Inferring user intention in stylus input systems

点击翻译按钮获取中文摘要

The personal computing revolution of the past two decades has moved to a new phase where the keyboard-mouse interface no longer dominates humans' interactions with computers.  While the long-term future may involve conversational interfaces, retinal projection, and computer vision recognition of hand gestures, the hardware technology currently being adopted most rapidly is directed toward pen or stylus input, on both bibliographic surfaces (i.e. personal-sized, normally horizontal, e.g. PDAs, Tablet PCs) and epigraphic surfaces (i.e. shared, normally vertical, e.g. SMARTBoard, iRoom). One attractive aspect of stylus-based interaction is the fluidity of interaction it appears to promise. Interactions between a pen and a computer application can include data input, annotations, free-hand drawings, the selection of content, and the interaction with user interface objects such as menus and buttons.  Users, using a stylus-based entry device such as a tablet computer, want an interaction experience that moves fluidly between these operations.  Pen-tablet computers, however, have typically not delivered the fluidity of interaction expected by users. These programs require frequent mode switches when switching tasks, and result in slow, pedantic input for all but the simplest of tasks. This research seeks to add intelligence to the activities performed by a stylus-based computer system when capturing and analyzing the actions of the user.  To do this, this research aims to build models of both the content created by the user with a pen and current actions performed by the user with the pen.  Given these models, the research aims to then develop computer programs that analyze these current actions in context to reason about the intention of the user when performing actions in interfaces.  The end result will be tablet interfaces in which users move fluidly though actions in the application interface. Onus is placed on the software application to understand users' goals, rather than on the user to understand the software application.