ABSTRACT
The development of interactive systems relies on an iterative process based on well identified phases such as requirement analysis, design, implementation, and evaluation. Tools are now available for supporting a number of these steps. In particular, task models such as the GOMS-like family (Card et al. 1983) and MAD (Scapin&Pierret-Goldreich 1989) provide methods and formalisms useful for structuring and conveying task analysis. Implementation is supported by even a wider variety of tools ranging from toolboxes to application skeletons and UIMSs. On the other hand, the evaluation process of user interfaces is still based on “craft techniques” (Long&Dowell 1989) and suffers from the lack of software tools. In this article, we propose a framework useful for classifying current and future evaluation techniques. We then present our own approach to the automatic evaluation of user interfaces.
