PUBLICATION
Automated Estimation of Web Interaction Complexity based on UI Tests
Type
Conference Paper
Year
2026
Authors
Dr.-Ing. Sebastian Heil
Julián Grigera
Ekaterina Pavlova
Prof. Dr.-Ing. Martin Gaedke
Research Area
Event
26th International Conference of Web Engineering (ICWE2026)
Published in
Web Engineering: 26th International Conference, ICWE 2026, Lyon, France, June 9 – June 12, 2026, Proceedings
ISBN/ISSN
tba
Download
Abstract
Evaluating the complexity of user interaction is crucial for engineering usable web application and to make informed design decisions. Automated approaches are increasingly being investigated due to the time, cost and expertise requirements of conducting empirical user studies. However, these approaches require creating extra artifacts otherwise not useful for software production, focus on static aspects of the user interface, or require specially instrumented running interface versions. Thus, we explore the feasibility of automated interaction complexity estimation through the analysis of existing web UI tests. We propose a novel software architecture for the analysis of UI tests and study the suitability of 11 different candidate UI test metrics from the domains of UI analysis, program understanding and cognitive modeling. By conducting an experiment with 38 participants, we compared human-delivered assessments to the UI test metrics. From this experiment, we obtained 2 linear and multiple linear regression models that can be used to predict interaction complexity. Additionally, we obtained a second set of multiple linear regression models based on a wider range of metrics, using Principal Component Analysis to narrow the input down to 3 components. These results suggest that automatically extracted UI test metrics can help determine interaction complexity, using existing artifacts. Our method is well-suited for integration into CI/CD pipelines, supporting the incremental and iterative nature of modern agile practices.
Reference
TBA
