Uploading PDFs yourself

Upload, then match

Uploading has two steps. The file is stored first, and an interrupted upload of a large file can continue where it stopped. Then the platform works out which paper the file belongs to.

Files are stored privately and downloaded only through short-lived links. Many of them are licensed copies obtained through your library, so they are never public.

When the match is automatic

A file is attached automatically only when a DOI found in the file or its name matches exactly one paper in the project. Anything less certain goes to the unmatched list for a person to confirm.

The rule is strict because a PDF attached to the wrong paper would give every value extracted from it the wrong citation, and that kind of mistake is very hard to find later.

Duplicate files

If an uploaded file is identical to one already attached to a different paper, nothing is changed until you decide. Either the same PDF really belongs to both papers, or one of the two is attached to the wrong paper.

Practical notes

A few things save time.

• There is an upload size limit; very large scanned volumes may need splitting.
• Prefer the publisher PDF to a scanned printout. Tables are read well from a digital PDF and poorly from a scan.
• A scanned document is recognised as a scan and read with optical character recognition. This is noted, so a reviewer knows why the text may be imperfect.
• Upload the supplementary file too when the tables you need are in it.