A Model-based Approach to Assess Regular, Constant, and Progressive User Interface Adaptivity

Adaptive user interfaces adapt their contents, presentation, or behavior mostly in a sudden, fluctuating, and abrupt way, which may cause negative effects on the end users, such as cognitive disruption. Instead, adaptivity should be regular, constant, and progressive. To assess these requirements, we present Taoist, a hidden Markov model-based approach and software environment that seek the longest repeating action subsequences in a task model. The interaction state space is discretely produced from a task model and the interaction observations are dynamically generated from a categorical distribution exploiting the subsequences. Parameters governing adaptivity and its results are centralized to support two scenarios: intra-session for the same user and inter-session for the same or any other user, even new ones. The end-user can control the adaptivity when initiated by accepting, declining, modifying, postponing,or reinitiating the process before propagating it to the next iteration. We describe the Taoist implementation and its algorithm for adaptivity. We illustrate its application with examples, including the W3C reference case study. We report the results of an experiment that evaluated Taoist with a representative group of ten practitioners who assessed the regular, constant, and progressive character of adaptivity after four intra-session iterations of the same task.