Automation in Model-based Usability Evaluation of Adaptive User Interfaces by Simulating User Interaction

Abstract

The goal of adaptive user interfaces (UI) is offering the opportunity to adapt to changes in the context of use and thus provide potentially improved interaction capabilities for different users in specific situations. But, this poses the challenge of evaluating usability aspects of many different variants of the resulting UI. Consequently, usability evaluations with real users or experts tend to become complex and time-consuming especially in the domain of adaptive UIs. Model-based usability evaluations and specifically automated tools and approaches have proven to correctly predict usability relevant aspects in early stages of development. However, the creation and provision of required models and information tends to be complex and time consuming as well and further requires a high degree of expertise for the specific tool and applied method. This thesis describes an integrated approach that provides automation in model-based usability evaluation based on already existing development models of adaptive UIs. The approach is based on required information for describing the UI surface information and the interaction capabilities of the UI. With the help of this information usability relevant criteria are predicted using specific tools of automated usability evaluation. The implementation of the approach presents integration of an existing runtime framework for adaptive UIs with a cognitive user behavior model for simulation. Information required for simulating interactions is created automatically with the help of the UI development models and by this means saves time and costs when preparing and running simulations. Additionally, with the help of two studies, the resulting predictions are further improved by directly using information encoded in the existing development models without requiring specific expertise from designers and usability experts.

Author:
Michael Quade
Category:
Dissertation
Year:
2015
Location:
Doctoral Thesis, Technische Universität Berlin, Berlin, Germany