02 What we publish
Eight rooms and an annex.
Every kind of research object the registration agencies serve has a room here. For each one: what it includes, what we check before it is published, which agency registers its identifier, and what its record carries.
2Choose a room
Books and chapters
Monographs, edited volumes, textbooks and reference works — and every chapter in them, as a citable object of its own.
Registered with Crossref book, chapter and reference-work records
What we check
- An editor reads the proposal; then the full manuscript is peer reviewed by at least two independent experts in its field.
- Originality: a similarity check against published work, read by a person, not just scored by software.
- Permission for every third-party figure, map, photograph or long quotation.
- Each chapter has its own authors, abstract and references, so it can stand on its own.
- Figures carry text descriptions, so the book can be read with a screen reader.
books & chapters · what it carries
- title
- title, subtitle, edition and series
- contributors
- every author, editor and translator, with an ORCID iD and a CRediT role
- affiliations
- organisations as ROR IDs
- abstract
- for the book and for each chapter
- references
- the full list, with DOIs where they exist
- funding
- funder ID and award number
- licence
- a licence URL that people and machines can read
- isbn + dates
- for each format, print and online
- languages
- title and abstract in a second language when you provide one
- updates
- any correction or new edition linked to the original
Conference proceedings, posters and slides
Proceedings volumes and the papers in them, and the posters and slide decks presented at the event.
Registered with Crossref conference-proceedings records for the volume and each paper
Registered with DataCite Poster and Presentation records (types added in DataCite Metadata Schema 4.7)
What we check
- That the conference ran as stated: organisers, dates, venue and programme committee.
- The review record for every paper — and an independent review where the committee’s is missing or thin.
- That each poster and slide deck was presented at the event, and its authors agree to publish it.
- A similarity check on every paper.
proceedings · what it carries
- event
- conference name, acronym, number, dates and place
- organisers
- organisers and sponsors as ROR IDs
- contributors
- authors with ORCID iDs and affiliations
- abstract + references
- for every paper
- relations
- each paper linked to its poster, slides, preprint and data
- funding + licence
- funder IDs, award numbers and a licence URL
Reports, theses and working papers
Technical and policy reports, working-paper series, standards and guidelines, and doctoral and master’s theses.
Registered with Crossref report and working-paper, dissertation and standard records
What we check
- Reports: an independent reviewer with the right expertise reads for accuracy, method and clarity.
- Theses: the awarding institution confirms the degree was awarded and that this is the examined version.
- Embargoes and permissions agreed in writing with the author and the institution.
- Working papers: editorial screening, and a clear label that they are not yet peer reviewed.
reports & theses · what it carries
- institution
- institution and department as ROR IDs
- series
- report or working-paper series and number
- degree
- degree, date awarded and supervisors with ORCID iDs (theses)
- abstract
- in the language of the work and in English
- funding
- funder IDs and award numbers
- relations
- the data, code and later articles behind or after the work
- licence + embargo
- licence URL and any embargo end date
Preprints and peer reviews
Early versions of work shared before review, and review reports published as citable objects in their own right.
Registered with Crossref posted-content (preprint) and peer-review records
What we check
- Preprints are screened, not peer reviewed — and labelled that way. We check scope, declared ethics approvals, plagiarism and harmful content.
- Review reports are published only with the consent of the reviewer and of the author reviewed.
- Signed reviews carry the reviewer’s ORCID iD; anonymous reviews stay anonymous.
preprints & reviews · what it carries
- versions
- every version of the preprint, in order
- isPreprintOf
- a link to the final version, wherever it is published
- isReviewOf
- for a review: the object it reviews
- review
- round, recommendation and competing interests
- licence
- a licence URL
Datasets and data papers
Research data of any size and discipline, and the data papers that describe them.
Registered with DataCite Dataset and DataPaper records
What we check
- Every file opens and matches its description.
- A README or codebook explains each variable, unit and code.
- Formats that can still be read in twenty years — or a stated reason why not.
- Personal data removed, or consent and approvals documented.
- How the data were collected and what they should and should not be used for — the questions asked in Datasheets for Datasets (Gebru et al.).
- A checksum recorded for every file.
