School of
Information Technology and Electrical Engineering

Speaker: Mr Tong Chen (Rocky)
Seminar Date: Tue, 28/11/2017 - 14:00
Venue: 78-631; Data Science MMLab
Host: Prof Xue Li

Seminar Type:  PhD Confirmation Seminar


With the rapid development of social media platforms, anomaly detection on social media becomes an important problem to tackle because anomalies exert threat to both cyber security and social stability. In regards to the task of anomaly detection on social media, we decompose it into three aspects: early rumor detection, opinion spamming behavior detection and outlier detection. Early rumor detection refers to the detection of rumor events at their emerging stage with users’ comments related to the same topic. Opinion spamming behavior detection requires us to distinguish abnormal behavioral patterns of malicious users from genuine users who deliberately give fake reviews towards products or services. Outlier detection is the discovery of rare samples within a large dataset, while in the context of social media, such rare samples usually provide evidence of malicious users or misinformation. With the goal of anomaly detection on social media, these three research topics complement each other, thus becoming indispensable components of our ultimate research target.

Following our work of early rumor detection on social media, we further seek the insight of opinion spamming behavior detection on social media via our current research on user behavior modelling. Inspired by Hawkes process, we design a novel deep attention based recurrent neural networks (RNNs) to jointly model users' behavior sequence as well as the contextual information. Furthermore, our model learns the mutual infectivity between different types of user behaviors to precisely predict users' future behavior with the occurrence time, and preliminary experiments have validated the effectiveness of our proposed model. Hence, the seminar will demonstrate both the necessity and feasibility of anomaly detection on social media.


Mr. Tong CHEN achieved his Bachelor Degree of Engineering on Software Engineering from Northwest A&F University, China in 2016. He is now a candidate of Doctor of Philosophy in his first 12-month-period under the joint supervision of Prof. Xue LI (principal), Dr. Lin WU (University of Queensland) and A.Prof. Jun ZHANG (Swinburne University of Technology).