Evaluation of Case-Based Classification to Support Automated Web Service Discovery and Brokering.

摘要

In this paper we evaluate the use of case-based classification to resolve a number of questions related to information sharing in the context of an Integrated Web Services Brokering System (IWB). We are developing the IWB to independently decompose and analyze ad hoc Web Services interface descriptions in order to identify Web Services of interest. Our approach is to have the IWB cache information about each service in order to support an autonomous mediation process. In this mediation process, the IWB independently matches the user's data request to the correct method within the appropriate web service, translates the user's request to the correct syntax and structure of the Web Service request, dynamically invokes the method on the service, and translates the Web Service response. We use casebased classification as a means of automating the IWB's analysis of relevant services and operations. Case-based classification retrieves and reuses decisions based on training data. We use sample Web Service Description Language (WSDL) files and schema from actual web services as training data in our approach and do not require the service to predeploy an OWL-S ontology. We present our evaluation of this approach and performance ratings in the context of meteorological and oceanographic (METOC) Web Services as it relates to the IWB.