Source & Confidence Criteria
What each classification tier means, and how many of the dataset's 11,887 records currently fall into each one.
Live breakdown
Computed just now, directly from events_portal_v1.jsonl in your browser — the same file the homepage timeline and map use. Not a cached or hand-typed figure.
What each tier means
- Documented — appears on a dedicated terrorist-incidents list (e.g. a Wikipedia "List of terrorist incidents in [year]" page). The strongest tier in this dataset, but still only as reliable as that source.
- — Wikidata links the item to terrorism-related categories without a dedicated list entry. Contextually connected, not independently confirmed here.
- Unverified candidate — surfaced from UCDP's organized-violence data via keyword matching. This is a lead, not a finding — treat it as "worth checking," not "confirmed terrorism."
- Unclassified — no classification was set at ingestion. Displayed as such rather than defaulted into any other tier.
Editorial status and review dates
Being direct about a limitation: individual dataset records do not currently carry a per-record "last reviewed by an editor" timestamp — the dataset is ingested from its three sources (see Methodology) with their classification preserved, not independently re-verified incident by incident. Where a record's classification_source field is present, it is shown in the record's detail view; where it is absent, the interface says so rather than inventing a date or reviewer. Published Reports, by contrast, do go through the editorial review described in the Editorial Policy before publication, and carry a real date and version.
See also: Methodology · Editorial Policy · Corrections Policy