Methodology for regression testing with open source tool

Abstract

The paper describes our methodology for optimizing regression testing that forms a major part of software maintenance. It necessitates the use of an automated testing tool, and we have selected Selenium, an open source tool. For simple projects, a formula is proposed that has been derived through data mining with Selenium. A genetic algorithm is added to this methodology for industry based projects, where the test cases are so large that they have to be grouped as Test Suites; this algorithm reconfigures Test suites in each cycle of regression testing. Commonly used APFD metric ignores fault severity but is included in our formula; this severity is determi ned by professional testers. The use of ANN to amend severity without manual intervention enhances the genetic algorithm. Tables presented in the paper are from both simple and industry projects. Comparison is made with IBM'S RFT, a proprietary tool for automated tes ting.