Modeling Email Networks and Inferring Leadership Using Self-Exciting Point Processes

• Activity Number: 470
• Type: Contributed
• Date/Time: Wednesday, August 6, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Defense and National Security
• Author(s): Eric Fox*+
• Companies: University of California, Los Angeles
• Keywords: conditional intensity ; epidemic-type aftershock sequence models ; Hawkes process ; IkeNet data ; social networks

Abstract

Self-exciting point process models are used to model a social network dataset consisting of email communications between officers at West Point Academy during a one year period beginning in May 2010. The models appear to adequately capture major clustering features in the data, and features of the model may be used to predict perceived leadership status within the social network. The results suggest that such models may be used for simulation, understanding basic properties of, and perhaps even prediction of underlying leadership status of social communication networks.