Empirical Studies of Privacy-Preserving Regression Test Selection Techniques for Web Services

In an end-to-end regression testing framework employing a safe regression test selection (RTS) technique that uses control-flow graphs (CFGs) to model Web service interactions, service providers must share their CFGs which are sensitive implementation details that service providers are unwilling to expose especially across autonomous systems. In a series of earlier works, we proposed two techniques: 1) A privacy-preserving RTS technique (PPT) which employs a series of privacy-preserving techniques to protect sensitive information contained within CFGs without reducing the functional effectiveness of the RTS technique and allows service providers to opt-out. 2) A privacy-aware RTS technique (PAT) which uses only local information to perform its work and requires no sharing. These techniques improve service provider privacy by trading either performance by employing privacy-preserving techniques in the PPT or selectivity by only using local information in the PAT. A series of empirical studies were undertaken to study the techniques to determine their feasibility and cost-effectiveness on a variety of systems. The results of the study show that while both techniques can be effective in reducing the costs of performing RTS and regression testing, the PPT is more cost-effective for at least the systems studied provided service providers participate in the framework.