DATA COLLECTION FOR TURKEY: MANUAL CODING

The Turkey dataset provides trajectories of grassroots politics since 1970, including the activities of the formal proletariat, the informal proletariat, the left, the Islamists, and the Kurds. To create a reliable empirical source on grassroots politics, newspaper articles have been collected from the archives of the Turkish daily Cumhuriyet (Republic). Cumhuriyet was chosen because it is one of the major national newspapers in Turkey that has extensively reported news about grassroots political activities since the 1970s. Events reported in Cumhuriyet were obtained by manually coding news articles published every other day based on a systematic sampling strategy. This dataset is not intended to produce an exhaustive count of all, or even most, incidences of political events, since Cumhuriyet reported on only a fraction of the events that occurred. During social movements, newspapers report social events more than usual (Silver 2003). Each incidence that an event is reported is included in the dataset, which enabled differentiation of events in terms of their importance based on the assumption that newspapers report important events several times. 

You can find the most detailed information on event feature definitions and training data creation process in our  ⮕ Annotation Manual