Social scientists and data journalists work with event data to investigate political conflicts and instability, social movements and protest mobilisation. The primary source for event data is textual news.
The collection of event data is closely related to the task of event extraction. The task is to identify in the text information about events and their key properties – roughly speaking, who did what where and when.
In computational social science, there has been a recent interest in automating extraction of complex event data using statistical learning techniques. Most effort has focused on event data for the public protest domain, which accounts for a large part of politically interesting events. Demonstrations, riots, industrial strikes, assassinations, terrorist attacks are examples of some common types of protest events. In addition to more commonplace information components (e.g. event type, location, time), researchers wish to categorise the claims and grievances of the participants in a protest action or know how many participants there were.
The POLCON research group studies recent protest activity and political conflict in Europe. We have been developing an annotated corpus of English-language newswire reports for the purpose of automated extraction of protest event data. Our work borrows from protest event analysis for the social sciences and linguistic annotation for NLP alike. Our goal is to advance the state of the art in automated extraction of politically relevant events.
The POLCON project is financed by the European Research Council (ERC-2013-ADG, project no.338875).
The project is based at the European University Institute, Florence. It is headed by Hanspeter Kriesi, Stein Rokkan Chair of Comparative Politics.
The sub-project "Semi-automated Media Content Analysis" is located at the Department of Computational Linguistics of the University of Zurich.