A Study On Technologies Of QoS Evaluation Oriented Service Selection

Currently, most of the research on the service selection is based on the principle ofQoS local optimum. For a group of services in the function node of the service compositonmodel,weighted and sort the QoS parameters information of the service,and select theservice having maximum weight to be the candidate service of the function node. In thesemethods, the service selection of each function node in service compositon model isindependent and can not solve the QoS global optimization problems of service composition.Therefore, this paper presents a objective and comprehensive QoS model of compositionservices based on basic services, containing the combination relationship among the basicservices. The QoS model of service composition contains combinational logic and datalogic information of the services. The introduction of directed graph concept converts theservice selection based on QoS evaluation to solve the single-source shortest path problemof the graph.Grey system theory is a new method that studies less data, poor information uncertainproblems. It makes the uncertainty system regarding "small sample",“poor information”,and “partial information is known, some of the information is unknown” as researchobject.Mainly through the generation and development of “partial” known information, itextracts valuable information and realizes the correct description and effective monitoringof the system operation behavior and the evolution law. At present in the research of theWeb Service Composition, parameters information of each service is obtained byexcavatting and analyzing the execution log of the Web services. But, as the relationship ofInternet Web services providers is relatively closed, external people are difficult to getlong-term and historical execution log data of the Web service, and instead small sample,incomplete, short timing characteristics. At the same time weight is very easy influnced byfactors of people, and difficult to be accurately defined. So the grey system theory provides non-weight evaluation with feasibility. With the grey system theory, application of greyprediction and grey relational analysis, this paper preliminarily studies QoS evaluationmethod oriented service selection.