Fake news mitigation architecture – Automatic reporting of incidents
Dec 8th, 2016 by Jesper Kråkhede
With the previous posts, I presented on a high level how fake news could be mitigated. However, there will always be some news that passes the filters, always some bully that has the technical skills to beat through. The mitigation for that is a reporting system. The architecture contains an automatic reporting component using text analysing to identify death threats, violence etc. that is punishable by law. With a fully authenticated user we can I real time collect as much information that is needed for the police to be able to start an investigation. This makes it possible for a user that receives such a comment to be able to quickly report it to authorities.
When it comes to fake news this is most often not illegal but should still be reported. Lets say that a major newspaper starts to report articles from Natural News, a well know source of fake news within alternative health care or Granskning Sverige, that has been exposed to be one of the largest troll factories geared at undermining Sweden. Most of the post under anonymous account and with few possibilities to follow-up on their articles. With authenticated users, I could disallow such articles to appear in my flow, and if they would appear I could easily send a notice to the publisher that this article by this users is possibly fake.