Optimal allocation of testing resources

As object-oriented technologies are getting used to build larger and larger software systems, the question of obtaining test plan and minimizing test cost and duration becomes more and more important. We propose an approach that allows the automatic production of an integration test plan for an object-oriented system that minimizes both the test effort and the test duration. We build on classical graph algorithms that have been customized to deal with test dependency graphs generated from UML design models. We illustrate the applicability of our approach on two real world case studies in the domains of telecommunication and compiler software. Results of our testing resources allocation strategy are compared with random strategies.