A knowledge-based multi-modal interface

The construction of the user interface is extremely costly, with some estimates as high as 60% of the development of a software system. Compounding the problem is the large number of interface modalities which exist (graphics, forms, natural language, etc.) and which the end-user would like to have available. The author chose a Navy command-and-control briefing task currently performed manually at the headquarters of the commander in chief of the US Pacific Fleet in Hawaii. The Integrated Interface (II) project developed a model of the domain of the naval application and of the capabilities of the various interface modalities available. With a relatively small collection of rules, a multi-modal interface to the task was generated. It responds to user requests dynamically, taking into account the specific data to be displayed and its representation of the semantics of the domain and the user´s request. Briefing maps which take hours to produce manually are generated by II in minutes