Reproducibility remains a major challenge in computational research, particularly for workflow-based analyses executed across heterogeneous data, software, and computing environments. While workflow management systems such as REANA support reproducible execution through containerisation and workflow orchestration, they provide limited support for workflow registration, service-oriented reuse, and structured provenance management. This paper presents ReProv, a provenance capture and management layer that extends REANA with Workflows-as-a-Service (WaaS) capabilities and automated provenance capture based on the W3C PROV-O standard. ReProv enables researchers to register, discover, execute, and reuse Common Workflow Language (CWL) workflows while automatically recording the relationships between workflows, executions, data artefacts, and responsible users. Provenance information is stored in a structured form and can be explored through APIs, a Python SDK, and graph-based visualisations. The system has been integrated with the European AI-on-Demand (AIoD) Platform, enabling workflows and datasets to be linked with a broader ecosystem of AI resources. Two use cases demonstrate the approach: a machine-learning workflow based on the MNIST benchmark and a larger-scale climate-data workflow for next-day Mediterranean heatwave prediction. Results show that ReProv captures detailed execution lineage, including failed workflow executions, while introducing only negligible runtime overhead. By combining workflow lifecycle management, provenance-aware execution, and service-oriented workflow reuse, ReProv provides a practical framework for transparent, reproducible, and FAIR-aligned computational research.
Troumpoukis et al. (Wed,) studied this question.