Shu-Wei ChengFollow


Automated test procedures to test mobile applications or other software may repeatedly revisit the same user interface screens of the app-under-test in an attempt to improve coverage. Complete traversal of all screens of the app-under-test for regression or exploratory testing invariably takes a long time, is inefficient, and is sometimes ineffective. This disclosure describes a crawler that leverages reinforcement learning (RL) to efficiently traverse the screens of a mobile app-under-test or other software, and to form a spanning tree interconnecting the screens of the mobile app-under-test. In an exploratory phase, newly added parts of the mobile app-under-test are discovered and mapped. In an exploitation phase, the mapped parts of the mobile app are explored and regression-tested. The crawler does not require prior knowledge or human pre-encoding of the app-under-test. Rather, it automatically discovers and maps the app-under-test, enabling rapid, scalable, and comprehensive testing of newly released software or applications.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.