Import extractions

Already have extraction JSON — from your own pipeline, an earlier run here, or a whole dataset export? Drop it here with the source PDFs (paired by filename, or assign them below) — the app renders the pages and highlights each citation from the JSON's evidence, so you get the full viewer. Papers without a PDF are imported too, just without page images. No model is called.

Drag & drop your .json + .pdf files here
or click to browse — pair by name, e.g. paper1.json + paper1.pdf. Accepted: a single-paper result, a workspace export, a dataset export (split into its papers), a .gemini.json suffix or an { "extraction": … } wrapper.
a new dataset shows only these uploads; pick an existing one to merge
the preset whose fields, tabs and evidence rules the review uses — an export's own schema id is picked up automatically