In the past four weeks, we have been attuned to the internet in Germany. Every tweet, hashtag, public comment and blog was carefully examined to gauge what people in Germany think as they head to the polls.
Overall, we checked more than 150,000 tweets and over 40,000 online articles. That is a lot of data. How does one make sense of it? We could count and interpret every message one-by-one, which admittedly is very painful. Or we could team-up with leading experts who apply the power of artificial intelligence (hint: we did the latter).
Together with researchers of the social media lab of EPFL Lausanne, we focused on four economic topics that may determine the outcome of this year's Bundestag election: climate change, unemployment, housing and inflation. We then modeled every statement in real time to reflect a traditional poll – all without asking a single question.
As a matter of fact, if social media voted for the German parliament today (19 September), it would look like below.