Protecting Data in Social Networks
In the era of social media, the integrity of social network data is paramount. MSI PIs Tianxi Li (Assistant Professor, Statistics) and Xuan Bi (Assistant Professor, Information and Decision Sciences ) are working on a project called “Advancing Security in Social Networks: Strategic Data Release Mechanisms to Counteract Adversarial Threats,” that emerges as a tool for both innovation and security. Tackling the dual challenge of privacy protection and utility retention in social data, this research introduces groundbreaking algorithms designed to safeguard against adversarial threats. At its core, the project aims to obfuscate sensitive information, protecting user identifies while ensuring the data's analytical value remains intact. By employing a sophisticated machine-learning methodology, the team seeks to understand and preserve the intricate web of relationships that define social networks, all while minimizing the risk of privacy breaches. This initiative not only addresses the immediate need for enhanced security measures but also lays the groundwork for research in several open areas. The project’s success promises to fortify the foundations of privacy in social networks, ensuring the safe use of these vital resources for research, marketing, and policy-making. Through a meticulous approach and dedication to privacy and utility, this work emphasizes the potential of data science in creating a safer, more secure digital world.
This project recently received a DSI Small Seed Grant. The Seed Grant program is intended to promote, catalyze, accelerate, and advance U of M-based data science research so that U of M faculty and staff are well prepared to compete for longer term external funding opportunities. The program was updated in Summer 2024 to include three focus areas: Foundational Data Sciences; Digital Health and Personalized Health Care Delivery; and Agriculture and the Environment. This project falls under the Foundational Data Sciences focus area.
Complete information about seed grants can be found on the DSI website.
Image description: Demonstration of the data protection procedure: Random split is first introduced to partition the dyads (into red and grey sets). Then node-embedding is applied, followed by a random perturbation mechanism. The regenerated entries are secure from the adversarial inference of membership and identities. A further improved security is introduced by summarizing a number of independent replicates to return the final A˜.
