Improvements to keystroke-based authentication by adding linguistic context

Abstract

Traditional keystroke-based authentication methods rely on timing of and between individual keystrokes, oblivious to the context in which the typing is taking place. By incorporating linguistic context into a keystroke-based user authentication system, we are able to improve performance, as measured by EER. Taking advantage of patterns in keystroke dynamics, we show that typists employ unique behavior relative to syntactic and lexical constructs, which can be used to help identify the typist.