Last month, 5 Mobsters participated in the Global AI Hackathon organized by our very own Eduardo Barbaro. Our team consisted of 4 data scientists (Jeroen van den Hoven, Ruben Sikkes, Ali el Hassouni and Tim Elfrink) and one Java developer (Mustafa Dagher). Their challenge? Make news real again!
.png?width=600&name=pasted%20image%200%20(2).png)
Fake news is taking the Internet by storm, making countless headlines in its own right. The rise of the 24-hour news cycle and the endless opportunity to publish on digital channels has left us overwhelmed by the vast number of news articles we come across. All too often, these articles are fake, but we currently have limited tools to help us discern between real news and fake news. Even then, these tools are often viewed with skepticism because it’s unclear to us why certain news is labelled as fake.
Our team set out to overcome this skepticism with a machine learning solution that helped to explain why certain news was called out as fake. We addressed this by finding articles that cover the same topic as an article whose veracity is questioned. By building a dataset of features based on the questionable article, the website where it was published, and the author who published it and then comparing these features to the features of other articles, we can not only indicate if the news is trustworthy but also allow users to verify our claim.
-1.png?width=600&name=pasted%20image%200%20(1)-1.png)
During the 2 and a half days of the hackathon, we created a working project, which can be found in this github repo. This great result earned us second place in the challenge for the Amsterdam event. Worldwide prizes will be announced soon, and we’re hopeful to be recognized at the global level.
