Overview

The GDELT Project (Global Database of Events, Language, and Tone) is the largest, most comprehensive, and highest-resolution open dataset of global human society ever created. Updated every 15 minutes, GDELT monitors broadcast, print, and web news from virtually every country in over 100 languages, translating and processing each article to identify events, entities, themes, emotions, and relationships. The project is supported by Google Jigsaw and hosted on Google Cloud Platform, leveraging BigQuery for analytical access to the full dataset.

GDELT maintains three primary datasets: the Event Database (structured records of who-did-what-to-whom events coded using the CAMEO taxonomy), the Global Knowledge Graph (people, organizations, locations, themes, emotions, and their interconnections), and the Global Entity Graph (entity-level analysis of news content with sentiment and context). Together, these datasets represent over a quarter-billion event records and billions of entity and theme references spanning decades of global media coverage.

For OSINT practitioners, GDELT provides unmatched media intelligence at a global scale. It enables real-time monitoring of geopolitical events, tracking of entities across news coverage, sentiment analysis around organizations and individuals, and early warning of emerging crises. The platform's 15-minute update cycle means analysts can detect breaking events nearly as quickly as newsrooms. When combined with targeted source intelligence from NewsAPI and social media monitoring via Social Searcher, GDELT provides the broadest possible media intelligence coverage.

The scale of GDELT is difficult to overstate: it processes hundreds of thousands of articles daily across virtually all countries, in all major languages, from news outlets ranging from global wire services to local newspapers. This breadth makes it uniquely suited for detecting weak signals that might appear only in regional media before escalating to international attention.

Data Coverage

Data Type	Description
Events	Structured who-did-what-to-whom records with CAMEO coding
Entities	People, organizations, locations extracted from articles
Themes	Hundreds of thematic categories (conflict, economics, health, etc.)
Sentiment	Tone scores (-100 to +100) per article and entity
Geolocation	Latitude/longitude for events and entities
Source URLs	Original article URLs for each processed story
Languages	100+ languages with machine translation
Imagery	Satellite, news photos analyzed via Google Cloud Vision
Counts	Aggregate mention counts for entities and themes
Relationships	Entity co-occurrence and relationship extraction

CAMEO Event Coding System

GDELT encodes events using the Conflict and Mediation Event Observations (CAMEO) taxonomy, which classifies actions into a hierarchical coding system:

CAMEO Root Code	Category	Examples
01	Make Public Statement	Press conferences, official statements
02	Appeal	Requests for cooperation, mediation
03	Express Intent to Cooperate	Trade agreements, diplomatic overtures
04	Consult	Meetings, negotiations, summits
05	Engage in Diplomatic Cooperation	Treaties, alliances, joint operations
06	Engage in Material Cooperation	Aid, loans, humanitarian assistance
10	Demand	Ultimatums, conditions, requirements
14	Protest	Demonstrations, strikes, boycotts
17	Coerce	Sanctions, embargoes, blockades
18	Assault	Armed attacks, bombings, occupations
19	Fight	Armed conflicts, military engagement
20	Engage in Mass Violence	Mass killings, ethnic cleansing

GDELT Processing Pipeline

Global News Sources (broadcast, print, web) in 100+ languages
    |
    v
Machine Translation --> English canonical text
    |
    v
NLP: Entity extraction, event coding, sentiment, themes
    |
    v
GDELT Event Database + Global Knowledge Graph + Entity Graph
    |
    v
BigQuery / REST API --> Prismatic OSINT Pipeline

Tone Scoring Model

GDELT's tone score measures the overall sentiment of an article on a scale from -100 (extremely negative) to +100 (extremely positive):

Tone Range	Interpretation	Typical Content
+10 to +100	Positive	Business growth, cooperation, achievements
+1 to +9	Slightly positive	Neutral-positive reporting
0	Neutral	Factual reporting without sentiment
-1 to -9	Slightly negative	Cautionary, concern-oriented
-10 to -100	Negative	Conflict, crisis, disaster, criticism

Integration with Prismatic

GDELT provides global media intelligence for the Prismatic platform's geopolitical risk assessment, brand monitoring, and entity tracking capabilities.

