Method of User Identification Based on Keystroke Behavior and its Application
Wang Yu-ji · 2015
Most researches on keyboard behavior have been limited in experimental environment,which force user to input a specific string or limite the number of participants,so the result of research can't apply into real production environment.To overcome this limitation and resolve the problem of the unbalanced samples between positive and negative and the problem of lacking negative samples,ten million keystroke behaviors were analyzed,and a method for user identification was introduced.The duration and the interval of keystroke describing the user's behavior are used,and the similarity of different keystrokes is measured by Mahalanobis distance.The result of experiment shows that our method gets an accuracy of 80%.