the National Institute for Materials Science (NIMS, Japan) has been developing and operating DICE, a materials data platform and a collection of databases/data-oriented systems. One of the DICE services is the Materials Data Repository (MDR), released in 2020. The public front-end of MDR is based on Hyrax by Samvera. Works deposited to MDR are assigned DataCite DOIs, allowing integration with external services such as Data Citation Index.

Significant effort was devoted to implement a metadata schema with a focus on PIDs and support for specimen, methods, and instrument metadata. The first schema was intended to be the common versatile format among multiple DICE services. Despite some of its success in inter-system data transfer, its complexity induced negative user feedback. It was superseded in favor of a leaner YAML-based schema, developed alongside a new system to natively support the new schema. Works deposited to this system are automatically transferred to the public-facing Hyrax. The new UI works in conjunction with NIMS’s research registration system and makes use of CrossRef API, minimizing metadata input from researchers. This led to a marked increase in the number of works deposited, highlighting the importance of usability in software solutions.

Using works deposited to MDR as instances, we can define an ontology for data linking and federation. An x-ray absorption fine structure (XAFS) spectrum in MDR XAFS DB is represented in RDF, wherein the specimen is defined using entities in our vocabulary system, MatVoc. This graph can be linked with the graph defining 'x-ray absorption edge'. The link can be swapped with 'core level' for hard x-ray photoelectron spectroscopy (HAXPES) while retaining the schema. This interoperability allows us to easily obtain the HAXPES spectrum for a sample in MDR XAFS DB, demonstrating the ability for cross-database queries.