Development and application framework for distribution of reference datasets

Content and exemplary usage of reference data of materials.

Link to DemonstratorLink to Publication

In IUC02, we propose and implement a framework for generating, distributing, and utilizing (reference) datasets in the materials science domain. The framework’s individual components ensured data functionality and usability, thereby aligning with the FAIR principles. For developing the framework we use creep data of Ni-based superalloys as an example, however this will be further developed and enhanced to better support non-reference research data, adding substantial value to the MSE community.

Our framework considers the definition of criteria for a dataset to qualify as a reference dataset and includes comprehensive material descriptions (including data schema, ontology, knowledge graph). New data schemas will be created for additional testing methods (e.g., hardness, fatigue) based on new or existing data, and existing schemas will be expanded to cover a broader range of processing conditions and materials (e.g., ceramics, polymers, additive manufacturing). Wherever possible, these schemas will be aligned with relevant ontologies. Large Language Models (LLMs) will support literature mining and data/schema annotation. Furthermore, we aim to test the framework’s generalizability across different methods and formats. Additionally, interfaces for the generation, operation and registration of FAIR Digital Objects (FAIR DOs) will be developed to further enhance materials data infrastructure.