Semantic Annotation Of WEB Services Based On Efficient Feedback

Semantic annotation plays an essential role in automatic service discovery and composition.Researchers match ontology knowledge with elements in Web services to make up of the lack of semantics in services.Testing the services is the standard of verifying whether the annotation results can accurately reflect the function of Web services.However,there are two problems in the current annotation methods: low quality and efficiency.Semantic annotation is to achieve better service discovery,recommendation and composition.Therefore,getting the accurate annotation to invoke the services better and propagate to other services is the key.This paper proposes a semi-automatic annotation framework with high quality and efficiency.Firstly,we define the annotation concept with restriction conditions to get the accurate semantics,which is easier to successfully to invoke the services.Secondly,we propose a pool of candidate annotation,which reduces the possibility of missing correct concepts.Thirdly,we propose a model of manual annotation,which mainly includes three case: when to introduce the manual annotation? How to select the service samples? And how to learn the annotation results? We focus on the research and implementation of manual annotation model.We monitor the service invocation efficiency and parameter propagation efficiency to determine when to introduce manual annotation.We adapt the cluster algorithm to select the service samples with the widest feedback range for manual annotation.The annotation results will be learned automatically to generate more annotation concepts with restriction conditions to enrich the local ontology,which can both improve the quality and efficiency.The experiments based on a collection of annotations for real-world web services shows that our method not only significantly improves the quality of semantic annotation,but also reduces a large number of unnecessary computer costs and labor costs in the whole annotation process,and achieves an efficient annotation process.