Online Social Media Networks and Cyber Security

Twitter botnets and spam attacks

Dr Shi Zhou’s team discovered two of the largest botnets on Twitter: (1) the Star Wars Botnet with >360,000 bots and (2) the Bursty Botnet with >500,000  bots.  Recently they uncovered a large-scale spam attack launched by the Bursty Botnet and traced down the botmaster responsible for both the botnet and the attack — perhaps the first time in this field of research. They also proposed a graphic method to detect fraudulent users and publishers in the online advertisement.

Critical hybrid epidemic spreading

His team introduced the theory of critical hybrid epidemics, where they used a meta-population model to explain how it is possible to combine basic, ineffective spreading mechanisms to produce extremely infectious and persistent epidemics. This theory explained the massive outbreak of Internet worm Conficker and led to the first mathematical model that reproduces the full course of HIV infection in the human body.

Information diffusion in social media

His team are currently using machine learning models to predict information diffusion in social media.

***  Here *** is a collection of news media reports on his research on Twitter bots detection and hybrid epidemic spreading.


Network Science and Complex Networks

Rich-club phenomenon

His team reported the rich-club pheromone in the Internet topology. Today, the rich-club coefficient has been widely used as a key metrics to quantify the structure of networks across a wide range of domains, including transport, social and biological systems.

Positive-feedback preference (PFP) model

His team proposed the Positive-Feedback Preference model, which is one of the most advanced Internet topology generators to date. The model provides insights into the evolution of the Internet topology, and it has been widely used in Internet research. The model can be generalised to create both disassortative and assortative networks.

Structural constraints in networks

His team introduced novel link rewiring algorithms to explore the structural constraints in complex networks, for example, they revealed that a network’s degree distribution has little constraint on the network’s rich-club structure.

Second-order assortative mixing in social networks

His team reported that many social networks exhibit the second-order assortative mixing, often significantly stronger than the first-order assortative mixing. This property revealed novel insights into how social networks were formed.

Network visualisation by the Bitmap of Sorted Adjacent Matrix (BOSAM)

His team introduced a novel network visualisation technique, called the bitmap of sorted adjacent matrix (BOSAM). It is an effective tool to study and compare complex networks.


Internet Routing and Cloud Computing

Inference of Internet AS relationships

His team introduced a novel approach to infer the Type of Relationships (ToR) between the Autonomous Systems (AS) on the Internet using the BGP Community attribute, including multilateral peering at IXPs. They produced so far the most accurate and reliable AS relationship data. They are currently studying the usage of multiple inter-domain border links in Internet multipath routing.

Universal cloud classification (UCC)

His team proposed the Universal Cloud Classification (UCC) as a universal scheme to uniquely identify each cloud provider, service and tenant across the global Internet. UCC IDs can be embedded in IP packet headers for traffic classification, which can support a variety of routing, QoS and security applications for data centres, inter-cloud routing and IoT. Cisco has filed a series of patents based on the concept of UCC.