Many network datasets allow us to observe not just the connections among a set of people, but also the activities and behaviors they engage in at different points in time. We consider two broad categories of questions on this kind of data. First, can we identify "hot spots" --- situations in which many people engage in the same activity at approximately the same location, the same point in time, or the same "region" of the network? Second, can we infer something about the processes that give rise to this kind of coordination, as changes in behavior spread between people across the links of the network? We provide examples of how the exploration of these questions can enable us to build annotated timelines and maps from raw data,assess the power of groups in recruiting new members, and trace the pathways that information follows as it travels through a large network. The talk will draw on joint work with many colleagues: Lars Backstrom, Dan Cosley, Dave Crandall, Cristian Danescu-Niculescu-Mizil, Dan Huttenlocher, Gueorgi Kossinets, Lillian Lee, Jure Leskovec, David Liben-Nowell, Brendan Meeder, Daniel Romero, Sid Suri, and Duncan Watts.
Keynote Lecture
Copyright. All Rights Reserved.