ICEWS: Aggregated Event Time Series from News Articles

ICEWS event aggregations transform automatically coded news events into numerical time series data, typically at a monthly level. The data is created by filtering events by parameters like country or sector and aggregating them via counts or intensity averages. Jennifer Lautenschlager of Lockheed Martin (Canada) created this dataset, which is periodically updated.

Use Cases

• Modeling political instability trends based on aggregated event intensity.
• Analyzing sector-specific conflict patterns based on filtered event counts.
• Forecasting international relations dynamics based on monthly country-level event aggregations.
• Creating derived indicators for social science research based on filtered news event parameters.

Strengths

• Data is derived from the ICEWS program, a known source for event data from news articles.
• Aggregations allow for creating numerical time series from complex event data.
• The dataset is planned for periodic updates.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count, file formats, and specific update frequency are unknown.

Provenance

Source

Lockheed Martin (Canada), via the ICEWS program.

Collection Method

Events were automatically extracted from news articles by the BBN ACCENT event coder and then aggregated.

Freshness

Update frequency is described as 'periodic', but the last update date is unknown.