Study Of Web Service QoS Evaluation Method Based On Unascertained Theory

The development of Web services technology provides a new technology for integration services. With the development of the network, there will appear various providers to provide the same or similar services, so QoS is the best measurement and evaluation of a service. However, QoS is uncertain and changeable. Firstly, the Web services run under the dynamic internet environment, the QoS fluctuates with the change of the network. Secondly, due to the autonomy of the Web services, the service provider cannot continue to provide its QoS they had committed for some sorts of problems. Therefore, it becomes a problem to be solved that getting a relatively reasonable evaluation results according to the uncertainty of QoS Web service.Considering the uncertainty evaluation of the QoS, the thesis puts forward a Web services QoS evaluation based on an unascertained theory. On a whole analysis of the uncertainty and the basic ability of the Web services QoS, by introducing the uncertainty of the Web services QoS processed by the unascertained theory, the definition of unascertained area is given in order to describe the uncertainty of the Web service QoS accurately. Moreover, by adoptting the rational Numbers, the unascertained rational number and unascertained interval number describing different indexes of the Web services QoS in unidentified fields, it sets up UBQoSM of Web services QoS, and works out the method of polymerization of the QoS model on the basis of the UBQoSM and, designs the service evaluation based on UBQoSM. It also includes calculation of the QoS according to how to fulfill the tasks. It contains two pars of calculations:UBQoSM parameter value calculation and UBQoSM parameter maintaining value increment parameters. It also gives a calculation in UBQoSM parameter value. In order to reflect the distribution of values in the execution information better in history, it adopts a multi-resolution and no density clustering method to cluster the data, and proposes an evaluation method based on the information entropy, getting the best state of clustering and then to get the partitioning model and generated UBQoSM parameter values based on the interval differentiate patterns. In the phase of UBQoSM parameter value incremental maintenance, in order to avoid the frequent update operations, the thesis adopted a new method to maintain the log data in UBQoSM parameter values respectively and single increment parameters of similar Web services and Web services with the original parameters and the calculation method of similarity and compatibility based on the increment parameter values. It proposes the incremental maintenance of aggregation algorithm as well. Finally, the Web services evaluation tool that based on UBQoSM is designed and implementen to achieve the service evaluation method that proposed in this thesis, and the applicability of this method was analysised by experiment in this thesis.