NATracker

Method

A record, not a prediction. NATracker organises published reporting about attacks in Nigeria for research and awareness. It is not built for real-time safety decisions, and publication does not prove completeness or accuracy. Open the cited sources before relying on a record.

How a report becomes a record

  1. 1

    Discover

    Read configured RSS feeds and source pages.

  2. 2

    Read

    Extract available article text and source metadata.

  3. 3

    Qualify

    Require a date and a state. Use the exact town or LGA when known, or state the wider area or state-level precision when not.

  4. 4

    Reconcile

    Compare incident fingerprints to reduce duplication.

  5. 5

    Publish

    Show reporting status, location precision, casualty precision and source evidence with each record.

Reporting status

Confirmed
Two or more cited sources, or an official statement plus a cited publisher.
Developing
Credible reporting exists, but important details or outcomes may still change.
Unconfirmed
A single source or incomplete corroboration. Use with additional caution.

What a record contains

Incident
Title, factual summary and incident date.
Location
State, LGA, town, surrounding area or state-level precision when available.
Attribution
The group named in reporting, or unknown. Never inferred for certainty.
Human impact
Killed, injured, abducted and displaced. Values are labelled as exact, estimated, reported ranges, unknown, or not reported.
Evidence
Publisher, source title, URL and review state.
Record history
Created and updated timestamps when returned by the database.

Known limits

Coverage is incomplete
Source availability, redirects, rate limits and missing local reporting create gaps.
Dates and places can conflict
Publication date is never treated as the incident date without supporting information.
Numbers can be exact or approximate
Figures may change. Ranges and estimates are displayed as such; unknown is reserved for reports that identify an impact without a reliable figure.
Post-April counts are not a trend
Lower stored counts after April can reflect earlier strict date, location and casualty rules rather than fewer incidents.
Automation has limits
Source discovery can miss events or confuse related stories. Publication is not a reliability percentage.

Corrections. Every record keeps its supplied sources and timestamps. When a source is corrected or a report changes, figures are reviewed against the cited source before the record is updated.