Improved word list ordering for text entry on ambiguous keypads
Jun Gong, Peter Tarasewich, Ian Scott MacKenzie · 2008
We present a design methodology for small ambiguous keypads, where input often produces a list of candidate words for a given desired word. The methodology uses context through semantic relatedness and a part-of-speech language model to improve the order of candidate words and, thus, reduce the overall number of keystrokes per character entered. Simulations yield improvements in text entry speed of about 10% and reductions in errors of about 20%, depending on the keypad size. We describe a user study with 32 participants entering text on a keypad with letters arranged on three keys. Entry speed was 9.6% faster, and error rates 21.2% lower, compared with standard disambiguation, as found on mobile phones.