Towards describing black-box testing methods as atomic rules

Abstract

Ideally, all black-box testing methods should be interpreted in the same way by different testers. In reality however, inconsistencies and ambiguities in original method descriptions may lead to differing interpretations and varying test set quality. In this paper, we decompose these methods into atomic rules for selecting test data and constructing test cases. We validate the rules via a worked example and discuss a pilot experiment to determine whether atomic rules are simpler to learn and use. Our approach also enables method tailoring and may simplify method comparison.