DOI (Digital Object Identifier)

A Digital Object Identifier (DOI) is a unique, persistent alphanumeric string assigned to a digital object, most commonly a journal article, but also a dataset, report, conference paper, book chapter, or other scholarly output. The DOI system, managed by the DOI Foundation (originally named the International DOI Foundation) and implemented by registration agencies such as Crossref and DataCite, provides a permanent link to the object's location on the internet, independent of changes to the publisher's website or URL structure.^[2] A DOI has two parts separated by a forward slash: a prefix, made of "10." and a registrant code, and a suffix chosen by the registrant (for example, in 10.1038/s41586-020-2012-7 the prefix is "10.1038" and the suffix is "s41586-020-2012-7").^[2]

Persistent identification is essential for the integrity of the scholarly record. Before DOIs, online citations relied on URLs that frequently broke as publishers reorganized their websites, changed platforms, or went out of business. A broken link makes a cited source inaccessible, undermining the ability to verify claims and reproduce findings. DOIs solve this problem by providing an identifier that remains valid regardless of where the object is hosted. When a publisher moves content to a new server, they update the DOI resolution record, and the DOI continues to point to the correct location. Citation indexes, reference managers, and systematic review tools rely on DOIs to link references to their sources. For datasets, DOIs enable formal citation, giving data creators credit for their contributions and making data discoverable and citable as first class research outputs.

The DOI system operates through a network of registration agencies. Crossref is the primary agency for journal articles, books, and conference proceedings. DataCite handles datasets, software, and other research outputs. When a publisher or data repository assigns a DOI to an object, they register it with the appropriate agency, providing metadata (title, authors, publication date, type) and the object's current URL. The DOI resolves to the object through a redirect service: entering https://doi.org/ followed by the DOI string in a browser redirects the user to the object's current location.^[2]

Publishers pay a fee to register DOIs, and they are responsible for maintaining the accuracy of the resolution URL. If a publisher moves their content, they must update the DOI resolution record to point to the new location.^[2] This obligation is why DOIs are more reliable than plain URLs: the DOI system creates an infrastructure of accountability that plain URLs lack.

DOIs can be registered for virtually any digital object: journal articles, preprints, datasets, software, protocols, conference presentations, reports, and even individual figures or supplementary files. Some objects receive DOIs at multiple levels of granularity; for example, a dataset may have one DOI for the overall collection and separate DOIs for individual data files within it.

In scholarly communication, DOIs serve multiple functions. They provide persistent links for citations, they enable automated metadata exchange between publishers and indexing services, they support reference linking (connecting a citation in one article to the cited article), and they facilitate metrics tracking (counting citations, downloads, and other measures of impact). Reference managers such as Zotero and Mendeley can automatically retrieve metadata from a DOI, simplifying the process of building reference lists.

DOIs should be included in scholarly citations wherever one is available and the citation style allows it. APA 7, for example, asks for the DOI of every source that has one, formatted as a URL (https://doi.org/10.xxxx/xxxx), while AMA style prints it as doi:10.xxxx/xxxx.^[1][3] When depositing data in a repository, requesting a DOI for the dataset makes it possible to cite and track it. When publishing articles in a style that asks for DOIs, authors should verify that all cited sources with DOIs include them in the reference list. In systematic reviews, DOIs facilitate deduplication (identifying the same article found in multiple databases) and full text retrieval.

A systematic review team uses DOIs to deduplicate search results from three databases. PubMed, Embase, and Web of Science return a combined 4,200 records, of which 3,100 have DOIs. Automated deduplication based on DOI matching removes 1,450 duplicates in seconds, a process that would take hours if done manually by comparing titles and authors. The 1,100 records without DOIs are then matched on title, author, and publication year, against each other and against the records already kept, which removes a further 250 duplicates. The final deduplicated set of 2,500 unique records proceeds to screening.