Ontology-based Modelling of Laboratory Processes

Kurzfassung

Laboratory processes are fundamental across many scientific and engineering domains, where experimental workflows underpin the generation, interpretation, and reuse of data. While digitalisation of modern laboratories has largely focused on data acquisition and storage, a central challenge remains the formal and shared modelling of laboratory processes and the systematic capture of execution records.

This work presents an ontology-based modelling approach that extends the W3C PROV Ontology (PROV-O) to represent laboratory process plans, executions, and associated provenance across different experimental domains. Building on PROV-O, the ontology provides a formally grounded and interoperable vocabulary for describing both intended workflows and realised executions, while introducing laboratory-specific constructs not covered by generic provenance models. Representations created with this approach are machine-interpretable and reusable across disciplines.

Captured process provenance supports key objectives in experimental sciences, including reproducibility, assessment of error impact, identification of outliers, and the generation of predictions and recommendations. Challenges in applying provenance models to existing and future laboratory workflows - scalability, understandability, and interoperability - are addressed through a deliberately simplified ontology design that captures essential process information while limiting complexity. Storing provenance in knowledge graphs instantiated from the ontology further enhances scalability.

The ontology-based modelling approach can be specialised for domain-specific paradigms, such as the processing-structure-properties-performance framework in materials science, while retaining a common PROV-O-aligned core for cross-domain laboratory process modelling. By providing a formal, reusable structure for process plans, executions, and provenance, this approach facilitates integration of laboratory data across domains and contributes to the FAIRification of experimental data.e of process knowledge and improved FAIRness of laboratory data.