| With the explosive growth of mobile applications in the smartphone market, the overall poor user experience and low satisfaction rate is an issue to be addressed. The usability of mobile applications is critical for their adoption particularly because of the relatively small screen and awkward (sometimes virtual) keyboard, despite the recent advances of smartphones. Traditional laboratory-based usability testing is often tedious, expensive, and does not reflect real life cases. It is therefore important to conduct usability research that is specially tailored to mobile applications. Recently, diary-based user studies of mobile applications in real-world settings have been increasingly adopted, but there are still few quantitative models that provide contextualized usability understanding and automatic usability inspection tools for mobile applications.;In this dissertation, we envisioned and implemented a framework that obtains usability insights and usage knowledge for mobile applications in a fully automated manner. We built a toolkit that embeds into mobile applications to collect fine-grained user interface (UI) events and readings of the built-in sensors as users interact with the applications. The data can be piped into an automata-based model which extracts rich and quantified metrics as the measurements of usability problems. We designed a web portal which presents these metrics with a clear and visualized approach to facilitate in-depth usability analysis.;We evaluated our framework by conducting user studies on multiple real-world Android applications across devices and platforms, which showed that the framework was robust and reliable in capturing detailed UI events and context information. By comparing it to traditional laboratory testing in terms of the number of detected usability problems, we showed that the framework is capable of uncovering many usability issues automatically. In addition, the framework advances the usability analysis by correlating context and interaction, which is not possible by traditional laboratory testing. We also collected feedback from usability experts and our pilot developers, who agreed that our framework is effective and of practical use in helping to improve the usability of mobile applications. |
