Political Mashup Position Paper

A mashup is a web application that combines data from more than one source into an integrated experience. A po-litical mashup as we envisage it brings together political in-formation produced by political parties (their promises and actions: election manifestos, party websites, parliamentary proceedings) and information on the reception of political promises and actions (news as well as user generated con-tent from blogs, discussion fora and reactions to online news-stories). Such a political mashup will enable novel histori-cal research, creates challenging opportunities for computa-tional linguists, and offers valuable and innovative testing-grounds for information retrieval researchers. We aim to collect, standardize in digital format, and inte-grate the four types of political data listed above. Integra-tion of these data sources consists of creating connections between them along three natural dimensions: temporal happening in the same period, being about the same event; political actors involving the same parties and/or mem-bers of parties and/or other politically influential ac-tors; political issue being about the same political issue. The elements in these dimensions link the data items to each other. The main challenge is to detect and normalize these elements in the textual data that we have. For this task we will use existing event- and news-story clustering, as well as named entity recognition and normalization techniques. The innovative and challenging aspect of our task is the type of data we use: rough and dirty user-generated content on the one hand and camouflaged content created by politicians on the other