CROSS-CUTTING

DOI

Persistent identifier for digital objects, defined by ISO 26324 and administered by the International DOI Foundation. The de facto standard in academic communication for stable citation of articles, datasets, chapters, and other research outputs.

Extended definition

A DOI (Digital Object Identifier) is a persistent, unique, machine-readable identifier assigned to digital objects — scientific articles, book chapters, datasets, preprints, software, technical reports, monographs. Defined by the ISO 26324 international standard and administered by the International DOI Foundation, the system guarantees that a registered object retains a stable reference even if hosts, URLs, or publishers change over time. Each DOI is structurally composed of a prefix (assigned to the registrant — typically a publisher, repository, or agency) and a suffix (chosen by the registrant to identify the specific object), separated by a slash. Resolution occurs via https://doi.org/{prefix}/{suffix}, redirecting the browser to the current URL of the object. Major registration agencies include Crossref (scholarly articles), DataCite (datasets), mEDRA (Italian and European publications), Airiti (East Asia), and the Multilingual European DOI Registration Agency.

When it applies

DOI is appropriate and increasingly required whenever a scientific object needs stable citation. For articles in indexed journals, DOI assignment is practically automatic via Crossref. For datasets, the hosting repository (Zenodo, Figshare, Dryad, OSF) assigns a DOI at deposit time. For preprints, arXiv and bioRxiv assign their own DOIs. Researchers should include DOI in all bibliographic references where available, and should register DOIs for any research output they intend to make citable (including software, protocols, and supplementary materials).

When it does not apply

DOI is not necessary for ephemeral content (social media posts, general web pages) or for informal communication between researchers. It does not replace ORCID — DOI identifies objects, ORCID identifies people. It does not confer any quality guarantee: predatory journals also assign DOIs, and the presence of an identifier does not imply serious peer review. To cite official institutional web pages without DOI, standard editorial practice uses URL with access date, not fabricated alternative identifiers.

Applications by field

Manuscript submission: all citable references should include DOI when available, an increasing requirement at indexed journals. — Data repositories: Zenodo, Figshare, OSF, Dryad assign DOI at deposit; linkage to depositor’s ORCID is standard practice. — Academic software: GitHub repositories can be deposited in Zenodo to receive a citable DOI for each release version. — Preprints and gray literature: arXiv, bioRxiv, SSRN, EngrXiv assign their own DOI, separate from the DOI of the final published version.

Common pitfalls

The first pitfall is confusing DOI with URL — the DOI is the permanent identifier; the URL may change, but the DOI continues to resolve to the correct location. Academic citation should use DOI, not publisher-specific URL. The second is assuming that DOI presence implies editorial quality — predatory journals with DOIs operate within the system without any serious review guarantee. The third is failing to assign DOI to one’s own outputs when possible: a dataset without DOI is practically invisible in modern bibliometrics. The fourth is using https://dx.doi.org/... (old format) when the canonical form is https://doi.org/... since 2015. The fifth is confusing the preprint DOI with the published version DOI — they are distinct entities, and citations should favor the final version when available.

Last updated —