Personal Identification by Keystroke Dynamics in Japanese Text Typing

Abstract

The Keystroke dynamics in Janpanses text typing are discussed. We developed CGI's typing software that collected the keystroke data on LAN. Experiments were conducted on a total of more than 100 subjects, and we classified them into the group who had the ability more than the beginner's class typing level, the mormal class level and the middle class typing level. After the feature measurements to Japanese texts had been extracted, identifications by the weighted Euclidean distance are calculated. For those analysis, we obtained high accuracy on the group who had the ability more than the norml class typing level.