Ed Pentz, our Executive Director, opened the 2026 Annual Meeting by introducing the theme: Good infrastructure is patient work. Crossref provides foundational infrastructure: an open, persistent scholarly record of metadata, relationships, and identifiers that thousands of tools and services build on. Maintaining that infrastructure and creating lasting value depends on sustained collective effort. True persistence requires active governance, long-term financial commitment, and organisational stewardship.
In 2022, we committed to balancing our working practices and environmental impact, and to openly reporting on those efforts. This is our third report on Crossref’s annual carbon footprint and reflects as much our emissions as it does our learning process. Travel remains the main area of our environmental impact. In 2025, our total emissions were 170 tCO2e, a 3% increase from 2024, with travel accounting for 85% of the total. A big change this year is that we have greater visibility into our AWS-related emissions. With the full transition of our metadata registry and content registration system to the cloud and significantly increased API use, we’re seeing higher carbon costs to host and run our services.
As we approach the end of 2026, research integrity remains a critical focus for the scholarly publishing ecosystem: fast technological advancements are constantly pushing the boundaries of integrity, impacting how everyone in this space interacts with the scholarly record.
Crossref members register around 37,000 records daily. All this metadata is openly shared with the global scholarly community through our free APIs. We receive over two billion calls to our API each month, signalling its use by hundreds of tools and services worldwide, supporting the discovery and assessment of scholarly metadata. As part of a new series in our blog, we will spotlight the diversity of tools and integrations that rely on Crossref metadata to serve the scholarly community. In this initial post, we chat to Luca Foppiano, the developer and maintainer of Grobid1, to learn about how Grobid uses Crossref metadata to improve the extraction and enrichment of bibliographic information from scholarly PDFs.
We maintain an expansive set of relationship types to support the various content items that a research object, like a journal article, might link to. For data and software, we ask you to provide the following information:
identifier of the dataset/software
identifier type: DOI, Accession, PURL, ARK, URI, Other (additional identifier types are also accepted beyond those used for data or software, including ARXIV, ECLI, Handle, ISSN, ISBN, PMID, PMCID, and UUID)
relationship type: isSupplementedBy or references (use the former if it was generated as part of the research results)
description of dataset or software
We and DataCite both use this kind of linking. Data repositories which register their content with DataCite follow the same process and apply the same metadata tags. This means that we achieve direct data interoperability with links in the reverse direction (data and software repositories to journal articles).
The possible relationship types between content items can be as varied as the items themselves. We use a controlled vocabulary to define these relationships, in order to construct an orderly mapped network of content.
This is achieved by (i) an implicit approach where the relation type is a function of a specific service and is declared in the structure of the deposited XML, and (ii) in an explicit approach where the relation type is selected as a value within the deposited metadata.
Reference linking and Cited-by: implicitly creates cites and isCitedBy relationships between a content item and the items in its bibliography
Crossmark: explicit creation of update relations between an item and other items that materially affect it (for example, a retraction)
Funding data: implicit creation of isFundedBy and hasAward relationships between an item and the funding source that supported the underlying research
Linked clinical trials: implicit creation of a belongsTo relationship between and item and a registered clinical trial
Components: implicit creation of a isChildOf relationship between an item and its elemental parts that are assigned their own DOI (limited parent relation typing)
General typed relations: explicitly typed relation between an item with a Crossref DOI and an item with one of several possible identifiers.
Relationship types for associated research objects: intra-work (within a work)
Description
Reciprocal relationship types
Expression
isExpressionOf, hasExpression
Format
isFormatOf, hasFormat
Identical
isIdenticalTo
Manifestation
isManifestationOf, hasManifestation
Manuscript
isManuscriptOf, hasManuscript
Preprint
isPreprintOf, hasPreprint
Replacement
isReplacedBy, Replaces
Translation
isTranslationOf, hasTranslation
Variant
isVariantFormOf, isOriginalFormOf
Version
isVersionOf, hasVersion
Relationship types for associated research objects: inter-work (between works)
Description
Reciprocal relationship types
Basis
isBasedOn, isBasisFor
Comment
isCommentOn, hasComment
Continuation
isContinuedBy, Continues
Derivation
isDerivedFrom, hasDerivation
Documentation
isDocumentedBy, Documents
Funding
finances, isFinancedBy
Part
isPartOf, hasPart
Peer review
isReviewOf, hasReview
References
references, isReferencedBy
Related material, such as a protocol
isRelatedMaterial, hasRelatedMaterial
Reply
isReplyTo, hasReply
Requirement
requires, isRequiredBy
Software compilation
isCompiledBy, compiles
Supplement, such as a dataset generated as part of research results
isSupplementTo, isSupplementedBy
General typed relations
This service allows for the creation of a typed relationship between an item with a Crossref DOI and another content item. The other item may be represented by another Crossref DOI, a DOI from some other Registration Agency, or an item not identified with a DOI. When DOIs are used, the deposit process will fail if the DOI does not exist. Non-DOI identifiers are not verified.
When DOIs are used, a bidirectional relation is automatically created by us when a relation is created in the deposit of one item in a pair. The DOI with metadata creating the relation is said to be the claimant, the other item does not need to have its metadata directly contain the relationship.
Example: translated article
A single journal article is published in two languages with each being assigned its own DOI. In this example, both are published in the same journal. The original language instance has metadata that contains no indication of the translation instance. The alternative language instance includes in its metadata a relation to the original language instance. Here is a screenshot of the relevant section in the code. Please refer to the code snippet below to see it in context.
<journal_article publication_type="full_text">
<titles>
<title>Um artigo na lÃngua original, que passa a ser o inglês</title>
<original_language_title language="en">An article in its original language which happens to be English</original_language_title>
</titles>
<contributors>
<person_name sequence="first" contributor_role="author">
<given_name>Daniel</given_name>
<surname>Stepputtis</surname>
<ORCID authenticated="true">http://orcid.org/0000-0003-4824-1631</ORCID>
</person_name>
</contributors>
<publication_date media_type="online">
<month>02</month>
<day>28</day>
<year>2013</year>
</publication_date>
<program xmlns="http://www.crossref.org/relations.xsd">
<related_item>
<description>Portuguese translation of an article</description>
<intra_work_relation relationship-type="isTranslationOf" identifier-type="doi">10.5555/original_language</intra_work_relation>
</related_item>
</program>
<doi_data>
<doi>10.5555/translation</doi>
<resource>https://www.crossref.org/</resource>
</doi_data>
</journal_article>
Example: book review
This example has a book review published as an article in the journal The Holocene. The article’s title, taken from the publisher’s site is “Book Review: Understanding the Earth system: compartments, processes and interactions” where this book has the DOI https://doi.org/10.1007/978-3-642-56843-5.
A: The current metadata for the review article gives no indication of the actual book being reviewed:
An article with a Crossref DOI identifies that data represented by a DataCite DOI was used in the research and was mentioned in the article’s acknowledgment section.