Review Policy

Why Nabu reviews are public

Nabu evaluates published research. A published paper is a public contribution to the scholarly record, and open discussion of published work — in reviews, journal clubs, cited critiques, post-publication commentary — is a foundation of how science corrects and improves itself. Nabu operates in that tradition.

For this reason, Nabu does not require author consent to evaluate a published paper, in the same way that a book review, a citation critique, or a PubPeer discussion does not require it. Any reader may request a review of any published article.

Reviews of unpublished manuscripts work differently. See Unpublished manuscripts below.

What we owe authors instead is fairness. Every Nabu review is:

• About the work, not the person. We evaluate the methodological quality of a specific paper against a published, versioned framework. We do not evaluate character, competence, or careers.
• Evidence-chained. Every score is supported by a traceable chain of evidence drawn from the paper itself. There are no unexplained judgements.
• Transparent about its own limits. Reviews state the framework version used, whether they were produced from full text or abstract only, and what could not be assessed.
• Open to reply, dispute, and correction — see below.

What a review contains

A Nabu review is attached to a paper, identified by its DOI, title, and standard bibliographic metadata as published. Author names appear only as part of that already-public bibliographic record. Nabu does not add commentary about authors as individuals, and reviews never speculate about intent, integrity, or ability.

Quality scores describe the paper as published and are frozen at the framework version used. Post-publication developments — retractions, corrections, replication evidence — are tracked separately and never silently alter a quality score.

Reviews of published research form part of Nabu’s public review record. The public record exists independently of the account that requested a review: if the requesting user deletes their account, their personal data is removed, but the review itself remains part of the record.

Unpublished manuscripts

You can upload a manuscript that has not been published, such as a manuscript in preparation, a submission under review, or a working paper. If the file has no registered DOI, Nabu gives it an internal identifier and the review is unlisted:

• it is not listed or searchable on Nabu;
• it is not included in our sitemap, and the page asks search engines not to index it;
• it can be viewed only by people who have its link. Anyone you share the link with can view it, and so can anyone they forward it to.

Unlisted reviews are not part of the public review record. The Right of reply, Disputes and Removal sections below apply to the public record, not to unlisted reviews.

Removal. The person who uploaded the manuscript can ask us to delete its review at any time by emailing hello@nabu.science from the email address used for the upload or, if no email was given, with the review’s link. We delete it within 7 days. Unlisted reviews are also deleted when the uploader deletes their Nabu account.

If the manuscript is later published. The unlisted review stays unlisted and is not linked to the published version. A review of the published paper, requested by anyone, is a separate review and joins the public record.

If you enter a DOI. A review is unlisted only when the manuscript has no registered DOI. If a DOI is found in the file or entered at upload, the review joins the public record like any other review of published research.

Right of reply

Any listed author of a paper with a Nabu review may publish a response, displayed alongside the review with equal prominence. A reply may address anything: disagreement with the assessment, context the review missed, constraints that shaped the work, or planned follow-up.

Replies are attributed and verified (via ORCID or institutional email), are not edited by Nabu, and may be updated or withdrawn by the author at any time.

Disputes

Where an author believes a review contains an error — a misreading of the paper, a factual mistake in the evidence chain, or a misapplied criterion — they may file a dispute.

• Disputes are assessed against the evaluation framework and the paper itself.
• If the dispute is upheld, the review is corrected and the correction is noted on the review page.
• Original scores are preserved in the review’s dispute history. We do not silently overwrite records; the trail of what changed and why is part of the review.
• Dispute outcomes and reasoning are visible on the review page.

A dispute challenges the evidence, not the existence of the review. “I disagree with the score” without a specific evidential claim is better served by the right of reply.

Corrections

We correct factual errors — wrong metadata, misattributed papers, errors in the evidence chain — as soon as they are verified, whether raised by an author, a reader, or our own checks. Corrections are timestamped and noted on the review page.

Removal

Because the public review record is only trustworthy if it is complete, we do not remove reviews of published research because an author or institution disagrees with them. Selective removal would bias the record towards favourable assessments and defeat the purpose of the platform.

A review is removed or amended only where:

• the paper was misidentified or the review was attached to the wrong work;
• the review is found to contain an error that cannot be remedied by correction;
• we are subject to a legal obligation to remove it; or
• the underlying content was evaluated in breach of our own content-access rules (see below), in which case it is re-issued on a compliant basis (for example, abstract-only) rather than deleted.

Retracted papers are not removed. Retraction is critical information for readers, and retracted papers remain visible with their retraction status prominently displayed.

How Nabu accesses papers

Nabu evaluates full text only where the paper is open access under a licence that permits this use. Where full text is not lawfully available to us, reviews are produced from the abstract and public metadata only, and are clearly labelled as abstract-based, with correspondingly limited scope. We respect publisher text-and-data-mining reservations.

Users may also submit papers for review where they hold the rights to do so, warranted at upload (see our Terms of Service). Uploaded files are processed to produce the review, are never published, displayed, or redistributed, and are deleted along with their extracted text within 7 days of upload. Reviews of uploaded manuscripts without a DOI are unlisted (see Unpublished manuscripts).

Authors’ data and your rights

Nabu processes a limited set of personal data in connection with reviews: author names, ORCID iDs, and affiliations as they appear in the public bibliographic record (sourced from the publisher and OpenAlex).

• Lawful basis. We process this data on the basis of legitimate interest (Article 6(1)(f) GDPR): enabling public, evidence-based assessment of published research. Our legitimate interest assessment is available on request.
• Right to object. Under Article 21 GDPR, you may object to this processing. We will consider every objection individually through the balancing test the law requires, taking into account that the data concerned is limited to the public bibliographic record of published work. An objection is not an automatic takedown, and outcomes may include data minimisation measures rather than removal of the review.
• Other rights. You may request access to, or rectification of, personal data we hold about you, and you may lodge a complaint with the Autoriteit Persoonsgegevens or your local supervisory authority.

To exercise any of these rights, contact privacy@nabu.science. See our Privacy Policy for how we handle personal data generally.

Contact

Questions about this policy, disputes, corrections, or replies: hello@nabu.science. Data protection requests: privacy@nabu.science.

This policy is versioned. Material changes will be announced in advance, and prior versions remain available.