What is salami slicing in academic publishing?

Salami slicing is splitting one body of research into several 'minimum publishable units' instead of publishing it as one study, usually to raise a publication count. Journals generally treat undisclosed overlap between the slices as redundant publication. Our check can surface an earlier slice with a near-identical title, or one in our local library whose abstract shares most of its distinctive words. It does not compare methods or datasets, so slices written up in different words will not match.

What's the difference between prior publication and duplicate publication?

Prior publication means your work — or substantial parts of it — has already appeared somewhere: a preprint, a conference paper or another journal. Duplicate publication is submitting the same paper to several venues at once, or republishing it without disclosure. Both are editorial issues, both can be acceptable when disclosed up front, and this check looks for both the same way: your title and abstract against 7 scholarly databases and our local library.

Does a preprint on arXiv or bioRxiv count as 'prior publication'?

At many journals, no — posting a preprint is widely accepted — but some still treat a preprint as prior publication, so check your target journal's policy. The result shows the matched title and, where there is one, a link, so you can see what was found; an exact title match on arXiv, bioRxiv or medRxiv, a match that CrossRef lists as posted content, or one whose DOI belongs to a known preprint server is labelled a preprint rather than a publication. The one exception is a preprint whose own record lists its published journal version: then the work is in a journal, and the check reports it as already published.

How does this compare to iThenticate or Turnitin?

Different scope, different stage. iThenticate / Turnitin are body-text similarity scanners that journals run on submitted manuscripts to detect copy-paste plagiarism. Our checker compares titles + abstracts BEFORE submission to find prior and duplicate publication. Use both: us for pre-submission strategy, the journal's iThenticate for post-submission body-text check.

Why several databases and not just CrossRef?

No single index covers everything. CrossRef holds the DOI metadata that publishers register; arXiv is searched directly for its preprints; PubMed and Europe PMC add biomedical and life-sciences records; Unpaywall confirms a DOI you already have; and our 4.5M-paper local library adds an abstract comparison that can surface a paper published under a different title.

What does the confidence score mean?

It is title similarity — the share of the shorter title's meaningful words that appear in the matched title — blended 60/40 with abstract word overlap when both abstracts are available, and shown as a percentage. It measures similarity, not the probability of a duplicate, and the verdict does not come from a threshold on it: an identical title, a DOI match or, in our local library, a close title whose abstract largely agrees is reported as published (or as a preprint, when it is on arXiv, bioRxiv or medRxiv, or CrossRef or the DOI marks it as one); a close title on its own is a possible match to check by hand.

Will running this check tip off the journal that I'm submitting to?

No. The searches go to scholarly indexes and library services, not to journals, and they come from our server: they carry your title (and, for some services, keywords from your abstract) but not your name or your target journal. On the web, your title and abstract are also saved to your own private run history so you can reopen the result; that copy is never shared with journals or publishers.

Can I run it again after revising the manuscript?

Yes. On the web each run is a Pre-Check at 15 credits per run, and credits never expire; through the API or the MCP tool it spends no credits, within the rate limit. A common pattern: once when the first draft is together, once after the literature review, and once before submission, in case a co-author published something related in the meantime. Your 25 welcome credits cover your first web run.

Duplicate Publication Checker

Duplicate Publication Checker for Academic Papers

Before you submit, find out whether your paper — or one very close to it — is already out there. The checker queries CrossRef, PubMed, arXiv, bioRxiv, medRxiv, Europe PMC and Unpaywall, plus our 4.5M-paper local library, in parallel, matching on your title and abstract — and, for Unpaywall and an exact library lookup, on a DOI if you give one. On the web it runs inside a Pre-Check: 15 credits per run, and a free account starts with 25 welcome credits. Through the API or our MCP tool, the same check is free and rate-limited.

Run a duplicate check — welcome credits cover your first Or find an open research gap to scope next

Scholarly databases

4.5M papers

Local library

15 credits

On the web, in Pre-Check

No credits

Through the API or MCP

What it checks

All 7 scholarly databases and our local library are queried at once — Unpaywall only when you give a DOI — and each source gets up to 12 seconds, so one slow index cannot hold up the result.

A title counts as an exact match when it is identical once case, punctuation and dashes are set aside, and as a close match when at least 90% of the shorter title's meaningful words appear in the other (titles need at least four words). Where a source returns abstracts, a close title match is dropped as a different paper when the two abstracts share under 40% of their meaningful words. Our local library goes one step further: an abstract sharing at least 70% of its meaningful words with yours flags a paper even under a different title.

CrossRef — the DOI metadata publishers register, searched by title (up to 60 results read)
PubMed — biomedical literature, searched for the exact title phrase, then by its keywords
arXiv — preprints in physics, mathematics, computer science and related fields, searched for the exact title phrase, then for the title's distinctive words
bioRxiv — biology preprints. bioRxiv's own API has no title search, so the check searches, by title, the records bioRxiv registers for its preprints with CrossRef
medRxiv — health-sciences preprints, searched by title the same way, through the records medRxiv registers with CrossRef
Europe PMC — life-sciences literature, searched by title
Unpaywall — looked up by DOI, so only when one is supplied: the API and the MCP tool accept a DOI, the web form does not ask for one
Our 4.5M-paper local library — an exact DOI lookup when you give one, then the exact title, then the title's distinctive words, then (for an abstract over 100 characters) the abstract's

What you get back

One verdict for the whole check: already published, a preprint of it, a possible match to check by hand, or nothing found (or could not check, when no source answered at all). It comes with the matched title (and its journal and year where the source gives them), a link where there is one, and a confidence percentage when the matching source computes one.

