Automated Test Data Generation using Search Based Software Engineering

Abstract

Generating test data is a demanding process. Without automation, the process is slow, expensive and error-prone. However, techniques to automate test data generation must cater for a bewildering variety of functional and non-functional test adequacy criteria and must either implicitly or explicitly solve problems involving state propagation and constraint satisfaction. This talk will show how optimisation techniques associated with search based software engineering (SBSE) have been used to automate test data generation. The talk will survey the area and present the results of recent work on characterising, transforming and eliding test data search landscapes.