Semantic-Guided Adaptive Exploration Testing for GUI Software

There are already a number of dynamic exploration techniques for GUI exploration that interact with elements automatically and explore the user interface state flow graphs of software. Some exhaustive exploration strategies, although they can explore the software in detail, require a lot of test time to achieve greater coverage. At the same time, there are also methods to improve the exploration efficiency according to semantic similarity, but it is difficult to explore more states. Therefore, this paper uses keyword describing specific functionalities to form adaptive GUI exploration strategies by calculating similarities between text and keyword in software interfaces, and guides exploration to discover user interface states and transitions related to specific functionalities. In the experimental results, our semantic-guided adaptive exploration strategy has stronger exploration ability and higher efficiency than ordinary keyword guided exploration strategy.