Ontology-based Test Generation And Mutation Testing On Web Services

Dependability has been a critical obstacle to the service-oriented applications because the service clients may not trust the services dynamically discovered by the service broker, such as correctness, performance, and reliability. Web services(WS) implement SOA, but traditional software verification and validation (V&V) techniques cannot satisfy the new challenges in WS testing. It is necessary to construct an automated, dynamic, collaborative WS V&V process, including test generation, test execution, test monitoring and test adequacy assessment etc. Semantic Web services employ standard ontologies to represent WS specifications, which are computer-interpretable, and enable users to locate, select, and compose services automatically. This thesis proposes an Ontology-based test model, a test data generation approach, and a mutation analysis method for WS testing.(1) It proposes a Test Ontology Model (TOM) which is compatible with UML 2.0 Test Profile (U2TP). The TOM specification can serve as test contracts among test components. TOM supports test assets such as test cases, test data, test plan, and test result. All stakeholders can construct, share and reuse the test assets based on TOM.(2) It proposes an automated test data generation approach for WS, based on the semantic specifications in OWL-S. Data pools are established for each parameter. Data partitions are derived by class property and relationship analysis. Data instances can be generated from predefined values, with different combination strategies. OWL DL reasoners and rule reasoners are used for classification. Completeness and consistency (C&C) checking can be performed on the test data.(3) It proposes an Ontology-based mutation analysis method for WS testing. Four categories of OWL-S mutant operators are specified. A mutation testing framework for WS is constructed to provide viable test adequacy assessment for WS testing.