More than ninety percent of published Jupyter notebooks do not state dependencies on external packages. This makes them non-executable and thus hinders reproducibility of scientific results. We present SnifferDog, an approach that 1) collects the APIs of Python packages and versions, creating a database of APIs; 2) analyzes notebooks to determine candidates for required packages and versions; and 3) checks which packages are required to make the notebook executable (and ideally, reproduce its stored results). In its evaluation, we show that SnifferDog precisely restores execution environments for the largest majority of notebooks, making them immediately executable for end users.
@article{arxiv.2103.02959,
title = {Restoring Execution Environments of Jupyter Notebooks},
author = {Jiawei Wang and Li Li and Andreas Zeller},
journal= {arXiv preprint arXiv:2103.02959},
year = {2021}
}
Comments
to be published in the 43rd ACM/IEEE International Conference on Software Engineering (ICSE 2021)