Data Dictionary
April 20, 2026 · View on GitHub
Top-level keys and the schema of their entries.
meta
Provenance block — sources, license, extraction date & method, editorial changes.
event_codes
Source: GDELT CAMEO.eventcodes.txt + CAMEO.goldsteinscale.txt
| Field | Type | Description |
|---|---|---|
code | string | 2-, 3-, or 4-digit CAMEO event code |
level | int | 2 = root, 3 = subcategory, 4 = specific |
parent | string or null | Code one level up (null for roots) |
root | string | 2-digit root code |
description | string | Human label |
goldstein | float (optional) | Goldstein conflict-cooperation scale, -10 to +10 |
country_codes
Source: GDELT CAMEO.country.txt, enriched via datasets/country-codes.
| Field | Type | Description |
|---|---|---|
code | string | 3-letter CAMEO country code (often matches ISO 3166-1 alpha-3) |
label | string | Country / entity name (editorial: Turkey → Türkiye) |
iso | object (optional) | ISO enrichment (alpha-2, numeric, currency, region, FIFA, IOC) — only present where ISO alpha-3 matched |
CAMEO also defines sub-national codes (e.g. WSB West Bank, GZS Gaza Strip, BAG Baghdad) which have no ISO equivalent and therefore no iso block.
known_groups, ethnic_groups_lookup, religions_lookup, actor_types_lookup
Simple {code, label} lookups from the corresponding GDELT files.
domestic_role_codes
Source: CAMEO Manual Table 3.1 (Generic Domestic Role Codes). Rich descriptions; grouped into Primary (COP, GOV, etc.), Secondary (AGR, BUS, etc.), and Tertiary (MOD, RAD) role codes.
| Field | Type | Description |
|---|---|---|
code | string | 3-letter role code |
description | string | Description (and grouping label when the row is a section divider) |
generic_codes
Generic non-geographic actor codes (e.g. IGO, NGO, UIS, MNC).
international_region_codes
Sub-continental and supranational region codes (e.g. EUR, NAM, SAM, MEA).
international_actors_special_codes
International organizations with custom CAMEO codes (UN, IMF, NATO, etc.).
religious_classification
Source: CAMEO Manual chapter "CAMEO Religious Classification". Hierarchical religion codes up to 9 chars (e.g. BUDMAH118). Extracted via regex fallback — may need review against the PDF.
| Field | Type | Description |
|---|---|---|
code | string | Uppercase alphanumeric hierarchical code |
description | string | Religion / denomination / sect label |
keds_project_actor_codes
Source: CAMEO Manual chapter "KEDS Project Actor Codes". The canonical KEDS actor dictionary — countries, insurgent groups, government sub-entities.
| Field | Type | Description |
|---|---|---|
code | string | KEDS actor code (varying length, often ISO3 or ISO3+ROLE+GROUP) |
actor | string | Actor label |
regional_dictionaries
Per-region sub-dictionaries extracted from the Regional Dictionaries chapter.
| Sub-key | Description |
|---|---|
main_ethnic_groups_a / _b | Ethnic group codes used across multiple regions (pages 2-3) |
middle_east | Middle East actor/region codes |
turkey | Türkiye-specific ambiguous and special actor codes |
nigerian_states | Nigerian states with custom codes |
liberian_counties | Liberian county codes |
west_african_actors | West African actors with special codes |
balkans | Balkan region actor codes |
second_three_letter_codes
Source: CAMEO Manual, Religious Coding Scheme page 4. Conventions for the second three-letter component of compound religion codes.
Conventions
- All code fields are case-preserved from the source (usually UPPERCASE).
- Empty/missing values are represented as empty strings, not
null, except where explicitly noted (e.g.event_codes.parent). - CSV exports under
data/lookups/anddata/tables/carry the same semantics as the corresponding JSON section.