At Curvenote, we prioritize the long-term preservation of scientific content, particularly as technologies and external services evolve. Computational articles, which often rely on live code execution, data access, and external services, can present challenges over time as these dependencies change. To ensure your content remains accessible and degrades gracefully, we implement a variety of strategies that preserve the original source documents, figures, code, and outputs, while offering multiple formats and fallbacks to enhance longevity.
The Challenges of Computational Articles¶
Computational articles often depend on:
- Data Access: APIs or data sources that might become unavailable or outdated.
- External Services: Services integrated with your computational work that can be deprecated or altered.
- Containerized Code: Even though tools like Docker containerize environments for reproducibility, changes to the underlying libraries, data, or services can still cause code to stop functioning over time.
Our approach ensures that even when these dependencies break or are no longer available, your article remains intact, readable, and useful to others.
Preserving Source Documents¶
At Curvenote, we store and publish the source documents used to create your content. By keeping these source materials, we ensure that your work remains editable, accessible, and reproducible long into the future. The source documents can currently be stored in the following formats:
- LaTeX: The go-to typesetting system for academic and scientific documents, especially in fields requiring complex equations or formatting.
- MyST Markdown: A modern, flexible markdown standard that integrates seamlessly with computational elements and documents.
- Jupyter Notebooks: Interactive notebooks that combine code, outputs, and explanations in a single, reproducible format.
By preserving these source formats, we allow for the future adaptation and re-execution of your work, ensuring that it remains usable even when the initial computational environment is no longer available.
Preserving Source Code and Outputs¶
For computational work, we store source code and ensure executed outputs are retained within Jupyter Notebooks. This allows for transparency, reproducibility, and a clear view of the original research methods. We preserve these notebooks in an executed form, meaning that:
- Code and Raw Outputs: The outputs generated by the code are saved alongside the source code, ensuring that even if the environment can’t be re-executed, the results remain accessible.
- Code Transparency: The original code is always available for future researchers to inspect, adapt, or reproduce.
We encourage authors to publish their code in a standard way (e.g. using PyPI) with clear versioning.
Handling Figures and Visuals¶
We take additional steps to preserve the integrity of figures and visuals:
- Snapshot Creation: For every figure in the main manuscript, we generate snapshots or custom placeholders (e.g., images or videos) in standard formats. This ensures that, even if the interactive or live computation components fail, static representations of the visuals remain accessible.
- Alternatives Block in JATS: When publishing in JATS (Journal Article Tag Suite) format, we utilize the “alternatives” block to store figures in both interactive and static formats. This ensures readers always have access to critical visuals, regardless of the state of external computations.
Equations and Mathematical Notation¶
Equations are a critical part of many scientific publications, and we ensure they are preserved in multiple formats:
- LaTeX Math: We store mathematical equations in their LaTeX form, widely used for typesetting complex mathematics in academic settings.
- MathML: In addition, we output MathML alongside LaTeX to ensure compatibility with modern web browsers and assistive technologies.
In the future, we will enhance our equation handling by automatically generating images of all equations. This will provide an additional layer of preservation, ensuring that mathematical notation remains readable even if LaTeX or MathML rendering fails.
JATS for Archiving and Publication¶
At Curvenote, we utilize the JATS (Journal Article Tag Suite) format for structured, long-term storage of your manuscripts. JATS is a widely accepted standard used by repositories like PubMed Central (PMC), ensuring that your work remains compatible with trusted archival systems. This structured format makes it easier to share and repurpose your work across platforms and institutions.
Large Data Storage in External Repositories¶
For large datasets or data directly useful for others beyond visualization or analysis in the article, we recommend using dedicated data repositories for long-term storage. This ensures both accessibility and discoverability of your datasets for future research.
We suggest trusted repositories like:
- Zenodo: An open-access repository for research outputs, including datasets, software, and publications.
- Source Cooperative: A platform focused on supporting open science projects, with an emphasis on transparency and accessibility for collaborative research.
If your data holds potential for reuse by the scientific community, depositing it in a data repository ensures its longevity and availability for future exploration.
Content as MyST Markdown and JSON¶
In addition to preserving the source documents and code, we store your content in JSON formats, following the MyST Spec (https://
Archiving with PDF¶
Finally, each publication includes a traditional PDF version of the manuscript. PDFs offer a long-term archival solution, widely accepted in academic institutions and repositories. The PDF format serves as a reliable, non-interactive snapshot of the manuscript, ensuring that your research remains accessible through traditional means, even when computational or interactive elements are no longer functional.
By preserving the source documents, offering alternatives for figures and equations, and supporting multiple archival formats like JSON, JATS and PDF, Curvenote ensures that your research degrades gracefully. Even as technologies an services evolve or fail, your content remains accessible, transparent, and available for future reuse, ensuring its impact for years to come.
