Using AI in Automated UI Localization Testing of a Mobile App
Ynion, Jose Cezar (2020)
Ynion, Jose Cezar
2020
All rights reserved. This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-202004084809
https://urn.fi/URN:NBN:fi:amk-202004084809
Tiivistelmä
Localization testing is seldom addressed in the scientific literature, especially on the mobile app domain. This thesis focused on the practical implementation of an automated localization testing system for an iOS mobile app.
I work as a Software Engineer in an international company that has mobile and desktop apps as main products. Each app is localized into multiple languages. Testing that each User Interface (UI) displays the right content per language is the most time-consuming part of the software development lifecycle. Due to the visual nature of the tests, this is done manually and repeatedly in different devices with various Operating Systems and screen resolutions.
Effectively testing the localized app is always a challenge for Quality Engineers because they are not language experts. The scope of the tests is somewhat limited to finding bugs like wrong layout, overlapping, untranslated texts, and wrongly represented characters.
The prototype system called NEAR is the outcome of this thesis. It was designed to automate most of the tasks in testing UI Localization. It integrates pre-trained cloud-based Artificial Intelligence models of Natural Language Processing (NLP) and Computer Vision from service providers like Google to add visual context to a test. As a result, the time required to run the regression test is less. The scope of the testing now includes finding bugs that need linguistic skills like mistranslation, text truncations, and locale violations.
I work as a Software Engineer in an international company that has mobile and desktop apps as main products. Each app is localized into multiple languages. Testing that each User Interface (UI) displays the right content per language is the most time-consuming part of the software development lifecycle. Due to the visual nature of the tests, this is done manually and repeatedly in different devices with various Operating Systems and screen resolutions.
Effectively testing the localized app is always a challenge for Quality Engineers because they are not language experts. The scope of the tests is somewhat limited to finding bugs like wrong layout, overlapping, untranslated texts, and wrongly represented characters.
The prototype system called NEAR is the outcome of this thesis. It was designed to automate most of the tasks in testing UI Localization. It integrates pre-trained cloud-based Artificial Intelligence models of Natural Language Processing (NLP) and Computer Vision from service providers like Google to add visual context to a test. As a result, the time required to run the regression test is less. The scope of the testing now includes finding bugs that need linguistic skills like mistranslation, text truncations, and locale violations.