Journalist-in-the-Loop (Round 5)

United Kingdom Project type: prototype University of Sheffield


Journalists follow social media alongside traditional channels however, social media is rife with rumours and disinformation. The problem is that fully manual rumour detection and analysis does not scale, while pure machine learning-based approaches often fall short on accuracy. Even though large amounts of new evidence accumulates continuously as a side effect of journalist analysis of online misinformation, journalists cannot use this at present to improve the accuracy of their tools over time.

The solution

This project will build an open-source web service for rumor analysis, where journalists and algorithms work together and learn from each other in an intuitive, scalable, modular and efficient manner. Journalists can both benefit from, and seamlessly help, improve the performance of machine intelligence algorithms for rumor analysis. From a technological perspective, the assumption to be tested is that human-in-the-loop machine learning can be used to underpin an easy to use misinformation analysis interface for journalists.