The AI match in Open Access Finder is a free journal suggester that works from your abstract. It compares your text with every journal in the Directory of Open Access Journals (DOAJ) and lists those whose scope is closest to your paper. By default it shows only the 14,571 diamond journals, which charge no author fee, each with its Scopus quartile, fees and the reason for the match. New to these terms? Read our guide to open access.
Paste your title and abstract; a small language model running in your browser ranks free journals by how well their scope fits. Nothing leaves your device.
How the AI journal finder works
- Paste your text. Open the AI match tab, enter your title in the first box and your abstract (up to 6,000 characters) in the second, then click Find journals with AI.
- The model loads once. Your browser downloads all-MiniLM-L6-v2, a small sentence-embedding model (about 23 MB), and runs it with Transformers.js. It stays cached afterwards.
- Your abstract becomes a vector. The model turns your text into 384 numbers that capture its topic. Long abstracts are split into chunks of about 150 words, then averaged.
- Every journal is compared. Each DOAJ journal has a precomputed vector. For more than 10,000 journals, close to half of DOAJ, it is built from the official aims and scope (80%), DOAJ keywords and subjects (15%) and SCImago categories (5%); the scope statements come from the journals' own websites or from PJIP. The other journals rely on title, keywords, subjects and categories, and are labelled "metadata only".
- Off-topic journals are removed. A discipline check drops journals from unrelated fields, and anything below a 20% match is hidden; the rest are ranked.
What each result shows
- The match score with a label: strong (70% or more), good (50–69%), possible (35–49%) or weak (20–34%, hidden until you ask).
- The evidence: "Official aims & scope" or "Official scope unavailable — metadata only", with a Verify aims & scope link to the source page or the journal's website.
- A two-part "why" line: the topical match (for example "Shares your topics: remote sensing, irrigation" and the SCImago categories), then what only affected the order, such as the quartile or "free to publish".
- The journal card: quartile, publisher, country, fees, SJR, H-index and weeks to publication. For about 3,500 journals you can expand the aims and scope text from PJIP (Practical Journal Insight Project) under CC BY-NC 4.0.
Filters limit which journals are compared, never their score: fee type, quartile, SCImago-indexed only, subject area, maximum weeks to publication and maximum APC. You can sort by match, quartile, SJR, H-index, turnaround, price or title, shortlist journals and run a live Scopus check.
Export CSV saves every scored match in the current order, with rank, match percentage, evidence level, the "why" text and shared keywords, plus the journal's ISSN, fees, quartile, SJR, H-index, publisher and links. The file is created in your browser.
What the match score means, and what it doesn't
The percentage measures topical similarity only: a cosine similarity between your abstract and the journal's scope, rescaled so that unrelated fields sit near 0% and the closest journals for a typical abstract land between about 50% and 95%. Quartile, fees and shared keywords never change it; they only reorder journals that already passed the threshold, and only slightly.
Only journals whose official aims and scope backed the match can be main recommendations. Metadata-only journals appear under "Possible matches — verify the journal's aims and scope". Texts under 40 words trigger a warning: a title alone gives unreliable results.
Privacy: nothing leaves your browser
The other tools in the comparison below process your abstract on their servers. Here the matching happens on your device:
- The model runs inside your browser. Your title and abstract are never sent to us, to an AI provider or to any API.
- Your browser downloads the Transformers.js library (jsDelivr), the model (Hugging Face) and the journal vectors (our data server). None of these requests contains your text, and the files are cached.
- There is no account. Your shortlist stays in your browser's storage, and the CSV is generated locally.
That suits unpublished manuscripts you would rather not paste into someone else's system.
How it compares with publisher journal finders
Publisher finders help you choose among one publisher's journals, with metrics such as Impact Factor or decision times. Open Access Finder answers another question: which open access journal, from any publisher, fits my paper without an author fee? The table reflects each tool's public pages on 15 October 2026.
| Tool | Covers all publishers? | Filters to no-APC journals? | Shows Scopus quartile? | Runs in your browser? | Sign-up needed? |
|---|---|---|---|---|---|
| Open Access Finder (AI match) | Yes, all DOAJ journals (fully open access only) | Yes, diamond by default | Yes (SCImago) | Yes | No |
| Elsevier JournalFinder | No, Elsevier journals indexed in Scopus | No; gold open access or subscription filter | Shows CiteScore and Impact Factor | No | No; only to save lists |
| Springer Nature Journal and funding finder (formerly Journal Suggester) | No, Springer Nature journals | Shows open access funding for your institution | — | No | No |
| Wiley Journal Finder | No, about 1,900 Wiley journals | No; fully open access, offers open access or subscription filter | Shows Impact Factor and CAS ranking | No | No |
| Taylor & Francis Journal Suggester | No, Taylor & Francis journals | No; flags open access options | Shows citation and speed metrics | No; abstracts are kept to improve the tool | No |
| JANE | Yes, journals in PubMed (biomedicine and life sciences) | No; can limit to open access journals | — | No | No |
| B!SON (TIB and SLUB) | Yes, all DOAJ journals | Yes, maximum APC slider | No | No; input is not stored | No |
"—" means the tool's public pages don't say. B!SON is the closest alternative: it also covers DOAJ, filters by APC and can compare your reference list with OpenCitations data. For subscription or hybrid journals, use a publisher finder: DOAJ lists only fully open access journals. Sources: Elsevier, Springer Nature, Wiley, Taylor & Francis, JANE, B!SON.
Tips for better matches
- Paste the title and the full abstract. 150 to 300 words give the model enough to work with; a title alone does not.
- Use English. The model was trained mainly on English text, including about 42 million title–abstract pairs from scientific papers. If your paper is in another language, paste an English version of the abstract.
- Add key terms to the title box: the method, material, organism, population or region. Specific terms help more than words like "analysis" or "system".
- Try variations. Run it again with only your methods and results, or stress the application rather than the technique; interdisciplinary papers often surface different journals.
- Widen, then narrow. If few journals appear, include APC or unranked journals and clear the subject area, then tighten quartile, turnaround and budget.
- Verify your shortlist. Open Verify aims & scope, read recent articles and confirm Scopus indexing if you need it.
Frequently asked questions
Is the AI journal finder really free?
Yes: no sign-up, no paid tier, no search limit. The model runs on your device, so searches cost us nothing.
Is my abstract uploaded or stored?
No. Your text is processed in your browser and is not sent anywhere. Only the model and the journal vectors are downloaded.
Does a high match score mean my paper will be accepted?
No. The score only measures how close your abstract is to the journal's stated scope. Acceptance depends on your work, the journal's standards and peer review. Treat the results as a shortlist to check.
How do I find Scopus-indexed journals with no APC?
Keep the default Diamond · free filter, switch on Indexed in SCImago only and pick your quartiles. 4,279 diamond journals are ranked in Scopus, 711 of them in Q1. For ready-made lists, see free Scopus-indexed journals or browse by subject.
Can I use it with an abstract in French, German or Arabic?
You can, but the matches will be weaker, because the model compares English text best. Paste an English version of your abstract, and filter the main finder by language to find journals that publish in yours.