Key points are not available for this paper at this time.
ISI is involved in user interface research aimed at bringing together multiple input and output modes in a way that handles mixed mode input (commands, menus, forms, natural language), interacts with a diverse collection of underlying software utilities in a uniform way, and presents the results through a combination of output modes including natural language text, maps, charts and graphs. Our system, Integrated Interfaces, derives much of its ability to interact uniformly with the user and the underlying services and to build its presentations, from the information present in a central knowledge base. This knowledge base integrates modes of: the application domain (Navy ships in the Pacific region, in the current demonstration version); the structure of visual displays and their graphical features; the underlying services (data bases and expert systems); and interface functions. The emphasis in this paper is on a presentation planner that uses the knowledge base to produce multi-modal output. There has been a flurry of recent work in user interface management systems (we list several recent examples in the references). Existing work is characterized by an attempt to relieve the software designer of the burden of handcrafting an interface for each application. The work has generally focused on intelligently handling input. In our paper we deal with the other end of the pipeline - presentations.
Arens et al. (Fri,) studied this question.