An adaptable scan-based text entry for mobile devices: Design, predictive modeling, and empirical evaluation

Authors

  • Sandi Ljubić University of Rijeka, Faculty of Engineering
  • Damir Arbula University of Rijeka, Faculty of Engineering
  • Krunoslav Smrekar University of Rijeka, Faculty of Engineering

Keywords:

human-computer interaction, text entry, assistive keyboard, mobile devices, predictive modeling, empirical evaluation

Abstract

This paper presents a highly customizable assistive on-screen keyboard for mobile devices, which supports several text entry methods based on row-column and bisection scanning techniques. Text entry can be accomplished using a zone based touch screen interface and/or via hardware keypads, involving configurable input control which can range from single-switch solution up to 5-key design. Apart from the presentation of a novel user interface, the paper contributions are as follows: development of movement models for all scan-based methods involved in text entry solution, computation of related upper-bound text entry speed predictions, and empirical investigation of their validity. In order to assess model predictions, a specific instance of row-column scanning technique was juxtaposed to bisection scanning principle in a user study involving 16 participants. Methods are evaluated against text entry performance, required workload, and general usability attributes. Although theoretical models predicted higher entry speed for bisection scanning, the results obtained from experiment demonstrated the row-column technique as significantly more efficient. This outcome discrepancy is specifically discussed by putting emphasis on factors that affect identified relation. 

Downloads

Published

2017-01-31