J.R. Birt and R. Sitte (Australia)
Software testing, software reliability, genetic algorithms, testing efficiency
This paper studies the effect of introducing random error seeding on the performance of Genetic Algorithms in the identification of error prone paths in software. This is based on our earlier research on identifying the potentially most error prone paths in a program. We use variable length Genetic Algorithms that optimize and select the software paths, which in turn are weighted with sources of error indexes. Although various methods have been applied for detecting and reducing errors in software, there is little research into partitioning a system into smaller error prone domains for testing. Our experiments with error seeding show that by selecting 80% of potential errors or 20% of most error prone paths we can detect on average greater than 65% of the randomly seeded errors.
Important Links:
Go Back