An exact title match on arXiv, bioRxiv or medRxiv is reported as a preprint rather than a publication, and so is a match that CrossRef lists as posted content or whose DOI belongs to a known preprint server. The exception is a preprint whose own record lists its published journal version (on arXiv, a journal DOI or journal reference; on bioRxiv and medRxiv, a link to the published article): then the work is in a journal, and it is reported as already published, linked to the journal DOI where the record gives one. A preprint is informational: unlike a journal match, it does not stop a Pre-Check. When nothing matches, the result names every source that did not answer, so a partial check never reads as a clean one.

How it works (4 steps)

On the web the check runs inside our Pre-Check tool (15 credits per run; a free account starts with 25 welcome credits). The API and the MCP tool run the same check free, within a rate limit.

Step 1. Paste your title and abstract — On the web, open the Pre-Check with the duplicate check first and paste the working title and abstract. Through the API or the MCP tool, send the title, the abstract and, if you have one, a DOI.
Step 2. Every source is queried at once — The check queries CrossRef, PubMed, arXiv, bioRxiv, medRxiv, Europe PMC and Unpaywall (Unpaywall only when a DOI is given) and searches our 4.5M-paper local library by title and abstract, all in parallel. Each source gets up to 12 seconds.
Step 3. Read the verdict — You get one verdict — already published, a preprint of it, a possible match to check by hand, or nothing found (or could not check, when no source answered) — with the matched title and, where there is one, a link and a confidence percentage. A nothing-found result names any source that did not answer, so a partial check never reads as a clean one.
Step 4. Decide what the match means — You decide whether a match is a true duplicate, your own preprint (disclose it), related work to cite, or a false positive.

Why it matters (the editorial reality)

Three situations this check is built for, and how far it goes on each:

Salami slicing — splitting one study into several 'least publishable units'. Journals generally treat undisclosed overlap between the slices as redundant publication. The check surfaces an earlier paper with a near-identical title, or one in our local library whose abstract shares most of its distinctive words. It does not compare methods or datasets, so slices written up under different titles and in different words will not match.

Forgotten preprint — a working draft you or a co-author posted months ago. Many journals accept preprinted work if you disclose it; some do not. The result shows the matched title and links to it, so you can see what is out there and disclose it.

Resubmission after a rejection — an earlier version may already be indexed, as a preprint or a conference paper. Knowing before the editor does lets you disclose it or explain the difference.

When to use it

Four moments in the submission cycle where running this check is high-leverage:

Pre-submission, when the manuscript is mostly done — it leaves time to add a citation or restructure if a match surfaces.
Before resubmitting to another journal after a rejection — the earlier version may already be indexed.
Before answering a reviewer's prior-publication concern — a named match, or a clean result that lists every source searched, says more than 'we don't think so'.
When taking over a manuscript from a lab member who has left — it can surface a preprint or paper they posted without telling the group.

What we DON'T do

Explicit limitations, because false confidence is worse than known gaps:

Not a full-text plagiarism scanner. We match titles and abstracts, not body paragraphs. If a paper paraphrases your entire methods section, we won't catch it.
Not iThenticate / Turnitin / Crossref Similarity Check. Those are body-text similarity tools that journals run internally; we're complementary to (not a replacement for) them.
Not a translation check. Matching is word-based, so a version of your paper published in another language will not match.
Not a substitute for institutional integrity review. If a match surfaces and you're unsure, talk to your research office.
On the web, your title and abstract are saved to your own private run history so you can reopen the result — never shared with journals or publishers, and never sold. The searches go out from our server to scholarly indexes and library services, carrying your title (and, for some, keywords from your abstract) but not your name.
Results are bounded by coverage. A duplicate in a venue none of these sources index will be missed, and a source that is down or slow is skipped for that run — a nothing-found result names it.

What does it cost?

Two ways to run the same check.

On the web, it runs inside a Pre-Check, which costs 15 credits per run with a free account. New accounts start with 25 welcome credits, which cover a first run. There is no subscription: credits come in one-time packs of 50 and never expire.

Through the API — POST /api/prior-publication with a title, and optionally an abstract and a DOI — or the MCP tool check_duplicate_publication, the check is free: it spends no credits. It is rate-limited to keep it that way: 20 checks an hour per IP address (at most 5 a minute), or 60 an hour (10 a minute) with an API key or a signed-in account. The MCP tool uses an API key from a free account.

Frequently asked questions

Through the API and the MCP tool, yes: it spends no credits, within a rate limit of 20 checks an hour per IP address (60 with an API key or a signed-in account). On the web, the check runs inside a Pre-Check, which costs 15 credits per run; a free account starts with 25 welcome credits.

Related tools

MCP server — run check_duplicate_publication from an MCP-capable AI assistant; it spends no credits.
AI Review — 8 specialist agents review the full manuscript, one of them this same prior-publication lookup — 30 credits per review.

Run a duplicate check — welcome credits cover your first Read the engineering blog