Variable Precision Rough Set-Based Fault Diagnosis for Web Services

Web service is the emergent technology for constructing more complex and flexible software system for business applications. However, some new features of Web service-based software such as heterogeneity and loose coupling bring great trouble to the latter fault debugging and diagnosis. In the paper, variable precision rough set-based diagnosis framework is presented. In such debugging model, SOAP message monitoring and service invocation instrument are used to record service interface information. Meanwhile, factors of execution context are also viewed as conditional attributes of knowledge representation system. The final execution result is treated as the decision attribute, and failure ontology is utilized to classify system´s failure behaviors. Based on this extended information system, variable precision rough set reasoning is performed to generate the probability association rules, which are the clues for locating the possible faulty services. In addition, the experiment on a real-world Web services system is performed to demonstrate the feasibility and effectiveness of our proposed method.