Influence Maximization in Networks: Spread Models and Optimization Methods
Time and place
12:30 PM on Thursday, April 19th, 2018; NAC 6/113
Gwen Spencer (Smith College)
Abstract
Sociologists first introduced the study influence in social networks during the 70's. This inspired mathematicians and computer scientists to propose a number of models about the spread of information and behavior in networks. Given a particular spread process, a natural question is how to choose a small set of individuals who are highly-influential (in the sense that their behavior change could cause a large cascade of behavior in the network). This optimization problem has attracted a great deal of theoretical study, but many "scientific facets" of the problem remain open. I'll mention several formal models of spread, mention algorithmic results were they are known, and point out contemporary challenges suggested by human studies in behavioral economics and noisy inputs. Better understanding of this planning problem has the potential to fuel industry efforts in viral marketing, increase the reach of campaigns to promote healthy behaviors, and inform management practices about how to make cooperation more robust.
