Structured design of an adaptive human-computer interface

Human beings and computers communicate in vastly different ways, so attempts to communicate with each other often break down. The key to improved human-computer interaction is the design of the human-computer interface, hereafter called the user interface. This research explores the design of adaptive user interface software systems, that intelligently and dynamically adjust their behavior to suit the goals and expertise of individual users.; A new design methodology for adaptive user interfaces has been developed. This methodology applies to a general class of applications for which interaction can be modeled abstractly as a network of high-level states. It identifies and addresses major facets of adaptive user interface design, and is consistent with principles and practices of software and knowledge engineering. At its core lies a highly-structured, three-pronged design process which considers the user, the interface, and the target application. The methodology is demonstrated through the design and implementation of the N-CHIME interface to support development of simulation models written in the HSL programming language.; A task-oriented analysis of potential HSL users and the modeling and simulation domain yielded an abstract interface model. This model consists of a state transition graph describing sequences of high-level activities that users undertake during an interactive session. Task plans are defined as directed paths through the graph to represent activity sequences conducive to achievement of specific goals. Each high-level interaction state is surrounded by supportive contextual structures, such as specific operations and assistance to be made available to the user. One such structure is a set of rules which control how the interface adapts to the user while in that state. Rule construction is based on an adaptation model consisting of adaptation criteria and methods which utilize individualized user models.; The abstract interface model is transformed directly into an object-oriented interface design. The state graph and its attached supportive structures form a network which controls both the flow of interaction and the display of visible objects with which the user interacts directly through keyboard and mouse devices. A new user interface software architecture is presented which details the roles and relationships of run-time interface objects.