When you click a DOI link and arrive reliably at an article years after it was published, when a reference list resolves into live links, when a funder can see which papers acknowledged its support, or when a reader is warned that a paper has been corrected or retracted, a single piece of infrastructure is usually working quietly in the background. That infrastructure is Crossref: the not-for-profit DOI registration agency through which a large share of the world’s scholarly literature is registered and described. Crossref is rarely visible to researchers, yet it underpins a great deal of how scholarship is found, cited and connected. This article explains what Crossref does and why it matters, drawing on the research information systems domain of the CASRAI Dictionary.
What Crossref actually is
Crossref is a membership organisation, founded by publishers, that operates as a registration agency for Digital Object Identifiers (DOIs). When a publisher registers an article, book chapter, conference paper or other scholarly work with Crossref, two things happen: the work receives a persistent DOI that will resolve to its current location indefinitely, and the publisher deposits structured metadata describing the work — its title, authors, publication venue, dates, references, and increasingly its funding, licences and identifiers for the people and organisations involved. The DOI is the visible part; the metadata is the substance. Crossref’s real product is not the identifier alone but the enormous, openly queryable corpus of interlinked metadata that the identifiers make navigable.
DOIs and persistent linking
The foundational service is persistent identification. A DOI is a stable handle that points to a work regardless of where the publisher later moves the content; update the target once in the registry and every citation of that DOI continues to resolve. This solves the link-rot problem that plagues ordinary web addresses, where URLs decay as sites are reorganised. Because the DOI is persistent and resolvable, it can serve as a reliable anchor for everything else: a reference can cite it, a dataset record can link to it, an author’s ORCID profile can list it, and all of those references remain valid over time. Persistent identification is the precondition for every other connection Crossref enables.
The metadata services that build on DOIs
On top of the DOI, Crossref runs a family of services that turn isolated records into a connected graph:
- Reference linking and Cited-by. When publishers deposit the reference lists of their articles, Crossref matches those references to registered DOIs. This makes reference lists resolvable, and it powers Cited-by, which lets a work surface the later works that cite it — citation relationships derived directly from deposited metadata.
- The Open Funder Registry. A controlled list of funding bodies that lets publishers record, in a standardised form, who funded the research behind a paper. Standardised funder names are what make it possible to ask reliably which outputs a given funder supported, rather than guessing from inconsistent free-text acknowledgements.
- Crossmark. A service that gives readers a way to check whether they are looking at the current version of a work and to see its update status — corrections, retractions, expressions of concern. Crossmark surfaces the scholarly record’s changes rather than letting a superseded version circulate unflagged.
- Grant linking. Funders can register their grants with Crossref and receive DOIs for them, and output metadata can reference those grants — connecting published works back to the awards that supported them in a machine-readable way.
Why open metadata matters
What gives Crossref its outsized importance is that its metadata is openly available through a public interface that anyone can query. This openness means the metadata is not locked inside one company’s product but is a shared resource that discovery services, repositories, research-information systems, analytics tools and researchers themselves can all build upon. A reference manager that resolves citations, an institutional system that ingests a researcher’s publications, a discovery layer that links related works — all of these draw on the same open Crossref corpus. The value compounds: each well-formed deposit enriches a commons that the whole community uses, which is why the quality of the metadata publishers deposit matters as much as the existence of the DOI.
Crossref alongside other identifier systems
Crossref does not work alone. It interoperates with the other persistent identifiers that describe the research world: ORCID for people, ROR for organisations, and DataCite for datasets and other research outputs. An article’s Crossref record can carry the ORCID iDs of its authors and the ROR identifiers of their institutions, while a DataCite record for a dataset can link to the Crossref-registered article that analyses it. Crossref and DataCite together cover much of the scholarly output graph — broadly, the literature and the data — and their alignment is part of what lets the whole network be traversed. Crossref’s place in this federated arrangement is described in our work on Crossref and federation.
Crediting the work behind the record
Rich Crossref metadata can carry more than bibliographic facts; it can carry information about contribution. As publishers increasingly deposit structured contributorship alongside author lists, the CRediT taxonomy — whose full set of contribution types is described in our overview of the CRediT roles — can travel with the article record itself, so that who-did-what becomes part of the persistent, openly queryable metadata rather than prose buried in a paper. For any of this to interoperate, a metadata element must mean the same thing wherever it appears. That consistency is what the CASRAI Dictionary exists to provide: a stable vocabulary so that the funding, contribution and relationship information flowing through Crossref is understood identically across the systems that consume it. Crossref supplies the connective metadata; a shared dictionary keeps it legible.