# Search for articles about an entity
{:ok, articles} = Gdelt.search_articles("Prismatic cybersecurity",
  mode: :article_list,
  max_records: 50,
  start_date: ~D[2025-01-01]
)
# => %{
#   total: 156,
#   articles: [
#     %{url: "https://example.com/article1",
#       title: "Czech Cybersecurity Startup Raises...",
#       source_country: "CZ", language: "English",
#       tone: 4.2, themes: ["CYBER_SECURITY", "BUSINESS"],
#       persons: ["John Doe"], organizations: ["Prismatic"],
#       locations: [%{name: "Prague", lat: 50.08, lon: 14.44}],
#       published: ~U[2025-06-15 10:00:00Z]}
#   ]
# }

# Get event records (CAMEO coded)
{:ok, events} = Gdelt.search_events(
  actor1: "CZE",
  event_type: "PROTEST",
  date_range: {~D[2025-01-01], ~D[2025-12-31]}
)

# Monitor entity mentions over time (tone timeline)
{:ok, timeline} = Gdelt.tone_timeline("Example Corp",
  resolution: :day,
  start_date: ~D[2025-01-01]
)

# Geographic analysis of events
{:ok, geo_events} = Gdelt.geo_search(
  latitude: 50.08,
  longitude: 14.44,
  radius_km: 100,
  theme: "CYBER_SECURITY"
)

# Theme search across all global media
{:ok, themed} = Gdelt.search_theme("TAX_FNCACT_SANCTIONS",
  source_country: "RU",
  start_date: ~D[2025-06-01]
)

# BigQuery direct query for complex analysis
{:ok, result} = Gdelt.bigquery("""
  SELECT SQLDATE, Actor1Name, Actor2Name, EventCode, AvgTone
  FROM `gdeltv2.events`
  WHERE Actor1CountryCode = 'CZE'
  AND SQLDATE >= 20250101
  ORDER BY SQLDATE DESC
  LIMIT 100
""")

# Volume timeline for entity media coverage
{:ok, volume} = Gdelt.volume_timeline("Example Corp",
  resolution: :week,
  start_date: ~D[2025-01-01]
)

Geopolitical Risk Monitor

The geopolitical risk monitor combines GDELT media intelligence with targeted news sources and social media signals for comprehensive entity risk assessment:

defmodule PrismaticIntelligence.GeoRisk.MediaMonitor do
  @moduledoc """
  Monitors global media for geopolitical risk signals affecting
  entities and regions relevant to Prismatic investigations.
  """

  def assess_entity_media_risk(entity_name) do
    tasks = [
      Task.async(fn -> Gdelt.search_articles(entity_name, mode: :article_list, max_records: 100) end),
      Task.async(fn -> Gdelt.tone_timeline(entity_name, resolution: :week) end),
      Task.async(fn -> NewsApi.search(entity_name, sort_by: :relevancy) end),
      Task.async(fn -> SocialSearcher.search(entity_name) end)
    ]

    [articles, tone, news, social] = Task.await_many(tasks, 30_000)

    {:ok, %{
      entity: entity_name,
      gdelt_articles: extract_count(articles),
      average_tone: extract_tone(tone),
      tone_trend: calculate_trend(tone),
      news_coverage: extract_count(news),
      social_sentiment: extract_sentiment(social),
      risk_signals: identify_media_risk_signals(articles, tone, news),
      country_exposure: extract_country_distribution(articles),
      theme_distribution: extract_themes(articles),
      monitored_at: DateTime.utc_now()
    }}
  end

  defp identify_media_risk_signals(articles, tone, news) do
    signals = []
    signals = if negative_tone_spike?(tone), do: [:negative_tone_spike | signals], else: signals
    signals = if coverage_volume_spike?(articles), do: [:coverage_spike | signals], else: signals
    signals = if sanctions_themes?(articles), do: [:sanctions_mentioned | signals], else: signals
    signals = if conflict_themes?(articles), do: [:conflict_context | signals], else: signals
    signals = if legal_themes?(articles), do: [:legal_proceedings | signals], else: signals
    signals
  end
end

Adverse Media Screening

For compliance-driven adverse media screening, GDELT provides the broadest global coverage of any single source:

defmodule PrismaticCompliance.Screening.AdverseMediaScreener do
  @moduledoc """
  Screens entities for adverse media coverage using GDELT
  global news analysis for KYC/AML compliance workflows.
  """

  @adverse_themes [
    "TAX_FNCACT_SANCTIONS", "CRISISLEX_CRISISLEXREC",
    "TAX_FNCACT_FRAUD", "KILL", "TAX_FNCACT_BRIBERY",
    "TAX_FNCACT_MONEY_LAUNDERING", "TERROR"
  ]

  def screen_entity(entity_name) do
    with {:ok, articles} <- Gdelt.search_articles(entity_name,
           mode: :article_list, max_records: 200),
         adverse <- filter_adverse_articles(articles) do
      {:ok, %{
        entity: entity_name,
        total_coverage: length(articles.articles),
        adverse_articles: length(adverse),
        adverse_themes: extract_adverse_themes(adverse),
        adverse_countries: extract_countries(adverse),
        risk_level: classify_adverse_risk(adverse),
        screening_date: DateTime.utc_now()
      }}
    end
  end

  defp filter_adverse_articles(articles) do
    Enum.filter(articles.articles, fn article ->
      Enum.any?(@adverse_themes, &(&1 in (article.themes || []))) or
      article.tone < -5.0
    end)
  end
end

