Interpreting Gestures For Text Entry On Touch Screen Devices

Text entry on touch screen devices is often performed through Soft keyboards. One of the latest research trends is to abandon the traditional tapping interaction in favor of more natural gesture-based interactions on these keyboards. The interpretation of the gestures is performed through sketch-based techniques.In this paper we present the sketch-based technology related to the interpretation of the gestures needed to enter text through KeyScretch, a novel text entry method which has been proven to be more efficient than the traditional tapping-based method. The method enables text entry through the input of gestures associated to word chunks. The gestures are guided by a menu which appears around a key of the keyboard as soon as the user presses it.The proposed procedures improve the interpretation of the gestures and consequently increase the accuracy of the method. This improvement is measured through a simulation in which the sketch-based recognition is compared to the target-based menu item selection through an estimation of the error rate.