Empirical networks demonstrate a level of self-organization which is difficult for random networks to model. Certain distinctive features of these real networks motivates the development of classifiers capable of separating real from random data. Literature in the field of network classification typically revolves around whole-network classification as a tool for studying network structure and developing realistic null models. In this thesis we present a classifier capable of operating at the node level, which enables the detection of fake nodes; opens possibilities for artificially expanding networks; and significantly optimizes highly accurate whole-network classification. We present this classifier through the lens of social networks, but suggest potential applications in any field where networks are involved.
"},{"label":"Degree","value":["Bachelor of Arts with Departmental Honors"]},{"label":"Administrative Note","value":"Drupal Node ID: 24173