Classification Based Parameter Association for Non-redundant Annotation

Semantic annotation of Web Services can facilitate the automated service discovery and composition. At present, however, many solutions suffer from redundant annotations or imprecise derived annotations. The fundamental task to address the issue is to find parameters that have same semantics in a large number of Web Services. This paper proposes a classification based approach for identifying parameters that are semantically equivalent, laying the foundation for non-redundant annotation. Subsequently the paper presents the non-redundant annotation, in which the parameter space is firstly reduced, and then semantic annotation is performed on the reduced parameter space, finally annotation results of the reduced parameter space are expanded to the original parameter space. To evaluate the final annotation results, the paper gives a methodology based on service discovery. The experimental results indicate that the parameter association approach can achieve outstanding accuracy in identifying the equivalence between parameters. Moreover, it is suggested that the non-redundant annotation can greatly reduce redundant annotations, and can augment the semantics of Web Services with adequate accuracy, which can achieve similar performance in service discovery as OWL-S services do.