Improving Android app exploratory testing with UI test cases using code change analysis

Several companies employ exploratory testing to enhance bug discovery in Android applications while minimizing testing session duration. Traditionally, this optimization relies on testers’ familiarity with problematic areas, identified through the review of change requests, release notes, and dogfooding reports. Our research focuses on refining the effectiveness of exploratory testing for Android applications based on a kind of model dressing exploratory testing. Model dressing maps an application under test to a model created for the domain of application. We achieve this by supplying testers with a more precise artifact related to areas where code changed. By scrutinizing internal code modifications, which include responses to bug reports and general software enhancements, we employ a reachability analysis to identify the corresponding user interface (UI) components, culminating in the creation of a targeted UI model. This UI model serves as the basis for formulating a collection of test scenarios, thoughtfully prioritized to maximize coverage of altered areas. To exemplify the practical advantages of our approach, we present an evaluation conducted in collaboration with our industry partner, Motorola Mobility, a subsidiary of Lenovo. The outcomes of this evaluation underscore the promise of our methodology, particularly in the realm of Android application testing on smartphone devices. In manual testing sessions, our approach achieved 4-10% higher code coverage compared to traditional charters. In automated testing, our tool achieved coverage comparable to experienced human testers. These results have led to the integration of our techniques into the real production testing process at Motorola.