Automatic Extraction of Agendas for Action from News Coverage of Violent Conflict

Words can make people act. Indeed, a simple phrase 'Will you, please, open the window?' can cause a person to do so. However, does this still hold, if the request is communicated indirectly via mass media and addresses a large group of people? Different disciplines have approached this problem from different angles, showing that there is indeed a connection between what is being called for in media and what people do. This dissertation, being an interdisciplinary work, bridges different perspectives on the problem and explains how collective mobilisation happens, using the novel term 'agenda for action'. It also shows how agendas for action can be extracted from text in automated fashion using computational linguistics and machine learning. To demonstrate the potential of agenda for action, the analysis of The NYT and The Guardian coverage of chemical weapons crises in Syria in 2013 is performed.