Sampling representative phrase sets for text entry experiments

Performance comparisons of phrase sets and presentation styles for text entry evaluations

“…Given that two new phrase sets have recently appeared that may be more suitable in certain situations, such as mobile text [7] or special use-cases [5], it is reassuring that the MacKenzie and Soukoreff [4] phrase set provides similar performance.…”

“…[4,7]), and on how well the phrases model the text end-users are likely to compose (e.g. [2,4,5,7]). In this paper we instead focus on the actual empirical performance implications of phrase sets and presentation styles.…”

“…We investigated five phrase sets which were selected based on the criterion that the set, or the source for the set, must be publicly available (this ruled out the children's phrase set [2] and the Twitter and Facebook n-gram phrase sets [5] MOBILEEMAIL This phrase set was released by Vertanen and Kristensson [7] in 2011. This phrase set contains phrases drawn from genuine mobile emails in the Enron email corpus.…”

“…NGRAM This phrase set was released by Peak and Hsu [5] in 2011. It consists of 4-grams sampled from the Enron email corpus.…”

“…MacKenzie and Soukoreff [4] raise the issue of external and internal validity of phrase sets and argue that phrase sets should be standardized. Paek and Hsu [5] follow up on MacKenzie and Soukoreff's argument of the importance of using representative phrase sets and propose a method for sampling representative phrase sets from large bodies of text. Recently, we released a new phrase set based on genuine mobile emails that have been validated for memorability [7].…”