Test Process Improvement with Documentation Driven Integration Testing

Improving the maturity of the test process in an organization, especially but not limited to integration testing, involves obstacles and risks, such as the additional work overhead of the new process. In addition, integration testing descriptions are often too technical not addressing the language needs of the domain. In research cooperations with companies from the insurance and banking domain it turned out that test descriptions and reports are one of the most useful testing artifacts, while doing adhoc testing. This paper presents a bottom up testing approach, which first helps the integration tester in producing a semi-formal test description and report, up to be an enabler for automatic model-based testing in the very end. The presented approach is based on a textual domain specific language that is able to evolve over time. This is done by analyzing the test descriptions and reports automatically with machine learning techniques as well as manually by integration testers. Often recurring test steps or used components are integrated into the test language, making it specially tailored for a specific organization. For each test step implementations can be attached, preparing it for the next iteration. In this paper the methodology and architecture of our integration testing approach are presented together with the underlying language concepts.