Paper Reading #7: Performance Optimizations of Virtual Keyboards for Stroke-Based Text Entry on a Touch-Based Tabletop

Performance Optimizations of Virtual Keyboards for Stroke-Based Text Entry on a Touch-Based Tabletop

By:
Jochen Rick

Presented at UIST 2010

• Jochen Rick has a Bachelors and Masters in Electrical Engineering and a PhD in Computer Science all from the Georgia Institute of Technology. He is currently an Assistant Professor at Saarland University.

Summary
Hypothesis
The Author hypothesises that in cases of equipment where a larger touch screen keyboard is the most viable text input option (such as Interactive Tabletops) and there is no available tactile feedback to register key strokes the best possible solution is a stroke-based virtual keyboard as opposed to a more traditional tap based keyboard. The Author states that such a keyboard would allow for quicker user inputs with an optimized keyboard.

Methods
To test and thus prove his hypothesis, the author had eight participants complete a series of 1152 stroke sequences through 4 points in a certain order, where each stroke had a beginning, a middle and an end. The participants had previous experience with the interactive tabletops and were asked to stand for the period of the testing so as to have better access to the tabletop. They targeted completing the tasks as quickly as possible. They did these tests for a large number of keyboard layouts as listed in the paper.

Results
The stroke based typing showed a minimum improvement in speed of 0% with Quikwriting and a maximum improvement almost 41% with Square ATOMIK. Speed increases were substantial for all other keyboard layouts tested, thus showing the obvious superiority of Stroke based typing.

Contents
The paper dealt with first proving the superiority of Stroke based typing and then later with developing an ideal virtual keyboard layout that maximized stroke based typing speeds. Using the results gathered from participant testing the author was able to see that OptiII was the layout that allowed for the greatest words per minute on stroke based typing. Using the results it was concluded that it would take far too long (10^20 years) to use an exhaustive algorithm to find the optimized keyboard layout. The author was however, able to use an annealing process followed by hill climbing based upon Square ATOMIK and Hex ATOMIK to come up with 2 optimised layouts that were 0.5% faster in the square category and 7% faster in the Hex category.

Discussion
Considering the fact that various phones are starting to use "stroke based" keyboard input (most notably the Swype feature for Android keyboards) it is exceedingly nice to see a paper about trying to optimising the keyboard layout. While technically the paper is limited to interactive tabletop applications, surely the basic concepts of stroke based typing are carried across platforms. While it is true that most users are highly used to and comfortable with QWERTY, I feel there would be widespread interest in any optimised keyboard that was put out there due to the speed offered by it.