Test-oriented Model Construction For Mobile Apps

With the fast increasing of mobile applications,there are more immediate pressure in testing of application for mobiles,and the important thing we need to determine is whether the App has flaws.One of the effective ways is Model-based testing(MBT).This method is a two-stage process:model building and testing.Compared to the traditional software world,such as C,C++,Java,the greatest difference is the way how to builde a model in mobile application.Existing research has experienced a dramatic development in model building,but there are still some problems to be resolved.On one hand,when facing the method of understanding a program's behavior,some researchers have sacrificed coverage for low modeling cost,while others chose the opposite.On the other hand,mobile application platform provides graphical applications for the user,so usability testing is crucial to please users.Nowadays the questionnaire based method is normal practice,and there is a lack of the study of modeling the user behavior.To address those issues,we researched Test-Oriented Model Construction for Mobile Apps,and proposed a mobile application behavior model generation method and a user behavior model generation method.The research includes mainly the following three aspects:? A mobile application behavior model generation method.Its biggest contribution is model generation theory with Smali Code,and Smali Code is obtained by reverse engineering.On the basis of Smali Code,we construct state machine model.In addition,the system layer in this method will be part of the test system,through scanning screen in the system layer we can determine whether a state change.This method is both in a high test coverage and in a low-overhead.? A user behavior model generation method.We classified events in the mobile application,and the event logging function was developed for each type of event.The event log system will be embedded in the program code and collect user action sequences at the running time.All the collected event sequences will be sent to the server automatically for the model generation.In addition,our study also perform usability test,and problems can be found by analysis of the user data in the model.? According to this article,we got the prototype tool,and the experimental study was carried out.The tool could directly from App generate mobile application behavior model and collect user events for user behavior generation.The data demonstrated the feasibility of this method.