Prepare to document each step you take either with screenshots or by recording your screen throughout the experiment. There is no need to record any audio unless you want to.
Pick a paragraph of text. Select the training stage ("Providing sample data") and type that text at least once for each typing method (both hands, only right, only left, single finger).
Then, switch to the testing stage ("Comparing to sample data") and type the text again at least once with just one of the methods. Click on "Analyze" and see if the result matches the typing method you chose.
Document the result, click on "Reset analysis" and repeat at least once for each of the other typing methods.
Then, experiment with how much you have to consciously alter your typing to make it impossible for the detection method to correctly determine the way in which you typed.
If you video-recorded the screen, please upload the video to a cloud service of your choice and provide the URL in your submission together with a written discussion of the process and your findings. If you took screenshots, please include a representative amount of them in your report.
Right-click to select "View source" on your browser to look at
the HTML code of the page that runs the example system. Then,
click click on the JavaScript link (file
extension .js) to look at the actual program
code. Find the code segment that records the keystrokes and
determine how the following to tasks are accomplished:
Document with code fragments and discussion in your own words what appears to be happening. You can provide fragments of the code to an AI model like ChatGPT or similar if you like to see what the AI thinks the code is doing and why.
Suppose that instead of varying how a single person types, different individuals provide training data for the system. (You could test this by typing with both hands yourself to train for the method "right hand only" and have a family member or a friend type normally to train for the methor "left hand only", if you like, and then testing whether the system is able to tell which of you is typing in the comparison stage by having one of you type, then analyze, and seeing whether the output is "right hand only" or "left hand only".)
The example system uses the time (in milliseconds) between pairs of consecutive keystrokes. Discuss under what circumstances (that is, what kind of a text does the person need to type and in what conditions) this way of identifying the individual who is typing could work. What are the limitations?
What benefits do you foresee in using this type of user authentication by itself? Are their additional benefits if it were combined with other authentication methods like passwords, facial recognition, voice recognition, etc.?
Search for scientific studies that identify people based on (rhytmic or other) behavioral phenomena, such as walking pace, pressure patterns while walking, seated, lying down, and others you can think of. Discuss potential applications you encounter for these methods in scientific literature or in commercial products, as well as original ideas that you yourself have on what could be done based on such recognition of individuals.