Several techniques exist for mobile test automation, from script-based techniques to automated test generation based on GUI models. Most techniques fall short in being adopted extensively by practitioners because of the very costly definition (and maintenance) of test cases. We present a novel testing framework for Android apps that allows a developer to write effective test scripts without having to know the implementation details and the user interface of the app under test. The main goal of the framework is to generate adaptive tests that can be executed on a significant number of apps, or different releases of the same app, without manual editing of the tests. The frameworks consists of: (1) a Test Scripting Language, that allows the tester to write generic test scripts tailored to activity and app categories; (2) a State Graph Modeler, that creates a model of the app’s GUI, identifying activities (i.e., screens) and widgets; (3) an app classifier that determines the type of application under test; (4) an activity classifier that determines the purpose of each screen; (5) a test adapter that executes test scripts that are compatible with the specific app and activity, automatically tailoring the test scripts to the classes of the app and the activities under test. We evaluated empirically the components of our testing framework. The classifiers were able to outperform available approaches in the literature. The developed testing framework was able to correctly adapt high-level test cases to 28 out of 32 applications, and to reduce the LOCs of the test scripts of around 90%. We conclude that machine learning can be fruitfully applied to the creation of high-level, adaptive test cases for Android apps. Our framework is modular in nature and allows expansions through the addition of new commands to be executed on the classified apps and activities.

Automated Test Selection for Android Apps Based on APK and Activity Classification / Ardito, Luca; Coppola, Riccardo; Leonardi, Simone; Morisio, Maurizio; Buy, Ugo. - In: IEEE ACCESS. - ISSN 2169-3536. - ELETTRONICO. - 8:(2020), pp. 187648-187670. [10.1109/ACCESS.2020.3029735]

Automated Test Selection for Android Apps Based on APK and Activity Classification

Luca, Ardito;Riccardo, Coppola;Simone, Leonardi;Maurizio, Morisio;
2020

Abstract

Several techniques exist for mobile test automation, from script-based techniques to automated test generation based on GUI models. Most techniques fall short in being adopted extensively by practitioners because of the very costly definition (and maintenance) of test cases. We present a novel testing framework for Android apps that allows a developer to write effective test scripts without having to know the implementation details and the user interface of the app under test. The main goal of the framework is to generate adaptive tests that can be executed on a significant number of apps, or different releases of the same app, without manual editing of the tests. The frameworks consists of: (1) a Test Scripting Language, that allows the tester to write generic test scripts tailored to activity and app categories; (2) a State Graph Modeler, that creates a model of the app’s GUI, identifying activities (i.e., screens) and widgets; (3) an app classifier that determines the type of application under test; (4) an activity classifier that determines the purpose of each screen; (5) a test adapter that executes test scripts that are compatible with the specific app and activity, automatically tailoring the test scripts to the classes of the app and the activities under test. We evaluated empirically the components of our testing framework. The classifiers were able to outperform available approaches in the literature. The developed testing framework was able to correctly adapt high-level test cases to 28 out of 32 applications, and to reduce the LOCs of the test scripts of around 90%. We conclude that machine learning can be fruitfully applied to the creation of high-level, adaptive test cases for Android apps. Our framework is modular in nature and allows expansions through the addition of new commands to be executed on the classified apps and activities.
File in questo prodotto:
File Dimensione Formato  
09217524.pdf

accesso aperto

Tipologia: 2a Post-print versione editoriale / Version of Record
Licenza: Creative commons
Dimensione 2.88 MB
Formato Adobe PDF
2.88 MB Adobe PDF Visualizza/Apri
Pubblicazioni consigliate

Caricamento pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11583/2848258