Abdelzaher Pairs with Ji, Kitware to Uncover Pathways of Influence on Social Media

8/25/2022 9:59:41 AM

In a world where social media have relatively recently altered many people's media consumption habits through new levels of accessibility and immediacy, Illinois Computer Science professor Tarek Abdelzaher seeks to alter the limited understanding of such a change.

Headshots of Illinois CS professors Tarek Abdelzaher (left) and Heng Ji.
Illinois CS professors Tarek Abdelzaher (left) and Heng Ji believe that efforts to better understand the way information is disseminated can lead to better interactions.

His most recent project, funded by the Defense Advanced Research Projects Agency (DARPA), will include a collaboration with fellow Illinois CS professor Heng Ji and Kitware – which is a custom software solutions company using open source technology and advanced artificial intelligence (AI).

This project, entitled “Multigranularity Influence Mapping in Social Terrain,” will span the next year and a half and received $1 million in funding from DARPA.

“Ultimately, what we would like to do through efforts like this is to help people live more peacefully and interact better with each other,” Abdelzaher said. “I feel that so much of the conflict we face in the world stems from not being able to understand each other better. In part, this stems from social media as an information source because it’s changed the way we perceive information – as many of us now only read part of the whole story.

“What we as researchers can look into and better understand is the manipulation of information, in the hopes that we can not only help the population empower themselves to better information but better connect themselves to others.”

A unique computing perspective allows Abdelzaher to take on an important role in this research effort, which he broadly calls social and information dynamics.

Stemming from his time as an embedded systems researcher, Abdelzaher delved into sensor networks – and still maintains that interest through another focus on the Internet of Things (IoT). His background allows for a unique perspective regarding social media’s influence over the way in which people receive and seek out information.

To him, social outlets like Twitter became some of the largest sensor networks in the world, because they served as entities through which “people were telling us about what they see around them in the world.”

With that as a background, Abdelzaher has sought to research the way social media and information dissemination works.

In this project, his focus boils down to two major prongs.

First, is graphing information flow.

“This Pathways project stems from a notion that disinformation campaigns start on a network where only a small sliver of a population picks up on it,” Abdelzaher said. “But once it gains acceptance there, then it moves on and gains traction elsewhere. Graphing that action as it becomes something more mainstream will help us understand how bizarre concepts gains acceptance.”

Second, Abdelzaher will attempt to define information flow through social networks in a similar way to that of physical sensors.

For example, models exist that track patterns of physical sensors, like a digital thermostat. Abdelzaher said it is fully understood how altering a thermostat sets off a series of actions that make the new setting a physical reality.

That, however, is not understood in social media. If an information flow becomes diverted or enhanced, how does that occur and what are the effects?

To best accomplish all of this, Abdelzaher paired with Ji – whose expertise in natural language processing, especially regarding multimedia multilingual information extraction, will serve an integral role in understanding how social information flows.

Kitware, which processes visual content, helps in compliment of Ji’s focus on textual information.

This project also ties in well with two other efforts Abdelzaher currently leads.

The first, also funded by DARPA, earned a $5.8 million award to characterize how different foreign populations respond to influence campaigns as a first step towards the development of effective countermeasures.

The second studies how information spreads and what the mathematical models are that dictate the action.

“The importance of understanding social media consumption habits has actually become one of my favorite topics, because it’s something that promises impact,” Abdelzaher said. “We know this touches a lot of people’s lives, but to best understand it we need to come together differently to study it. It hasn’t been all too common in the past for researchers in the social and cognitive sciences to pair with computer scientists.

“I’m excited about these projects I’m included in, because we are looking seriously at a problem that touches a lot of people’s lives.”