A Machine Learning Based Methodology for Web Systems Codeless Testing with Selenium

Web system testing is a crucial software development cycle. However, though there are real needs for testing these complex systems, it often requires specific skills in testing and/or technical programming. Moreover, the lifecycles of web systems are today very dynamic. They are often modified, updated, integrating new data, links, widgets, etc. Therefore, the testing processes and scripts for these systems have to be modified as well which can be very costly in terms of time and resources. Based on that context, this paper aims at reducing these prerequisites and constraints for tester in proposing a codeless testing automation framework. Our approach is based on Selenium and a machine learning technique to propose generic testing scripts that can be automatically tuned to the tested use cases. Experiments are provided leading to relevant results demonstrating the success of our methodology.