The Research Nexus is Crossref’s framework for connecting the scholarly record through declared relationships in metadata, rather than treating each registered work — a journal article, a preprint, a dataset, a grant, a protocol — as an isolated record with its own DOI. Crossref describes it as an effort to build “a rich and reusable open network of relationships connecting research organisations, people, things, and actions.” The underlying mechanism is not new (Crossref’s relationships schema has existed for years); what the Research Nexus adds is a unifying name, a stated priority, and ongoing investment in making those connections more complete, more consistently used, and more discoverable across the whole graph of registered content.
What problem the Research Nexus is meant to solve
A single research output rarely exists in isolation. A published article may have been preceded by a preprint, funded by one or more grants, built on a dataset and a registered protocol, later corrected, translated, or reviewed in a separate peer-review report. Each of those items can independently have a DOI or other persistent identifier, but without explicit, machine-readable links between them, that context is invisible to anyone consuming the metadata — a funder checking research outputs, an institution compiling a researcher’s record, or a reader trying to find the data behind a figure.
Crossref’s position is that metadata quality, not identifier assignment alone, is what makes this connective tissue usable at scale. The Research Nexus is the organizing concept under which Crossref frames relationship metadata, funding metadata, and related infrastructure as pieces of one connected system rather than separate features.
How relationships are declared in Crossref metadata
Technically, a relationship is declared when a Crossref member deposits metadata for a work that names another item and states how the two are related. Crossref’s schema distinguishes two kinds of relationship:
- Intra-work relationships — different versions, formats, or manifestations of what is essentially the same work. Types include
isVersionOf/hasVersion,isPreprintOf/hasPreprint,isManuscriptOf/hasManuscript,isTranslationOf/hasTranslation,isFormatOf/hasFormat, andisReplacedBy/Replaces. - Inter-work relationships — links between genuinely distinct research objects. Types include
isSupplementTo/isSupplementedBy(an article and its supplementary dataset),references/isReferencedBy,isPartOf/hasPart,isReviewOf/hasReview(peer-review reports),isCommentOn/hasComment,isDerivedFrom/hasDerivation,requires/isRequiredBy, andfinances/isFinancedByfor funding links.
Both are declared in a member’s deposit XML using a <related_item> element (in the relations.xsd namespace) that names the relationship type and the identifier type of the related item. Crossref does not require the related item to itself be identified by a Crossref DOI — the schema accepts a range of identifier types, including a DOI from another registration agency, an ISSN, ISBN, URI, PMID, PMCID, arXiv ID, ARK, handle, or accession number. This matters specifically for connecting to research protocols registered on platforms like protocols.io, or to datasets registered with DataCite, both of which sit outside Crossref’s own registration but can still be linked from a Crossref record. There is no additional fee for members to deposit relationship metadata — it is part of standard content registration.
Grants and funding as part of the graph
Funding information is one of the Research Nexus’s more actively developed strands. Crossref members can declare funding relationships for a registered work — identifying the funding body (via a funder identifier drawn from Crossref’s Funder Registry, which has been consolidating with the Research Organization Registry, ROR) and, where available, an award or grant identifier. Crossref has also worked on making its own grant records — where a funder registers a DOI for a specific grant or award — linkable to the outputs that grant produced, which is what lets a funder or institution trace a straight line from an award through to the articles, datasets, and preprints it funded. This is distinct from simply naming a funder in free text: a declared relationship is a structured, queryable link.
Relationship to Crossref Event Data and other Crossref infrastructure
Crossref Event Data and Research Nexus relationships metadata address related but distinct problems. Event Data captures third-party signals about a DOI — a tweet, a Wikipedia citation, a Newsfeed mention — happening outside Crossref’s own registered metadata, gathered from external sources after the fact. Research Nexus relationships, by contrast, are first-party metadata that a member deliberately declares at deposit (or update) time about how their own registered content connects to other scholarly objects, including ones with no Crossref DOI at all. Put another way: Event Data is about detecting mentions; Research Nexus relationships are about members asserting structured provenance. Crossref frames Event Data, funding metadata, ORCID integration, and CrossMark update tracking as complementary pieces of the same broader Research Nexus vision, rather than as unrelated services.
Who benefits, and how
- Researchers get a more complete, machine-traceable record of a body of work — a reader following a published article can find the preprint it evolved from, the dataset and protocol behind it, and any subsequent corrections or reviews, without relying on the article text itself to mention them.
- Funders can trace outputs back to specific grants and awards using structured funding relationships instead of manually matching acknowledgment text, which supports research-output reporting and impact assessment.
- Institutions and research offices compiling a researcher’s or department’s full output — articles, preprints, datasets, protocols — benefit from being able to pull a connected set of records via API rather than reconciling separate, unlinked identifiers by hand.
- Publishers and repositories gain a mechanism to make provenance (a version history, a correction, a data or protocol dependency) explicit and durable in the permanent record, rather than relying on a landing page that may change or disappear.
The practical value of all of this depends on member participation: relationships have to actually be deposited to exist. Coverage is uneven across publishers and content types, which is a known limitation of the current state of the Research Nexus rather than a hypothetical one — a work with no declared relationships is not necessarily unconnected to other research outputs, only undocumented as such in Crossref’s metadata.
How this differs from simple citation linking
Reference lists and citation counts (what most researchers think of when they hear “linking” between articles) are one narrow relationship type — references/isReferencedBy — among the many the Research Nexus schema supports. Citation linking answers “what did this work cite, and what has cited it” — and matching a raw, unstructured reference-list citation to the DOI it actually refers to (the practical first step before a references relationship can even be declared) is what CASRAI’s Crossref Simple Text Query entry covers. Research Nexus relationships additionally answer questions citation linking cannot: which preprint preceded this article, which grant funded it, which dataset and protocol it depends on, and which peer-review report accompanies it. Institutions relying only on citation counts for research assessment are working with a narrower slice of the same underlying metadata infrastructure.
Frequently asked questions
Is the Research Nexus a separate product from Crossref’s core DOI registration service?
No. It is not a separate paid service or a distinct system members sign up for — it is Crossref’s framing for how existing and evolving metadata (relationships, funding data, ORCID links, update records) fit together as one connected graph. Declaring relationships uses the same deposit XML mechanism members already use to register a DOI.
Do I need a Crossref DOI on both items to declare a relationship?
No. The related item can be identified by a non-Crossref DOI or by another identifier type entirely (ISSN, ISBN, URI, PMID, arXiv ID, handle, accession number, among others), which is what allows a Crossref-registered article to declare a relationship to, for example, a protocol or dataset registered elsewhere.
Does declaring relationship metadata cost anything extra?
No. Relationship metadata deposit is part of standard Crossref content registration, with no additional fee.
How is this different from Crossref Event Data?
Event Data collects external, third-party signals about a DOI after the fact (mentions in social media, Wikipedia, news). Research Nexus relationships are first-party metadata a member deliberately declares about their own content’s connections to other research objects at deposit or update time.
For the underlying deposit mechanics — what metadata to submit and when — see Crossref Metadata Deposit Workflow. For how Crossref’s DOI registration compares to the other major registration agency, see Crossref vs. DataCite.
