Communications of the ACM
Detecting Software Errors sing Genetic Algorithms
Saarland University professor Andreas Zeller and his colleagues have developed software that tests other programs -- even the sensor functions in cars.
Credit: Oliver Dietze
Saarland University professor Andreas Zeller and his colleagues have developed XMLMATE, software that automatically tests other programs for errors.
XMLMATE generates test cases and tests code automatically, but program input must be structured in a certain way because the researchers use it to generate the initial set of test cases.
Zeller says testing is based on a genetic algorithm that functions similarly to biological evolution, with chromosomes being analogous to input. "It is not easy to detect a real error, and the more code we are covering, the more sure we can be that more errors will not occur," says XMLMATE co-developer Nikolas Havrikov.
In testing XMLMATE on actual open source programs, the team detected nearly twice as many fatal errors as similar test methods that work only with randomly generated input. "But the best thing is that we are completely independent from the application area," Zeller says. "With our framework, we are not only able to test computer networks, the processing of datasets, websites, or operating systems, but we can also examine software for sensors in cars."
From Saarland University
View Full Article
Abstracts Copyright © 2014 Information Inc., Bethesda, Maryland, USA
No entries found