Abstract

Today it is possible to judge the quality of user interfaces by using automatic testing agents, simulating user behavior for given tasks. Furthermore, such evaluations can be improved by giving these agents required knowledge to fulfill these tasks. Currently, such user task knowledge is more or less a string comparison between given knowledge that is represented as strings and the readable labels of the graphical user interface. Unfortunately, this approach leads to ambiguity problems, where real users cannot intuitively fulfill a task, even if the automatic testing found no obstacles. We propose to represent knowledge not in such a simple way, but using language resources and ontologies to represent user knowledge. We show that such representation helps detect and avoid ambiguous labels in graphical user interfaces.

@inproceedings{Proceedings of the Computational
Linguistics-Applications Conference},
title={Towards An Enhanced Semantic Approach For Automatic Usability Evaluation},
author={Steinn{"o}kel, P. and Scheel, C. and Quade, M. and Albayrak, S.},
year = {2011},
isbn = {N 978-83-60810-47-7},
pages = {85-91},
}