An integrated data pipeline for semantic data representation of the flame spray pyrolysis process.

Ongoing digitalization and data-driven developments in materials science and engineering (MSE) emphasize the growing importance of reusing research data and enabling machine accessibility, which requires robust data management and consistent semantic data representation. Ontologies have emerged as powerful tools for establishing interoperable and reusable data structures from inconsistent data structures. Despite advancements in semantic data representation for specific applications, integrating application ontologies with primary data repositories, such as electronic lab notebooks (ELNs), to feed world data remains an open task. As a use case in the MSE domain, this work presents a system based on semantic technologies from the point of view of engineers, developed with the help of information scientists, and unraveled on a small scale. The development of an application ontology (AO) was elaborated for flame spray pyrolysis (FSP) processes with the implementation of a data pipeline. The proposed FSP application ontology emerges from experimental in-house best-practice procedures and is adapted to the mid-level Project Material Digital core ontology (PMDco) to allow interoperability within the MSE domain. The pipeline retrieves manually acquired experimental data from an ELN, translates it into a machine-actionable format, and converts it into a Resource Description Framework (RDF) format to support semantic interoperability. The latter was stored in a triple store with a SPARQL interface, enabling findable and accessible datasets that are searchable and traceable. By creating semantically linked data structures in line with FAIR principles, this approach allows traceable and findable experimental results between stakeholders through both human-readable and machine-actionable formats. Seamless integration of the modular microservices of the data pipeline within established lab practices minimizes disruption while maintaining the software framework. The present work demonstrates the practical implementation of a FAIR data pipeline within a laboratory setting, paving the way for future data-centric science.