Automatic Identification of Persons by Using Keyboard Typing Metrics
The way of typing on a keyboard can vary from person to person. On the one hand, there are different typing strategies like the ten- finger touch typing system or two- finger touch typing system. On the other hand, the movement between two key presses differs tremendously and leads to so called features, which could give some hints who is currently writing. The idea of this project is to implement a key listener/logger and to calculate a variety of features. These features will be used to train a machine learning system in order to automatically identify the current typer.
The main goal of this project is to find out a person from a list of test writers with a meaningful success rate. In a first step, the system should be able to verify whether the currently typing person is the one s/he claims to be. In a further analysis it should be investigated, whether it is possible – and to which extent – to attribute the current typer to one out of a bigger possible set of persons.