Description of field of research:

Fake news detection methods need to be capable of identifying events from an unseen event or a different domain, since the emerging events would differ from that of existing data of the system. However, recent studies are confined to extracting domain or event-specific features which are not transferable across domains or adaptive to a new event. We will develop a continuous GNN based graph domain adaptive learning that transfers knowledge from a labelled source to the unlabelled target domain. GNN has become very powerful in representing news, and news-social context connections as different types of graphs for fake news detection. Continuous GNN naturally avoids over-smoothing and enjoys better interpretability. In this regard, we aim to exploit continuous GNN for domain adaptive fake news detection. 

School

Computer Science and Engineering

Research areas

Social Media, Cyber Security

  1. This project will be conducted under the supervision of Dr Jiang (expertise in misinformation detection) and Prof Jha (expertise in distributed systems). You will be able to get help from PhDs and honours students who are also working on fake news detection.
  2. You will have access to powerful GPU servers/workstations to conduct the research.

The project will generate novel research outcomes, and represent significant advancements over existing methods and techniques. The research output will be written as a research article and submitted to a good venue.