Automated Extraction of Augmented Models for Android Apps

Mobile software development involves significant challenges to developers such as device fragmentation (i.e., enormous hardware and software diversity), event-driven programming (i.e., programming based on user interactions, sensor readings and other events where the program must react) and continuous evolving platforms (i.e., fast changing mobile frameworks and technologies). This can lead programmers to error-prone code, because of the multiple combinations of external variables that must be taken into account in an app development process. Thus, testing is an underlying necessity in mobile applications to deliver high quality apps. However, defining tests suites for app development is a difficult task that requires a lot of effort, because it must consider all the possible states of an app, its context (e.g., device in which is running, sensors, touch gestures, screen proportions, connectivity), and a large combination of mobile devices and operating systems. Previous efforts have been done to extract models that support automated testing. However, as of today there is not a single model that synthesizes different aspects in mobile apps such as domain, usage, context and GUI-related information. These aspects represent complementary information that can be mixed into a single and enriched model. In this paper, we propose a multi-model representation that combines information extracted statically and dynamically from Android apps. Our approach allows practitioners to automatically extract augmented models that combine different types of information, and could help them during comprehension and testing tasks.