Text entry is essential for interacting with humans as well as computers. Physical keyboards are still the main input method in desktop settings and typing is an important skill in many jobs. However, many people never undergo a formal typing training but learn to type on a computer keyboard autodidactically: they develop different typing strategies simply by using the keyboard for different computer tasks. While in principle, self-taught typists can reach speeds of 80 or 90 words per minute (WPM) similar to trained typists, many people cannot even type at half that speed. There is no way for these people to improve their typing performance, since all training programs are designed to teach users the 10-finger touch typing method. However, switching to a new strategy would require many hundreds of hours of dedicated practice before they can even reach the same typing speed again, let alone exceed it.
Our goal is to develop an intelligent tutoring system that is independent of the finger-to-key mapping. It first ana-lyzes the user’s typing performance and weaknesses from their keystroke dynamics and identifies optimal exercises for them to improve their typing speed. Thus, the program helps users to improve their existing strategy instead of requiring them to learn the 10-finger typing system.
The project requires to come up with methods for modeling the current skill of the user from their keystroke dy-namics and selecting the best exercises (e.g. Bayesian knowledge tracing).
The goal is to develop an intelligent tutoring system for teaching people how to improve their typing speed. It does not require users to learn the touch typing system. Instead it analyzes the user’s typing problems from keystroke dynamics and chooses optimal exercises.
• Literature review on intelligent tutoring systems and typing on physical keyboards
• Develop different types of exercises to improve typing speed
• Analyze keystroke dynamics to identify areas for improvement and select suitable exercises
• Develop a system that automatically selects exercises, displays them to the user, and analyzes the results
• Test the system in a user study
Dr. Anna Maria Feit
intelligent tutoring systems
CLS Student Project (MPG ETH CLS)
ETH Organization's Labels (ETHZ)
Information, Computing and Communication Sciences
Engineering and Technology