Dynamics of information spread on networks

Date: 
Monday, May 2, 2011 - 9:21am

UCSB COMPUTER SCIENCE DEPARTMENT PRESENTS:

Monday, May 16, 2011
11:00 AM – 12:00 PM
Computer Science Conference Room, Harold Frank Hall Rm. 1132

HOST: Amr El Abbadi

SPEAKER: Kristina Lerman
USC, Information Sciences Institute

Title: Dynamics of information spread on networks

Abstract:

The recent proliferation of data about human behavior in social media offers a natural laboratory for studying social dynamics, specifically, the structure of social networks and the flow of information on them. I begin by describing the framework for quantitative analysis of information cascades on networks that captures the details of their microscopic dynamics, while also allowing us to compute their macroscopic properties, such as size and diameter. By applying this framework to the popular social media site Digg, we have observed information cascades that spread fast enough for one initial seed to affect hundreds of people, yet end up spreading to less than 1% of the entire Digg social network. This conflicts with our understanding of epidemic processes on networks, which suggests the existence of an epidemic threshold below which epidemics die out and above which they spread to a significant fraction of the network. We demonstrate that the mechanism for social contagion on Digg deviates from standard social contagion models, and this severely curtails the final size of information cascades. In the second part of the talk, I argue that many dynamic processes on social networks, such as the spread of information or disease, are fundamentally different from the random walk, and show how these differences impact the choice of metrics used for network analysis. I show that Alpha-Centrality, which mathematically describes spreading processes, leads to better, empirically verifiable insights into the structure of online social networks than random walk-based metrics such as PageRank. I conclude with detailed investigation of the properties of Alpha-Centrality.

Bio:

Kristina Lerman is a Project Leader at the USC Information Sciences Institute and holds a joint appointment as a Research Assistant Professor in the Computer Science Department at University of Southern California. She joined ISI in 1998, shortly after receiving a Ph.D. in physics from University of California at Santa Barbara. Her research builds on the fascination with complex systems she developed at UCSB, which she applies to the study of complex networks and social dynamics.