The Use of Data Mining to Indicate Virtual (Email) Bullying

In this paper we describe how a novel application of Data Mining techniques can be used to provide the engine for a tool which can be used to identify email correspondence which may be an early indication of virtual bullying or harassment. The approach which we have taken makes use of linear discriminant approaches to classify normal, and non-normal, style of email correspondence for each sender. This change in email style could be used to provide an early indication of virtual harassment/bullying.. This approach has great potential for use in large organization where it is often appears to be hard to identify unacceptable information transmission between two colleagues. By identifying indicative behavior it should be possible to instigate company anti bullying processes in a more timely manner. This should ensure a more effective work force in terms of work place efficiency and reduction of stress related absence resulting from harassment or bullying. We show that it is possible to improve the identification of the number of sender signature styles contained within the email pages, irrespective of the number of pages concerned. The implications of the use of SSSs, and ASSSs, for identification of future email interactions are discussed.