Using Dependency Information to Select the Test Focus in the Integration Testing Process

Existing software systems consist of thousands of software components realizing countless requirements. To fulfill these requirements, components have to interact with or depend on each other. The goal of the integration testing process is to test that the interactions between these components are correctly realized. However, it is impossible to test all dependencies because of time and budget constraints. Therefore, error-prone dependencies have to be selected as tests. This paper presents an approach to select the test focus in the integration test process. It uses dependency and error information of previous versions of the system under test. Error-prone dependency properties are identified by statistical approaches and used to select dependencies in the current version of the system. The results of two case studies with real software systems are presented.