Towards Effective Bug Reproduction for Mobile Applications

This thesis tackles the challenge of reproducing bugs in mobile apps, a task that is often slow and labor-intensive when user-written bug reports are vague or incomplete. To overcome these barriers, this thesis introduces automated tools that interpret natural language bug descriptions, analyze bug screen recordings, and adaptively replay actions based on GUI screen rendering state. Together, these tools reduce developer effort, improve reproduction accuracy, and accelerate bug resolution, ultimately contributing to the development of higher-quality mobile applications.