The power of a network
Despite applications of network science running through biology and physics, computer scientists were some of the most attracted to network science. Dr John Lui was one of those researchers drawn in by the world of network science.
The basis of network science involves the analysis of nodes and their links to one another. These nodes can be anything, from individual cells in a colony of bacteria to children in a sociologist’s study of classroom dynamics. But when it comes to computer science, we, as users, are the nodes.
The study of these nodes and their interconnectedness reveals a lot about the spread of information through the network. For example, if A is friends with B and C, then it can be assumed that B and C must share some level of contact.
Using this information, users (whether it be companies spreading advertisements or media outlets spreading news) can more accurately predict where and how to release ads and news in a way that would spread most effectively.
Because of the intense complexity of the interconnectedness of users, what matters more than the sheer number of nodes is the way that they are connected.
Further complicating things in the world of social networks are the unbalanced distribution of links. Some nodes, for example a celebrity on twitter, have a disproportionately large number of links to other users. But when it comes to something like viral marketing, it’s more about the reach of the connections between users, not the number of links themselves.
For example, say you’re having a garage sale and want to hand out flyers. Of course you wouldn’t give one to each member of the family next door, but rather would spread them out to as many people as possible. It would be a waste to give a flyer to someone who would already hear about your garage sale from his or her sister.
The social aspect
Network science is also used by social media sites like Facebook to make more accurate suggestions for people you may know or targeted advertisement. If A and B are both friends with C, it could reasonably be assumed that A and B may know each other as well.
But for such large networks like Facebook, studying these networks involves analysing huge amounts of data. Searching through all of Facebook’s connections to find any friends of friends that may possibly be linked up would be incredibly impractical. To remedy this, Lui is working on an algorithm to more quickly scan these huge amounts of data for any potential nodes that could be linked.
However, things can get even more complicated once multiple networks are laid over each other, tracking users’ connections across Facebook and Twitter for example.
Lui highlights the popularity of mobile cell phone games in China, where game playing ‘celebrities’ are used to entice more players. But these gamers often play multiple mobile games, so the networks they are a part of often overlap, leading to interplay not just within the network but across a series of networks like graphs laid upon graphs.
By tracking the links between social media nodes, like game players or Facebook users, companies can make more educated decisions about how to spread advertisements or suggest links between users.
Tracing a target
But an understanding of the network system can also be used in reverse. After the outbreak of a virus, programmers can pick their way backwards through the links connecting different affected users to see from which node the virus originated.
It was network system techniques like this that enabled the tracing of the ‘patient zero’ in Hong Kong’s 2003 SARS epidemic. By working backwards through infected patients, the origin of the disease could be located.
But back then, it was done manually. Now, finding algorithms to more quickly and accurately digest and glean information from these complex networks is one of the main goals of researchers like Lui.
But these large collections of data require a large amount of storage. Creating new, efficient storage solutions is another of Lui’s goals. While large companies like Facebook and Google have the resources to store and process these huge quantities of data, small and medium sized companies interested in analysing this type of large network data are left without a safe way to secure their data.
After finishing his PhD, Lui started working at IBM, taking a departure from his PhD work and exploring new topics and tackling new issues. But after a few years there, he felt he needed a change. Lui says of himself that he gets bored of things after a few years, always curious and craving something new to learn about. While working at IBM as a researcher enabled him to be part of some really great projects, he much preferred the greater degree of freedom in his research that being a professor afforded, along with the chance to teach and cultivate the next generation of scientists.
It’s so much fun to be a professor in computer science, the field is just so dynamic
Professor Lui received his PhD in computer science from the University of California, Los Angeles in 1992. He worked at the IBM Almaden Research Laboratory in California before joining the faculty of CUHK in 1994. He was appointed to a professorship in the Department of Computer Science and Engineering in 2002.
To view Lui’s personal Croucher profile, please click here.