Fault-oriented automated test data generation
Bogdan Korel, Ali M. Al-Yami · 1996
In this research, a novel method for automated test data generation has been developed. This method is called Fault-Oriented Automated Test Data Generation. The main objective of this method is to automatically generate test data that cause the program to behave incorrectly. Fault-oriented automated test data generation shifts the process of test data generation toward revealing more faults. The main goal of this approach is to automatically generate test cases that will uncover faults, i.e., if a test case is generated then a fault has been uncovered. Based on this novel concept of fault-oriented automated test data generation, we have developed two automatic test data generation methods: Assertion-Oriented Automated Test Data Generation and Automated Test Data Generation for Regression Testing. The objective of assertion-oriented test data generation is to generate test cases that cause the violation of one or more assertions in a given program. If such test case(s) is found then a fault has been uncovered in the program under test. The objective of automated test data generation for regression testing, on the other hand, is to generate test cases that cause a modified version of a program P to produce different output than that produced by the original version of this program with respect to some output variable v. If such test case(s) is found then a fault has been detected. This fault may be in the original version, the modified version, or in both. Our experimental studies in both these new methods have shown that fault-oriented test data generation is very effective as compared to traditional test data generation methods. However, fault-oriented automated test data generation is not meant to replace existing testing methods; instead, it is an additional step toward better utilization of system resources and more effective and automatic test data generation.