Northeastern University College of Computer and Information Science - CCIS

"Two recent accomplishments are near and dear to my heart. First, I've worked on role discovery in complex networks. A complex network is an ecosystem with the nodes (i.e., entities) playing multiple functional roles over time. Automatically discovering the underlying roles of nodes in a network is a challenging problem. Given a network, intuitively two nodes belong to the same role (or behavioral cluster) if they have similar structural behavior. Examples of roles that can be automatically determined from data include'broker,' 'clique-member,' 'periphery-node,' etc. Roles enable numerous novel and useful tasks, such as transfer learning, sense making, and anomaly detection. "In 2012, together with colleagues at LLNL and CMU, I introduced RolX (short for Role eXtraction).