One of the critical parts of a tutoring system is its user interface (UI), which must neither constrain an author in developing lessons, nor impede a student during practice. A system providing training over the Web must also address issues of interface transport, providing feedback and managing local context. We have developed a system, MUSE, that applies a model-based technology to address the above requirements. It supports a wide range of interface styles. Resulting Uls can be customized and capture enough application semantics to provide local feedback and manage the context required for evaluating a student's work and providing coaching.

INTRODUCTION Model-based user interface design refers to a paradigm which uses an explicit, largely declarative representation capturing application semantics and other knowledge needed to specify the appearance and behavior of an interactive system. In this paradigm, an application developer instead of writing a large procedural program defines a model of facts, which controls behavior of reusable code, and a much smaller procedural program. The goal of the model-based UI design is to identify reusable components of a UI and to capture more knowledge in the model, while reducing the amount of new (procedural) code that has to be written for each new application.

We describe MASTERMIND, a step toward our vision of a knowledge-based design-time and run-time environment where human-computer interfaces development is centered around an all-encompassing design model. The MASTERMIND approach is intended to provide integration and continuity across the entire life cycle of the user interface. In addition it facilitates higher quality work within each phase of the life cycle. MASTERMIND is an open framework, in which the design knowledge base allows multiple tools to come into play and makes knowledge created by each tool accessible to the others.

The paper describes TACTICS, the model and a modelbased tool capable of supporting a wide range of design decisions and providing assistance in the design process. The TACTICS tool automatically generates a user interface for an application and assists in refining it and in detecting and resolving design inconsistencies. The TACTICS model of human-computer interaction integrates a compositional model of UIs and a transformational model of the UI design space. A user interface is viewed as a composition of primitives structured based on the application and on the desired dialogue style, and the model identifies user interface components and structuring principles for assembling components into a coherent interface. The model also defines transformations that modify UI structures to achieve a desired look-and-feel and enable designers to easily explore different UI designs. The paper describes the knowledge structure of the model and the TACTICS approach to generating user interfaces. UI components are discussed and examples of UI structures given.

The iterative and exploratory nature of user interface (UI) design determines the requirements for UI design tools: to provide good coverage of the UI design space and to support exploration. That is, the tools should facilitate going from one design to another, providing guidance and assistance in the process. The motivation for this research is to show that the model-based approach to UI design addresses all of the above requirements. Improved models can not only expand the range of interfaces a UI tool can produce, but can also capture the knowledge needed for the tool to assist in the design process. This research develops a compositional model of human-computer interaction as the basis for UI design tools. The model integrates the UI representation based on the compositional view of user interfaces with the domain-independent representation of application semantics. The UI representation identifies and classifies components of a UI and structuring principles for composing them into a UI. Both the components and the structuring principles are reusable across applications and different user interfaces; application specificity and different UI designs are achieved by configuring components and applying structuring principles. The UI representation can explain behavior and the look-and-feel of a UI in terms of properties of UI components and the way the components are structured. The compositional model establishes an explicit relation between the application semantics and the corresponding UI, captured in two sets of rules: composition rules that explicitly represent the mapping from the structure representing the application semantics to the structure representing the application's UI, and consistency rules that evaluate the UI structure with respect to the application semantics according to a given set of criteria. An integral part of the model is design transformations--abstractions for changing the application conceptual model and the user conceptual model of an application while preserving the application functionality. The compositional model provides a framework for a fully-functional UI tool. A prototype tool based on the model is implemented as a proof of concept. It automatically generates a UI from a high-level specification, supports a wide range of different UI designs, and can provide assistance and guidance in the process of exploring alternative designs.

One of the newest user-interface management systems, the User-Interface Development Environment, is described by its creators. UIDE is built around a knowledge-based representation of the conceptual design of a user interface. UIDE supports design at a high-level specification of the interface from information provided by the designer. By using the same knowledge base, the interface developer can generate a new interface design with the same functionality as the original design. This lets users try many functionally equivalent interfaces for the same application.<<ETX>>

The user-computer interface is often the major impediment to successful use of interactive computer graphics systems. Design of the interface is often thought of as art rather than science, and suffers from lack of formalisms, models, tools, and methodical design approaches. Slowly, the design process is becoming more structured, and more formal tools are becoming available.This paper describes the User Interface Design Environment (VIDE), a knowledge-based system to assist in user interface design and implementation. At the heart of UIDE is a representation of the conceptual design of a user interface. This design includes: the class hierarchy of objects which exist in the system (only single inheritance is currently supported), properties of the objects, actions which can be performed on the objects, units of information required by the actions, and pre- and post-conditions for the actions. The conceptual design and related information is used to: check the interface design for consistency and completeness via the Knowledge Acquisition In a Man-machine Interface system (KAIMI), transform the knowledge base, and hence the user interface it represents into a different but functionally-equivalent interface via the Transformation Algorithms system (TA), evaluate the interface design with respect to speed of use, with the keystroke Analysis Tool as input to a Simple User Interface Management System (SUIMS) which implements the user interface, automatically generate intelligent run-time help to the end-user via the Help Support system (HS), and produce a printed description of the design in the Interface Definition Language (IDL). IDL has been developed to present the conceptual design of the user interface in a structured and readable representation.UIDE goes beyond the capabilities of the typical User Interface Management System (UIMS), which requires the designer to work at the syntactic and lexical levels of design and hence to focus on command names, screen and icon design, menu organization, sequencing rules, and interaction techniques. UIDE provides a higher-level conceptual design tool that facilitates the iterative design process of specification, generation, and evaluation.

A knowledge base which defines a user-computer interface is described. The knowledge base serves as input to a user interface management system, which implements the user interface. However, the knowledge base represents user interface design knowledge at a level of abstraction higher than is typical of user interface management systems. In particular, it represents objects, actions, attributes of objects, an object class hierarchy, and pre- and post-conditions on the actions. The knowledge base can be algorithmically transformed into a number of functionally equivalent interfaces, each of which is slightly different from the original interface. The transformed interface definition can be input to the UIMS, providing a way to quickly experiment with a family of related interfaces.