Dataset Card for dwb2023/gdelt-event-2025-v3

This dataset contains global event records from the GDELT (Global Database of Events, Language, and Tone) Project, capturing real-world events and their characteristics across the globe through news media coverage.

Dataset Details

Dataset Description

The GDELT Event Database is a comprehensive repository of human societal-scale behavior and beliefs across all countries of the world, connecting every person, organization, location, count, theme, news source, and event across the planet into a single massive network. The database records what happens in every corner of the world, containing over a quarter-billion event records in over 300 categories.

• Curated by: The GDELT Project
• Funded by: Google Ideas, supported by Google Cloud Platform
• Language(s) (NLP): Multi-language source data, processed into standardized English format
• License: All GDELT event data is available for free download and use with proper attribution
• Updates: Every 15 minutes, 24/7

Dataset Sources

• Repository: http://gdeltproject.org/
• Paper: Leetaru, Kalev and Schrodt, Philip. (2013). "GDELT: Global Data on Events, Language, and Tone, 1979-2012." International Studies Association Annual Conference, April 2013. San Francisco, CA.
• Documentation:
  • http://data.gdeltproject.org/documentation/GDELT-Event_Codebook-V2.0.pdf
  • https://www.gdeltproject.org/data/documentation/CAMEO.Manual.1.1b3.pdf

Uses

Direct Use

• Monitoring global events and conflicts in real-time
• Analyzing international relations and diplomatic activities
• Tracking social movements and protests
• Studying media coverage patterns
• Research in political science, international relations, and social sciences
• Crisis early warning systems
• Geopolitical risk assessment

Out-of-Scope Use

• Real-time emergency response (due to potential reporting delays)
• Individual-level surveillance or tracking
• Definitive source of truth for events (should be cross-referenced with other sources)
• Prediction of future events without careful consideration of limitations

Dataset Structure

The dataset consists of tab-delimited files with 61 fields per event record, including:

1. Event Identification

   • GlobalEventID: Unique identifier for each event
   • DATEADDED: Timestamp in YYYYMMDDHHMMSS format
   • Day, MonthYear, Year: Various date formats

2. Actor Information (for both Actor1 and Actor2)

   • ActorCode: CAMEO-coded actor identifier
   • ActorName: Name of the actor
   • ActorCountryCode: 3-character country code
   • Various actor attribute codes (ethnic, religious, type)

3. Event Details

   • EventCode: CAMEO action code
   • EventBaseCode: Root event category
   • QuadClass: Primary event classification (Verbal/Material Cooperation/Conflict)
   • GoldsteinScale: Event impact score (-10 to +10)
   • AvgTone: Average tone of coverage (-100 to +100)

4. Geographic Information

   • Multiple geographic fields for each actor and action
   • Includes country codes, feature IDs, and lat/long coordinates

Dataset Creation

Curation Rationale

GDELT was created to capture and analyze global human society in real-time, providing a platform for understanding global behavior through media coverage. It processes news media from around the world to identify and code events using the CAMEO (Conflict and Mediation Event Observations) coding system.

Curation Method

• Prefect based python extract script: https://gist.github.com/donbr/704789a6131bb4a92c9810185c63a16a

Source Data

Data Collection and Processing

• Continuous monitoring of news media worldwide
• Automated processing using natural language processing
• Event coding using CAMEO taxonomy
• Geographic coding using full-text geocoding
• 15-minute update cycle
• Machine translation for non-English sources

Who are the source data producers?

Primary sources include:

• International news media
• Web news
• Broadcast transcripts
• Print media
• Various online platforms

Personal and Sensitive Information

While the dataset primarily focuses on public events and public figures, it may contain:

• Names of public figures and officials
• Locations of public events
• Public statements and actions
• Media coverage details

Bias, Risks, and Limitations

1. Media Bias

   • Over-representation of English-language media
   • Varying media coverage across regions
   • Event selection bias based on news worthiness

2. Technical Limitations

   • Machine coding errors
   • Translation inaccuracies
   • Geographic coding challenges
   • Duplicate event reporting

3. Coverage Gaps

   • Limited coverage in media-restricted regions
   • Potential missed events in less-covered areas
   • Varying detail levels across events

Recommendations

1. Users should:

   • Cross-reference critical events with other sources
   • Consider media bias in coverage
   • Account for regional differences in coverage
   • Use appropriate statistical methods for aggregation
   • Be aware of potential duplicates and coding errors

2. Best Practices:

   • Aggregate data over appropriate time periods
   • Consider confidence scores in analysis
   • Use multiple GDELT fields for validation
   • Account for regional and temporal variations in coverage

Citation

BibTeX:

@inproceedings{leetaru2013gdelt,
  title={GDELT: Global Data on Events, Language, and Tone, 1979-2012},
  author={Leetaru, Kalev and Schrodt, Philip},
  booktitle={International Studies Association Annual Conference},
  year={2013},
  address={San Francisco, CA}
}

APA: Leetaru, K., & Schrodt, P. (2013). GDELT: Global Data on Events, Language, and Tone, 1979-2012. Paper presented at the International Studies Association Annual Conference, San Francisco, CA.

Dataset Card Contact

dwb2023