Data field · Event

Media extraction confidence

A source-produced confidence value for automated extraction of an event mention; scale, direction, and model version must accompany coverage.

Who carries it ↓

1Data sources

1Providers

1Free to access

Sources that carry it

1 source

GDELT Project

GDELT 2.0 Event Database

Open worldwide machine-coded news-event metadata beginning with the GDELT 2.0 raw stream on 2015-02-18. The coordinated product comprises the 61-field Events table and 16-field Mentions table, published as 15-minute tab-delimited ZIP files and queryable in Google BigQuery. Events encode two actors, a CAMEO action, three geographies, media-coverage measures, ingest time, and first source URL; Mentions encode each article mention, extraction confidence, article position, source identity, tone, and translation metadata. It is a media-derived event index, not a verified incident-observation database. Event-table legacy coverage measures are fixed to the first 15-minute update; later discussion is emitted as new Mentions rows rather than revisions to the Event row.

Free APIFull file Global 20 fields

Questions it helps answer

Area political conflict and unrest events

Need Media extraction confidence at scale?

We compare every source that carries it on your own records — fill rate, freshness, definition — and bring the winner into production for you.