A Cross-Domain Analysis for Actor Identification in Multiple Networks

Abstract

This paper proposes a technique for actor identification in any individual’s multiple communications network. This problem is specially an important problem for personal overview, intelligence gathering and countering terrorism but can also be applied on automatic actor identification for social network managements of any individual. The technique performs the correlation of different features using the feature set consisting of basic features, text segment feature and authorship attribution features. The features are extracted from datasets of three different ego-centric networks (Facebook, Messages and Email), collected over a period of three months. The preliminary result shows effectiveness of the technique in identification of actors in a communication network