| Web service is the main implementation of Service Oriented Architecture,we can use it to access to the resources which provided by service providers ignoring the difference of platforms.For the effective integration of web services,the web services composition coming up and widely researching because of the features of semantics.OWL is a basic technology for the research of ontology-based semantic Web.OWL-S,which is based on OWL,is one of the most important standards.As for Web service technology,besides the needs for functions,more factors in performance is very important.Based on OWL-S and semantic Web service process,performance analysis can improve the performance of Web service request,because it can help service provider adjust their Web service architecture and enable them to handle complex Web service request.In the same time,performance analysis helps users choose the best Web service which fits better for their actual needs.In this work,we propose a Petri-net-based formal framework for OWL-S processes using non-markovianstochastic-petri-nets(NMSPN)as the intermediate representation.So that we can analysis the performance parameter with the framework。Then we illustrate an example to check the accuracy of our model.Finally,we do research on Web service performance metrics and use a prediction method based on time series to predict the trend of some metrics.The framework in our paper,not only can describe the processes of Web service composition in OWL-S,but also can analysis the performance parameter quantitatively. |
