A Bayesian Approach to Software Testing Evaluation

This paper proposes a Bayesian approach to the quantification of some of the subjectivity that is intrinsic to the testing process (i.e., the decision on when to stop). The goal of the paper is to put this subjectivity on clear and firm bases, by showing how a Bayesian approach can be used to model the degree of belief that testers have on the fact that an input of a program results in a failure. We also define a model for the increase of this degree of belief as a program is run with inputs that do not result in failures, and we show how the model can be used to decide when to stop testing.