This gated package is the complete dated 2026-08-30 metadata snapshot from Plethora Solutions. It contains 965,792 unique scholarly-document records in 10 Parquet shards with normalized discovery fields, source links, operational rights tiers, NASA-specific rights evidence, duplicate flags, and historical link-health fields.

The package contains metadata and links only. It contains no PDF files and no extracted document text. Rights in linked documents remain with their authors, publishers, agencies, or licensors.

Purchase and access

The standard commercial snapshot license is $1,250 USD for one purchasing legal entity and up to five authorized users. The first ten customers may purchase the same license for the $499 USD founding-customer price.

Secure founding-customer checkout: https://www.paypal.com/ncp/payment/4F6QSYDG5DBQJ

Access is approved manually after payment. Submit the PayPal transaction ID and license acknowledgment through this repository's access-request form. Team, custom validation, customer-cloud delivery, and enterprise arrangements start at $2,500 and are quoted separately.

Read TERMS.md, RIGHTS.md, and PRICING.md before purchasing or requesting access.

Snapshot composition

Rights tier	Rows	Operational meaning
`A`	430,214	Recognized commercial CC/CC0 URL in upstream metadata
`N`	284,371	Official NASA public-use permission evidence; not CC-BY
`B`	230,637	Plausible rights assertion without strong machine-checkable evidence
`C`	7,503	Manual rights review required
`D`	13,067	Restrictive or adverse rights evidence detected

Under the release's historical link-status criteria, 316,314 tier-A rows and 266,400 tier-N rows were strong document-acquisition candidates. These counts are metadata classifications, not a guarantee that every URL remains live.

Completed full-catalog PDF-link validation

The fresh two-worker validation pass is complete. Between 2026-08-31 and 2026-09-02, it checked all 776,075 distinct non-empty PDF URLs associated with 783,073 catalog records. 611,609 URLs (78.81%) began serving content with a PDF signature at check time.

The worker followed redirects, requested only bytes 0-16,383 when supported, read no more than 16 KiB, and required %PDF- within the first 1 KiB. It did not download or store complete PDFs. See FULL_LINK_VALIDATION.md and full_link_validation_summary.json for status totals, rights-tier outcomes, integrity checks, and hashes of both final evidence databases.

The snapshot's pdf_url_status and pdf_url_checked_at columns still preserve the pipeline's earlier historical values. The separate full-validation report is the current dated network evidence and has not been written back into those Parquet columns. A successful prefix check establishes neither full-file integrity nor document identity, and it does not change a document's rights.

The separate public 10,000-row evaluation sample has stronger current evidence: all 8,500 ready-tier sample rows were freshly checked on 2026-08-31 using a bounded streamed GET and PDF-signature test. That test established that a URL began serving PDF content at that time; it did not establish full-file integrity or document identity.

Evaluate the public sample first: https://huggingface.co/datasets/PlethoraSolutions/open-scholarly-document-catalog-10k

Load with Python

from datasets import load_dataset

catalog = load_dataset(
    "PlethoraSolutions/open-scholarly-document-catalog-full",
    split="train",
    token=True,
)
print(catalog.num_rows)  # 965792

Or download and read the Parquet shards directly with PyArrow, DuckDB, Polars, Spark, or another Parquet-compatible tool.

Package contents

data/catalog-*.parquet: 10 schema-identical Parquet shards.
FIELD_DICTIONARY.md: all 37 fields and their interpretation.
dataset_stats.json: exact counts and source coverage.
source_release_validation.json: row, uniqueness, schema, and source-release QA.
package_validation.json: validation of the packaged commercial copy.
FULL_LINK_VALIDATION.md: completed fresh PDF-link validation report.
full_link_validation_summary.json: machine-readable link outcomes and evidence hashes.
package_corrections.json: transparent record of mechanical package corrections.
MANIFEST.json and CHECKSUMS.csv: file sizes, row counts, and SHA-256.
RIGHTS.md: operational rights-methodology notice.
TERMS.md: commercial catalog license terms.
PRICING.md: launch pricing and product boundaries.
CHANGELOG.md: dated release notes.
SUPPORT.md: delivery and support policy.

Important limitations

Rights tiers are operational evidence classifications, not legal opinions.
The catalog does not grant rights to any linked document.
Link availability and upstream metadata can change after the snapshot date.
Tier A still requires compliance with the identified Creative Commons terms.
Tier N is NASA public-use evidence and is never represented as CC-BY.
Tiers B, C, and D should not be treated as redistribution authorization.
Coverage is source-driven, predominantly English, and not exhaustive.

Downloads last month: 25

Total file size:

70.2 MB