@book{JRC46168, editor = {}, address = {East Stroudsburg (United States of America)}, year = {2008}, author = {Pouliquen B and Steinberger R and Deguernel O}, isbn = {}, abstract = {The Europe Media Monitor gathers and aggregates a daily average of 40.000 newspaper articles in currently 42 languages. It is used as a source of information for various services of the European Commission, for institutions in the European Union (EU) member states, as well as for any world citizen via the publicly accessible website http://press.jrc.it/NewsExplorer/. To overcome the 'information overflow' problem, it was decided to group similar articles per day and per language into clusters and to link daily clusters over time into 'sto-ries'. A story automatically comes into existence when related groups of articles occur within a 7-day window. The automatically computed stories are listed on the public website. A visualisation tool shows the development of the story over time and allows users to explore and browse past and present news items be-longing to each story. While cross-lingual links across 19 languages for indi-vidual news clusters have been displayed online for some time, the newest de-velopment is (to date still experimental) work on linking entire stories across languages. The evaluation of the monolingual aggregation of historical clusters into stories and of the linking of stories across languages shows that the overall performance is rather good for a fully automatic system. }, title = {Story Tracking: Linking Similar News over Time and Across Languages}, url = {}, volume = {}, number = {}, journal = {}, pages = {49-56}, issn = {}, publisher = {The Association for Computational Linguistics}, doi = {} }