A MULTIMODAL APPROACH TO MAN-MACHINE DIALOG WITH THE UNIX OPERATING SYSTEM

The Unix system interface is often criticised for its complexity, and for the length of the training period necessary to acquire adequate proficiency with the system. This paper presents a multimodal approach to man-machine dialogue with Unix, so as to paliate some of the difficulties users have in acquiring rapid, effective use of the operating system. The multimodal interface presented here combines graphical, mouse-based, voice-driven and keyboard dialogue pathways. Interface system software architecture is described, together with the role of the various communication modes, their complementarity, and an appreciation of the difficulties in conceiving a multimodal dialogue architecture of this kind. Finally, details are explored in respect of the user request interpretation mechanism, which comprises a syntactic-semantic analyser and task model.