[Hackathon] High Tech Storytelling: How To Predict Elections, Using Robot Journalism?
,Deutsche Welle Innovation aims to find out, based on a collaboration with Ludwig-Maximilians-University (Munich) and AX Semantics.
The (virtual) event on December 17th will provide an opportunity for journalists, scientists and students to learn and play with the next wave of technologies usable in newsrooms. The focus will be on prediction models and robot journalism, supported by design thinking as a method to develop impactful new offerings on digital platforms.
In the hackathon participants have a chance to learn about prediction models (PollyVote). How accurate are such models? How are they set-up prior to an election? Another topic will be the use of the much debated “robot journalism” where automated programs write timely reports when values change. But, who will actually write the base texts for such robots, who determines the variables when the robot will sent of the next update? Or, in other words, can newsrooms train robots to write cool texts?
As data is increasingly available, prediction models using statistical methods are becoming more agile. This was first demonstrated by the big public success of Nate Silver’s blog “FiveThirtyEight”. Using a statistical prediction process where newly available data was ingested into the prediction models Silver managed to predict the outcomes of the US presidential elections in the 2008 and 2012. Silver “called out the outcomes of 49 of 50 state in the 2008 election” (Source: Wikipedia).
Robot journalism, on the other hand, is a much debated area, where software would write updates on topics where human editing is not really needed. Current examples are reports about local sport games, the rise and fall of public stocks can already be created in this way. Elections are just another area where text automation could provide additional benefits. So far only very few editors have actually worked with such software. The goal of the hackathon is to let journalists explore the benefits and the current limitations, by being able to write base texts and learning about the variables that would trigger off a new article or alert.
The initiative is funded by Volkswagen Stiftung, which is a foundation owned by the German state of Lower-Saxony. The goal of their recent call for proposals aims to let scientists work closer with newsrooms, adding in-depth know-how on technologies and scientific methods. A one day event where the winning projects where announced asked the right question: “Is this still journalism or already science?” As newsrooms have multiple options to deploy big data analytics, there is a growing need to check whether the methods to reach conclusions are actually reliable and meaningful. The project of DW Innovation with colleagues from AX Semantics and Ludwig-Maximilians-University aims to do just that.
More details about the planned hackathon and a registration form can be found here:
