2/5/2018 1:38:03 PM
CS @ ILLINOIS Professor Tarek Abdelzaher is leading a group of researchers studying how information propagates through social media and the effects that information has on people’s beliefs as it moves.
The project started in late 2017 when the group was awarded a five-year, $4 million grant from the Defense Advanced Research Projects Agency, or DARPA, under its SocialSim program.
The spread of information by social media and how it shapes the understanding of events—and even influences their outcomes—has been a hot-button topic in many major news stories the past few years, from the influence of terrorist groups such as ISIS to the 2016 presidential election.“How do we react to events? Are social media creating echo chambers where we only hear what people who will think like us say, or do we become more accepting of other people’s beliefs?” Abdelzaher said, articulating key questions he hopes to address. “There are some fundamental questions about societies and information that we can answer by doing this research.”
Abdelzaher is leading the work, joined by Abel Bliss Professor of Engineering Jiawei Han; David Nicol, who is the Franklin W. Woeltge Professor of Electrical and Computer Engineering and Director of the U of I’s Information Trust Institute; and Professors Boleslaw Szymanski and Gyorgy Korniss at Rensselaer Polytechnic Institute. Szymanski and Korniss are known for their accomplishments in social network modeling and analysis.
Abdelzaher says the project brings together Han’s expertise in data mining, Nicol’s extensive knowledge of high-speed simulation, and his own deep experience with embedded systems, together with RPI’s social network modeling expertise. They plan to create a tool called Social Cube to model, analyze, and simulate social media information propagation and how it both affects and is affected by people’s beliefs.
Much like you might detect and track a tank moving through a field or a river valley by the sound it creates and the vibrations that propagate from the tank through the ground, the team will model the reverberations caused by information as it travels through a platform such as Twitter to develop the new information physics.
“An event happens and a response propagates through social media.” Abdelzaher said. “And I want to study it the way we studied sound, the way we studied vibrations. It’s just a new type of signal propagating in a new type of medium.”
A number of factors have to be understood in order to model how people behave as information relays moving those signals—everything from the degree to which someone likes or agrees with a given piece of information, how much they trust the source of the information, and whether those and other factors make it more or less likely that the person consuming that information will share it with others, Abdelzaher said.Nicol will apply his experience in assessing the reliability of the simulations used in that analysis.
“I am interested in developing means of assessing how trustworthy these models are in capturing characteristics of information propagation,” Nicol said. “The challenge is that the causes of this propagation might be unseen. For example, communication between people by cellphone not captured in social media. Yet that communication can affect behavior, and those effects will be seen in the data from which the models are derived.”
And Han will provide the means of sifting the vast amounts of data that are the project’s raw material.“I hope to develop new technologies to mine phrases from text, recognize entities and relationships from social media and build a dual structure: a heterogeneous information network and a multi-dimensional text cube to facilitate Social Cube analysis,” Han said.
Abdelzaher’s team began by modeling information spread on Twitter and Instagram, and will work on modeling other platforms.
Abdelzaher says he has long been fascinated by the way social media is fundamentally changing the ways people communicate, in giving anyone the ability to spread their ideas but also the ability to both find others with similar ideas and block or ignore anything that may be unfamiliar or uncomfortable.
“People have started filtering information differently, people have started finding other people who think like them. How is that affecting how information spreads?” he asked.
“We want to study that - how do beliefs form in that future where social media rule? How do they collide? How do they stabilize? I think those are fundamental, very, very interesting questions about that new reality that didn’t exist 10 years ago.”