Optimizing Keyboard Configuration Using Single and Multi-Objective Evolutionary Algorithms
Ahmer Khan, Kalyanmoy Deb · 2023
An English language keyboard configuration involves 30 keys - 26 alphabets and four punctuation marks: comma, period, semi-colon, and question mark. Based on the sequence of letters that frequently appear in typing activities, one or more keyboard configurations will be optimal for a specific performance metric. Here, we consider the cumulative distance between two consecutive keystrokes to type out a large piece of text as the optimization objective, as it directly relates to the typing efficiency. The shorter the distance the fingers needs to move between strokes, the faster is the typing speed. When considering such a scenario, researchers often focus on a specific single objective one at a time and come up with the respective optimal keyboard configuration. The complete dissimilarity of these optimal, yet esoteric, keyboards from the current in-practice QWERTY configuration makes the new optimal keyboard configuration harder to learn and deploy on a larger scale. Hence in this study, we look at a multi-objective version of such a problem and consider two objectives - maximizing typing efficiency and maximizing similarity to QWERTY configuration. We make use of the NSGA-II algorithm and produce a Pareto set of keyboard configurations ranging from the most efficient to the most similar configurations. The range of solutions can potentially provide a design innovation path for gradually moving from the popular QWERTY configuration to more efficient configurations as a series of improvements.