Source-linked AI summary
Using the Jupyter Notebook as a Tool for Open Science: An Empirical Study
Bernadette M. Boscoe, Irene V. Pasquetto, Milena S. Golshan, Christine L. Borgman
TL;DR
Complex computational research can be difficult to interpret without access to the data and procedures behind it. The authors examine notebook mentions and links in astronomy and physics publications and find both open notebooks with detailed research materials and mentions without access.
Problem
For complex computational research, data alone is rarely sufficient to interpret findings; the software, workflows, and documentation behind the work also matter.
Source: §I · p. 1Method
The authors identified and coded 91 notebook-mentioning publications in the Astrophysics Data System for links to notebooks and related research objects.
Source: §II · p. 1Results
37 papers linked to openly accessible notebooks with detailed research materials, while 54 mentioned notebooks without providing access.
Source: §III · p. 2Takeaways & Limitations
Jupyter notebooks can connect code, data, workflows, and documentation as part of open science and scholarly communication.
Source: §I · p. 1Motivations for using specific tools such as Jupyter notebooks to make science more open remain little known.
Source: §IV · p. 2
Abstract
from arXiv · showhide
As scientific work becomes more computational and data intensive, research processes and results become more difficult to interpret and reproduce. In this poster, we show how the Jupyter notebook, a tool originally designed as a free version of Mathematica notebooks, has evolved to become a robust tool for scientists to share code, associated computation, and documentation.
I. INTRODUCTION AND BACKGROUND
Jupyter notebooks bring code, data, workflows, and documentation together as a machine- and human-readable means of scholarly communication. The poster examines their relationship to FAIR principles against the need for access to more than data alone.
I. INTRODUCTION AND BACKGROUND
- The authors examine Jupyter notebooks as a way to support open science and scholarly communication and assess their utility in relation to FAIR principles.Notebooks can connect to research objects such as datasets, code, methods documents, workflows, and publications.Source: §I · p. 1
- The preliminary work investigated how scientists had begun citing Jupyter notebooks in published papers in astronomy and related fields.The authors identify astronomy as a field that has made progress toward open science and note requirements for data release by funders and journals.Source: §I · p. 1
- Data alone is rarely sufficient to interpret results from complex computational processes; software, workflows, instrumentation details, and documentation also matter.The poster describes these resources together as the “recipe” for how research was accomplished.Source: §I · p. 1
- The poster uses computational reproducibility to mean availability of the data and code used in computation, and replicability to mean obtaining the same results with independently collected data.The definitions are attributed to Peng.Source: §I · p. 1
II. METHODS
The study identified Jupyter notebook mentions in astronomy and physics publications, then coded their links to notebooks and related research objects for contributions to open science.
II. METHODS
- The authors identified 91 relevant publications in the Astrophysics Data System and coded mentions, citations, and explicit links to notebooks and research objects.Coded research objects included datasets, code, and publications.Source: §II · p. 1
III. RESULTS
The study found varied ways of mentioning and providing access to Jupyter notebooks, including open notebooks with detailed research procedures and a case illustrating links among research objects.
III. RESULTS
- 37 papers linked to openly accessible notebooks with research procedures, code, analytical methods, and results, while 54 mentioned notebooks without providing access.The authors identified eight ways of mentioning notebooks and five ways of providing access, only some of which appeared to facilitate open science.Source: §III · p. 2
- Authors used notebooks for tasks ranging from data construction to analysis and manuscript preparation, with most papers mentioning them in multiple ways.The most popular access links were to NBviewer or GitHub repositories; 23 papers referenced Perez and Granger, and 21 referenced the main Jupyter website.Source: §III · p. 2
- For the case scenario, the authors traced mentions and hyperlinks to associated notebooks to examine their connections to research objects, focusing on discoverability.They used this tracing to identify how the paper’s authors used notebooks to make their science more open.Source: §III · p. 2
- In a case study of an astronomy article on binary-star masses, notebooks linked data, code, and documented refinement and analysis procedures across research objects.The authors concluded that the case applied FAIR principles by providing open access to the study’s research objects.Source: §III · p. 2
IV. CONCLUSIONS
The poster identifies unresolved questions about notebook contents and stability, alongside limited knowledge about why researchers use specific tools to make science more open.
IV. CONCLUSIONS
- The authors leave open what information an ideal Jupyter notebook should contain to support FAIR principles effectively.They also ask how software and other tools should be stabilized or preserved when access is provided through published papers.Source: §IV · p. 2
- The authors identify persistence mechanisms such as Object Reuse and Exchange or ResourceSync as a future question for supporting notebook stability.Source: §IV · p. 2
- Motivations for using specific tools such as Jupyter notebooks to make science more open remain little known, while sustaining access to research objects and their relationships remains a continuing challenge.The authors suggest that studying components of open science may help address these larger questions.Source: §IV · p. 2