A research team led by Chae Wi Song, a researcher at the Faculty of Distance Education, has developed an authentication system in which users are authenticated by analysis of keystroke dynamics.
Biometric authentication systems identify individuals by their physiological features (fingerprint, skin, retina, iris, etc.).
Each individual’s keystroke pattern is unique and a unique profile can be constructed by their typing speed, key press release timing, pressure applied, and finger positions on a keyboard.
The user authentication system based on this keystroke pattern is appealing for many reasons: simple and transparent authentication and no need for any extra equipment for feature capture.
The research team chose key press time and key release time among many keystroke patterns.
Key press time and key release time enable extraction of four features ― key hold time, key press latency, intervals between key press and release, and release and press. They used random forest algorithm for user authentication. The basic idea is to construct many trees using random vectors sampled from a data set.
This system guarantees fair online exams by automatically blocking several kinds of cheating attempts such as taking exams in place of examinees, keying in answers read by others, etc.
© 2021 KumChaek University of Technology