Detection of Roles of Actors in Social Networks Using the Properties of Actors' Neighborhood Structure

Members of the societies form many complex social relations that are often the subject of study in order to classify the members according to the roles they play in the considering society. Classification of roles played by the actors can be very informative because it gives us the complete overview of the network allowing for understanding how the network is organized and how it is going to behave in front of certain external or internal events. In the paper we present the roles identification method that can be useful in social environments about which we have limited knowledge mostly constrained to local relations in regard to the considered set of actors. The method uses the properties of social networks that can be derived from neighborhood of actors whose role we are going to determine. The method uses social relation graph, roles identification algorithm and a set of relations pattern graphs supporting the roles identification process. The method has been tested with local society of Internet e-mail users of one of the Faculty of Gdansk University of Technology to classify the role of each e-mail user as spammer or regular user.