Data Dictionary
A complete reference for every field, classification, and category used in the Epstein Exposed database. For researchers building on our data or journalists verifying our methodology.
Persons
Each person in the database is identified by a unique slug and categorized by their primary role relevant to the Epstein case. Person data includes biographical information, document associations, flight records, and known connections.
Fields
| Field | Type | Description |
|---|---|---|
| id | string | Unique identifier (e.g., 'jeffrey-epstein') |
| slug | string | URL-safe identifier used in page URLs |
| name | string | Full display name |
| aliases | string[] | Known alternative names or spellings |
| category | enum | Primary classification (see categories below) |
| description | text | Detailed biography and Epstein case involvement |
| shortBio | string | One-line summary for listings and cards |
| nationality | string | Primary nationality |
| notablePositions | string[] | Professional titles and roles |
| blackBookEntry | boolean | Whether they appear in Epstein's contact book |
| blackBookPhones | number | Count of phone numbers in the black book |
| tags | string[] | Freeform labels (e.g., 'victim', 'inner-circle') |
| imageUrl | string? | Headshot URL (Wikimedia Commons with attribution) |
| flightCount | computed | Number of flight log appearances |
| documentCount | computed | Number of associated documents |
| connectionCount | computed | Number of known connections to other persons |
| emailCount | computed | Number of associated emails |
Person Categories
Elected officials, government appointees, diplomats, political operatives
Business executives, entrepreneurs, financial professionals, hedge fund managers
Members of royal families and aristocratic lineages
Entertainers, actors, musicians, media personalities, public figures
Known associates of Epstein, staff members, personal contacts, recruiters
Attorneys, judges, prosecutors, law enforcement involved in Epstein cases
Professors, researchers, university administrators, scientists
High-society figures, philanthropists, socialites
Military officers, intelligence operatives, national security officials
Persons who don't fit neatly into other categories
Documents
Documents range from single-page court orders to multi-thousand-page depositions. Each document is categorized by source, linked to mentioned persons, and full-text searchable via OCR extraction.
Fields
| Field | Type | Description |
|---|---|---|
| id | string | Unique document identifier (e.g., 'gov.uscourts.nysd.447706.1090.0') |
| title | string | Document title or filename |
| category | string | Document type classification |
| source | enum | Origin source (see sources below) |
| date | string? | Document date (ISO format when available) |
| summary | text | AI-generated or manually written summary |
| personIds | string[] | IDs of persons mentioned (auto-linked + manual) |
| pdfUrl | string? | Direct URL to original PDF |
| pageCount | number? | Number of pages (when known) |
| ocrText | text? | Full OCR-extracted text (156K+ documents) |
| redactionScore | float? | Percentage of document that is redacted (0-100) |
Document Sources
Primary court documents filed in federal and state proceedings
Sworn testimony taken under oath during pre-trial discovery
FBI investigative reports and memoranda obtained via FOIA or court releases
Department of Justice official releases, indictments, plea agreements
Freedom of Information Act responses from federal agencies
Financial records, wire transfers, bank statements, tax documents
Journalistic investigations, news reports, interview transcripts
Victim impact statements, survivor testimonies
Law enforcement investigation reports and records
Grand jury and trial subpoenas (257 subpoenas indexed)
Electronic File Transfer Agreement documents from SDNY prosecution (28,942 PDFs)
Documents that don't fit standard categories
Flights
Flight records from Epstein's private aircraft, primarily the Boeing 727 ("Lolita Express") and various Gulfstream jets. Sourced from FAA records and pilot logbooks entered into court evidence.
Fields
| Field | Type | Description |
|---|---|---|
| id | string | Unique flight identifier |
| date | string | Flight date (ISO format) |
| aircraft | string | Aircraft type (e.g., 'Boeing 727-31') |
| tailNumber | string | FAA registration number (e.g., 'N908JE') |
| origin | string | Departure location |
| destination | string | Arrival location |
| passengers | array | List of passengers with name and person ID |
| source | string | Data source (e.g., 'pilot-log', 'faa-record') |
Connections
Connections represent relationships between persons, derived from shared documents, co-flights, court testimony, and manual research. Each connection has an evidence-based strength rating.
Connection Strength Levels
Direct evidence from court documents: co-defendants, employer/employee, victim/perpetrator, family, or 3+ shared documents
Circumstantial evidence: shared flights, phone contacts, mentioned together in depositions, 1-2 shared documents
Tangential connections: same social circles, one-time mentions, disputed or unconfirmed links
Fields
| Field | Type | Description |
|---|---|---|
| personId1 | string | First person in the pair |
| personId2 | string | Second person in the pair |
| strength | enum | strong | moderate | weak |
| summary | text | Description of the relationship and evidence |
| coFlights | number | Number of shared flights |
| coDocuments | number | Number of shared documents |
| relationshipType | string? | Typed classification (e.g., traveled_with, associated_with) |
| weight | number? | Numeric weight for network graph rendering |
| dateFirst | string? | Earliest evidence date |
| dateLast | string? | Latest evidence date |
Trust Levels
Every piece of data is assigned a trust level based on its source reliability and verification status. This helps researchers assess the confidence level of any claim.
Primary sources filed with courts — highest reliability. Includes unsealed filings, exhibits, and judicial rulings.
Government records: DOJ releases, FBI reports, FOIA responses, flight logs. Authenticated by issuing agencies.
Depositions and trial testimony given under oath. Subject to perjury penalties but represents one party's account.
Journalist investigations and researcher findings. Cross-referenced where possible but not primary sources.
Uncorroborated claims, anonymous tips, or single-source information. Included for completeness with clear labeling.
Person Matching Methodology
Documents are automatically linked to persons using a multi-step matching pipeline:
- Multi-word name matching — Names with 2+ words are matched using word-boundary regex against document OCR text. This prevents false positives (e.g., "Prince" alone matching Prince Andrew).
- Single-word name exclusion — Single-word names (e.g., "Maxwell") are never auto-matched due to legal liability. These are only linked through manual review.
- Co-occurrence analysis — When two persons appear in the same document, a co-occurrence record is created. The database tracks 25,700+ person pairs.
- Mention context extraction — For each person-document link, a snippet of surrounding text is stored to show how the person is referenced.
Data Access
All data is available for researchers and journalists:
This data dictionary was last updated on February 17, 2026. If you have questions about our methodology or data structure, please visit our contribute page or reach out through the community forum.