Rate Limits and Access

Access Method	Rate/Cost	Features
DOC API	Free, rate-limited	Article search, tone timeline, geo
GKG API	Free, rate-limited	Knowledge graph search
BigQuery	1TB/month free	Full dataset SQL queries
Raw Files	Free download	CSV/JSON master files every 15 min

Data Access Methods

Method	Description	Best For
GDELT DOC 2.0 API	REST API for article and event search	Quick searches, integration
Google BigQuery	Full SQL access to all GDELT tables	Complex analysis, aggregation
Raw Data Files	Updated every 15 minutes as downloadable CSVs	Offline analysis, data lake
GDELT Analysis Service	Pre-built visualizations and dashboards	Quick visualization

BigQuery Tables

Table	Description	Update Frequency
`gdeltv2.events`	CAMEO-coded event records	Every 15 minutes
`gdeltv2.gkg`	Global Knowledge Graph records	Every 15 minutes
`gdeltv2.gkg_partitioned`	Partitioned GKG for efficient queries	Every 15 minutes
`gdeltv2.eventmentions`	Event mention details with source context	Every 15 minutes

Use Cases

Geopolitical Risk Assessment

Monitor political instability signals in target countries using CAMEO event codes
Track sanctions-related media coverage for compliance intelligence
Early warning system for crises affecting business operations
Country risk scoring based on conflict event density and sentiment trends

Entity Monitoring

Track global media coverage of investigation targets across 100+ languages
Detect sentiment shifts around companies and individuals before they become trends
Monitor for negative coverage that may indicate emerging risks
Build media exposure timelines for due diligence reports

Due Diligence Enhancement

Assess media reputation of companies during M&A review with adverse media screening
Identify adverse media coverage for KYC/AML compliance requirements
Cross-reference with ARES entity data for Czech companies
Verify claims against global media record for consistency

Supply Chain Intelligence

Monitor media coverage of key suppliers across all operating geographies
Detect regional instability that may disrupt supply chains
Track regulatory changes and compliance events in supplier jurisdictions
Feed media risk signals into NIS2 supply chain assessments

Market Intelligence

Track industry trends through theme and entity co-occurrence analysis
Monitor competitor media footprint and sentiment trajectory
Identify emerging markets through geographic event density analysis
Correlate media sentiment with market movements for financial intelligence

NewsAPI - Targeted news article aggregation from 80K+ sources
Social Searcher - Social media monitoring across platforms
EU Sanctions - European sanctions for correlation with media signals
OFAC - US sanctions for correlation with media intelligence
Crunchbase - Company intelligence to enrich media findings
UN Sanctions - Global sanctions for adverse media correlation

OSINT Core - Core OSINT framework and adapter layer
Prismatic Perimeter - Media intelligence in security ratings

Connect & Contribute

Created by Tomáš Korcak (korczis) | Open Source under GHL

GitHub | GitLab | LinkedIn | Contact
Developer Portal | Architecture | Meet the Creator

GDELT Project

Capabilities

Try GDELT Project

Overview

Data Coverage

CAMEO Event Coding System

GDELT Processing Pipeline

Tone Scoring Model

Integration with Prismatic

Geopolitical Risk Monitor

Adverse Media Screening

Rate Limits and Access

Data Access Methods

BigQuery Tables

Use Cases

Geopolitical Risk Assessment

Entity Monitoring

Due Diligence Enhancement

Supply Chain Intelligence

Market Intelligence

Connect & Contribute

Intelligence Network

Related Content

Agent Ecosystem

Applications

Doctrine

GDELT Project

Capabilities

Try GDELT Project

Overview

Data Coverage

CAMEO Event Coding System

GDELT Processing Pipeline

Tone Scoring Model

Integration with Prismatic

Geopolitical Risk Monitor

Adverse Media Screening

Rate Limits and Access

Data Access Methods

BigQuery Tables

Use Cases

Geopolitical Risk Assessment

Entity Monitoring

Due Diligence Enhancement

Supply Chain Intelligence

Market Intelligence

Related Sources

Related Platform Components

Connect & Contribute

Intelligence Network

Related Content

Agent Ecosystem

Applications

Doctrine