ProvONE+: A Provenance Model for Scientific Workflows

Abstract

The provenance of workflows is essential, both for the data they derive and for their specification, to allow for the reproducibility, sharing and reuse of information in the scientific community. Although the formal modelling of scientific workflow provenance was of interest and studied, in many fields like semantic web, yet no provenance model has existed, we are aware of, to model control-flow driven scientific workflows. The provenance models proposed by the semantic web community for data-driven scientific workflows may capture the provenance of control-flow driven workflows execution traces (i.e., retrospective provenance) but underspecify the workflow structure (i.e., workflow provenance). An underspecified or incomplete structure of a workflow results in the misinterpretation of a scientific experiment and precludes conformance checking of the workflow, thereby restricting the gains of provenance. To overcome the limitation, we present a formal, lightweight and general-purpose specification model for the control-flows involved scientific workflows. The proposed model can be combined with the existing provenance models and easy to extend to specify the common control-flow patterns. In this article, we inspire the need for control-flow driven scientific workflow provenance model, provide an overview of its key classes and properties, and briefly discuss its integration with the ProvONE provenance model as well as its compatibility to PROV-DM. We will also focus on the sample modelling using the proposed model and present a comprehensive implementation scenario from the agricultural domain for validating the model.