Mapping World Events

Abstract

This paper describes new methods used for mapping news events gathered from around the world. Web based graphical map displays are used to monitor both the real time situation, and longer term historical trends. The results are derived from a synthesis of world events based on 20,000 news reports collected from the Internet each day by the Europe Media Monitor (EMM) (http://emm.jrc.org). EMM has invented a powerful method of rapidly classifying multilingual articles by matching weighted combinations of phrase and word patterns. The articles are cross-classified according to the countries mentioned and to general themes like “Conflict” “FoodSecurity” “Natural Disaster” “Ecology” and so on. The Alert system runs 24 hours per day keeping continuous hourly statistics on article populations. This then allows to generate real-time graphical presentations of “news maps”, time series of Indicators, and animations of crisis developments.

In a second step, a more detailed geolocation of news articles based on town names identification in text and a reference gazetteer is performed. This has been coupled with a daily “Top News” clustering algorithm and results in 24 hourly News distributions for each EMM alert topic. This technique can also link clusters through different language versions and track stories in time. Geospatial representation of news developments enables decision makers to better assess a complex situation in a fast and objective fashion.