Data flow analysis and testing for OWL-S semantic web service compositions

Semantic web services SWSs, which integrate the concept of ontology to support incorporation of machine understandable semantics into web services, have drawn much attention in recent years. By introducing semantics into web services, the quality and robustness of web service discovery, selection, and invocation can be greatly improved. However, web ontology language for service OWL-S, the most widely accepted standard used to describe SWS, is extremely complex and hard to understand. This makes OWL-S compositions difficult to analyse and test. In view of this, this paper proposes a data flow testing approach for web service compositions based on OWL-S. The data flow test artefacts introduced by OWL-S are identified and thoroughly analysed. A test model that considers a variety of control constructs and semantics of OWL-S is proposed to abstract these test artefacts. Based on the test model, test paths can be derived to effectively uncover defects caused by improper data handling and message exchanges of OWL-S compositions.