Object reuse and exchange for publishing and sharing workflows

Abstract

The workflow paradigm can provide the means to describe the complete functional pipeline for a scientific experiment and therefore expose the underlying scientific processes for enabling the reproducibility of results. However, current means for exposing such information are tied closely to the individual workflow engines and there is no existing method that provides a common way to share this information. In this paper, we discuss a lightweight approach that can be used to expose such information, using the Open Archives Initiative Object Reuse and Exchange (ORE) standard, to provide a common format for representing and sharing workflows and their associated metadata required for their execution. We describe how workflows can be mapped to the ORE format using RDF and how they can be stored using bundles for sharing with others. We discuss tooling we have developed that provides a mechanism for existing workflow engines to conveniently export workflows as ORE bundles. We present three use cases for Triana, ASKALON and MOTEUR, where such integration has already been undertaken, and conclude the paper by providing a short study showing that the overhead implications of adopting the proposed ORE bundling format are minimal.