---
license: cc-by-4.0
pretty_name: Geotone Global News Signals
language:
  - en
tags:
  - news
  - geopolitics
  - gdelt
  - time-series
  - international-relations
size_categories:
  - 10K<n<100K
---

# Geotone Global News Signals

Daily media-intelligence series derived from the [GDELT Project](https://www.gdeltproject.org/) across 65+ languages, published by [Geotone](https://geotone.news) - live dashboards for every series in this dataset, plus a [free JSON API](https://geotone.news/api-docs) and an MCP server for agents (`https://geotone.news/mcp`).

## Files

- **topic_daily.csv** - daily coverage share (% of all GDELT-monitored articles matching each topic's fixed, published query by title) and average tone for tracked geopolitical topics (conflicts, flashpoints, risk themes). Columns: `topic_slug`, `topic_name`, `date`, `coverage_share_pct`, `matched_articles`, `tone`.
- **country_daily.csv** - daily share of world news coverage that is ABOUT each of 197 countries (extracted article locations - media attention about a place, not press output from it), with average tone, CAMEO-coded event counts located in the country (`protest_events`, `diplomacy_events`, `coercion_events`), and the coverage-origin split (`articles_known_home`, `articles_local` - how much coverage about the country came from outlets based in it). Columns: `iso2`, `country_name`, `date`, `coverage_share_pct`, `tone`, `protest_events`, `diplomacy_events`, `coercion_events`, `articles_known_home`, `articles_local`.
- **relations_daily.csv** - daily bilateral co-coverage (share of articles mentioning both countries) for ~50 strategic pairs. Columns: `country_a`, `country_b`, `date`, `co_coverage_share_pct`, `tone`.
- **source_countries.csv** - mapping of 4,300+ news domains to each outlet's home country (from GDELT source-country data plus Geotone's archive). Columns: `domain`, `home_country_iso2`, `mapping_source`.

Tone is GDELT's standard document tone (positive minus negative words as a share of all words; typically -10 to +10). War coverage reads negative for everyone, so trends matter more than levels.

## Method and caveats

Full methodology with formulas: https://geotone.news/methodology. The entire range is computed with one consistent methodology (bulk GKG 15-minute files, title matching). Coverage measures media ATTENTION, not ground truth.

## License and attribution

CC BY 4.0 - cite "Geotone / GDELT" with a link to https://geotone.news. Refreshed monthly; the live API updates every 2 hours.