Discovering links between political debates and media

Abstract

Politics and media are heavily intertwined and both play a role in the discussion on policy proposals and current affairs. However, a dataset that allows a joint analysis of the two does not yet exist. In this paper we take the first step by discovering links between parliamentary debates in a political dataset and newspaper articles in a media dataset. Our approach consists of 3 steps. We first discover topics discussed in the debates. Second, we query a newspaper archive for relevant articles using a combination of debate elements: dates, actors, topics, and named entities of the debates. Finally, we discover links, represent them in RDF, and make them available for download. An evaluation of various versions of this approach shows that the topic detection adds to the quality of the discovered links, as well as the use of the semantic structure of the debate, such as headers and a division into smaller events. © 2013 Springer-Verlag Berlin Heidelberg.