WordTree: Results of a word prediction system presented thanks to a tree

Abstract

For mobility and portability reasons, written communication devices are getting smaller and physical keyboards are replaced with virtual or on-screen one (PDA, pocket pc, GPS, etc.). Text entry has become a tiring job for people (ordinary and motor disabled people) because of the low entry speed. A list of predicted words can be presented to reduce the number of clicks to complete the word. However, these techniques present a disadvantage: let's consider the situation where the word the user wants to enter is not presented in the list. He has then to continue his writing. The prediction system narrows the possibilities in order to display the wanted word. In this paper we present a novel interaction technique in which the user can select a substring of the word. The user clicks on any letter, and the substring to this letter is inserted in the text. We also present the simulated experiments to compare our system to the classical list. The results are then discussed and analyzed. © 2009 Springer Berlin Heidelberg.