Importing reference files

Identifiers and pasted references

Paste DOIs, PubMed IDs or PMC IDs, one per line or separated by commas or spaces (up to 500). Each is checked first: anything that is not one of the three is listed as not valid and is never guessed. The valid ones are looked up, and each comes back as found, not found, or could not be looked up — which means the database did not answer, not that the paper does not exist. You choose which of the found papers to add.

You can also paste a reference list (up to 100 references). For each reference the closest match is shown with its title, authors, year and DOI, and how closely it matches. Nothing is accepted for you: tick the matches that are right. "No confident match" is a normal result for an unusual or incomplete reference.

Reference files

The importer reads RIS, BibTeX, CSV and Excel files (the first sheet of a workbook, with the same columns as a CSV file). Imported papers join the same list as searched papers and go through the same duplicate check, screening and scoring.

The file name is kept on every paper it produced, and the import appears in the project search history marked as manual, so you can always tell where a paper came from.

Export settings that matter

The most common problem is a file with no abstracts. Screening works from the title and abstract, so a paper without an abstract cannot be judged and waits for a person. Include abstracts when you export.

• Scopus — run the search, select all, Export → CSV, and tick Abstract, Keywords, DOI and Citation information. A subscriber may prefer to export from the Scopus interface rather than switch the API connector on. Both routes are legitimate; without abstracts, screening cannot make a decision.
• Web of Science — Export → Tab delimited or RIS, record content Full Record. Web of Science caps an export at 1000 records, so a large search exports in batches; import each file. Choosing Full Record is what carries the abstract.
• Google Scholar — search → Save to My Library → Export → BibTeX. Scholar exports carry no abstracts, so those records need full text before they can be screened.
• ResearchGate — open the paper → copy the citation, or export it, and bring the file here. A DOI is enough for this platform to fetch verified metadata from the DOI registry, so a copied DOI is usually the fastest route.
• MDPI (a publisher, not a database) — open the article on the publisher's site → Export citation → RIS or BibTeX, and bring the file to the Imports page. A publisher export covers that publisher's own articles and nothing else, so it is a way to bring in papers you already have, never a way to survey a literature.
• Any other database — export RIS / BibTeX / CSV with abstracts included. Screening classifies from title and abstract, so tick the abstract box in the export dialog.
• Institutional access — open the library link → sign in with your own institution → download the PDF → upload it here. The upload is authorised by the person who made it, and is recorded that way.

What the importer fixes and what it leaves

Identifiers are tidied on the way in. A DOI can be written in many ways and is stored in one, so two copies of the same paper from different databases match.

The importer does not invent anything. A missing year stays missing, an author list is stored as it arrived, and a record with no title is rejected.

Import first, then remove duplicates

Import everything you plan to screen, then run the duplicate check before anyone screens. Screening the same paper twice wastes time and can produce two decisions that disagree. Done in this order, each paper is screened once.