A Cooperative Problem Solving System for User Interface Design

Abstract

Designing a user interface is an ill-defined problem making cooperative problem solving systems a promising approach to support user interface designers. Cooperative problem solving systems are modular systems that support the human designer with multiple, independent system components. We present a system architecture and an implemented system, FRAMER, that demonstrate the cooperative problem solving approach. FRAMER represents design knowledge in formal, machine-interpretable knowledge sources such as critics and dynamic specification sheets, and in semi-formal knowledge sources such as a palette of user interface building blocks and a checklist. Each of these components contributes significantly to the overall usefulness of the system while requiring only limited resources to be designed and implemented. ’ Cooperative Problem Solving Artificial intelligence research has traditionally focused on building systems that autonomously solve complex problems (e.g., Rl/XCON (McDermott, 1982) and MYCIN (Buchanan, Shortliffe, 1984)). This approach is however not easily applicable in ill-defined problem domains, such as user interface design. Consistency (Grudin, 1989), learnability, and many other concepts of user interface design cannot be adequately formalized in a precise way. Alternatively, one can design cooperative problem solving systems (Fischer, 1988) that work in conjunction with human problem solvers rather than replacing them. Cooperative problem solving systems are located between systems that design with human guidance (e.g., UofA*, (Singh, Green, 1989)) and passive CAD tools (e.g., MENULAY, (Buxton et al., 1983)). A desirable characteristic of practical cooperative problem solving systems is a modular, incremental architecture with simple but extensible components. In contrast, many intelligent support systems that have been