tuandunghcmut/multipage-dla-pdf-corpus

Generated at: 2026-05-01T10:42:46+00:00

Scope

Private PDF corpus export from the multipage DLA pdf_preparation subproject. This dataset is a candidate training pool and audit archive, not reviewed gold data.

Export policy

• Catalog source: /mnt/project/productivity/dungvpt/dla_workspace_2/multipage_dla/artifacts/reports/pdf_file_catalog_latest.jsonl
• Status filter: accepted
• Include Mixed-Verify: True
• Repository visibility: private

Aggregate

Metric	Value
PDFs	1457
Pages	53057

By family

Family	PDFs	Pages
A	546	16160
B	92	4341
C	91	3298
D	94	2741
E	634	26517

By status

Status	PDFs
accepted	1457

By license tag

License tag	PDFs
PD	622
Free-Standard-IETF	499
PD-US-Gov	310
Mixed-Verify	26

Important limitations

• Per-file metadata is authoritative in metadata/catalog.jsonl.
• Mixed-Verify files need license review before public redistribution.
• This private dataset may include candidate annotations and should not be described as reviewed gold data.