Combining P-Plan and the REPRODUCE-ME Ontology to Achieve Semantic Enrichment of Scientific Experiments Using Interactive Notebooks

Abstract

End-to-end reproducibility of scientific experiments requires scientists to share their experimental data along with the computational environment. Interactive notebooks have recently gained widespread popularity among scientists because they allow users to document their experiments along with the code, visualize the results inline and selectively execute the code. In a multi-user environment where users can run and modify the shared notebooks, it becomes essential to capture the provenance of notebooks along with the experiments which used them. In this paper, we propose a way to capture provenance of these interactive notebooks and convert them into semantic descriptions so that a user can query the difference between the results, steps, errors and the execution environment of the code. We use the REPRODUCE-ME ontology extended from PROV-O and P-Plan to describe the provenance of notebook execution. We evaluate our prototype in a multi-user environment provided by JupyterHub.