Design problems involve issues of stylistic preference and flexible standards of success; human designers often proceed by intuition and are unaware of following any strict rule-based procedures. These features make design tasks especially difficult to automate. Adaptation is proposed as a means to overcome these challenges. We describe a system that applies an adaptive algorithm to automated user interface design within the framework of the MOBI-D (Model-Based Interface Designer) interface development environment. Preliminary experiments indicate that adaptation improves the performance of the automated user interface design system. Keywords Model-based interface development, machine learning, decision trees, theory refinement, user interface development tools, interface models, theory refinement
Jacob Eisenstein, Angel R. Puerta