Automatic Usability Evaluation of Model-based Interactive Systems in Ambient Assisted Living (DFG)
Software developers in the area of Ambient Assisted Living (AAL) are facing a multitude of requirements due to diverse user groups and different interactive devices. In order to simplify the development process, model-based user interfaces are increasingly used. In such an approach, the results from an analysis of tasks, users and context, which preceed the development, are formalized as models. Such models can then be executed dynamically by suitable frameworks. The evaluation of such systems so far requires extensive user testing, where results from earlier design iterations can hardly be utilized.
In this project, model-based development is combined with model-based evaluation, aiming at an integrated development environment. In this integrated approach, user errors and recovery strategies when using AAL systems are simulated. To do so, the behavior of real users is observed, classified and formally described. Next, a simple simulation approach based on path search is iteratively extended and improved. In each iteration, the simulation results are evaluated with respect to how realistic the interactions are, which Usability problems could be found, and how accurately performance measures could be predicted. Finally, a development process making use of the proposed method as well as occasional user tests is outlined.
The project is conducted in cooperation with the DAI-Labor at TU Berlin and draws on prior work at both institutions, namely the MASP framework for model-based user interface development developed at DAI-Labor, and the MeMo workbench for semi-automatic usability evaluation, developed at DAI-Labor and the Quality and Usability Lab.