Epstein Files
Onbekend
Publicly released U.S. Department of Justice records from the Jeffrey Epstein investigations, searchable page by page.
More about this case
3.173 court-filings (v1) + 15.130 nieuwe pilot-docs (clusters 01-08). Streefdoel: 3,5 miljoen documenten publiek doorzoekbaar.
What do you want to know?
Two routes to an answer.
An answer from the original documents, with its sources.
Reading is free. An account is needed to post, comment or fact-check. Create free account →
New DoJ dataset 09: 533k docs — what is the scope of cluster 10-12?
Cluster 09 = 533.468 docs (completely downloaded).
Cluster 10 = ~154k of presumed 500k (29% within).
Cluster 11 and 12 = just started, scope unknown.
Extract-status pipeline: 312.597 .txt files (~58% of cluster 09).
1 Sources
ORIGINALQuotes and page numbers are public. The full source text opens with a free account; the original scan or PDF is a Pro feature.
- 1
Handwritten docs in the DoJ set: how accurate is the OCR?
According to NPO Radio 1 there are handwritten documents included. PaddleOCR (currently in use on our pipeline) is trained on printed text — handwritten accuracy is considerably lower (30-60% instead of >90%). For critic…
1 Sources
ORIGINALQuotes and page numbers are public. The full source text opens with a free account; the original scan or PDF is a Pro feature.
- 1
Cross-reference: which known names appear in cluster 08?
Cluster 08 (11.034 docs) contains significantly more names from the political-economic field than clusters 01-07 (together 4.096 docs). Examples: references to Sultan Ahmed bin Sulayem (DP World Ports) and Peter Mandelso…
1 Sources
ORIGINALQuotes and page numbers are public. The full source text opens with a free account; the original scan or PDF is a Pro feature.
- 1