Direct comparison
Data Paper Vs Software Paper: Key Differences & Comparison | CASRAI
Data papers and software papers are two peer-reviewed publication types that give researchers citable academic credit for non-traditional research outputs — datasets and software tools respectively. Both address the same underlying gap: conventional journal articles do not reward the creation of research infrastructure that others build upon.
Side-by-side comparison
| Dimension | Data paper | Software paper |
|---|---|---|
| Primary output described | A research dataset: observational data, experimental measurements, curated databases, survey data, or processed data products | A research software tool, code library, or computational workflow developed to support or conduct research |
| Key journals | Earth System Science Data (Copernicus), Scientific Data (Nature Portfolio), Data in Brief (Elsevier), GigaScience (Oxford University Press), Geoscience Data Journal | Journal of Open Source Software (JOSS, free, GitHub-based review), SoftwareX (Elsevier), Journal of Open Research Software (JORS), Bioinformatics Applications Notes |
| Peer review of the output itself | Scientific Data and ESSD require reviewers to directly examine the deposited data files; Data in Brief reviews the manuscript only; standards vary by journal | JOSS requires reviewers to examine the software code, documentation, and tests in the GitHub repository; SoftwareX and JORS review the article text and the software; review standards vary |
| Key metadata standard | DataCite Metadata Schema (DOI registration for datasets); domain-specific schemas such as ISO 19115 (geospatial), DDI (social science survey data), or CF Conventions (climate data) | CITATION.cff (YAML, for citation metadata) and/or CodeMeta (JSON-LD, for rich software registry metadata); schema.org SoftwareSourceCode type |
| Persistent identifier | DataCite DOI assigned to the dataset deposit in a trusted repository (Zenodo, Dryad, UK Data Service, PANGAEA, domain repositories) | Zenodo DOI (for the specific version) and/or SWHID (SoftWare Heritage persistent IDentifier for content-based archival reference) |
| FAIR alignment | Data papers directly implement FAIR for datasets: DOI for Findability, open repository deposit for Accessibility, standard formats for Interoperability, methods description for Reusability | Software papers support FAIR4RS (FAIR principles for Research Software): open repository and DOI for Findability and Accessibility; standard formats and documentation for Interoperability and Reusability |
| Citation benefit | Dataset creators receive citations when the dataset is reused; the data paper is indexed in Scopus/Web of Science and cited in the reference lists of papers using the data | Software developers receive citations when the tool is used; JOSS papers are indexed and citable; the preferred-citation in CITATION.cff can direct users to the software paper |
| Funder and REF recognition | UKRI, Wellcome, and ERC all accept dataset outputs for recognition; UKRI REF 2021 and REF 2029 accept data outputs as submittable outputs | UKRI recognises software outputs; JOSS papers are accepted as REF outputs; EPSRC and MRC policies support software as a research output |
Common questions
FAQ
Can the same output have both a data paper and a software paper?+
Yes, where an output has both a significant dataset and a significant software tool associated with it — for example, a bioinformatics pipeline that also produces a curated genomic database — a researcher could publish both a software paper describing the pipeline and a data paper describing the database. They should be cross-referenced. This is uncommon but legitimate.
Is a JOSS paper a full peer-reviewed journal article?+
JOSS papers are peer reviewed, open access, and assigned DOIs, so they meet the formal criteria for a citable journal article. However, JOSS is specifically designed for short papers (typically around 1,000 words) whose primary purpose is to describe the software rather than present research findings. The review focuses on the software itself — its functionality, documentation, and tests in the GitHub repository — rather than on the prose of the article. JOSS papers are indexed by Crossref and are citable in any reference list.
What is the relationship between a data paper and a research article about the same data?+
A data paper and a research article are separate, independently citable publications. The data paper describes the dataset without interpreting what the data means. The research article uses the dataset to test hypotheses or generate findings, typically citing the data paper (and the dataset DOI) rather than reproducing the methodological description. This separation allows the dataset to be reused and cited independently of any particular analysis.
Going deeper
