Automatic Annotation Approach of Events in News Articles

Daily, several news agencies publish thousands of articles concerning many events of all types (political, economic, cultural, etc.). The decision-makers find themselves in front of a great number of events, a few of which concern them. The automatic treatment of such events becomes increasingly necessary. Thus, we propose a machine learning-based approach that allows annotating news articles to generate an automatic summary of the events. We propose a new similarity measurement between events and we validate our approach by the development of the "AnnotEv" system.