- Data papers are peer reviewed like any article.
datasets · what it carries
- all 20 properties
- every DataCite property that applies, not just the six required
- Creator
- names with ORCID iDs; organisations with ROR IDs
- Description
- abstract, methods and technical notes
- GeoLocation
- points, boxes or polygons for where
- Date
- collected, valid, updated
- Size / Format / Version
- what is in the release
- RelatedIdentifier
- the article it supports, the software that made it, earlier versions
- FundingReference
- funder and award number
- Rights
- a licence URL — CC0 or CC BY
AI models and model cards
Trained machine-learning models, their weights, and the model cards that explain them.
Registered with DataCite a Model or Software record, whichever describes it, with the model card linked to it
What we check
- A model card that answers the questions in Model Cards for Model Reporting (Mitchell et al.): model details, intended use, factors, metrics, evaluation data, training data, quantitative analyses, ethical considerations, caveats and recommendations.
- The training data are identified, and their licences allow this release.
- The evaluation can be repeated: its data and code are linked.
- The weights are archived with checksums, so the identifier names exactly one model.
- Known limits and risks are stated plainly. We decline releases whose risks outweigh their use.
ai models · what it carries
- Version
- the exact release, and the base model it IsDerivedFrom
- training data
- each dataset linked, by DOI where one exists
- evaluation
- the evaluation data and results
- Requires
- the code needed to run it
- IsDocumentedBy
- the model card
- Size / Format
- parameters, files and formats
- Rights
- a licence URL
- FundingReference
- funders, including compute grants
Software, notebooks and protocols
Research software releases, computational notebooks, workflows and step-by-step protocols.
Registered with DataCite Software, ComputationalNotebook and Workflow records
What we check
- The code installs and runs by following its own instructions.
- Its tests run, if the project has them.
- An exact release is archived, so the identifier always points to the same code.
- An open-source licence approved by the Open Source Initiative.
- A citation file and every contributor credited, following the Software Citation Principles.
- Notebooks run top to bottom in a stated environment; protocols list every step, material and safety note.
software & protocols · what it carries
- Version
- release number and date, and the repository it came from
- Creator / Contributor
- with ORCID iDs and roles
- Description
- what it does, its environment and dependencies
- RelatedIdentifier
- the papers that describe it and the data it needs
- FundingReference
- funder and award number
- Rights
- a licence URL
Learning and audiovisual works
Courses and open educational resources, recorded lectures, research and documentary films, sound recordings and images.
Registered with DataCite Audiovisual, Sound, Image, InteractiveResource and Presentation records
What we check
- A subject expert checks the content for accuracy.
- Captions and transcripts, so the work can be used by people who cannot hear or see it.
- Consent from everyone who appears or speaks; permission for music and images.
- Courses state their level and learning outcomes.
- Files in formats suited to long-term preservation.
learning & audiovisual · what it carries
- Size / Format
- running time, files and formats
- Language
- spoken language and caption languages
- Contributor
- every contributor and role — director, presenter, producer
- RelatedIdentifier
- transcripts, lecture notes and the work it is based on
- Description
- learning outcomes and level, for courses
- Rights
- a licence URL
New kinds of objects
Objects research now produces that have no record type of their own yet: AI evaluation and audit reports, records of what AI contributed to a work, versions of a digital twin, field trials that run across seasons and sites, restoration and monitoring sites, disaster impact assessments, and cultural-heritage objects.
Registered with Crossref or DataCite whichever offers the nearest record type today — for example a Report, Dataset or PhysicalObject record — with the record saying exactly what the object is
What we check
- We agree the review with you, object by object, and publish what we checked alongside the object.
- No records about identifiable private people: no personal credentials and no decisions about an individual.
- Knowledge held by a community only with that community’s informed consent.
- We decline objects whose publication could cause serious harm.
new kinds of objects · what it carries
- type
- the nearest record type, plus a plain statement of what the object really is
- version
- every version, linked in order
- relations
- the data, reports and systems it belongs to
- everything else
- all the fields the record type allows
What we do not publish.
- A second identifier for something that already has one. We link to the existing DOI instead, or publish a new version that points back to it.
- Records about identifiable private people — personal credentials, health records, decisions about an individual.
- Knowledge held by a community without that community’s informed consent.
- Work you do not have the right to publish, and objects whose publication could cause serious harm.