WEB Service Clustering And Selection Based On OWL-S

In the Internet,more and more resources are wrapped by services with the widely application of the SOC.Service composition satisfies the requirements of the users by combining different resources.Service selection is an important part of service composition,and the result of selection will affect the credibility of composite service largely.The existing service selection methods mainly consider the Qos and function attributes of the service,and rarely consider the specific process of the service.The accuracy of the methods that calculating service process similarity is not high in a few service selection methods considering service process.And the current methods mostly are in order when selecting services,leading to the low efficiency.Aiming at improving the low time efficiency and precision in current service selection methods,this thesis proposes a service selection method based on OWL-S.The main works in this thesis are as follows.(1)This thesis formalizes the input and output parameters in service functional description part of OWL-S as vector model.On the basic of formalizing parameters,this thesis defines the method of calculating function similarity of service.Aiming at improving the precision of the methods that calculating the service process similarity.This thesis formalizes the process part of service as Petri net,and defines a method that can accurately calculate the process similarity considering the structure and semantic information of Petri net.(2)Based on Web service similarity calculating method,this thesis adopts clustering methods such as k-means.Dbscan?Squeezer to cluster services in the service library from function and process aspects.Because the selection of clustering algorithm and parameters can affect the clustering results largely,this thesis selects optimal parameters for each clustering algorithm via cluster validity index.Based on the cluster preprocessing,this thesis selects proper clustering algorithm for Web service via recall ratio?precision ratio and the value of F.(3)After dividing the services in the library into several subclasses that are accordance in function and process,we can select services that satisfy user's demands.A Web service selection method based on OWL-S proposed by this thesis solves the problem that the low recall ratio?precision ratio and time efficiency.of the existing service selection methods and the achieved research results have a certain significant influence on service selection and service composition.