Detecting False Identity through Behavioural Patterns

Combating identity fraud is prominent and urgent since false identity has become the common denominator of all serious crime. Typical approaches to detecting false identity rely on the similarity measure of text-based identity attributes, which are usually not applicable to falsely-defined and unknown identities. This paper presents a novel link-based approach that can efficiently overcome such barrier. Its experimental evaluation against well-known link-oriented and text-based methods significantly indicates the great potential towards an effective verification system.