ABSTRACT

Software applications with Graphical User Interfaces (GUIs) have become essential in people's daily lives, and sufficient testing is a necessity to ensure their quality. When performed manually, GUI testing is a costly and tedious process requiring many human testers to explore the user interface and check whether the output is as expected. In contrast, existing automated testing techniques are less effective due to the lack of availability of domain knowledge that human testers typically possess. In this project, the investigators will explore the reuse and migration of manual GUI tests, an alternative route to complement existing automatic GUI testing research. The intuitive observation behind the project is that developers tend to use similar GUI designs in different platform versions of a same application or different applications within the same domain. Therefore, it is possible to reuse the exploration sequences, input values, and expected output with proper adaptations taking into account the subtle implementation differences between applications. The project is expected to enhance the coverage and productivity of GUI-testing processes, leading to GUI applications with higher quality and fewer defects. Additionally, the incorporated training and education activities will provide opportunities for participants to acquire research experience and become highly qualified researchers and practitioners.

In this project, the PIs are going to answer the research question: whether and how existing GUI tests can be reused in automatic GUI testing with necessary adaptation. In particular, the investigators will work on the generation of GUI-code embeddings to represent the semantics of GUI views and develop novel GUI-view mapping techniques to map GUI views among different applications. The investigators will also study how input-value constraints and event-sequence constraints in existing GUI tests can be extracted as domain knowledge, how such knowledge can be translated across platform and application boundaries, as well as how the translated knowledge can be incorporated into the automatic GUI-test generation process of the target application. Moreover, the investigators will develop techniques to identify the potential reusability of existing test oracles based on measuring their fitness with the new context, and techniques to create new test oracles by summarizing common behaviors of software applications in the same domain. The findings of this project are intended to shed light on the more general problem of reusing and migrating any test cases such as unit tests and integration tests, as well as the solution to the open problem of creating meaningful test oracles.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

Please report errors in award information by writing to: awardsearch@nsf.gov.