Automated Behaviour Model Construction For Android Applications

Automated testing is an important method to guarantee the quality of software. Re-cently, Android applications have gained an amazing increasement. As a widely used testing method, model-based testing (MBT) is suitable for Android testing and can per-form well. For MBT, the basic step is to construct the behaviour model. A novel approach is proposed for the automation in behaviour model construction.The contributions of this paper are as follows:Combined static and dynamic analysis of the invokable events In this paper, the dynamic analysis and static analysis are combined to obtain the invokable events for Android applications. Dynamic process can allocate events for componetns ac-cordingly and the static process will extract the registration interface and inherited interface from source code.Weighted bias random search algorithm This paper presented a novel weighted bias search algorithm.This algorithm can consider the weight of the components and the latent weight as well.With this novel algorithm,the construction of be-haviour model can be more effecient and effective.Automated behaviour model construction With the execution of the invokable events, the state transitions can be constructed automatically.Overall, this paper proposed an approch for combining dynamic and static analysis, as well as a novel algorithm to automated constructing the behaviour model of Android applications. The experimental results show the effectiveness of this approach.