The construction and maintenance of interactive user interfaces have been simplified by the development of a generation of software tools.The tools range from window managers, toolkits, and widget sets to user interface management systems and knowledge-based design assistants.However, only a small number of the tools attempt to incorporate principles of good design.They offer no help with decisions regarding the variety of input devices and methods available.In this paper we briefly describe a methodology for interaction technique selection based on natural physical analogs of the application tasks.Special emphasis is given to the physical characteristics of input devices and the pragmatics of their use.The methodology is incorporated in a software environment named Toto which includes knowledge acquired from a variety of disciplines such as: semiotics, ergonomics, and industrial design.Toto also incorporates a set of interactive tools for modifying the knowledge and for supporting the selection of natural interaction techniques.A two phased design process (matching followed by sequencing) is embedded in the Toto rule base.Examples of the use of Toto tools are provided to illustrate the design process.

Background: Traditional paint programs can be criticized on the grounds that they provide fewer degrees of freedom than conventional artistic media. Mouse or tablet input devices usually provide only (x, y) spatial information. Thus most paint programs allow continuous control of the path of a stroke, but render it with a single “brush shape” of fixed size and orientation and use menus, either stationary or appearing, for such tasks as changing “brush” shapes. This produces two distinct discontinuities for the artist. First is a physical discontinuity: stopping the act of drawing and making a brush selection using a different physical device, or using the same device to point at a menu selection. A second discontinuity is the cognitive one involved with deciphering the menu contents and making a selection. Although the second may be minimized by using iconic menus, these interruptions of the drawing act make it difficult or impossible for the artist to maintain the kind of continuous control over his or her medium that is required. A particular example of this level of control is the continuous and immediate control that an artist maintains when sketching with a piece of charcoal or a pencil. By manipulating the implement, the artist can change the shape of its contact point with the drawing surface. By changes in pressure exerted with the implement against the surface, it is possible to control both the darkness (value) and width (weight) of the line. All of these factors are controlled by the artist in real time and with continuous feedback. Both Greene [2] and Strassman [4] have presented methods for capturing more of the artist's “gesture” in the rendered stroke. However, Greene's technique depends on a special-purpose input device and Strassman's does not work in real time. Hill [3] describes a two-handed paintbrush with the off hand controlling a valuator that varies the 'brush width. The technique described here represents another way to provide such control to the user of an interactive paint program. It uses off-the-shelf hardware, is easy to implement, and works in real time. As a single-handed technique, it takes advantage of the artist's skill with existing media.

The George Washington User Interface Management System (GWUIMS) has been designed as a test bed for comparing user interface models, as a tool for rapidly prototyping highly interactive graphic user interfaces, and as a vehicle for investigating the applicability of knowledge-based technology to user interface design. The GWUIMS was designed and implemented using the object-oriented programming paradigm and consists of a variety of object classes representing different levels of abstraction, Responsibility for lexical, syntactic, and semantic levels of both input parsing and feedback are distributed throughout these classes. We include a description of the GWUIMS and a brief scenario to demonstrate its capabilities. A description of the implementation is followed by a discussion of the future application of knowledge representation techniques and the evolution towards an intelligent assistant for the user interface designer.

Despite the current interest in user-computer interfaces, the design of a good interface remains to a great extent an art, with much argument over guidelines and principles for interface design. Pertinent information, scattered throughout the literature of psychology, graphic design. linguistics, hardware design, and under the general umbrella of computer science, is only gradually being gathered into survey publications for application by computer scientists [1,4,12,15]. Our purpose is to design a specification language which will serve as a vehicle for the design of and experimentation with user-computer interfaces. The specification language not only defines the external characteristics of the interface, but can also be analyzed to determine whether the interface meets a set of generally-accepted human factors guidelines. We believe the language is general enough to be used with principles of interface design other than those we have proposed.