Skip to contents

Split AI extraction/QA output into per-reviewer batch folders.

Source: R/split_for_review.R
split_for_review.Rd

Reads the AI pipeline's two .xlsx deliverables (extraction + QA) and the source PDFs, and writes one self-contained batch folder per reviewer, ready to hand off (see run_app).

Usage

split_for_review(
  extraction,
  qa,
  papers_dir,
  out,
  manifest = NULL,
  n_batches = 6
)

Arguments

extraction

Path to the extraction .xlsx (paper_id, pipeline, one column per extraction field, Flags).

qa

Path to the QA .xlsx (paper, pipeline, subitem, verdict, evidence).

papers_dir

Folder containing the source PDFs.

out

Destination folder for the batch_N/ subfolders. No default on purpose – this data may live nowhere near wherever this package is installed, so the destination is always explicit.

manifest

Optional path to a manifest.json mapping paper_id -> pdf_file (and, optionally, doi) for when PDF filenames don't reliably match paper_id. Without it, assumes "<paper_id>.pdf" under papers_dir, and every paper's DOI is left blank.

n_batches

Number of reviewer batches to split into (even split, no overlap). Default 6.

Value

Invisible; called for its side effect of writing batch_N/ folders under out. Prints a per-batch and total summary.