Enhancing the Automation of GUI Testing

GUI testing is one of the most important and significant testing approaches among all different software testing techniques. Most software errors are captured and detected through the software GUI layer. Manual testing for GUIs has its problems. It lacks in capturing all different cases and takes a huge time from the software tester to plan, design and re-design the testing suites in case of UI change. Old techniques in the area of test-case generation are not fully-automated or dependent on human inputs. This paper presents a test-case generation model to build a testing suite for webpages using its HTML file. The proposed model has two branches. The first one focuses on generating test cases for each web-element individually based on its type. The other branch focuses on generating test cases based on different paths between web-elements in the same webpage. It is also concerned with eliminating redundant test-cases using a supervised learning, feed-forward, dynamic artificial neural network that changes number of inputs according to generated cases per web page. The proposed system has been evaluated using several datasets. Results show a significant enhancement in the test-case generation procedure.