I am a graduate student in Computational Linguistics with a keen interest in fake news detection. I propose to build a classifier for detecting objective vs. sensationalist news, that would rely primarily on non-lexical features to get at the stylistic differences between the two genres. It would be interesting to combine this model with other approaches to fake news detection. For example, do statements tend to co-occur across sensationalist news? Are there mismatches of dates, times, and numbers between objective and dramatic news? A prototype of a detector that I have built can be seen here: https://github.com/MashaIvenskaya/fake_news_analyzer.git


Masha Ivenskaya


  • Guy De Pauw
  • Vincent Merckx
  • Tom De Smedt
  • Walter Daelemans