Social Influence Diffusion and How Amo Tong is Managing it

Social media defines the modern era. To be online is to be informed and up-to-date. But the question arises then, how exactly does information online spread? How does it reach us, and why? The answer: social influence diffusion. Behind the scenes of any post, article, or advertisement is a complex map of algorithms, designed to facilitate the rapid spread of media to the public eye.

Guangmo “Amo” Tong, associate professor of computer and information sciences at the University of Delaware, is currently working on research in the Computational Data Science Lab that not only seeks to make sense of these diffusion networks, but also to harness them. This research has been Tong’s focus for 10 years, beginning when he was studying for his doctorate. He is supported by a $590,000 grant from the National Science Foundation (NSF) Core Programs small project grant, which will support his research until 2027. The project has three main aims: understanding how social influence diffusion tasks work, developing AI and machine learning models that can facilitate these tasks, and using these models to help manage the spread of information. This research, which bridges the gap between machine learning and social sciences, can help control the damage of misinformation and keep online platforms trustworthy.

Simply put, social influence diffusion is the study of how information spreads from one person to another or from a source to the public. Tong’s current project aims to analyze understand the patterns of how information moves and why those patterns exist. Hidden behind every direct action — whether clicking “retweet” or “add to story” is a vast expanse of coded formulas managing where and to whom the information goes next. Scaling these actions up from one person to a hundred, a thousand, a million, the formulas become exponentially more complex. “It offers new theoretical tools to study this type of learning and optimization problems,” Tong said. He and his team are working on the use of theoretical computing and mathematical models that use AI to better understand the patterns in these algorithms and exactly how they work.

One possible method to solve the proposed research problem

To understand how information spreads from one place to another, Tong is developing new machine learning models that can process the surplus of data that comes from information being moved around online. These new models rely on first principles to analyze exactly how the diffusion of information occurs, how quickly, and at what success rate. They are also used to simulate the diffusion process, providing data about the structures of social media and how various types of information are differently spread.

Sampling methods that will be explored in the project

 

This type of model analysis is necessary in Tong’s research because when applied, it can actually help address “social influence management tasks.” These tasks help AI models use their collected analyses of social influence diffusion to help minimize the spread of misinformation online and break the echo chamber effect. People tend to get stuck in cycles on social media where their existing beliefs are reinforced by only seeing posts or other content that agrees with them. Other perceptions and opinions become minimized, which can prevent the public from staying well-informed. “This project is to develop new algorithmic and machine learning tools to help us better solve these kinds of tasks,” Tong said. He hopes that the application of his research will reduce the impact of the echo chamber effect and offer greater clarity in information shared online.

A poster created by undergraduate students and Amo Tong showing the community patterns derived from social influence cascades.

Tong has a positive outlook on the long-term goals of his research—not only will it work to make social media more trustworthy, but it will also offer educational opportunities to K-12 students and UD undergraduates. One branch of the project is geared towards creating training opportunities for high school students, specifically those in underrepresented groups, and helping them become familiar with and potentially sparking an interest in computer science. Additionally, the project offers research opportunities to undergraduate students through the Vertically Integrated Projects program (VIP), with the Algorithms: Programming Contest and Research team.

Social media is a continuous flow of information which shapes the modern era. With the help of Tong’s research, however, we can work efficiently toward better understanding how information spreads on social media. By solving social influence management tasks with a better understanding of social influence diffusion, this research is projected to help slow the spread of misinformation and improve the social media landscape.

Article by Gabrielle London