Identifying and Investigating Global News Coverage of Disasters and Terrorist Attacks

This release contains two parts of data: (1) 6,468 annotations of news articles to see if each is an actual match to the event identified by large language model; labels are binary -- 1 for actual match and 0 otherwise;(2) thousands of observations of (event_country, reporting country) pair with corresponding relational country factors, average death number per event, and average number of news that the reporting country have for each event in the event_country.