Date of Award
Nathan Sturtevant, Ph.D.
In this thesis we explore alternative keyboard layouts in hopes of finding one that increases the accuracy of text input on mobile touchscreen devices. In particular, we investigate if a single swap of 2 keys can significantly improve accuracy on mobile touchscreen QWERTY keyboards. We do so by carefully considering the placement of keys, exploiting a specific vulnerability that occurs within a keyboard layout, namely, that the placement of particular keys next to others may be increasing errors when typing. We simulate the act of typing on a mobile touchscreen QWERTY keyboard, beginning with modeling the typographical errors that can occur when doing so. We then construct a simple autocorrector using Bayesian methods, describing how we can autocorrect user input and evaluate the ability of the keyboard to output the correct text. Then, using our models, we provide methods of testing and define a metric, the WAR rating, which provides us a way of comparing the accuracy of a keyboard layout. After running our tests on all 325 2-key swap layouts against the original QWERTY layout, we show that there exists more than one 2-key swap that increases the accuracy of the current QWERTY layout, and that the best 2-key swap is i ↔ t, increasing accuracy by nearly 0.18 percent.
Kirk, Amanda, "Improving the Accuracy of Mobile Touchscreen QWERTY Keyboards" (2018). Electronic Theses and Dissertations. 1512.
Received from ProQuest