Scanned 7,288 pages of native text-layer extraction (xberg + pdf_oxide) across the published data-no-ocr corpus. The signatures below are structural artifacts of native PDF text extraction (encoding shifts, missing inter-word spaces, junk text layers); genuinely scanned pages that fell to the OCR fallback cannot exhibit them.
45 page(s) flagged across 3 class(es):
class
severity
pages
noise-garbage
high
6
concatenation
medium
3
cmap-encoding-shift
high
2
noise-garbage (high) — 6 page(s)
file
page
chars
snippet
aaro/aaro-hrr-vol1-2024.pdf
4
4375
International Security Environment and Technological Surprise ......................................