Artificial Intelligence / Machine Learning for Understanding Misinformation in Social Media

Joseph Harris


Supervised by Alun D Preece; Moderated by Neetesh Saxena

Modern artificial intelligence (AI) has made huge advances due to machine learning on large amounts of data, usually collected from the internet. Current AI systems excel at making predictions or classifications for unseen inputs, based on previously-seen examples. Many researchers, companies and agencies are applying these sorts of data-driven predictive AI to the processing of social media data, e.g., to determine what content to push into a user's feed, or to work out "what's happening".

The goal of this project is to examine this kind of application of AI to perform situation understanding (SU) tasks on streams of social media data, e.g., from Twitter. There are two aspects to SU tasks: working out what's currently happening (insight), and working out what might happen next (foresight). The basic outcome of the project would be a demonstration system that provides insight; a more advanced outcome would be a system that provides foresight.

One particular area of current interest for SU on social media data is the widespread problem of misinformation, sometimes called "fake news". The preferred application area of this project would be some kind of "dashboard" to visualise current trends (insight) and possibly predictions (foresight) generated from analysis of large volumes of social media data collected with a focus on misinformation.

This project doesn’t require specific background knowledge but you’ll need to have a keen interest in AI and machine learning - you can choose the technique you want to apply, though you'd be encouraged to apply a deep neural network based approach - and to be willing to take on a research-focussed rather than software development project. Support and large datasets will be available from PhD students in the Crime & Security Research Institute DAIS group: https://www.csri-dais.org

Initial Plan (03/02/2020) [Zip Archive]

Final Report (19/05/2020) [Zip Archive]

Publication Form