EVENTI corpus

The EVENTI corpus collects news articles and stories annotated with temporal information at different levels (i.e. events, temporal expressions, signals and temporal relations) following the It-TimeML Annotation Guidelines (https://sites.google.com/site/eventievalita2014/file-cabinet). <p>The corpus is divided into 3 datasets: the Main task training data (103,593 tokens), the Main task test data (26,686 tokens) and the Pilot task test data (5000 tokens from newspaper articles written by Alcide De Gasperi in 1914). <p>The dataset has been used in the context of the EVENTI task (https://sites.google.com/site/eventievalita2014/) during the EVALITA 2014 evaluation campaign (http://www.evalita.it/2014). <p>