User Classification by Keystroke Dynamics using Text Retrieval Methods

To verify the identity of users, the majority of computer systems employ conventional authentication schemes such as Personal Identification Number (PIN), password and token. Over the years, these schemes have become less robust as they can be guessed, cracked, stolen or shared. Biometric-based authentication methods have positioned themselves as better options as they do not suffer from the same limitations as conventional methods. The reason being that biometrics exploit the uniqueness of a subject with regards to what they are or how they behave. In this paper, we investigate the use of a behavioural biometric, Keystroke Dynamics (KSD) as a means of identity verification for online persons. We present the proof-of-concept of a novel keystroke dynamics authentication method that is based on text retrieval concepts and methods. We test our algorithm on a classification task and achieve promising results. Furthermore, we show experimentally that a person's typing behaviour is susceptible to the environment in which they are typing.